diff --git a/.flake8 b/.flake8 index 13d7fdec..db676f94 100644 --- a/.flake8 +++ b/.flake8 @@ -15,7 +15,8 @@ exclude = # This contains our built documentation build, # This contains builds of flake8 that we don't want to check - dist + dist, + ./src/app/baml_client max-complexity = 12 max-line-length = 88 per-file-ignores = diff --git a/.github/workflows/ci.yml b/.github/workflows/ci.yml index 5ddffcaf..bc6264b0 100644 --- a/.github/workflows/ci.yml +++ b/.github/workflows/ci.yml @@ -26,20 +26,20 @@ jobs: registry-username: ${{ secrets.DOCKER_PROD_USERNAME }} registry-password: ${{ secrets.DOCKER_PROD_PASSWORD }} - lint-and-test: - uses: ./.github/workflows/lint-and-test.yml - with: - registry-name: ${{ vars.DOCKER_PROD_REGISTRY }} - image-name: welearn-api - image-tag: ${{ github.sha }} - secrets: - registry-username: ${{ secrets.DOCKER_PROD_USERNAME }} - registry-password: ${{ secrets.DOCKER_PROD_PASSWORD }} - needs: - - build-docker + # lint-and-test: + # uses: ./.github/workflows/lint-and-test.yml + # with: + # registry-name: ${{ vars.DOCKER_PROD_REGISTRY }} + # image-name: welearn-api + # image-tag: ${{ github.sha }} + # secrets: + # registry-username: ${{ secrets.DOCKER_PROD_USERNAME }} + # registry-password: ${{ secrets.DOCKER_PROD_PASSWORD }} + # needs: + # - build-docker tag-deploy: needs: - build-docker - - lint-and-test + # - lint-and-test uses: CyberCRI/github-workflows/.github/workflows/tag-deploy.yaml@main diff --git a/.gitignore b/.gitignore index c68be4b1..13e92a9c 100644 --- a/.gitignore +++ b/.gitignore @@ -245,4 +245,5 @@ http-client.private.env.json .vscode/* charts/ -test.yaml \ No newline at end of file +test.yaml +src/app/baml_client \ No newline at end of file diff --git a/Dockerfile b/Dockerfile index c90eb7b3..b8683c2c 100644 --- a/Dockerfile +++ b/Dockerfile @@ -21,6 +21,7 @@ COPY --from=requirements-stage ./tmp/requirements.txt ./requirements.txt RUN pip install --no-cache-dir --upgrade -r requirements.txt COPY . . +RUN baml-cli generate --from /app/src/app/ --no-tests COPY devops-toolbox/scripts/secrets-entrypoint.sh /app/secrets-entrypoint.sh diff --git a/k8s/welearn-api/secrets.dev.yaml b/k8s/welearn-api/secrets.dev.yaml index dbfefd6d..b5f6c22a 100644 --- a/k8s/welearn-api/secrets.dev.yaml +++ b/k8s/welearn-api/secrets.dev.yaml @@ -3,11 +3,17 @@ config: AZURE_API_KEY: ENC[AES256_GCM,data:vU5sozgleY904OU1Fb0yEaRzWYNHnCD2ILjthvtSfk4=,iv:b7ItArThdtIsz1qvgpptbOq96GHa1q6VWAweHYWw2Po=,tag:emJB+azc6vcIWCgv8idtyg==,type:str] AZURE_MISTRAL_API_KEY: ENC[AES256_GCM,data:FjMSqC6wQM4/1cg5hwM3qAZiKmDEEU0rz6l44XDr0HQ0WLXRYKxDcm/0Ii1khG693fyQJalH7lvf8CRVdD4mnIm3KOaw5Q5MCkqrqoQo2GGm+7ZF,iv:h157m7hUESIOMC81yrnxzG//fInFLfsR8o2zYTsGcUs=,tag:pMOQn7+ijCbvYh1AksjYbQ==,type:str] AZURE_APIM_API_KEY: ENC[AES256_GCM,data:GKEHlKS5GgzSO+rHYYvrdL0e7WDZDyByD6mYbDmSNZQ=,iv:k/SHzVKSvMr0ibz3O19yQc3Hz4rzFRALxINlNJRLAwk=,tag:9sgSAnfTKEL5SDPyyw2uhA==,type:str] + AZURE_LLM_FRANCE_API_KEY: ENC[AES256_GCM,data:IjnwPKTBYrZdOlunF664qXfPl9vUqelU7Kk03cY7sZhoiLz3FEPMyuLpoG1T6TzEdWs8uU3UuETCYhOMUiF5pJZC7FPILDHZfVr65ahGaPzwYNbJ,iv:c/GC+sKXUrkPmpx3thLE92xj3Uv2DqYq3jEMhCfpG5U=,tag:6Wt0WjwG4Qv3Il3EFEe6rg==,type:str] + AZURE_LLM_SWEDEN_API_KEY: ENC[AES256_GCM,data:y4lrjKkDaqvaIZy1XYrnaHq0q9hCB3xp6WTu9C6i3sKVmCHSz3WPIAG1kh9JvtNGZHWr2V6K6F/ykHERO06HXn6P3LSIeAva+Db2zc2FWx4deRn2,iv:1o16QNoQYT9/NXMRT8zgVNF/D4y9nXkG71vsbK5KDa8=,tag:YVoY0XjTMZzPqkolwUfRsg==,type:str] PG_PASSWORD: ENC[AES256_GCM,data:8GsISVAuJyIh0WUFwb9mpqyWcNT04fnmpH4ru5CA+aw=,iv:oGKLDiadbhRQWVSE5J/4QJbppiTIcHhNr9fFY7XI0jY=,tag:Ksk0uMrB9fmVT8thQODFJA==,type:str] azureSecrets: azurestorageaccountname: ENC[AES256_GCM,data:vhsGuyM8IYCRua2BQtuYpLI/Dzdx,iv:kPcuAWarV4RXujfwe+9rKe/YWARoeOcFp789WwrI+z8=,tag:Hg8JJmYW0Zuz/2brM/z8yQ==,type:str] azurestorageaccountkey: ENC[AES256_GCM,data:xssnW7Kk3LGmF9dpeCiiV2Z7vyNo4n0aSMcA3aXJYd7uvx7B1HmmSRnc47y3pBS8Hg0dc+6Lf0AIod/Yr2mMC+Dfoosfj2oiY4P/7FQZNMD2xVuZCebTYA==,iv:wlEwj0M1U/EIgVbqxiT0aLgtcwmVGfuRgTaOdQYs6IU=,tag:pLiJujwzEMZOg1xEM22Ojg==,type:str] sops: + kms: [] + gcp_kms: [] + azure_kv: [] + hc_vault: [] age: - recipient: age1dvu205sflecfu6uewgwmgs83nkm2c2fzfxs55zfkjs2gmqjya50qmm6lck enc: | @@ -63,7 +69,8 @@ sops: Q1lsakJjbW5rTGRIYUt0M3ArN2xjMW8KyUV4W7zltLlQ3Bdp4U1OgU5hRlWLa5RU hxOYlyFSCKkn/hXTx2AAwyTVcwVGt/OoUujuRN1A1/pGrO1sDY0uFg== -----END AGE ENCRYPTED FILE----- - lastmodified: "2025-12-05T11:12:47Z" - mac: ENC[AES256_GCM,data:FL4ZyGT2BoO/xUYY4ZEvQqf1nTuLMbX8wE0tvghsFlnOp7bhINuZBA9Mc5xK5dsy5LkVesVV60CtMXu6yhNJUxGRcyrZHkrQnvUX+0qVh4GqRPxhFRP++e8oOFbsVs+XTwZSpJNN740VV27B5NbsM6lXbWi4DGH1q+aL2qE2EuY=,iv:lVWWYPAwkfNg9dJJtTRvD3mYmLcOnvId6fDsndNE6qc=,tag:T4sAj8BPtVzjABw++ejy7Q==,type:str] + lastmodified: "2026-07-01T12:43:28Z" + mac: ENC[AES256_GCM,data:bXYUdTgXnuizCXt89q4JXGZMl+zOSW9QjDnUXpMZT9xb/zzI+/h9ctz/mTJX1u15PrjVZfuxPLDew8/g6DRvuXxhrOkJJJG4TDTjNvQvhHKIidhjg/mvYGcOR1Y8g+zA8Wg764t4mZYAz7TfPclT0p1AvvbyNhqkeybznGGdA4w=,iv:BVzVbKotP9BtbUvPlGcASw+XjI/oFF501H1OxM/x8cQ=,tag:Hu1ggPC3+scuFx2RSXTAJA==,type:str] + pgp: [] unencrypted_suffix: _unencrypted version: 3.9.4 diff --git a/k8s/welearn-api/values.yaml b/k8s/welearn-api/values.yaml index 7db37766..f539a541 100644 --- a/k8s/welearn-api/values.yaml +++ b/k8s/welearn-api/values.yaml @@ -38,6 +38,8 @@ config: AZURE_API_BASE: "https://welearn-openai.openai.azure.com/openai/" AZURE_API_VERSION: "2024-08-01-preview" LLM_MODEL_NAME: gpt-oss-120b + AZURE_BAML_LLM_FRANCE_BASE_URL: https://welearn-mistral.services.ai.azure.com/openai/v1/ + AZURE_BAML_LLM_SWEDEN_BASE_URL: https://welearn-mistral-sweden.services.ai.azure.com/openai/v1/ QDRANT_HOST: "http://qdrant.qdrant" QDRANT_PORT: "6333" MODELS_FOLDER_ROOTS: "/models" diff --git a/output.txt b/output.txt new file mode 100644 index 00000000..ec4dbac3 --- /dev/null +++ b/output.txt @@ -0,0 +1,2 @@ +{ + "description": \ No newline at end of file diff --git a/poetry.lock b/poetry.lock index 2da2046e..dce3daf7 100644 --- a/poetry.lock +++ b/poetry.lock @@ -583,6 +583,42 @@ files = [ {file = "backoff-2.2.1.tar.gz", hash = "sha256:03f829f5bb1923180821643f8753b0502c3b682293992485b0eef2807afa5cba"}, ] +[[package]] +name = "baml-cli" +version = "0.1.0" +description = "The BAML CLI, distributed in Python" +optional = false +python-versions = "<4.0,>=3.8" +groups = ["main"] +markers = "sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\"" +files = [ + {file = "baml_cli-0.1.0-py3-none-any.whl", hash = "sha256:3e0630a10bf940c539e57fc37daf70ea6c097a7feda423f01822a936c2439cdb"}, + {file = "baml_cli-0.1.0.tar.gz", hash = "sha256:ebc3c325f167282918d397e0597dadf2323f463be7be1249b8664a6d2c48166c"}, +] + +[package.dependencies] +baml_py = ">=0.1.0" + +[[package]] +name = "baml-py" +version = "0.220.0" +description = "BAML python bindings (pyproject.toml)" +optional = false +python-versions = "*" +groups = ["main"] +markers = "sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\"" +files = [ + {file = "baml_py-0.220.0-cp38-abi3-macosx_10_12_x86_64.whl", hash = "sha256:edf4112d1b1da5765ebfcadab7f9eee48ac337e5b06ff8286c970e27cd87e28e"}, + {file = "baml_py-0.220.0-cp38-abi3-macosx_11_0_arm64.whl", hash = "sha256:5e72466c12d9b6597d3dbe16a364699307911483bffc07e0b66d701cdefecf6d"}, + {file = "baml_py-0.220.0-cp38-abi3-manylinux_2_17_x86_64.manylinux2014_x86_64.whl", hash = "sha256:1da0548281e0d21d56886a46aaeef0680e660fc1b36a9bdbc11d04efad96cfcc"}, + {file = "baml_py-0.220.0-cp38-abi3-manylinux_2_24_aarch64.whl", hash = "sha256:0895a3633b061505a3d48d259ca6a480dd37f2a9dade6ea3bf5b8fb1f7632d11"}, + {file = "baml_py-0.220.0-cp38-abi3-musllinux_1_1_aarch64.whl", hash = "sha256:52ba7755026322fd73dd70087c09745aae86df47306b1dde06cf731266b55c34"}, + {file = "baml_py-0.220.0-cp38-abi3-musllinux_1_1_x86_64.whl", hash = "sha256:a240cdc76455f8bd8aa5a5b3ad805369edb27320111bb915df543c55a846b916"}, + {file = "baml_py-0.220.0-cp38-abi3-win_amd64.whl", hash = "sha256:a5b6d1374eec1225bc76444fc2e5dbb0b7f14ed5a1a043e20beef778475ff8d4"}, + {file = "baml_py-0.220.0-cp38-abi3-win_arm64.whl", hash = "sha256:76fe51528ac5cd07ea2bd8ab2ccd0fd302cd9fc1f5c5c3236c35edfede2d8b48"}, + {file = "baml_py-0.220.0-pp39-pypy39_pp73-manylinux_2_24_aarch64.whl", hash = "sha256:04d286c6772c2d8f20e58e994ac0e8dfc6b08254a85cc9603f7daf09538feeea"}, +] + [[package]] name = "beautifulsoup4" version = "4.14.3" @@ -914,7 +950,7 @@ files = [ {file = "cffi-2.0.0-cp39-cp39-win_amd64.whl", hash = "sha256:b882b3df248017dba09d6b16defe9b5c407fe32fc7c65a9c69798e6175601be9"}, {file = "cffi-2.0.0.tar.gz", hash = "sha256:44d1b5909021139fe36001ae048dbdde8214afa20200eda0f64c068cac5d5529"}, ] -markers = {main = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and platform_python_implementation != \"PyPy\"", metrics = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and (platform_python_implementation == \"CPython\" or implementation_name == \"pypy\") and (implementation_name == \"pypy\" or sys_platform == \"win32\")"} +markers = {main = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and platform_python_implementation != \"PyPy\"", metrics = "(implementation_name == \"pypy\" or sys_platform == \"win32\") and (sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and (platform_python_implementation == \"CPython\" or implementation_name == \"pypy\")"} [package.dependencies] pycparser = {version = "*", markers = "implementation_name != \"PyPy\""} @@ -4295,61 +4331,61 @@ xml = ["lxml (>=4.9.2)"] [[package]] name = "pandas" -version = "3.0.0" +version = "3.0.3" description = "Powerful data structures for data analysis, time series, and statistics" optional = false python-versions = ">=3.11" -groups = ["metrics"] +groups = ["main", "metrics"] markers = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and python_version >= \"3.11\"" files = [ - {file = "pandas-3.0.0-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:d64ce01eb9cdca96a15266aa679ae50212ec52757c79204dbc7701a222401850"}, - {file = "pandas-3.0.0-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:613e13426069793aa1ec53bdcc3b86e8d32071daea138bbcf4fa959c9cdaa2e2"}, - {file = "pandas-3.0.0-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0192fee1f1a8e743b464a6607858ee4b071deb0b118eb143d71c2a1d170996d5"}, - {file = "pandas-3.0.0-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:f0b853319dec8d5e0c8b875374c078ef17f2269986a78168d9bd57e49bf650ae"}, - {file = "pandas-3.0.0-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:707a9a877a876c326ae2cb640fbdc4ef63b0a7b9e2ef55c6df9942dcee8e2af9"}, - {file = "pandas-3.0.0-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:afd0aa3d0b5cda6e0b8ffc10dbcca3b09ef3cbcd3fe2b27364f85fdc04e1989d"}, - {file = "pandas-3.0.0-cp311-cp311-win_amd64.whl", hash = "sha256:113b4cca2614ff7e5b9fee9b6f066618fe73c5a83e99d721ffc41217b2bf57dd"}, - {file = "pandas-3.0.0-cp311-cp311-win_arm64.whl", hash = "sha256:c14837eba8e99a8da1527c0280bba29b0eb842f64aa94982c5e21227966e164b"}, - {file = "pandas-3.0.0-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:9803b31f5039b3c3b10cc858c5e40054adb4b29b4d81cb2fd789f4121c8efbcd"}, - {file = "pandas-3.0.0-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:14c2a4099cd38a1d18ff108168ea417909b2dea3bd1ebff2ccf28ddb6a74d740"}, - {file = "pandas-3.0.0-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d257699b9a9960e6125686098d5714ac59d05222bef7a5e6af7a7fd87c650801"}, - {file = "pandas-3.0.0-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:69780c98f286076dcafca38d8b8eee1676adf220199c0a39f0ecbf976b68151a"}, - {file = "pandas-3.0.0-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:4a66384f017240f3858a4c8a7cf21b0591c3ac885cddb7758a589f0f71e87ebb"}, - {file = "pandas-3.0.0-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:be8c515c9bc33989d97b89db66ea0cececb0f6e3c2a87fcc8b69443a6923e95f"}, - {file = "pandas-3.0.0-cp312-cp312-win_amd64.whl", hash = "sha256:a453aad8c4f4e9f166436994a33884442ea62aa8b27d007311e87521b97246e1"}, - {file = "pandas-3.0.0-cp312-cp312-win_arm64.whl", hash = "sha256:da768007b5a33057f6d9053563d6b74dd6d029c337d93c6d0d22a763a5c2ecc0"}, - {file = "pandas-3.0.0-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b78d646249b9a2bc191040988c7bb524c92fa8534fb0898a0741d7e6f2ffafa6"}, - {file = "pandas-3.0.0-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:bc9cba7b355cb4162442a88ce495e01cb605f17ac1e27d6596ac963504e0305f"}, - {file = "pandas-3.0.0-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:3c9a1a149aed3b6c9bf246033ff91e1b02d529546c5d6fb6b74a28fea0cf4c70"}, - {file = "pandas-3.0.0-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:95683af6175d884ee89471842acfca29172a85031fccdabc35e50c0984470a0e"}, - {file = "pandas-3.0.0-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:1fbbb5a7288719e36b76b4f18d46ede46e7f916b6c8d9915b756b0a6c3f792b3"}, - {file = "pandas-3.0.0-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:8e8b9808590fa364416b49b2a35c1f4cf2785a6c156935879e57f826df22038e"}, - {file = "pandas-3.0.0-cp313-cp313-win_amd64.whl", hash = "sha256:98212a38a709feb90ae658cb6227ea3657c22ba8157d4b8f913cd4c950de5e7e"}, - {file = "pandas-3.0.0-cp313-cp313-win_arm64.whl", hash = "sha256:177d9df10b3f43b70307a149d7ec49a1229a653f907aa60a48f1877d0e6be3be"}, - {file = "pandas-3.0.0-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:2713810ad3806767b89ad3b7b69ba153e1c6ff6d9c20f9c2140379b2a98b6c98"}, - {file = "pandas-3.0.0-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:15d59f885ee5011daf8335dff47dcb8a912a27b4ad7826dc6cbe809fd145d327"}, - {file = "pandas-3.0.0-cp313-cp313t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:24e6547fb64d2c92665dd2adbfa4e85fa4fd70a9c070e7cfb03b629a0bbab5eb"}, - {file = "pandas-3.0.0-cp313-cp313t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:48ee04b90e2505c693d3f8e8f524dab8cb8aaf7ddcab52c92afa535e717c4812"}, - {file = "pandas-3.0.0-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:66f72fb172959af42a459e27a8d8d2c7e311ff4c1f7db6deb3b643dbc382ae08"}, - {file = "pandas-3.0.0-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:4a4a400ca18230976724a5066f20878af785f36c6756e498e94c2a5e5d57779c"}, - {file = "pandas-3.0.0-cp313-cp313t-win_amd64.whl", hash = "sha256:940eebffe55528074341a5a36515f3e4c5e25e958ebbc764c9502cfc35ba3faa"}, - {file = "pandas-3.0.0-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:597c08fb9fef0edf1e4fa2f9828dd27f3d78f9b8c9b4a748d435ffc55732310b"}, - {file = "pandas-3.0.0-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:447b2d68ac5edcbf94655fe909113a6dba6ef09ad7f9f60c80477825b6c489fe"}, - {file = "pandas-3.0.0-cp314-cp314-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:debb95c77ff3ed3ba0d9aa20c3a2f19165cc7956362f9873fce1ba0a53819d70"}, - {file = "pandas-3.0.0-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:fedabf175e7cd82b69b74c30adbaa616de301291a5231138d7242596fc296a8d"}, - {file = "pandas-3.0.0-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:412d1a89aab46889f3033a386912efcdfa0f1131c5705ff5b668dda88305e986"}, - {file = "pandas-3.0.0-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:e979d22316f9350c516479dd3a92252be2937a9531ed3a26ec324198a99cdd49"}, - {file = "pandas-3.0.0-cp314-cp314-win_amd64.whl", hash = "sha256:083b11415b9970b6e7888800c43c82e81a06cd6b06755d84804444f0007d6bb7"}, - {file = "pandas-3.0.0-cp314-cp314-win_arm64.whl", hash = "sha256:5db1e62cb99e739fa78a28047e861b256d17f88463c76b8dafc7c1338086dca8"}, - {file = "pandas-3.0.0-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:697b8f7d346c68274b1b93a170a70974cdc7d7354429894d5927c1effdcccd73"}, - {file = "pandas-3.0.0-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:8cb3120f0d9467ed95e77f67a75e030b67545bcfa08964e349252d674171def2"}, - {file = "pandas-3.0.0-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:33fd3e6baa72899746b820c31e4b9688c8e1b7864d7aec2de7ab5035c285277a"}, - {file = "pandas-3.0.0-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a8942e333dc67ceda1095227ad0febb05a3b36535e520154085db632c40ad084"}, - {file = "pandas-3.0.0-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:783ac35c4d0fe0effdb0d67161859078618b1b6587a1af15928137525217a721"}, - {file = "pandas-3.0.0-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:125eb901e233f155b268bbef9abd9afb5819db74f0e677e89a61b246228c71ac"}, - {file = "pandas-3.0.0-cp314-cp314t-win_amd64.whl", hash = "sha256:b86d113b6c109df3ce0ad5abbc259fe86a1bd4adfd4a31a89da42f84f65509bb"}, - {file = "pandas-3.0.0-cp314-cp314t-win_arm64.whl", hash = "sha256:1c39eab3ad38f2d7a249095f0a3d8f8c22cc0f847e98ccf5bbe732b272e2d9fa"}, - {file = "pandas-3.0.0.tar.gz", hash = "sha256:0facf7e87d38f721f0af46fe70d97373a37701b1c09f7ed7aeeb292ade5c050f"}, + {file = "pandas-3.0.3-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:455f6f8139d4282188f526868dbc3c828470e88a3d9d59a891bd46a455f21b98"}, + {file = "pandas-3.0.3-cp311-cp311-macosx_11_0_arm64.whl", hash = "sha256:4e15135e2ee5df1063313e2425ceef8ac0f4ae775893815b0923651b806a5639"}, + {file = "pandas-3.0.3-cp311-cp311-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:05f1f1752b8533ea03f7f39a9c15b1a058d067bb48f4748948e7a8691e0510f2"}, + {file = "pandas-3.0.3-cp311-cp311-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:8a1e45c80cceb3b4a21bc5939d52e8cbd8d9b7305309219d59e9754d9ce09e27"}, + {file = "pandas-3.0.3-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:14da8316da4d0c5a77618425996bfb1248ca87fc2c1486e6fde4652bd18b5824"}, + {file = "pandas-3.0.3-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:a55066a0505dae0ba2b50a46637db34b46f9094c65c5d4800794ef6335010938"}, + {file = "pandas-3.0.3-cp311-cp311-win_amd64.whl", hash = "sha256:6674ab18ad8c57802867264b00e15e7bb904700cdd9046e3b2fa1fce237439ea"}, + {file = "pandas-3.0.3-cp311-cp311-win_arm64.whl", hash = "sha256:5cc09a68b3120e0f54870dede8287a7bb1fa463907e4fcec1ea77cab6179bf7a"}, + {file = "pandas-3.0.3-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:fed2ff7fd9779120e388e285fc029bd5cf9490cdd2e4166a9ee22c0e49a9ab09"}, + {file = "pandas-3.0.3-cp312-cp312-macosx_11_0_arm64.whl", hash = "sha256:b168fc218fd80a6cbdbdbc1a97ddc7889ed057d7eb45f50d866ceab5f39904c4"}, + {file = "pandas-3.0.3-cp312-cp312-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:0383c72c75cdcca61a9e116e611143902dbfd08bff356829c2f6d1cf40a9ca8c"}, + {file = "pandas-3.0.3-cp312-cp312-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:6dc0b3fd2169c9157deed50b4d519553a3655c8c6a96027136d654592be973a9"}, + {file = "pandas-3.0.3-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:7e65d5407dc0b394f509699650e4a2ec01c0514f21850f453fa60f3be79a5dbf"}, + {file = "pandas-3.0.3-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:f8894dc474d648fe7b6ff0ca9b0bd73950d19952bc1a6534540762c5d79d305c"}, + {file = "pandas-3.0.3-cp312-cp312-win_amd64.whl", hash = "sha256:c7be265b62cef88e253a941e4698604973736dcfe242fdb5198f0f7bc473cdcc"}, + {file = "pandas-3.0.3-cp312-cp312-win_arm64.whl", hash = "sha256:557409bc4178e70ee8d9ddb494798e51ebf6ea59330f6be22c51bab2a7db6c49"}, + {file = "pandas-3.0.3-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:67b3b64c11910cfa29f4e94a14d3bff9ee693b6fc76055e7cad549cee0aec5fa"}, + {file = "pandas-3.0.3-cp313-cp313-macosx_11_0_arm64.whl", hash = "sha256:39436b377d56d2a2e52d0395bdbee171f01068e99af5250509aceeb929f765c7"}, + {file = "pandas-3.0.3-cp313-cp313-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:d4be06d68f9ddcfc645b87534911da79a8fbffc7573c80e0edcf42a5020624d8"}, + {file = "pandas-3.0.3-cp313-cp313-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:a4eeb6830daf35a71cc09649bd823e2b542dac246cdee9614c6e4bd65028cd6a"}, + {file = "pandas-3.0.3-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:1928e07221f82db493cd4af1e23c1bfca524a19a4699887975bff68f49a72bfb"}, + {file = "pandas-3.0.3-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:51b1fe551acb77dac643c6fda86084d8d446c10fe64b06a9cc29c4cc8540e7f2"}, + {file = "pandas-3.0.3-cp313-cp313-win_amd64.whl", hash = "sha256:a82d532a3351d435432cd913edbccaf8b8e01d4dd0e5ced5a8d2e8ecd94c7e44"}, + {file = "pandas-3.0.3-cp313-cp313-win_arm64.whl", hash = "sha256:275c14e0fce14a2ec20eee474aecd305478ea3c1e6f6a9d8fe219a165542717e"}, + {file = "pandas-3.0.3-cp313-cp313t-macosx_10_13_x86_64.whl", hash = "sha256:46997386d528eb40376ecd6b033cf4a8a1e5282580f68f43de875b78cba2199d"}, + {file = "pandas-3.0.3-cp313-cp313t-macosx_11_0_arm64.whl", hash = "sha256:261e308dfb22448384b7580cf719d2f998fe2966c92893c3e77d14008af1f066"}, + {file = "pandas-3.0.3-cp313-cp313t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:dd1a5d1def6a46002e964510bdc67c368aa0951df5d1d9f8365336f5a1f490cd"}, + {file = "pandas-3.0.3-cp313-cp313t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:d72828c20c6d6e83e1e22a6a3b47b326b71664112fa9705dcbccfd7a39b62085"}, + {file = "pandas-3.0.3-cp313-cp313t-musllinux_1_2_aarch64.whl", hash = "sha256:d26cbe1fcfc12e8fd900e2454163e466b2d3af84f7c75481df7683ffc073d870"}, + {file = "pandas-3.0.3-cp313-cp313t-musllinux_1_2_x86_64.whl", hash = "sha256:3e91cec1879ada0624fc3dc9953c5cbd60208e59c0db28f540c5d6d47502422f"}, + {file = "pandas-3.0.3-cp313-cp313t-win_amd64.whl", hash = "sha256:08d789b41f87e0905880e293cedf6197ce71fe67cc081358b1e148a491b9bd13"}, + {file = "pandas-3.0.3-cp314-cp314-macosx_10_15_x86_64.whl", hash = "sha256:3650109c0f22879df8bd6179ab9ee3d7f1d1d4e7e0094a3f0032d9f51e2e64ac"}, + {file = "pandas-3.0.3-cp314-cp314-macosx_11_0_arm64.whl", hash = "sha256:bab900348131a7db1f69a7309ef141fd5680f1487094193bcbbb61791573bf8f"}, + {file = "pandas-3.0.3-cp314-cp314-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:ba7e08b9ac1d54569cd1e256e3668975ed624d6826f7b68df0342b012007bddb"}, + {file = "pandas-3.0.3-cp314-cp314-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:9d71c63ae4ebdbf70209742096f1fc46a83a0613c99d4b23766cced9ff8cd62a"}, + {file = "pandas-3.0.3-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:e3a2ec42c98ffa2565a67e08e218d06d72576d758d90facb7c00805194d8f360"}, + {file = "pandas-3.0.3-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:335f62418ed562cfc3c49e9e196375c28b729dcef8543abf4f9438e381bf3c76"}, + {file = "pandas-3.0.3-cp314-cp314-win_amd64.whl", hash = "sha256:3c20a521bbb85902f79f7270c80a59e1b5452d96d170c034f207181870f97ac5"}, + {file = "pandas-3.0.3-cp314-cp314-win_arm64.whl", hash = "sha256:a2d2dff8a04f3917b55ab3910c32990f8ddf7eceba114947838cefa976a68977"}, + {file = "pandas-3.0.3-cp314-cp314t-macosx_10_15_x86_64.whl", hash = "sha256:0d589105b3c14645af1738ff279b2995102d8f7a03b0a66dc8d95550eb513e04"}, + {file = "pandas-3.0.3-cp314-cp314t-macosx_11_0_arm64.whl", hash = "sha256:13fc1e853d9e04743d11ba75a985ccbc2a317fe07d8af61e445a6fd24dacd6a6"}, + {file = "pandas-3.0.3-cp314-cp314t-manylinux_2_24_aarch64.manylinux_2_28_aarch64.whl", hash = "sha256:819959dab7bbd0049c15623fbac4e29a191b9528160a61fb1032242d8ced2d9c"}, + {file = "pandas-3.0.3-cp314-cp314t-manylinux_2_24_x86_64.manylinux_2_28_x86_64.whl", hash = "sha256:60ae316d3fd75d1858d450d0db0103ea2be3e7d4a95ec2f064f7e2ae63f7b028"}, + {file = "pandas-3.0.3-cp314-cp314t-musllinux_1_2_aarch64.whl", hash = "sha256:bd3a518890b400d32f9023722dc9a9a5c969f00b415419a3c06c043f09bb5d7d"}, + {file = "pandas-3.0.3-cp314-cp314t-musllinux_1_2_x86_64.whl", hash = "sha256:9c39be2d709d01fa972a0cabc522389fceca4f3969332ba25a7d6c5802cf976a"}, + {file = "pandas-3.0.3-cp314-cp314t-win_amd64.whl", hash = "sha256:4db8c527972a821cf5286b40ccc57642a39bc62e62022b42f99f8a67fca8c3a1"}, + {file = "pandas-3.0.3-cp314-cp314t-win_arm64.whl", hash = "sha256:b2c95f8bfc1ee412bf482605d7bfd30c12d1d26bd59fdd91efeef1d4718decb1"}, + {file = "pandas-3.0.3.tar.gz", hash = "sha256:696a4a00a2a2a35d4e5deb3fc946641b96c944f02230e4f76137fe35d806c4fc"}, ] [package.dependencies] @@ -4358,7 +4394,7 @@ python-dateutil = ">=2.8.2" tzdata = {version = "*", markers = "sys_platform == \"win32\" or sys_platform == \"emscripten\""} [package.extras] -all = ["PyQt5 (>=5.15.9)", "SQLAlchemy (>=2.0.36)", "adbc-driver-postgresql (>=1.2.0)", "adbc-driver-sqlite (>=1.2.0)", "beautifulsoup4 (>=4.12.3)", "bottleneck (>=1.4.2)", "fastparquet (>=2024.11.0)", "fsspec (>=2024.10.0)", "gcsfs (>=2024.10.0)", "html5lib (>=1.1)", "hypothesis (>=6.116.0)", "jinja2 (>=3.1.5)", "lxml (>=5.3.0)", "matplotlib (>=3.9.3)", "numba (>=0.60.0)", "numexpr (>=2.10.2)", "odfpy (>=1.4.1)", "openpyxl (>=3.1.5)", "psycopg2 (>=2.9.10)", "pyarrow (>=13.0.0)", "pyiceberg (>=0.8.1)", "pymysql (>=1.1.1)", "pyreadstat (>=1.2.8)", "pytest (>=8.3.4)", "pytest-xdist (>=3.6.1)", "python-calamine (>=0.3.0)", "pytz (>=2024.2)", "pyxlsb (>=1.0.10)", "qtpy (>=2.4.2)", "s3fs (>=2024.10.0)", "scipy (>=1.14.1)", "tables (>=3.10.1)", "tabulate (>=0.9.0)", "xarray (>=2024.10.0)", "xlrd (>=2.0.1)", "xlsxwriter (>=3.2.0)", "zstandard (>=0.23.0)"] +all = ["PyQt5 (>=5.15.9)", "SQLAlchemy (>=2.0.36)", "adbc-driver-postgresql (>=1.2.0)", "adbc-driver-sqlite (>=1.2.0)", "beautifulsoup4 (>=4.12.3)", "bottleneck (>=1.4.2)", "fastparquet (>=2024.11.0)", "fsspec (>=2024.10.0)", "gcsfs (>=2024.10.0)", "html5lib (>=1.1)", "hypothesis (>=6.116.0)", "jinja2 (>=3.1.5)", "lxml (>=5.3.0)", "matplotlib (>=3.9.3)", "numba (>=0.60.0)", "numexpr (>=2.10.2)", "odfpy (>=1.4.1)", "openpyxl (>=3.1.5)", "psycopg2 (>=2.9.10)", "pyarrow (>=13.0.0)", "pyiceberg (>=0.8.1)", "pymysql (>=1.1.1)", "pyreadstat (>=1.2.8)", "pytest (>=8.3.4)", "pytest-xdist (>=3.6.1)", "python-calamine (>=0.3.0)", "pytz (>=2020.1)", "pyxlsb (>=1.0.10)", "qtpy (>=2.4.2)", "s3fs (>=2024.10.0)", "scipy (>=1.14.1)", "tables (>=3.10.1)", "tabulate (>=0.9.0)", "xarray (>=2024.10.0)", "xlrd (>=2.0.1)", "xlsxwriter (>=3.2.0)", "zstandard (>=0.23.0)"] aws = ["s3fs (>=2024.10.0)"] clipboard = ["PyQt5 (>=5.15.9)", "qtpy (>=2.4.2)"] compression = ["zstandard (>=0.23.0)"] @@ -4380,7 +4416,7 @@ pyarrow = ["pyarrow (>=13.0.0)"] spss = ["pyreadstat (>=1.2.8)"] sql-other = ["SQLAlchemy (>=2.0.36)", "adbc-driver-postgresql (>=1.2.0)", "adbc-driver-sqlite (>=1.2.0)"] test = ["hypothesis (>=6.116.0)", "pytest (>=8.3.4)", "pytest-xdist (>=3.6.1)"] -timezone = ["pytz (>=2024.2)"] +timezone = ["pytz (>=2020.1)"] xml = ["lxml (>=5.3.0)"] [[package]] @@ -4904,8 +4940,10 @@ files = [ {file = "psycopg2_binary-2.9.11-cp310-cp310-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:c47676e5b485393f069b4d7a811267d3168ce46f988fa602658b8bb901e9e64d"}, {file = "psycopg2_binary-2.9.11-cp310-cp310-manylinux2014_ppc64le.manylinux_2_17_ppc64le.whl", hash = "sha256:a28d8c01a7b27a1e3265b11250ba7557e5f72b5ee9e5f3a2fa8d2949c29bf5d2"}, {file = "psycopg2_binary-2.9.11-cp310-cp310-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:5f3f2732cf504a1aa9e9609d02f79bea1067d99edf844ab92c247bbca143303b"}, + {file = "psycopg2_binary-2.9.11-cp310-cp310-manylinux_2_38_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:865f9945ed1b3950d968ec4690ce68c55019d79e4497366d36e090327ce7db14"}, {file = "psycopg2_binary-2.9.11-cp310-cp310-musllinux_1_2_aarch64.whl", hash = "sha256:91537a8df2bde69b1c1db01d6d944c831ca793952e4f57892600e96cee95f2cd"}, {file = "psycopg2_binary-2.9.11-cp310-cp310-musllinux_1_2_ppc64le.whl", hash = "sha256:4dca1f356a67ecb68c81a7bc7809f1569ad9e152ce7fd02c2f2036862ca9f66b"}, + {file = "psycopg2_binary-2.9.11-cp310-cp310-musllinux_1_2_riscv64.whl", hash = "sha256:0da4de5c1ac69d94ed4364b6cbe7190c1a70d325f112ba783d83f8440285f152"}, {file = "psycopg2_binary-2.9.11-cp310-cp310-musllinux_1_2_x86_64.whl", hash = "sha256:37d8412565a7267f7d79e29ab66876e55cb5e8e7b3bbf94f8206f6795f8f7e7e"}, {file = "psycopg2_binary-2.9.11-cp310-cp310-win_amd64.whl", hash = "sha256:c665f01ec8ab273a61c62beeb8cce3014c214429ced8a308ca1fc410ecac3a39"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-macosx_10_9_x86_64.whl", hash = "sha256:0e8480afd62362d0a6a27dd09e4ca2def6fa50ed3a4e7c09165266106b2ffa10"}, @@ -4913,8 +4951,10 @@ files = [ {file = "psycopg2_binary-2.9.11-cp311-cp311-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:2e164359396576a3cc701ba8af4751ae68a07235d7a380c631184a611220d9a4"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-manylinux2014_ppc64le.manylinux_2_17_ppc64le.whl", hash = "sha256:d57c9c387660b8893093459738b6abddbb30a7eab058b77b0d0d1c7d521ddfd7"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:2c226ef95eb2250974bf6fa7a842082b31f68385c4f3268370e3f3870e7859ee"}, + {file = "psycopg2_binary-2.9.11-cp311-cp311-manylinux_2_38_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:a311f1edc9967723d3511ea7d2708e2c3592e3405677bf53d5c7246753591fbb"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-musllinux_1_2_aarch64.whl", hash = "sha256:ebb415404821b6d1c47353ebe9c8645967a5235e6d88f914147e7fd411419e6f"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-musllinux_1_2_ppc64le.whl", hash = "sha256:f07c9c4a5093258a03b28fab9b4f151aa376989e7f35f855088234e656ee6a94"}, + {file = "psycopg2_binary-2.9.11-cp311-cp311-musllinux_1_2_riscv64.whl", hash = "sha256:00ce1830d971f43b667abe4a56e42c1e2d594b32da4802e44a73bacacb25535f"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-musllinux_1_2_x86_64.whl", hash = "sha256:cffe9d7697ae7456649617e8bb8d7a45afb71cd13f7ab22af3e5c61f04840908"}, {file = "psycopg2_binary-2.9.11-cp311-cp311-win_amd64.whl", hash = "sha256:304fd7b7f97eef30e91b8f7e720b3db75fee010b520e434ea35ed1ff22501d03"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-macosx_10_13_x86_64.whl", hash = "sha256:be9b840ac0525a283a96b556616f5b4820e0526addb8dcf6525a0fa162730be4"}, @@ -4922,8 +4962,10 @@ files = [ {file = "psycopg2_binary-2.9.11-cp312-cp312-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:ab8905b5dcb05bf3fb22e0cf90e10f469563486ffb6a96569e51f897c750a76a"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-manylinux2014_ppc64le.manylinux_2_17_ppc64le.whl", hash = "sha256:bf940cd7e7fec19181fdbc29d76911741153d51cab52e5c21165f3262125685e"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:fa0f693d3c68ae925966f0b14b8edda71696608039f4ed61b1fe9ffa468d16db"}, + {file = "psycopg2_binary-2.9.11-cp312-cp312-manylinux_2_38_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:a1cf393f1cdaf6a9b57c0a719a1068ba1069f022a59b8b1fe44b006745b59757"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-musllinux_1_2_aarch64.whl", hash = "sha256:ef7a6beb4beaa62f88592ccc65df20328029d721db309cb3250b0aae0fa146c3"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-musllinux_1_2_ppc64le.whl", hash = "sha256:31b32c457a6025e74d233957cc9736742ac5a6cb196c6b68499f6bb51390bd6a"}, + {file = "psycopg2_binary-2.9.11-cp312-cp312-musllinux_1_2_riscv64.whl", hash = "sha256:edcb3aeb11cb4bf13a2af3c53a15b3d612edeb6409047ea0b5d6a21a9d744b34"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-musllinux_1_2_x86_64.whl", hash = "sha256:62b6d93d7c0b61a1dd6197d208ab613eb7dcfdcca0a49c42ceb082257991de9d"}, {file = "psycopg2_binary-2.9.11-cp312-cp312-win_amd64.whl", hash = "sha256:b33fabeb1fde21180479b2d4667e994de7bbf0eec22832ba5d9b5e4cf65b6c6d"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-macosx_10_13_x86_64.whl", hash = "sha256:b8fb3db325435d34235b044b199e56cdf9ff41223a4b9752e8576465170bb38c"}, @@ -4931,8 +4973,10 @@ files = [ {file = "psycopg2_binary-2.9.11-cp313-cp313-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:8c55b385daa2f92cb64b12ec4536c66954ac53654c7f15a203578da4e78105c0"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-manylinux2014_ppc64le.manylinux_2_17_ppc64le.whl", hash = "sha256:c0377174bf1dd416993d16edc15357f6eb17ac998244cca19bc67cdc0e2e5766"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:5c6ff3335ce08c75afaed19e08699e8aacf95d4a260b495a4a8545244fe2ceb3"}, + {file = "psycopg2_binary-2.9.11-cp313-cp313-manylinux_2_38_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:84011ba3109e06ac412f95399b704d3d6950e386b7994475b231cf61eec2fc1f"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-musllinux_1_2_aarch64.whl", hash = "sha256:ba34475ceb08cccbdd98f6b46916917ae6eeb92b5ae111df10b544c3a4621dc4"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-musllinux_1_2_ppc64le.whl", hash = "sha256:b31e90fdd0f968c2de3b26ab014314fe814225b6c324f770952f7d38abf17e3c"}, + {file = "psycopg2_binary-2.9.11-cp313-cp313-musllinux_1_2_riscv64.whl", hash = "sha256:d526864e0f67f74937a8fce859bd56c979f5e2ec57ca7c627f5f1071ef7fee60"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-musllinux_1_2_x86_64.whl", hash = "sha256:04195548662fa544626c8ea0f06561eb6203f1984ba5b4562764fbeb4c3d14b1"}, {file = "psycopg2_binary-2.9.11-cp313-cp313-win_amd64.whl", hash = "sha256:efff12b432179443f54e230fdf60de1f6cc726b6c832db8701227d089310e8aa"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-macosx_10_13_x86_64.whl", hash = "sha256:92e3b669236327083a2e33ccfa0d320dd01b9803b3e14dd986a4fc54aa00f4e1"}, @@ -4940,8 +4984,10 @@ files = [ {file = "psycopg2_binary-2.9.11-cp314-cp314-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:9b52a3f9bb540a3e4ec0f6ba6d31339727b2950c9772850d6545b7eae0b9d7c5"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-manylinux2014_ppc64le.manylinux_2_17_ppc64le.whl", hash = "sha256:db4fd476874ccfdbb630a54426964959e58da4c61c9feba73e6094d51303d7d8"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:47f212c1d3be608a12937cc131bd85502954398aaa1320cb4c14421a0ffccf4c"}, + {file = "psycopg2_binary-2.9.11-cp314-cp314-manylinux_2_38_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:e35b7abae2b0adab776add56111df1735ccc71406e56203515e228a8dc07089f"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-musllinux_1_2_aarch64.whl", hash = "sha256:fcf21be3ce5f5659daefd2b3b3b6e4727b028221ddc94e6c1523425579664747"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-musllinux_1_2_ppc64le.whl", hash = "sha256:9bd81e64e8de111237737b29d68039b9c813bdf520156af36d26819c9a979e5f"}, + {file = "psycopg2_binary-2.9.11-cp314-cp314-musllinux_1_2_riscv64.whl", hash = "sha256:32770a4d666fbdafab017086655bcddab791d7cb260a16679cc5a7338b64343b"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-musllinux_1_2_x86_64.whl", hash = "sha256:c3cb3a676873d7506825221045bd70e0427c905b9c8ee8d6acd70cfcbd6e576d"}, {file = "psycopg2_binary-2.9.11-cp314-cp314-win_amd64.whl", hash = "sha256:4012c9c954dfaccd28f94e84ab9f94e12df76b4afb22331b1f0d3154893a6316"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-macosx_10_9_x86_64.whl", hash = "sha256:20e7fb94e20b03dcc783f76c0865f9da39559dcc0c28dd1a3fce0d01902a6b9c"}, @@ -4949,8 +4995,10 @@ files = [ {file = "psycopg2_binary-2.9.11-cp39-cp39-manylinux2014_aarch64.manylinux_2_17_aarch64.whl", hash = "sha256:9d3a9edcfbe77a3ed4bc72836d466dfce4174beb79eda79ea155cc77237ed9e8"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-manylinux2014_ppc64le.manylinux_2_17_ppc64le.whl", hash = "sha256:44fc5c2b8fa871ce7f0023f619f1349a0aa03a0857f2c96fbc01c657dcbbdb49"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-manylinux2014_x86_64.manylinux_2_17_x86_64.whl", hash = "sha256:9c55460033867b4622cda1b6872edf445809535144152e5d14941ef591980edf"}, + {file = "psycopg2_binary-2.9.11-cp39-cp39-manylinux_2_38_riscv64.manylinux_2_39_riscv64.whl", hash = "sha256:2d11098a83cca92deaeaed3d58cfd150d49b3b06ee0d0852be466bf87596899e"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-musllinux_1_2_aarch64.whl", hash = "sha256:691c807d94aecfbc76a14e1408847d59ff5b5906a04a23e12a89007672b9e819"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-musllinux_1_2_ppc64le.whl", hash = "sha256:8b81627b691f29c4c30a8f322546ad039c40c328373b11dff7490a3e1b517855"}, + {file = "psycopg2_binary-2.9.11-cp39-cp39-musllinux_1_2_riscv64.whl", hash = "sha256:b637d6d941209e8d96a072d7977238eea128046effbf37d1d8b2c0764750017d"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-musllinux_1_2_x86_64.whl", hash = "sha256:41360b01c140c2a03d346cec3280cf8a71aa07d94f3b1509fa0161c366af66b4"}, {file = "psycopg2_binary-2.9.11-cp39-cp39-win_amd64.whl", hash = "sha256:875039274f8a2361e5207857899706da840768e2a775bf8c65e82f60b197df02"}, ] @@ -5040,7 +5088,7 @@ files = [ {file = "pycparser-3.0-py3-none-any.whl", hash = "sha256:b727414169a36b7d524c1c3e31839a521725078d7b2ff038656844266160a992"}, {file = "pycparser-3.0.tar.gz", hash = "sha256:600f49d217304a5902ac3c37e1281c9fe94e4d0489de643a9504c5cdfdfc6b29"}, ] -markers = {main = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and platform_python_implementation != \"PyPy\" and implementation_name != \"PyPy\"", metrics = "sys_platform == \"win32\" and platform_python_implementation == \"CPython\" and implementation_name != \"PyPy\" or implementation_name == \"pypy\""} +markers = {main = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and platform_python_implementation != \"PyPy\" and implementation_name != \"PyPy\"", metrics = "(implementation_name == \"pypy\" or sys_platform == \"win32\") and (sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and (platform_python_implementation == \"CPython\" or implementation_name == \"pypy\") and implementation_name != \"PyPy\""} [[package]] name = "pydantic" @@ -5433,12 +5481,12 @@ version = "2.9.0.post0" description = "Extensions to the standard Python datetime module" optional = false python-versions = "!=3.0.*,!=3.1.*,!=3.2.*,>=2.7" -groups = ["metrics"] -markers = "sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\"" +groups = ["main", "metrics"] files = [ {file = "python-dateutil-2.9.0.post0.tar.gz", hash = "sha256:37dd54208da7e1cd875388217d5e00ebd4179249f90fb72437e91a35459a0ad3"}, {file = "python_dateutil-2.9.0.post0-py2.py3-none-any.whl", hash = "sha256:a8b2bc7bffae282281c8140a97d3aa9c14da0b136dfe83f850eea9a5f7470427"}, ] +markers = {main = "(sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\") and python_version >= \"3.11\"", metrics = "sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\""} [package.dependencies] six = ">=1.5" @@ -7209,7 +7257,7 @@ files = [ {file = "tzdata-2025.3-py2.py3-none-any.whl", hash = "sha256:06a47e5700f3081aab02b2e513160914ff0694bce9947d6b76ebd6bf57cfc5d1"}, {file = "tzdata-2025.3.tar.gz", hash = "sha256:de39c2ca5dc7b0344f2eba86f49d614019d29f060fc4ebc8a417896a620b56a7"}, ] -markers = {main = "sys_platform == \"win32\"", metrics = "sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\" and python_version == \"3.10\""} +markers = {main = "(sys_platform == \"win32\" or sys_platform == \"emscripten\") and python_version >= \"3.11\" or sys_platform == \"win32\"", metrics = "sys_platform == \"win32\" or sys_platform == \"emscripten\" or sys_platform != \"win32\" and sys_platform != \"emscripten\" and python_version == \"3.10\""} [[package]] name = "urllib3" @@ -7298,7 +7346,7 @@ description = "Fast implementation of asyncio event loop on top of libuv" optional = false python-versions = ">=3.8.1" groups = ["main"] -markers = "platform_python_implementation != \"PyPy\" and sys_platform != \"win32\" and sys_platform != \"cygwin\"" +markers = "sys_platform != \"win32\" and sys_platform != \"cygwin\" and platform_python_implementation != \"PyPy\"" files = [ {file = "uvloop-0.22.1-cp310-cp310-macosx_10_9_universal2.whl", hash = "sha256:ef6f0d4cc8a9fa1f6a910230cd53545d9a14479311e87e3cb225495952eb672c"}, {file = "uvloop-0.22.1-cp310-cp310-macosx_10_9_x86_64.whl", hash = "sha256:7cd375a12b71d33d46af85a3343b35d98e8116134ba404bd657b3b1d15988792"}, @@ -8226,4 +8274,4 @@ cffi = ["cffi (>=1.17,<2.0) ; platform_python_implementation != \"PyPy\" and pyt [metadata] lock-version = "2.1" python-versions = ">=3.10,<3.13" -content-hash = "9dc235ee46a73295928c93bd9429eb30a9ca29e3d49c3f60688a025410f6a8b3" +content-hash = "b128aad02a83e4c1fe5f9003b4cdc680b42d6b487b53d4d648c9ac8be49bfabb" diff --git a/pyproject.toml b/pyproject.toml index 10106074..f00a245d 100644 --- a/pyproject.toml +++ b/pyproject.toml @@ -50,6 +50,9 @@ aiohttp = "^3.13.3" httpx = "^0.28.1" beautifulsoup4 = "^4.14.3" json-repair = "^0.55.0" +baml-py = "^0.220.0" +baml-cli = "^0.1.0" +pandas = { version = "^3.0.3", markers = "python_version >= '3.11'" } [tool.poetry.group.dev.dependencies] pytest = "^7.4.0" diff --git a/src/app/baml_src/agents/tutor/activity_guide_generator.baml b/src/app/baml_src/agents/tutor/activity_guide_generator.baml new file mode 100644 index 00000000..3d0f70f4 --- /dev/null +++ b/src/app/baml_src/agents/tutor/activity_guide_generator.baml @@ -0,0 +1,109 @@ +// Activity Guide Generator Agent +// Personalizes activity template for specific course context + +function GenerateActivityGuide( + activity_template: ActivityTemplate, + session: SessionPlan, + metadata: CourseMetadata, + sustainability_map: SustainabilityIntegration, + output_language: string +) -> ActivityGuide { + client CustomFast + prompt #" + + You are a pedagogical designer who creates ready-to-use activity guides for professors. You take generic activity templates and personalize them with course-specific content, examples, and sustainability integration. + + + + Transform this generic activity template into a detailed, ready-to-use guide for this specific course. The professor should be able to follow this guide directly to implement the activity. + + + + + Name: {{ activity_template.name }} + Description: {{ activity_template.description }} + Estimated duration: {{ activity_template.estimated_duration }}h + + + + Session {{ session.session_number }} + Objectives: {{ session.objectives | join(", ") }} + Duration: {{ session.duration }}h + Class size: {{ session.class_size }} + + + + Discipline: {{ metadata.discipline }} + Topic: {{ metadata.topic }} + Level: {{ metadata.level }} + Format: {{ metadata.session_type }} - {{ metadata.session_mode }} + + + + {% for conn in sustainability_map.connections %} + Objective {{ conn.objective_number }}: {{ conn.connection_explanation[:200] }}... + {% endfor %} + + + + + Create a detailed guide with: + + 1. Activity description: Adapted to this course context (mention discipline, topic) + + 2. Learning objectives: Which objective numbers (from session.objectives) this addresses + + 3. Detailed steps: Concrete procedure professor can follow + - Be specific (not "discuss the topic" but "pose question X, give 5min for pairs to discuss, then collect 3 key points") + - Include timing for each step + - Use course-specific examples from the discipline + + 4. Teacher role: What professor does during activity + + 5. Student role: What students do, how they participate + + 6. Resources needed: Specific materials, handouts, tools + - Course-specific resources (not generic) + - Include digital tools if mode is Distanciel/Hybride + + 7. Timing breakdown: Detailed time allocation + - Example: {"introduction": 10, "pair_work": 20, "group_discussion": 30, "debrief": 10} + - Total should match activity template's estimated_duration + + 8. Evaluation method: How to assess if learning happened + - Formative assessment integrated into activity + + 9. Sustainability integration: HOW sustainability themes are woven into this activity + - Use sustainability connections for relevant objectives + - Be specific and natural (not tacked on) + + + + GENERIC: "Students discuss the topic in groups" + SPECIFIC: "Students analyze the case study of [discipline-specific example] in groups of 4. Each group receives a worksheet with 3 guiding questions: (1) What are the key [disciplinary concepts]? (2) How do [sustainability themes] factor into the analysis? (3) What trade-offs exist? Groups have 20 minutes to discuss and prepare a 2-minute summary." + + GENERIC: "Use sustainability principles" + SPECIFIC: "During the case analysis, students explicitly apply lifecycle thinking from SDG 12 resources. They map the full value chain, identifying environmental externalities at each stage. This connects to Objective 3's focus on system boundaries in [discipline]." + + + + Return a JSON object with: + { + "activity_name": "Adapted name if needed", + "description": "Course-specific description...", + "learning_objectives": [1, 3], // Objective numbers + "steps": ["Step 1...", "Step 2...", ...], + "teacher_role": "What professor does...", + "student_role": "What students do...", + "resources_needed": ["Resource 1", "Resource 2", ...], + "timing_breakdown": {"phase1": 15, "phase2": 30, ...}, + "evaluation_method": "How to assess...", + "sustainability_integration": "How sustainability is woven in..." + } + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/activity_recommendation.baml b/src/app/baml_src/agents/tutor/activity_recommendation.baml new file mode 100644 index 00000000..40b914f4 --- /dev/null +++ b/src/app/baml_src/agents/tutor/activity_recommendation.baml @@ -0,0 +1,105 @@ +// Activity Recommendation Agent +// Ranks and selects 2-4 pedagogical activities for session based on learning outcomes + +function RecommendActivities( + session: SessionPlan, + learning_outcomes: LearningOutcomes, + filtered_activities: ActivityTemplate[], + greencomp_methods: string[], + output_language: string +) -> ActivityRecommendations { + client CustomFast + prompt #" + + You are a pedagogical activity design expert. You select and rank activities based on learning outcomes, pedagogical fit, variety, and feasibility. + IMPORTANT: Activities should help students achieve the LEARNING OUTCOMES (what students will be able to demonstrate) rather than just covering objectives. + + + + From the pre-filtered activities, select 2-4 that best enable students to achieve the learning outcomes for this session. Ensure variety (mix of active/reflective, individual/group) and feasibility (total time fits session duration). + + + + + Session {{ session.session_number }} + Objectives covered: {{ session.objectives | join(", ") }} + Type: {{ session.type }} + Mode: {{ session.mode }} + Duration: {{ session.duration }} hours + Class size: {{ session.class_size }} + + + + Learning outcomes for this session (select activities to help achieve these): + {% for outcome in learning_outcomes.outcomes %} + {% if outcome.related_objectives | length > 0 %} + {% for obj_num in outcome.related_objectives %} + {% if obj_num in session.objectives %} + - Outcome {{ outcome.number }}: {{ outcome.text }} + Assessment method: {{ outcome.assessment_method }} + {% endif %} + {% endfor %} + {% endif %} + {% endfor %} + + + + GreenComp framework recommends the following teaching methods for these outcomes: + {% if greencomp_methods | length > 0 %} + {% for method in greencomp_methods %} + - {{ method }} + {% endfor %} + + Consider prioritizing activities that align with these evidence-based pedagogical methods from the GreenComp sustainability education framework. These methods have been specifically matched to the learning outcomes for this session. + {% else %} + No specific GreenComp method recommendations available for this session. + {% endif %} + + + + {% for activity in filtered_activities %} + {{ loop.index }}. {{ activity.name }} (ID: {{ activity.id }}) + Description: {{ activity.description }} + Estimated duration: {{ activity.estimated_duration }}h + Compatible with: {{ activity.compatible_types | join(", ") }} - {{ activity.compatible_modes | join(", ") }} + {% endfor %} + + + + + Selection criteria: + 1. Pedagogical fit: Activities that directly help students achieve the learning outcomes (not just cover objectives) + - Match activities to assessment methods specified in outcomes + - Ensure activities allow students to demonstrate competencies + 2. Variety: Mix of activity types (not all lecture, not all group work) + - Balance active (discussions, exercises) with reflective (analysis, writing) + - Balance individual, pair, and group activities + 3. Feasibility: Total estimated time ≤ session duration (leave buffer for transitions) + 4. Pedagogical progression: Activities build on each other within session + + Number to select: + - Short sessions (<2h): 2-3 activities + - Long sessions (≥2h): 3-4 activities + + Provide clear rationale: + - WHY each activity fits the LEARNING OUTCOMES (be specific about which outcomes) + - HOW activities complement each other + - TIME breakdown showing feasibility + + + + Return a JSON object with: + { + "session_number": {{ session.session_number }}, + "recommended_activities": [ + // Copy ActivityTemplate objects for selected activities + ], + "rationale": "Detailed explanation of selections, variety, and time management..." + } + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/competency_mapping.baml b/src/app/baml_src/agents/tutor/competency_mapping.baml new file mode 100644 index 00000000..ced50fb6 --- /dev/null +++ b/src/app/baml_src/agents/tutor/competency_mapping.baml @@ -0,0 +1,83 @@ +// Competency Mapping Agent +// Maps learning outcomes to GreenComp competencies + +function MapCompetencies( + outcomes: LearningOutcomes, + greencomp_framework: map?, + output_language: string +) -> CompetencyMappings { + client CustomFast + prompt #" + + You are an expert in the GreenComp framework (European sustainability competences). You map learning outcomes to relevant GreenComp competencies with clear pedagogical rationale. + + + + For each learning outcome, identify 1-3 relevant GreenComp competencies and provide brief rationale for the mapping. + + + + + {% for outcome in outcomes.outcomes %} + {{ outcome.number }}. {{ outcome.text }} + (Related to objectives: {{ outcome.related_objectives | join(", ") }}) + {% endfor %} + + + + Area 1: Embodying sustainability values + - C1: Valuing sustainability (reflect on personal values, identify and explain how they guide actions) + - C2: Supporting fairness (support equity and justice for current and future generations) + - C3: Promoting nature (acknowledge humans are part of nature, respect needs of other species) + + Area 2: Embracing complexity in sustainability + - C4: Systems thinking (approach sustainability challenges as complex systems) + - C5: Critical thinking (assess information and arguments, identify assumptions and limitations) + - C6: Problem framing (formulate current and future sustainability challenges as problems) + + Area 3: Envisioning sustainable futures + - C7: Futures literacy (envision alternative sustainable futures) + - C8: Adaptability (manage transitions and challenges in complex sustainability situations) + - C9: Exploratory thinking (adopt relational thinking, explore and link different disciplines) + + Area 4: Acting for sustainability + - C10: Political agency (navigate political systems, identify political responsibility) + - C11: Collective action (act for change collaboratively) + - C12: Individual initiative (identify own potential for sustainability, take action) + + + + + - Map each outcome to 1-3 GreenComp competencies (most relevant) + - Provide brief rationale (1-2 sentences) explaining WHY this competency fits + - Use official GreenComp codes and names (e.g., "C4: Systems thinking") + - Ensure diversity across outcomes (don't map everything to C4, distribute across framework) + - Be pedagogically honest - only map if there's genuine alignment + + + + Good mapping: + Outcome: "Analyze supply chain efficiency using lifecycle thinking and environmental impact metrics" + Competencies: ["C4: Systems thinking", "C9: Exploratory thinking"] + Rationale: "This outcome develops systems thinking (C4) by requiring students to view supply chains as interconnected systems with environmental feedback loops. It also promotes exploratory thinking (C9) by linking business operations (supply chain) with environmental science (lifecycle analysis)." + + + + Return a JSON object with: + { + "mappings": [ + { + "outcome_number": 1, + "greencomp_competencies": ["C4: Systems thinking", "C9: Exploratory thinking"], + "rationale": "Brief explanation..." + }, + ... + ] + } + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/course_description.baml b/src/app/baml_src/agents/tutor/course_description.baml new file mode 100644 index 00000000..bd08caa8 --- /dev/null +++ b/src/app/baml_src/agents/tutor/course_description.baml @@ -0,0 +1,58 @@ +// Course Description Agent +// Generates course description (150-250 words) balancing disciplinary focus with sustainability + +function GenerateCourseDescription( + mode: string, + metadata: CourseMetadata, + context_text: string, + output_language: string +) -> CourseDescription { + client CustomFast + prompt #" + + You are an expert course designer for French higher education, specializing in creating compelling course descriptions that balance disciplinary rigor with sustainability integration. + + + + Generate a course description (150-250 words) for the following course. The description should be academically rigorous, discipline-specific, and naturally integrate sustainability themes without sounding promotional or generic. + + + + {{ mode }} + + - Discipline: {{ metadata.discipline }} + - Topic: {{ metadata.topic }} + - Level: {{ metadata.level }} + - Number of sessions: {{ metadata.num_sessions }} + - Session format: {{ metadata.session_type }} ({{ metadata.session_mode }}) + - Class size: {{ metadata.class_size }} + + + {{ context_text }} + + + + + - Length: 150-250 words exactly + - Academic tone appropriate for French university context (enseignement supérieur) + - Be specific to the discipline and topic (avoid generic language) + - Naturally integrate sustainability themes relevant to the discipline + - Frame sustainability as enrichment, not compliance + - Focus on what students will learn and why it matters + - For Mode 2B with existing description: use provided description as-is + - For Mode 3 (metadata only): if context is sparse or generic, you may use general disciplinary knowledge but be specific + + + + Return a JSON object with: + { + "text": "The course description (150-250 words)", + "word_count": + } + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/learning_objectives.baml b/src/app/baml_src/agents/tutor/learning_objectives.baml new file mode 100644 index 00000000..fac95c54 --- /dev/null +++ b/src/app/baml_src/agents/tutor/learning_objectives.baml @@ -0,0 +1,87 @@ +// Learning Objectives Agent +// Generates 3-8 learning objectives (teacher-centered: what will be taught/covered) + +function GenerateLearningObjectives( + description: string, + context_text: string, + metadata: CourseMetadata, + mode: string, + output_language: string +) -> LearningObjectives { + client CustomFast + prompt #" + + You are a pedagogical expert specializing in learning objective design. You create clear learning objectives that describe what will be taught and covered in the course from the instructor's perspective. + + + + Generate learning objectives for this course. The number and complexity should match the course level and duration. + + For Mode 2B: Parse the provided objectives and structure them (do NOT generate new ones). + For all other modes: Generate new objectives. + + + + {{ mode }} + {{ description }} + {{ context_text }} + + - Discipline: {{ metadata.discipline }} + - Topic: {{ metadata.topic }} + - Level: {{ metadata.level }} + - Number of sessions: {{ metadata.num_sessions }} + + + + + - Number of objectives: + * Short courses (<6 sessions): 3-4 objectives + * Medium courses (6-12 sessions): 4-6 objectives + * Long courses (>12 sessions): 6-8 objectives + + - Each objective should be: + * Written from TEACHER PERSPECTIVE (what instructor intends to teach/cover) + * Use teacher-centered verbs: "Introduce", "Familiarize students with", "Provide an overview of", "Explore", "Cover", "Examine", "Present" + * Focused on instructional intent and content coverage + * Specific enough to guide course design + * Achievable within course duration + * Focused on disciplinary content (sustainability comes in next step) + + - Examples of good learning objectives: + * "Introduce students to the fundamentals of sustainable development" + * "Familiarize students with economic theories of market failures" + * "Provide an overview of climate change policy frameworks" + * "Explore the relationship between resource consumption and environmental impact" + + - AVOID student-centered language like: + * "Students will be able to..." + * "Students will demonstrate..." + * "Learners will..." + + - For Mode 2B: Keep objectives exactly as provided, just parse and structure them + + + + Return a JSON object with: + { + "objectives": [ + { + "number": 1, + "text": "Introduce students to sustainable development principles" + }, + { + "number": 2, + "text": "Familiarize students with environmental policy frameworks" + }, + ... + ] + } + + IMPORTANT: Write objectives from teacher perspective describing instructional intent, NOT student outcomes. Objectives should never be an empty list. + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/learning_outcomes.baml b/src/app/baml_src/agents/tutor/learning_outcomes.baml new file mode 100644 index 00000000..ab936d62 --- /dev/null +++ b/src/app/baml_src/agents/tutor/learning_outcomes.baml @@ -0,0 +1,79 @@ +// Learning Outcomes Agent +// Generates measurable learning outcomes (1-3 per objective) + +function GenerateLearningOutcomes( + objectives: LearningObjectives, + sustainability_map: SustainabilityIntegration, + metadata: CourseMetadata, + output_language: string +) -> LearningOutcomes { + client CustomFast + prompt #" + + You are a pedagogical assessment expert. You create measurable, observable learning outcomes that specify what students will be able to demonstrate upon completing the course. + + + + For each learning objective (including any suggested sustainability objectives), generate 1-3 specific, measurable outcomes. Naturally integrate sustainability themes informed by the sustainability mapping. + + + + + {% for obj in objectives.objectives %} + {{ obj.number }}. {{ obj.text }} ({{ obj.bloom_level }}) + {% endfor %} + + + + {% for conn in sustainability_map.connections %} + Objective {{ conn.objective_number }}: {{ conn.sdg_themes | join(", ") }} + Connection: {{ conn.connection_explanation }} + {% endfor %} + + + + - Discipline: {{ metadata.discipline }} + - Level: {{ metadata.level }} + - Duration: {{ metadata.num_sessions }} sessions + + + + + Each outcome should be: + - Measurable and observable (use action verbs: demonstrate, analyze, create, evaluate, design, etc.) + - Specific enough to guide activity design + - Achievable within course duration + - Related to 1-2 objectives (can span multiple) + - Include assessment method (how to measure: exam, project, presentation, quiz, etc.) + + Integration of sustainability: + - Weave sustainability naturally into outcomes (don't create separate "sustainability outcomes") + - Use the sustainability connections to inform outcome design + - Example: "Analyze supply chain efficiency using lifecycle thinking and environmental impact metrics" + vs separate: "Analyze supply chains" + "Understand environmental impacts" + + Number of outcomes: + - 1-3 outcomes per objective (more complex objectives may need more outcomes) + - Total outcomes: typically 1.5x-2x number of objectives + + + + Return a JSON object with: + { + "outcomes": [ + { + "number": 1, + "text": "Students will demonstrate...", + "related_objectives": [1, 2], + "assessment_method": "Case study analysis and written report" + }, + ... + ] + } + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/pedagogical_engineer.baml b/src/app/baml_src/agents/tutor/pedagogical_engineer.baml new file mode 100644 index 00000000..ccaec63c --- /dev/null +++ b/src/app/baml_src/agents/tutor/pedagogical_engineer.baml @@ -0,0 +1,130 @@ +// Pedagogical Engineer Agent +// Validates pedagogical coherence and quality + +function ValidatePedagogicalFramework( + description: string, + objectives: LearningObjectives, + outcomes: LearningOutcomes, + competencies: CompetencyMappings, + sustainability_map: SustainabilityIntegration, + metadata: CourseMetadata, + output_language: string +) -> ValidationReport { + client CustomFast + prompt #" + + You are an experienced pedagogical engineer (ingénieur pédagogique) providing constructive peer review of course frameworks. Your role is to validate overall coherence, quality, and feasibility. + + You should identify issues at three severity levels: + - MINOR: Formatting, wording, small inconsistencies (can be auto-fixed) + - MAJOR: Missing elements, serious misalignment, pedagogical errors (requires regeneration) + - SUBJECTIVE: Pedagogical choices that are valid but could be improved (recommendations for user) + + + + Review this complete pedagogical framework for overall coherence, quality, and feasibility. Provide specific, actionable feedback. + + + + {{ description }} + + + {% for obj in objectives.objectives %} + {{ obj.number }}. {{ obj.text }} ({{ obj.bloom_level }}) + {% endfor %} + + + + {% for outcome in outcomes.outcomes %} + {{ outcome.number }}. {{ outcome.text }} + Related objectives: {{ outcome.related_objectives | join(", ") }} + Assessment: {{ outcome.assessment_method }} + {% endfor %} + + + + {% for mapping in competencies.mappings %} + Outcome {{ mapping.outcome_number }}: {{ mapping.greencomp_competencies | join(", ") }} + {% endfor %} + + + + {% for conn in sustainability_map.connections %} + Objective {{ conn.objective_number }}: {{ conn.sdg_themes | join(", ") }} + {% endfor %} + Integration strategy: {{ sustainability_map.integration_strategy }} + + + + - Discipline: {{ metadata.discipline }} + - Level: {{ metadata.level }} + - Duration: {{ metadata.num_sessions }} sessions of {{ metadata.session_duration }}h + - Format: {{ metadata.session_type }} + + + + + Check for: + + 1. Coverage: + - All objectives have at least one outcome? ✓ + - All outcomes mapped to competencies? ✓ + + 2. Coherence: + - Outcomes align with objectives? ✓ + - Assessment methods appropriate for outcomes? ✓ + - Competency mappings make pedagogical sense? ✓ + + 3. Feasibility: + - Objectives achievable in given number of sessions? ✓ + - Outcomes realistic for course level and duration? ✓ + - Assessment methods practical for class size and format? ✓ + + 4. Quality: + - Objectives specific and measurable? ✓ + - Outcomes observable and assessable? ✓ + - Sustainability integration authentic (not forced or tacked on)? ✓ + + 5. Red flags: + - Overly ambitious for duration? ✗ + - Misalignment between objectives and outcomes? ✗ + - Generic or vague language? ✗ + - Too many or too few objectives/outcomes? ✗ + + + + - Be constructive and specific (peer review tone, not judgmental) + - Reference pedagogical best practices + - Classify severity: minor, major, or subjective + - For major issues: explain what needs to change AND provide corrected version + - For minor issues: provide auto-fix suggestions + - For subjective: frame as recommendations, not requirements + - If no issues: passed = true, empty issues list + - IMPORTANT: Always provide specific corrections that can be applied automatically + + + + MINOR issue: "Outcome 3 uses passive voice ('will be understood'). Rephrase with active, observable verb." + + MAJOR issue: "Objective 1 has no corresponding outcomes. At least one outcome must be created to measure this objective." + + SUBJECTIVE recommendation: "Consider balancing Bloom levels. Currently 5/6 objectives are 'Analyze' level. Adding one 'Create' level objective could enhance higher-order thinking for M2 students." + + + + Return a JSON object with: + { + "passed": true/false, + "issues": ["Issue 1...", "Issue 2..."], // Empty if passed + "suggestions": ["Suggestion 1...", "Suggestion 2..."], // Actionable feedback + "severity": "minor" | "major" | "subjective" // Highest severity if multiple issues + } + + If everything is valid and high quality: {"passed": true, "issues": [], "suggestions": [], "severity": "minor"} + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/agents/tutor/sustainability_integration.baml b/src/app/baml_src/agents/tutor/sustainability_integration.baml new file mode 100644 index 00000000..cb8985b3 --- /dev/null +++ b/src/app/baml_src/agents/tutor/sustainability_integration.baml @@ -0,0 +1,109 @@ +// Sustainability Integration Agent +// CRITICAL: Creates authentic, disciplinarily-relevant sustainability connections (core value proposition) + +function IntegrateSustainability( + description: string, + objectives: LearningObjectives, + sdg_resources: Document[], + metadata: CourseMetadata, + mode: string, + output_language: string +) -> SustainabilityIntegration { + client CustomFast + prompt #" + + You are an expert in sustainability education and disciplinary pedagogy. Your role is to create AUTHENTIC, DEEP connections between course objectives and sustainability themes using SPECIFIC examples from provided SDG resources. + + This is the CORE VALUE PROPOSITION of WeLearn. You must avoid generic SDG labels and instead explain HOW sustainability enriches the disciplinary concept with concrete examples. + + + + For EACH learning objective, identify relevant sustainability themes from the provided SDG resources and explain the SPECIFIC connection using examples from those resources. + + For Mode 2B ONLY: Also suggest 2-3 ADDITIONAL sustainability-focused objectives that complement (don't replace) the original objectives. + + + + {{ mode }} + {{ description }} + + {% for obj in objectives.objectives %} + {{ obj.number }}. {{ obj.text }} (Bloom: {{ obj.bloom_level }}) + {% endfor %} + + + + {% for doc in sdg_resources %} + Document {{ loop.index }}: + Title: {{ doc.title | default("Untitled") }} + Relevance: {{ doc.relevance_score }} + URL: {{doc.url}} + Content excerpt: {{ doc.text[:500] }}... + {% endfor %} + + + + - Discipline: {{ metadata.discipline }} + - Topic: {{ metadata.topic }} + - Level: {{ metadata.level }} + + + + + CRITICAL - This is what separates good from mediocre output: + + ❌ BAD (generic): "Objective 2 relates to SDG 13 (Climate Action)" + + ✅ GOOD (authentic): "Objective 2 explores market failures. SDG resources on carbon pricing (Document 3, pp. 15-23) provide concrete examples where markets fail to price environmental externalities. Use the European carbon tax case study from Document 5 to illustrate how economic instruments can correct this market failure, directly connecting microeconomic theory to climate policy." + + For each objective: + 1. Identify 1-3 relevant SDG themes from the resources + 2. Cite SPECIFIC documents and examples (document titles, page numbers if available) + 3. Explain HOW sustainability connects to the disciplinary concept (mechanism, not just label) + 4. Use academic language appropriate for the discipline + 5. Frame as enrichment that makes the learning deeper, not as compliance + + For Mode 2B: Suggest 2-3 additional sustainability-focused objectives that: + - Are appropriate for the course level + - Build on existing objectives + - Are achievable in course duration + - Feel natural, not forced + + Overall integration strategy: + - Write a brief paragraph (3-5 sentences) explaining the overarching approach to integrating sustainability in this course + - How does it fit the discipline? Why is it pedagogically valuable? + + + + Good connection explanation: + "This objective on analyzing supply chain efficiency (Objective 3) connects to SDG 12 (Responsible Consumption) through concrete examples in Document 2. The case study of circular economy in manufacturing (Document 2, Section 4.2) demonstrates how lifecycle thinking transforms traditional supply chain analysis. Students can apply the same analytical frameworks they learn for cost optimization to evaluate environmental impacts, using the methodology described in Document 7 (Environmental Supply Chain Management Framework)." + + + + Return a JSON object with: + { + "connections": [ + { + "objective_number": 1, + "sdg_themes": ["SDG 12: Responsible Consumption", "SDG 13: Climate Action"], + "connection_explanation": "Detailed, specific explanation using examples from resources...", + "key_resources": [{title: title, url: url, corpus: corpus_name, sdg: [1, 2], description: description, text: slice, relevance_scrore: 0.6}] + }, + ... + ], + "suggested_objectives": [ // ONLY for Mode 2B + { + "number": 99, // Use high numbers to distinguish from original + "text": "Introduce different theories of...", + } + ], + "integration_strategy": "Overall narrative paragraph...", + "resources_used": [...] // Copy from sdg_resources input + } + + + + Generate all outputs in {{ output_language }}. + + "# +} diff --git a/src/app/baml_src/clients.baml b/src/app/baml_src/clients.baml new file mode 100644 index 00000000..00d417f6 --- /dev/null +++ b/src/app/baml_src/clients.baml @@ -0,0 +1,60 @@ +// Learn more about clients at https://docs.boundaryml.com/docs/snippets/clients/overview + + +// Example Azure OpenAI client (uncomment to use) +client CustomAzure_sweden { + provider openai + options { + model "gpt-oss-120b" + api_key env.AZURE_LLM_SWEDEN_API_KEY + base_url env.AZURE_BAML_LLM_SWEDEN_BASE_URL + } +} + +// client CustomAzure_uk { +// provider azure-openai +// options { +// model "gpt-5" +// api_key env.AZURE_LLM_SWEDEN_API_KEY +// base_url env.AZURE_LLM_UK_BASE_URL +// api_version "2024-05-01-preview" +// } +// } + +client CustomAzure_france { + provider openai + options { + model "gpt-oss-120b" + api_key env.AZURE_LLM_FRANCE_API_KEY + base_url env.AZURE_BAML_LLM_FRANCE_BASE_URL + } +} + + // load balancer client that alternates between the two Azure clients + client CustomFast { + provider round-robin + options { + // This will alternate between the two clients + strategy [CustomAzure_france, CustomAzure_sweden] + } + } + + +// https://docs.boundaryml.com/docs/snippets/clients/retry +retry_policy Constant { + max_retries 3 + strategy { + type constant_delay + delay_ms 200 + } +} + +retry_policy Exponential { + max_retries 2 + strategy { + type exponential_backoff + delay_ms 300 + multiplier 1.5 + max_delay_ms 10000 + } +} \ No newline at end of file diff --git a/src/app/baml_src/generators.baml b/src/app/baml_src/generators.baml new file mode 100644 index 00000000..cd86297c --- /dev/null +++ b/src/app/baml_src/generators.baml @@ -0,0 +1,18 @@ +// This helps use auto generate libraries you can use in the language of +// your choice. You can have multiple generators if you use multiple languages. +// Just ensure that the output_dir is different for each generator. +generator target { + // Valid values: "python/pydantic", "typescript", "go", "rust", "ruby/sorbet", "rest/openapi" + output_type "python/pydantic" + + // Where the generated code will be saved (relative to baml_src/) + output_dir "../" + + // The version of the BAML package you have installed (e.g. same version as your baml-py or @boundaryml/baml). + // The BAML VSCode extension version should also match this version. + version "0.220.0" + + // Valid values: "sync", "async" + // This controls what `b.FunctionName()` will be (sync or async). + default_client_mode sync +} diff --git a/src/app/baml_src/resume.baml b/src/app/baml_src/resume.baml new file mode 100644 index 00000000..5a37c5aa --- /dev/null +++ b/src/app/baml_src/resume.baml @@ -0,0 +1,42 @@ +// Defining a data model. +class Resume { + name string + email string + experience string[] + skills string[] +} + +// Create a function to extract the resume from a string. +function ExtractResume(resume: string) -> Resume { + // Specify a client as provider/model-name + // You can also use custom LLM params with a custom client name from clients.baml like "client CustomGPT5" or "client CustomSonnet4" + client CustomFast // Set OPENAI_API_KEY to use this client. + prompt #" + Extract from this content: + {{ resume }} + + {{ ctx.output_format }} + "# +} + + + +// Test the function with a sample resume. Open the VSCode playground to run this. +test vaibhav_resume { + functions [ExtractResume] + args { + resume #" + Vaibhav Gupta + vbv@boundaryml.com + + Experience: + - Founder at BoundaryML + - CV Engineer at Google + - CV Engineer at Microsoft + + Skills: + - Rust + - C++ + "# + } +} diff --git a/src/app/baml_src/types.baml b/src/app/baml_src/types.baml new file mode 100644 index 00000000..470fe608 --- /dev/null +++ b/src/app/baml_src/types.baml @@ -0,0 +1,131 @@ +// BAML Type Definitions for WeLearn MAS +// These types match the Pydantic models in models.py + +// Enums +enum SessionMode { + PRESENTIEL @alias("Présentiel") + DISTANCIEL @alias("Distanciel") + HYBRIDE @alias("Hybride") +} + +// Core Types +class CourseMetadata { + discipline string + topic string + level string + num_sessions int + session_duration float + session_type string + session_mode SessionMode + class_size int + output_language string + user_description string? +} + +class Document { + text string + corpus string + description string + sdg int[] + url string + title string + relevance_score float? +} + +class CourseDescription { + text string + word_count int +} + +class LearningObjective { + number int + text string + bloom_level string? +} + +class LearningObjectives { + objectives LearningObjective[] +} + +class SustainabilityConnection { + objective_number int + sdg_themes string[] + connection_explanation string + key_resources Document[] +} + +class SustainabilityIntegration { + connections SustainabilityConnection[] + suggested_objectives LearningObjective[]? + integration_strategy string + resources_used Document[] +} + +class LearningOutcome { + number int + text string + related_objectives int[] + assessment_method string +} + +class LearningOutcomes { + outcomes LearningOutcome[] +} + +class CompetencyMapping { + outcome_number int + greencomp_competencies string[] + rationale string +} + +class CompetencyMappings { + mappings CompetencyMapping[] +} + +class ValidationReport { + passed bool + issues string[] + suggestions string[] + severity string +} + +class SessionPlan { + session_number int + objectives int[] + type string + mode SessionMode + duration float + class_size int +} + +class ActivityTemplate { + id string + name string + description string + compatible_types string[] + compatible_modes SessionMode[] + min_duration float + max_duration float + min_size int + max_size int + estimated_duration float +} + +class ActivityGuide { + activity_name string + description string + learning_objectives int[] + steps string[] + teacher_role string + student_role string + resources_needed string[] + timing_breakdown map + evaluation_method string + sustainability_integration string +} + +class ActivityRecommendations { + session_number int + recommended_activities ActivityTemplate[] + rationale string +} diff --git a/src/app/tutor/api/router.py b/src/app/tutor/api/router.py index faae374f..b9d9b7b6 100644 --- a/src/app/tutor/api/router.py +++ b/src/app/tutor/api/router.py @@ -11,6 +11,7 @@ UploadFile, ) +from src.app.baml_client.async_client import b, types from src.app.core.config import Settings from src.app.search.helpers.search_helpers import search_multi_inputs from src.app.search.models.search import EnhancedSearchQuery @@ -21,17 +22,27 @@ from src.app.shared.utils.dependencies import get_settings from src.app.shared.utils.requests import extract_session_cookie from src.app.shared.utils.utils import get_files_content + +# from src.app.tutor.service.agents import TEMPLATES from src.app.tutor.service.agents import TEMPLATES from src.app.tutor.service.models import ( + CompetencyMappingRequest, + CourseDescriptionRequest, ExtractorOutputList, + IntegrateSustainabilityRequest, + LearningObjectivesRequest, + LearningOutcomesRequest, SummariesList, SyllabusFeedback, + SyllabusGenerationRequest, SyllabusResponse, SyllabusResponseAgent, SyllabusUserUpdate, TutorSearchResponse, TutorSyllabusRequest, + UserInput, ) +from src.app.tutor.service.orchestrator import SyllabusOrchestrator from src.app.tutor.service.prompts import ( extractor_system_prompt, extractor_user_prompt, @@ -291,3 +302,212 @@ async def register_syllabus_user_update( ) return {"message": "user update registered"} + + +@router.post("/syllabus/description") +async def baml_test(body: CourseDescriptionRequest) -> types.CourseDescription: + b_response = await b.GenerateCourseDescription( + mode=body.mode, + metadata=types.CourseMetadata( + discipline=body.course_metadata.discipline, + topic=body.course_metadata.topic, + level=body.course_metadata.level, + num_sessions=body.course_metadata.num_sessions, + session_duration=body.course_metadata.session_duration, + session_type=body.course_metadata.session_type, + session_mode=body.course_metadata.session_mode, # type: ignore + class_size=body.course_metadata.class_size, + output_language=body.course_metadata.output_language, + user_description=body.course_metadata.user_description, + ), + context_text=body.context_text, + output_language=body.course_metadata.output_language, + ) + return b_response + + +@with_backoff() +@router.post("/syllabus/learning_objectives") +async def baml_learning_objectives( + body: LearningObjectivesRequest, +) -> types.LearningObjectives: + b_response = await b.GenerateLearningObjectives( + description=body.description, + mode=body.mode, + metadata=types.CourseMetadata( + discipline=body.course_metadata.discipline, + topic=body.course_metadata.topic, + level=body.course_metadata.level, + num_sessions=body.course_metadata.num_sessions, + session_duration=body.course_metadata.session_duration, + session_type=body.course_metadata.session_type, + session_mode=body.course_metadata.session_mode, # type: ignore + class_size=body.course_metadata.class_size, + output_language=body.course_metadata.output_language, + ), + context_text=body.context_text, + output_language=body.course_metadata.output_language, + ) + + return b_response + + +@with_backoff() +@router.post("/syllabus/sustainability_integration") +async def integrate_sustainability( + body: IntegrateSustainabilityRequest, +) -> types.SustainabilityIntegration: + print(body.course_metadata) + # session_id = request.headers.get("X-Session-ID") + + metadata = types.CourseMetadata( + discipline=body.course_metadata.discipline, + topic=body.course_metadata.topic, + level=body.course_metadata.level, + num_sessions=body.course_metadata.num_sessions, + session_duration=body.course_metadata.session_duration, + session_type=body.course_metadata.session_type, + session_mode=body.course_metadata.session_mode, # type: ignore + class_size=body.course_metadata.class_size, + output_language=body.course_metadata.output_language, + ) + + b_response = await b.IntegrateSustainability( + description=body.description, + objectives=types.LearningObjectives( + objectives=[ + types.LearningObjective( + number=obj.number, + text=obj.text, + bloom_level=obj.bloom_level, + ) + for obj in body.objectives.objectives + ] + ), + metadata=metadata, + sdg_resources=[ + types.Document( + text=res.text, + metadata=res.metadata, + relevance_score=res.relevance_score if res.relevance_score else None, + ) + for res in body.sdg_resources + ], + mode=body.mode, + output_language=body.course_metadata.output_language, + ) + + return b_response + + +@with_backoff() +@router.post("/syllabus/learning_outcomes") +async def baml_learning_outcomes( + body: LearningOutcomesRequest, +) -> types.LearningOutcomes: + b_response = await b.GenerateLearningOutcomes( + metadata=types.CourseMetadata( + discipline=body.course_metadata.discipline, + topic=body.course_metadata.topic, + level=body.course_metadata.level, + num_sessions=body.course_metadata.num_sessions, + session_duration=body.course_metadata.session_duration, + session_type=body.course_metadata.session_type, + session_mode=body.course_metadata.session_mode, # type: ignore + class_size=body.course_metadata.class_size, + output_language=body.course_metadata.output_language, + ), + output_language=body.course_metadata.output_language, + sustainability_map=body.sustainability_map, + objectives=types.LearningObjectives( + objectives=[ + types.LearningObjective( + number=obj.number, + text=obj.text, + bloom_level=obj.bloom_level, + ) + for obj in body.objectives.objectives + ] + ), + ) + + return b_response + + +@with_backoff() +@router.post("/syllabus/competency_map") +async def generate_competency_map( + body: CompetencyMappingRequest, +) -> types.CompetencyMappings: + # session_id = request.headers.get("X-Session-ID") + + competencies = await b.MapCompetencies( + outcomes=types.LearningOutcomes( + outcomes=[ + types.LearningOutcome( + number=outcome.number, + text=outcome.text, + related_objectives=[], # This field is not used in the BAML function, so we can leave it empty + assessment_method="", # This field is not used in the BAML function, so we can leave it empty + ) + for outcome in body.outcomes + ] + ), + output_language=body.output_language, + greencomp_framework=body.framework, + ) + + return competencies + + +@router.post("/api/generate") +async def generate_syllabus(body: SyllabusGenerationRequest): + + try: + metadata = types.CourseMetadata( + discipline=body.course_metadata.discipline, + topic=body.course_metadata.topic, + level=body.course_metadata.level, + num_sessions=body.course_metadata.num_sessions, + session_duration=body.course_metadata.session_duration, + session_type=body.course_metadata.session_type, + session_mode=body.course_metadata.session_mode, + class_size=body.course_metadata.class_size, + output_language=body.course_metadata.output_language, + ) + + user_input = UserInput( + metadata=metadata, + mode=body.mode, + rag_resources=body.rag_resources, + provided_description=( + body.course_metadata.user_description + if body.course_metadata.user_description + else None + ), + ) + orchestrator = SyllabusOrchestrator() + output = await orchestrator.run(user_input) + + return { + "status": "success", + "validation": { + "passed": True, + "major_issues": [], + "minor_issues": [], + "suggestions": [], + }, # Placeholder for validation results + **output, + } + except Exception as e: + logger.error(f"Error generating syllabus: {e}") + return { + "status": "error", + "message": str(e), + "validation": { + "passed": False, + "major_issues": [str(e)], + "minor_issues": [], + "suggestions": [], + }, + } diff --git a/src/app/tutor/service/b_agents/__init__.py b/src/app/tutor/service/b_agents/__init__.py new file mode 100644 index 00000000..e01a7f1b --- /dev/null +++ b/src/app/tutor/service/b_agents/__init__.py @@ -0,0 +1,26 @@ +""" +Agent wrappers - Python interface to BAML agents +""" + +from .activity_guide_generator_agent import ActivityGuideGeneratorAgent +from .activity_recommendation_agent import ActivityRecommendationAgent +from .base import AgentError, BaseAgent +from .competency_mapping_agent import CompetencyMappingAgent +from .course_description_agent import CourseDescriptionAgent +from .learning_objectives_agent import LearningObjectivesAgent +from .learning_outcomes_agent import LearningOutcomesAgent +from .pedagogical_engineer_agent import PedagogicalEngineerAgent +from .sustainability_integration_agent import SustainabilityIntegrationAgent + +__all__ = [ + "BaseAgent", + "AgentError", + "CourseDescriptionAgent", + "LearningObjectivesAgent", + "SustainabilityIntegrationAgent", + "LearningOutcomesAgent", + "CompetencyMappingAgent", + "PedagogicalEngineerAgent", + "ActivityRecommendationAgent", + "ActivityGuideGeneratorAgent", +] diff --git a/src/app/tutor/service/b_agents/activity_guide_generator_agent.py b/src/app/tutor/service/b_agents/activity_guide_generator_agent.py new file mode 100644 index 00000000..c6af28c6 --- /dev/null +++ b/src/app/tutor/service/b_agents/activity_guide_generator_agent.py @@ -0,0 +1,47 @@ +"""Activity Guide Generator Agent wrapper""" + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class ActivityGuideGeneratorAgent(BaseAgent): + """Generates personalized activity guides""" + + def __init__(self): + super().__init__("activity_guide_generator") + + async def generate( + self, + activity_template: types.ActivityTemplate, + session: types.SessionPlan, + metadata: types.CourseMetadata, + sustainability_map: types.SustainabilityIntegration, + output_language: str = "Français", + ) -> types.ActivityGuide: + """Generate activity guide""" + self.logger.info(f"Generating guide for activity {activity_template.name}") + + async def _call_baml(): + result = await b.GenerateActivityGuide( + activity_template=activity_template, + session=session, + metadata=metadata, + sustainability_map=sustainability_map, + output_language=output_language, + ) + # Convert BAML types to Pydantic models + return types.ActivityGuide( + activity_name=result.activity_name, + description=result.description, + learning_objectives=result.learning_objectives, + steps=result.steps, + teacher_role=result.teacher_role, + student_role=result.student_role, + resources_needed=result.resources_needed, + timing_breakdown=result.timing_breakdown, + evaluation_method=result.evaluation_method, + sustainability_integration=result.sustainability_integration, + ) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/b_agents/activity_recommendation_agent.py b/src/app/tutor/service/b_agents/activity_recommendation_agent.py new file mode 100644 index 00000000..38635486 --- /dev/null +++ b/src/app/tutor/service/b_agents/activity_recommendation_agent.py @@ -0,0 +1,92 @@ +"""Activity Recommendation Agent wrapper""" + +from typing import List + +from src.app.baml_client.async_client import b, types +from src.app.tutor.tools import ( + get_suitable_methods_for_outcome_text, + load_greencomp_methods, +) + +from .base import BaseAgent + + +class ActivityRecommendationAgent(BaseAgent): + """Recommends activities for sessions using GreenComp learning methods guidance""" + + def __init__(self): + super().__init__("activity_recommendation") + # Pre-load GreenComp methods for reference + self.greencomp_methods = load_greencomp_methods() + + async def recommend( + self, + session: types.SessionPlan, + learning_outcomes: types.LearningOutcomes, + filtered_activities: List[types.ActivityTemplate], + output_language: str = "Français", + ) -> types.ActivityRecommendations: + """ + Recommend activities based on learning outcomes + + Args: + session: Session plan with objectives + learning_outcomes: All learning outcomes (filtered to session objectives in prompt) + filtered_activities: Pre-filtered activity templates + output_language: Output language + + Returns: + ActivityRecommendations with selected activities and rationale + """ + self.logger.info( + f"Recommending activities for session {session.session_number} " + f"based on learning outcomes" + ) + + # Get GreenComp-recommended methods for session outcomes + # This helps guide activity selection based on pedagogical best practices + session_outcomes = [ + out + for out in learning_outcomes.outcomes + if any(obj in session.objectives for obj in out.related_objectives) + ] + + greencomp_suggestions = [] + for outcome in session_outcomes[:3]: # Top 3 outcomes + methods = get_suitable_methods_for_outcome_text(outcome.text) + if methods: + greencomp_suggestions.extend([m["name"] for m in methods[:2]]) + + self.logger.info(f"GreenComp suggests methods: {greencomp_suggestions}") + + async def _call_baml(): + # Pass GreenComp method suggestions to guide activity selection + result = await b.RecommendActivities( + session=session, + learning_outcomes=learning_outcomes, + filtered_activities=filtered_activities, + greencomp_methods=greencomp_suggestions, + output_language=output_language, + ) + # Convert BAML types to Pydantic models + return types.ActivityRecommendations( + session_number=result.session_number, + recommended_activities=[ + types.ActivityTemplate( + id=activity.id, + name=activity.name, + description=activity.description, + compatible_types=activity.compatible_types, + compatible_modes=activity.compatible_modes, + min_duration=activity.min_duration, + max_duration=activity.max_duration, + min_size=activity.min_size, + max_size=activity.max_size, + estimated_duration=activity.estimated_duration, + ) + for activity in result.recommended_activities + ], + rationale=result.rationale, + ) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/b_agents/base.py b/src/app/tutor/service/b_agents/base.py new file mode 100644 index 00000000..8e940a0b --- /dev/null +++ b/src/app/tutor/service/b_agents/base.py @@ -0,0 +1,165 @@ +""" +Base agent class with retry logic and error handling +""" + +import logging +import time +from typing import Any, Callable, Dict, TypeVar + +logger = logging.getLogger(__name__) + +T = TypeVar("T") + + +class AgentError(Exception): + """Custom exception for agent errors""" + + pass + + +class BaseAgent: + """Base class for all agents with common functionality""" + + def __init__(self, name: str, max_retries: int = 3): + """ + Initialize agent + + Args: + name: Agent name for logging + max_retries: Maximum number of retry attempts + """ + self.name = name + self.max_retries = max_retries + self.logger = logging.getLogger(f"agent.{name}") + self.total_tokens = 0 # Track total tokens used by this agent + self.call_count = 0 # Track number of calls + + def with_retry(self, func: Callable[..., T], *args, **kwargs) -> T: + """ + Execute function with retry logic and token tracking + + Args: + func: Function to execute + *args, **kwargs: Arguments to pass to function + + Returns: + Function result + + Raises: + AgentError: If all retries fail + """ + attempt = 1 + last_error = None + + while attempt <= self.max_retries: + try: + start_time = time.time() + self.logger.info(f"Attempt {attempt}/{self.max_retries}") + + result = func(*args, **kwargs) + + duration = time.time() - start_time + self.logger.info(f"Success in {duration:.2f}s") + + # Estimate token usage (rough approximation) + # BAML doesn't expose token counts directly, so we estimatebased on input/output sizes + # 1 token ≈ 4 characters for English text + estimated_tokens = self._estimate_tokens(args, kwargs, result) + self.total_tokens += estimated_tokens + self.call_count += 1 + + self.logger.info(f"Estimated tokens for this call: {estimated_tokens}") + self.logger.info( + f"Total tokens used by {self.name}: {self.total_tokens}" + ) + + return result + + except Exception as e: + last_error = e + duration = time.time() - start_time + + self.logger.warning( + f"Attempt {attempt}/{self.max_retries} failed after {duration:.2f}s: {str(e)}" + ) + + if attempt < self.max_retries: + # Exponential backoff + wait_time = 2 ** (attempt - 1) + self.logger.info(f"Retrying in {wait_time}s...") + time.sleep(wait_time) + + attempt += 1 + + # All retries failed + error_msg = ( + f"{self.name} failed after {self.max_retries} attempts: {str(last_error)}" + ) + self.logger.error(error_msg) + raise AgentError(error_msg) from last_error + + def _estimate_tokens(self, args, kwargs, result) -> int: + """ + Estimate token usage based on input/output sizes + + Rough approximation: 1 token ≈ 4 characters + + Args: + args: Function args + kwargs: Function kwargs + result: Function result + + Returns: + Estimated token count + """ + + def count_chars(obj): + """Recursively count characters in an object""" + if isinstance(obj, str): + return len(obj) + elif isinstance(obj, (list, tuple)): + return sum(count_chars(item) for item in obj) + elif isinstance(obj, dict): + return sum(count_chars(v) for v in obj.values()) + elif hasattr(obj, "__dict__"): + return count_chars(obj.__dict__) + else: + return len(str(obj)) + + input_chars = sum(count_chars(arg) for arg in args) + input_chars += sum(count_chars(v) for v in kwargs.values()) + output_chars = count_chars(result) + + # Total chars / 4 ≈ tokens + # Multiply by 1.5 to account for prompt structure, formatting, etc. --> TODO:NOT SURE ABOUT THIS + estimated = int((input_chars + output_chars) / 4 * 1.5) + + return estimated + + def log_call( + self, + inputs: Dict[str, Any], + outputs: Any, + duration: float, + tokens: Dict[str, int] = None, + ): + """ + Log agent call details + + Args: + inputs: Input parameters + outputs: Output result + duration: Execution duration in seconds + tokens: Token usage (prompt, completion) + """ + log_data = { + "agent": self.name, + "duration_s": duration, + "inputs_preview": {k: str(v)[:100] for k, v in inputs.items()}, + "outputs_type": type(outputs).__name__, + } + + if tokens: + log_data["tokens"] = tokens + + self.logger.info(f"Call completed: {log_data}") diff --git a/src/app/tutor/service/b_agents/competency_mapping_agent.py b/src/app/tutor/service/b_agents/competency_mapping_agent.py new file mode 100644 index 00000000..a19627bf --- /dev/null +++ b/src/app/tutor/service/b_agents/competency_mapping_agent.py @@ -0,0 +1,43 @@ +"""Competency Mapping Agent wrapper""" + +from typing import Dict + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class CompetencyMappingAgent(BaseAgent): + """Maps outcomes to GreenComp competencies""" + + def __init__(self): + super().__init__("competency_mapping") + + async def map_competencies( + self, + outcomes: types.LearningOutcomes, + greencomp_framework: Dict[str, str], + output_language: str = "Français", + ) -> types.CompetencyMappings: + """Map competencies""" + self.logger.info("Mapping competencies") + + async def _call_baml(): + result = await b.MapCompetencies( + outcomes=outcomes, + greencomp_framework=greencomp_framework, + output_language=output_language, + ) + # Convert BAML types to Pydantic models + return types.CompetencyMappings( + mappings=[ + types.CompetencyMapping( + outcome_number=mapping.outcome_number, + greencomp_competencies=mapping.greencomp_competencies, + rationale=mapping.rationale, + ) + for mapping in result.mappings + ] + ) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/b_agents/course_description_agent.py b/src/app/tutor/service/b_agents/course_description_agent.py new file mode 100644 index 00000000..b7db8be8 --- /dev/null +++ b/src/app/tutor/service/b_agents/course_description_agent.py @@ -0,0 +1,50 @@ +"""Course Description Agent wrapper""" + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class CourseDescriptionAgent(BaseAgent): + """Generates course descriptions""" + + def __init__(self): + super().__init__("course_description") + + async def generate( + self, + mode: str, + metadata: types.CourseMetadata, + context_text: str = "", + output_language: str = "Français", + ) -> types.CourseDescription: + """ + Generate course description + + Args: + mode: Input mode (mode_1, mode_2a, mode_2b, mode_3) + metadata: Course metadata + context_text: Contextual text (documents, description) + output_language: Output language + + Returns: + CourseDescription object + """ + self.logger.info(f"Generating course description (mode={mode})") + + async def _call_baml(): + result = await b.GenerateCourseDescription( + mode=mode, + metadata=metadata, + context_text=context_text, + output_language=output_language, + ) + # Convert BAML types back to Pydantic models + return types.CourseDescription( + text=result.text, word_count=result.word_count + ) + + return await self.with_retry(_call_baml) + + +# Additional agent wrappers follow similar pattern diff --git a/src/app/tutor/service/b_agents/learning_objectives_agent.py b/src/app/tutor/service/b_agents/learning_objectives_agent.py new file mode 100644 index 00000000..1bd50fc1 --- /dev/null +++ b/src/app/tutor/service/b_agents/learning_objectives_agent.py @@ -0,0 +1,44 @@ +"""Learning Objectives Agent wrapper""" + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class LearningObjectivesAgent(BaseAgent): + """Generates learning objectives""" + + def __init__(self): + super().__init__("learning_objectives") + + async def generate( + self, + description: str, + context_text: str, + metadata: types.CourseMetadata, + mode: str, + output_language: str = "Français", + ) -> types.LearningObjectives: + """Generate learning objectives""" + self.logger.info(f"Generating learning objectives (mode={mode})") + + async def _call_baml(): + result = await b.GenerateLearningObjectives( + description=description, + context_text=context_text, + metadata=metadata, + mode=mode, + output_language=output_language, + ) + # Convert BAML types to Pydantic models and ensure sequential numbering + objectives = [] + for i, obj in enumerate(result.objectives, start=1): + objectives.append( + types.LearningObjective( + number=i, # Force sequential numbering starting from 1 + text=obj.text, + ) + ) + return types.LearningObjectives(objectives=objectives) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/b_agents/learning_outcomes_agent.py b/src/app/tutor/service/b_agents/learning_outcomes_agent.py new file mode 100644 index 00000000..3600ce4e --- /dev/null +++ b/src/app/tutor/service/b_agents/learning_outcomes_agent.py @@ -0,0 +1,44 @@ +"""Learning Outcomes Agent wrapper""" + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class LearningOutcomesAgent(BaseAgent): + """Generates learning outcomes""" + + def __init__(self): + super().__init__("learning_outcomes") + + async def generate( + self, + objectives: types.LearningObjectives, + sustainability_map: types.SustainabilityIntegration, + metadata: types.CourseMetadata, + output_language: str = "Français", + ) -> types.LearningOutcomes: + """Generate learning outcomes""" + self.logger.info("Generating learning outcomes") + + async def _call_baml(): + result = await b.GenerateLearningOutcomes( + objectives=objectives, + sustainability_map=sustainability_map, + metadata=metadata, + output_language=output_language, + ) + # Convert BAML types to Pydantic models + return types.LearningOutcomes( + outcomes=[ + types.LearningOutcome( + number=outcome.number, + text=outcome.text, + related_objectives=outcome.related_objectives, + assessment_method=outcome.assessment_method, + ) + for outcome in result.outcomes + ] + ) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/b_agents/pedagogical_engineer_agent.py b/src/app/tutor/service/b_agents/pedagogical_engineer_agent.py new file mode 100644 index 00000000..f16b1adc --- /dev/null +++ b/src/app/tutor/service/b_agents/pedagogical_engineer_agent.py @@ -0,0 +1,45 @@ +"""Pedagogical Engineer Agent wrapper""" + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class PedagogicalEngineerAgent(BaseAgent): + """Validates pedagogical framework""" + + def __init__(self): + super().__init__("pedagogical_engineer") + + async def validate( + self, + description: str, + objectives: types.LearningObjectives, + outcomes: types.LearningOutcomes, + competencies: types.CompetencyMappings, + sustainability_map: types.SustainabilityIntegration, + metadata: types.CourseMetadata, + output_language: str = "Français", + ) -> types.ValidationReport: + """Validate framework""" + self.logger.info("Validating pedagogical framework") + + async def _call_baml(): + result = await b.ValidatePedagogicalFramework( + description=description, + objectives=objectives, + outcomes=outcomes, + competencies=competencies, + sustainability_map=sustainability_map, + metadata=metadata, + output_language=output_language, + ) + # Convert BAML types to Pydantic models + return types.ValidationReport( + passed=result.passed, + issues=result.issues, + suggestions=result.suggestions, + severity=result.severity, + ) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/b_agents/sustainability_integration_agent.py b/src/app/tutor/service/b_agents/sustainability_integration_agent.py new file mode 100644 index 00000000..c8c4f577 --- /dev/null +++ b/src/app/tutor/service/b_agents/sustainability_integration_agent.py @@ -0,0 +1,64 @@ +"""Sustainability Integration Agent wrapper""" + +from typing import List + +from src.app.baml_client.async_client import b, types + +from .base import BaseAgent + + +class SustainabilityIntegrationAgent(BaseAgent): + """Creates sustainability connections""" + + def __init__(self): + super().__init__("sustainability_integration") + + async def integrate( + self, + description: str, + objectives: types.LearningObjectives, + sdg_resources: List[types.Document], + metadata: types.CourseMetadata, + mode: str, + output_language: str = "Français", + ) -> types.SustainabilityIntegration: + """Create sustainability integration""" + self.logger.info(f"Integrating sustainability (mode={mode})") + + async def _call_baml(): + result = await b.IntegrateSustainability( + description=description, + objectives=objectives, + sdg_resources=sdg_resources, + metadata=metadata, + mode=mode, + output_language=output_language, + ) + # Convert BAML types to Pydantic models + return types.SustainabilityIntegration( + connections=[ + types.SustainabilityConnection( + objective_number=conn.objective_number, + sdg_themes=conn.sdg_themes, + connection_explanation=conn.connection_explanation, + key_resources=conn.key_resources, + ) + for conn in result.connections + ], + suggested_objectives=( + [ + types.LearningObjective( + number=obj.number, + text=obj.text, + bloom_level=obj.bloom_level, + ) + for obj in result.suggested_objectives + ] + if result.suggested_objectives + else None + ), + integration_strategy=result.integration_strategy, + resources_used=result.resources_used, + ) + + return await self.with_retry(_call_baml) diff --git a/src/app/tutor/service/models.py b/src/app/tutor/service/models.py index e8ceaf7a..48386099 100644 --- a/src/app/tutor/service/models.py +++ b/src/app/tutor/service/models.py @@ -1,10 +1,13 @@ import uuid from dataclasses import dataclass -from typing import Dict, List +from enum import Enum +from typing import Any, Dict, List, Optional from pydantic import BaseModel from qdrant_client.models import ScoredPoint +from src.app.baml_client import types + class SummariesList(BaseModel): summaries: list[str] @@ -83,3 +86,263 @@ class MessageWithFeedback(BaseModel): class TaskResponse: task_id: str result: str + + +# create enum sessionmode with values PRESENTIEL, DISTANCIEL, HYBRIDE + + +class SessionMode(str, Enum): + PRESENTIEL = "PRESENTIEL" + DISTANCIEL = "DISTANCIEL" + HYBRIDE = "HYBRIDE" + + +class CourseDescriptionRequest(BaseModel): + mode: str + course_metadata: types.CourseMetadata + context_text: str + + +class LearningObjectivesRequest(BaseModel): + description: str + course_metadata: types.CourseMetadata + mode: str + context_text: str + + +class SustainabilityIntegrationRequest(BaseModel): + connections: list[Dict] + suggested_objectives: types.LearningObjectives + integration_strategy: str + key_resources: list[str] + + +class LearningOutcomesRequest(BaseModel): + description: str + objectives: types.LearningObjectives + course_metadata: types.CourseMetadata + sustainability_map: types.SustainabilityIntegration + + +class Document(BaseModel): + text: str + metadata: Dict[str, Any] + relevance_score: float | None = None + url: Optional[str] = None # Link to resource + + +class CourseDescription(BaseModel): + text: str + word_count: int + + +class SustainabilityConnection(BaseModel): + objective_number: int + sdg_themes: List[str] + connection_explanation: str + key_resources: List[types.Document] + + +class SustainabilityIntegration(BaseModel): + connections: List[SustainabilityConnection] + suggested_objectives: list[types.LearningObjective] | None = None + integration_strategy: str + resources_used: List[types.Document] + + +class IntegrateSustainabilityRequest(BaseModel): + description: str + objectives: types.LearningObjectives + course_metadata: types.CourseMetadata + sdg_resources: List[types.Document] + mode: str + + +class LearningOutcome(BaseModel): + number: int + text: str + related_objectives: List[int] + assessment_method: str + + +class LearningOutcomes(BaseModel): + outcomes: List[LearningOutcome] + + +class CompetencyMappingRequest(BaseModel): + outcomes: List[LearningOutcome] + output_language: str + framework: Optional[Dict[str, str]] = None + + +class CompetencyMapping(BaseModel): + outcome_number: int + greencomp_competencies: List[str] + rationale: str + + +class CompetencyMappings(BaseModel): + mappings: List[CompetencyMapping] + + +class ValidationReport(BaseModel): + passed: bool + issues: List[str] + suggestions: List[str] + severity: str + + +class SessionPlan(BaseModel): + session_number: int + objectives: List[int] + type: str + mode: SessionMode + duration: float + class_size: int + + +class ActivityTemplate(BaseModel): + id: str + name: str + description: str + compatible_types: List[str] + compatible_modes: List[SessionMode] + min_duration: float + max_duration: float + min_size: int + max_size: int + estimated_duration: float + + +class ActivityGuide(BaseModel): + activity_name: str + description: str + learning_objectives: List[int] + steps: List[str] + teacher_role: str + student_role: str + resources_needed: List[str] + timing_breakdown: Dict[str, float] + evaluation_method: str + sustainability_integration: str + + +class ActivityRecommendations(BaseModel): + session_number: int + recommended_activities: List[ActivityTemplate] + rationale: str + + +class CourseLevel(str, Enum): + L1 = "L1" + L2 = "L2" + L3 = "L3" + M1 = "M1" + M2 = "M2" + + +# Core Models +class CourseMetadata(BaseModel): + """Course metadata provided by user""" + + discipline: str + topic: str + level: str # Free-form to allow flexibility + num_sessions: int + num_objectives: Optional[int] = None # Target number of objectives + session_duration: float + session_type: str # Free-form (CM, TD, TP, Séminaire, etc.) + session_mode: SessionMode + class_size: int + output_language: str = "Français" # Default to French + + +class LearningObjective(BaseModel): + """Single learning objective (teacher-centered: what will be taught/covered)""" + + number: int + text: str + + +class LearningObjectives(BaseModel): + """Collection of learning objectives""" + + objectives: List[LearningObjective] + + +class ValidationIssue(BaseModel): + """Single validation issue with fix applied""" + + description: str + fix_applied: Optional[str] = None # What was done to fix it + + +class OrchestratorState(BaseModel): + """Complete state of the orchestrator""" + + mode: str + metadata: types.CourseMetadata + rag_resources: Optional[List[types.Document]] = None + provided_objectives: Optional[List[str]] = None + + description: Optional[CourseDescription] = None + objectives: Optional[LearningObjectives] = None + sustainability_integration: Optional[SustainabilityIntegration] = None + outcomes: Optional[LearningOutcomes] = None + competencies: Optional[CompetencyMappings] = None + sessions: Optional[List[SessionPlan]] = None + activities: Optional[Dict[int, List[ActivityGuide]]] = None + + retrieved_resources: List[types.Document] = [] + validation_history: List[ValidationReport] = [] + current_phase: str = "input" + + +class SyllabusOutput(BaseModel): + """Final complete syllabus output""" + + description: Optional[CourseDescription] = None + objectives: Optional[LearningObjectives] = None + outcomes: Optional[LearningOutcomes] = None + competencies: Optional[CompetencyMappings] = None + sustainability: Optional[SustainabilityIntegration] = None + sessions: Optional[List[SessionPlan]] = None + # activities: Optional[Dict[int, List[ActivityGuide]]] = None + + +# Checkpoint Models for Interactive Flow +class CheckpointData(BaseModel): + """Data for a checkpoint in the orchestration flow""" + + phase: str # "description", "framework", "sessions", "session_N", "complete" + data: Dict[str, Any] + actions: List[str] # Available actions like ["accept", "edit", "regenerate"] + + +class UserAction(BaseModel): + """User action at a checkpoint""" + + action_type: str # "accept", "edit", "regenerate", "restart" + edited_data: Optional[Dict[str, Any]] = None # For direct edits + feedback: Optional[str] = None # For regeneration with feedback + + +class UserInput(BaseModel): + """Initial user input""" + + metadata: types.CourseMetadata + mode: Optional[str] = None + documents: Optional[List[bytes]] = None + documents_summaries: Optional[List[str]] = None + rag_resources: Optional[List[types.Document]] = None + provided_objectives: Optional[List[str]] = None + provided_description: Optional[str] = None + objective_processing: Optional[str] = None # "augment" or "transform" for Mode 2B + + +class SyllabusGenerationRequest(BaseModel): + mode: str + course_metadata: types.CourseMetadata + rag_resources: Optional[List[types.Document]] = None + provided_objectives: Optional[List[str]] = None + provided_description: Optional[str] = None diff --git a/src/app/tutor/service/orchestrator.py b/src/app/tutor/service/orchestrator.py new file mode 100644 index 00000000..08eceea9 --- /dev/null +++ b/src/app/tutor/service/orchestrator.py @@ -0,0 +1,1013 @@ +""" +Syllabus Orchestrator - Coordinates all agents and phases +""" + +import json +import logging +from enum import Enum +from typing import List, Optional + +from src.app.tutor.tools import distribute_objectives + +from .b_agents import ( + ActivityGuideGeneratorAgent, + ActivityRecommendationAgent, + CompetencyMappingAgent, + CourseDescriptionAgent, + LearningObjectivesAgent, + LearningOutcomesAgent, + PedagogicalEngineerAgent, + SustainabilityIntegrationAgent, +) +from .models import ( + CompetencyMappings, + CourseDescription, + LearningObjectives, + LearningOutcomes, + OrchestratorState, + SessionPlan, + SustainabilityIntegration, + SyllabusOutput, + UserInput, +) + +logger = logging.getLogger(__name__) + + +class Phase(str, Enum): + """Orchestrator phases""" + + INPUT = "input" + DESCRIPTION = "description" + OBJECTIVES = "objectives" + SUSTAINABILITY = "sustainability" + OUTCOMES = "outcomes" + COMPETENCIES = "competencies" + VALIDATION = "validation" + SESSIONS = "sessions" + ACTIVITIES = "activities" + COMPLETE = "complete" + + +class SyllabusOrchestrator: + """ + Main orchestrator that coordinates all agents and tools through sequential phases + """ + + def __init__(self): + """Initialize orchestrator with agents and configuration""" + # self.config = load_config() + + # Initialize all agents + self.course_description_agent = CourseDescriptionAgent() + self.learning_objectives_agent = LearningObjectivesAgent() + self.sustainability_agent = SustainabilityIntegrationAgent() + self.learning_outcomes_agent = LearningOutcomesAgent() + self.competency_mapping_agent = CompetencyMappingAgent() + self.pedagogical_engineer_agent = PedagogicalEngineerAgent() + self.activity_recommendation_agent = ActivityRecommendationAgent() + self.activity_guide_generator_agent = ActivityGuideGeneratorAgent() + + # Load activity templates (CAG) + # load_activity_templates() + + # State + self.state: Optional[OrchestratorState] = None + + logger.info("Orchestrator initialized") + + def get_total_tokens(self) -> int: + """ + Get total token usage across all agents + + Returns: + Total estimated tokens used + """ + agents = [ + self.course_description_agent, + self.learning_objectives_agent, + self.sustainability_agent, + self.learning_outcomes_agent, + self.competency_mapping_agent, + self.pedagogical_engineer_agent, + self.activity_recommendation_agent, + self.activity_guide_generator_agent, + ] + + total = sum(agent.total_tokens for agent in agents) + logger.debug(f"Total tokens across all agents: {total}") + return total + + # ======================================== + # Phase 0: Input & Intent Clarification + # ======================================== + + def start(self, user_input: UserInput) -> OrchestratorState: + """ + Start orchestration with user input + + Args: + user_input: Initial user input with metadata and optional content + document_filenames: Optional list of filenames for documents (for better error messages) + + Returns: + Initial state + """ + logger.info("Starting orchestration - Phase 0: Input") + + # Detect mode + mode = self._detect_mode(user_input) + logger.info(f"Detected mode: {mode}") + + # TODO: content of the user input doc == summary + context_text = "" + + # Add provided description to context (Mode 2A) + if user_input.provided_description: + if context_text: + context_text = ( + user_input.provided_description + "\n\n---\n\n" + context_text + ) + else: + context_text = user_input.provided_description + logger.info("Added provided description to context") + + # Store context_text in state for later use + self._context_text = context_text + + # Initialize state + self.state = OrchestratorState( + mode=mode, + metadata=user_input.metadata, + rag_resources=user_input.rag_resources, + provided_objectives=user_input.provided_objectives, + current_phase=Phase.INPUT, + ) + + return self.state + + def _detect_mode(self, user_input: UserInput) -> str: + """ + Detect input mode from user input: + - Mode 1: Documents present, no provided_description + - Mode 2A: Documents present + provided_description present (from uploaded existing syllabus) + - Mode 2B-Augment: provided_objectives present, objective_processing="augment" + - Mode 2B-Transform: provided_objectives present, objective_processing="transform" + - Mode 3: Only metadata (default) + """ + if user_input.documents: + if user_input.provided_description: + return "mode_2a" # Existing syllabus → generate new + else: + return "mode_1" # Documents → generate new + elif user_input.provided_objectives: + # Mode 2B fork: augment vs transform + if user_input.objective_processing == "transform": + return "mode_2b_transform" # Rewrite objectives with sustainability + else: + return "mode_2b_augment" # Keep objectives, add sustainability + else: + return "mode_3" # Metadata only → generate from scratch + + # ======================================== + # Phase 1: Pedagogical Framing + # ======================================== + + async def phase_1_generate_description( + self, context_text: str = "" + ) -> CourseDescription: + """ + Phase 1.1: Generate course description + + Args: + context_text: Optional context (extracted documents, provided description) + + Returns: + Generated course description + """ + logger.info("Phase 1.1: Generating course description") + self.state.current_phase = Phase.DESCRIPTION + + description = await self.course_description_agent.generate( + mode=self.state.mode, + metadata=self.state.metadata, + context_text=context_text, + output_language=self.state.metadata.output_language, + ) + + self.state.description = description + logger.info(f"Description generated ({description.word_count} words)") + + return description + + async def phase_1_generate_objectives( + self, context_text: str = "" + ) -> LearningObjectives: + """ + Phase 1.2: Generate learning objectives + + Args: + context_text: Optional context + + Returns: + Generated objectives + """ + logger.info("Phase 1.2: Generating learning objectives") + self.state.current_phase = Phase.OBJECTIVES + + objectives = await self.learning_objectives_agent.generate( + description=self.state.description.text, + context_text=context_text, + metadata=self.state.metadata, + mode=self.state.mode, + output_language=self.state.metadata.output_language, + ) + + self.state.objectives = objectives + logger.info(f"Generated {len(objectives.objectives)} objectives") + + return objectives + + async def phase_1_integrate_sustainability(self) -> SustainabilityIntegration: + """ + Phase 1.3: Integrate sustainability + + Returns: + Sustainability integration + """ + logger.info("Phase 1.3: Integrating sustainability") + self.state.current_phase = Phase.SUSTAINABILITY + + # Build RAG query + # query = ( + # self.state.description.text + # + " " + # + " ".join([obj.text for obj in self.state.objectives.objectives]) + # ) + + # # Retrieve SDG resources + # filters = {"discipline": self.state.metadata.discipline} + # # sdg_resources = rag_retrieve(query, filters, top_k=15) + + # # Check if RAG results sufficient + # high_relevance_docs = [ + # doc + # for doc in sdg_resources + # if doc.relevance_score and doc.relevance_score > 0.7 + # ] + + # self.state.retrieved_resources.extend(sdg_resources) + + # Call sustainability agent + sustainability = await self.sustainability_agent.integrate( + description=self.state.description.text, + objectives=self.state.objectives, + sdg_resources=self.state.rag_resources, + metadata=self.state.metadata, + mode=self.state.mode, + output_language=self.state.metadata.output_language, + ) + + # Handle Mode 2B variants + if self.state.mode == "mode_2b_augment": + # Augment mode: Keep original objectives, add suggested ones + if sustainability.suggested_objectives: + logger.info( + f"Mode 2B-Augment: Adding {len(sustainability.suggested_objectives)} suggested objectives" + ) + all_objectives = ( + self.state.objectives.objectives + + sustainability.suggested_objectives + ) + self.state.objectives.objectives = all_objectives + elif self.state.mode == "mode_2b_transform": + # Transform mode: Objectives already rewritten with sustainability + logger.info( + "Mode 2B-Transform: Using rewritten objectives with sustainability" + ) + # No need to merge - objectives are already transformed + + self.state.sustainability_integration = sustainability + logger.info("Sustainability integration complete") + + return sustainability + + async def phase_1_generate_outcomes(self) -> LearningOutcomes: + """ + Phase 1.4: Generate learning outcomes + + Returns: + Generated outcomes + """ + logger.info("Phase 1.4: Generating learning outcomes") + self.state.current_phase = Phase.OUTCOMES + + outcomes = await self.learning_outcomes_agent.generate( + objectives=self.state.objectives, + sustainability_map=self.state.sustainability_integration, + metadata=self.state.metadata, + output_language=self.state.metadata.output_language, + ) + + self.state.outcomes = outcomes + logger.info(f"Generated {len(outcomes.outcomes)} outcomes") + + return outcomes + + async def phase_1_map_competencies(self) -> CompetencyMappings: + """ + Phase 1.5: Map GreenComp competencies + + Returns: + Competency mappings + """ + logger.info("Phase 1.5: Mapping competencies") + self.state.current_phase = Phase.COMPETENCIES + + # GreenComp framework (could be loaded from config) + # competency_mapping_agent already has the competencies in its prompt, so no need to repeat, centralise in 1 place TODO + greencomp = { + "C1": "Valuing sustainability", + "C2": "Supporting fairness", + "C3": "Promoting nature", + "C4": "Systems thinking", + "C5": "Critical thinking", + "C6": "Problem framing", + "C7": "Futures literacy", + "C8": "Adaptability", + "C9": "Exploratory thinking", + "C10": "Political agency", + "C11": "Collective action", + "C12": "Individual initiative", + } + + competencies = await self.competency_mapping_agent.map_competencies( + outcomes=self.state.outcomes, + greencomp_framework=greencomp, + output_language=self.state.metadata.output_language, + ) + + self.state.competencies = competencies + logger.info(f"Mapped {len(competencies.mappings)} competencies") + + return competencies + + async def phase_1_validate_framework(self) -> bool: + """ + Phase 1.6: Validate pedagogical framework + + Returns: + True if validation passed (after any corrections) + """ + logger.info("Phase 1.6: Validating pedagogical framework") + self.state.current_phase = Phase.VALIDATION + + validation = await self.pedagogical_engineer_agent.validate( + description=self.state.description.text, + objectives=self.state.objectives, + outcomes=self.state.outcomes, + competencies=self.state.competencies, + sustainability_map=self.state.sustainability_integration, + metadata=self.state.metadata, + output_language=self.state.metadata.output_language, + ) + + self.state.validation_history.append(validation) + + if validation.passed: + logger.info("Validation passed") + return True + + logger.warning(f"Validation failed: {validation.severity}") + + if validation.severity == "minor": + # Auto-apply minor fixes + logger.info("Minor issues detected - auto-applying suggestions") + for suggestion in validation.suggestions: + logger.info(f"Applied fix: {suggestion}") + # Minor issues don't block - continue with current state + return True + + elif validation.severity == "major": + # Log major issues but continue (user will see in checkpoint) + logger.info( + "Major issues detected - providing corrected version and will be shown to user at checkpoint" + ) + for issue in validation.issues: + logger.info(f"Corrected: {issue}") + return True + + else: # subjective + # Log subjective suggestions + logger.info("Subjective recommendations - will be shown to user") + for suggestion in validation.suggestions: + logger.info(f"Recommendation: {suggestion}") + return True + + async def run_phase_1(self, context_text: str = "") -> bool: + """ + Run complete Phase 1: Pedagogical Framing + + Args: + context_text: Optional context + + Returns: + True if phase completed successfully + """ + logger.info("========== PHASE 1: PEDAGOGICAL FRAMING ==========") + + # try: + await self.phase_1_generate_description(context_text) + + await self.phase_1_generate_objectives(context_text) + + await self.phase_1_integrate_sustainability() + + await self.phase_1_generate_outcomes() + + await self.phase_1_map_competencies() + + logger.info("Phase 1 complete") + return True + + # except AgentError as e: + # logger.error(f"Phase 1 failed: {e}") + # return False + + # ======================================== + # Phase 2: Session Planning + # ======================================== + + def phase_2_distribute_sessions(self) -> List[SessionPlan]: + """ + Phase 2: Distribute objectives across sessions (deterministic) + + Returns: + List of session plans + """ + logger.info("========== PHASE 2: SESSION PLANNING ==========") + self.state.current_phase = Phase.SESSIONS + + sessions = distribute_objectives( + objectives=self.state.objectives.objectives, metadata=self.state.metadata + ) + + self.state.sessions = sessions + logger.info(f"Created {len(sessions)} session plans") + + return sessions + + # ======================================== + # Complete Orchestration + # ======================================== + + async def run( + self, + user_input: UserInput, + context_text: str | None = None, + document_filenames: Optional[List[str]] = None, + ): + # -> SyllabusOutput: + """ + Run complete orchestration + + Args: + user_input: User input + context_text: Optional context (if None, will be built from documents) + document_filenames: Optional list of filenames for documents + + Returns: + Complete syllabus output + """ + logger.info("========== STARTING ORCHESTRATION ==========") + + # Phase 0 + # user description + documents content + self.start(user_input) + + # Use context_text from state if not provided explicitly + if context_text is None: + context_text = getattr(self, "_context_text", "") + + # Phase 1 + if not await self.run_phase_1(context_text): + raise Exception("Phase 1 failed") + + # Phase 2 + # self.phase_2_distribute_sessions() + + # # Phase 3 + # self.phase_3_design_activities() + + # Complete + self.state.current_phase = Phase.COMPLETE + logger.info("========== ORCHESTRATION COMPLETE ==========") + + return self.get_output() + + # def get_output(self) -> SyllabusOutput + def get_output(self): + """Get final syllabus output""" + if not self.state: + raise Exception("No state - orchestration not started") + + data = { + "description": self.state.description, + "objectives": self.state.objectives, + "outcomes": self.state.outcomes, + "competencies": self.state.competencies, + "sustainability": self.state.sustainability_integration, + "sessions": self.state.sessions, + } + + return data + + print( + "Final output data:", + json.dumps(data, default=lambda o: o.__dict__, indent=2), + ) + + return SyllabusOutput( + description=self.state.description, + objectives=self.state.objectives, + outcomes=self.state.outcomes, + competencies=self.state.competencies, + sustainability=self.state.sustainability_integration, + sessions=self.state.sessions, + # activities=self.state.activities, + ) + + # ======================================== + # Checkpoint-Based Orchestration + # ======================================== + + # def run_with_checkpoints( + # self, user_input: UserInput, document_filenames: Optional[List[str]] = None + # ): + # """ + # Generator-based orchestration that yields checkpoints for user approval. + + # Usage: + # orchestrator = SyllabusOrchestrator() + # gen = orchestrator.run_with_checkpoints(user_input) + + # checkpoint = next(gen) # Get first checkpoint + # action = UserAction(action_type="accept") + # checkpoint = gen.send(action) # Send action and get next checkpoint + + # Yields: + # CheckpointData: Data for user to review at each checkpoint + # """ + # logger.info("========== STARTING CHECKPOINT-BASED ORCHESTRATION ==========") + + # # Phase 0: Initialize + # self.start(user_input, document_filenames=document_filenames) + # context_text = getattr(self, "_context_text", "") + + # # ======================================== + # # CHECKPOINT 1: Course Description + # # ======================================== + # logger.info("Phase 1.1: Generating course description") + # self.state.current_phase = Phase.DESCRIPTION + + # description = self.course_description_agent.generate( + # mode=self.state.mode, + # metadata=self.state.metadata, + # context_text=context_text, + # output_language=self.state.metadata.output_language, + # ) + # self.state.description = description + + # # Yield checkpoint 1 + # checkpoint = CheckpointData( + # phase="description", + # data={ + # "description": description.text, + # "word_count": description.word_count, + # }, + # actions=["accept", "edit", "regenerate"], + # ) + # action = yield checkpoint + + # # Handle user action + # if action.action_type == "edit": + # # User edited description directly + # self.state.description = CourseDescription( + # text=action.edited_data["description"], + # word_count=len(action.edited_data["description"].split()), + # ) + # logger.info("User edited description") + # elif action.action_type == "regenerate": + # # Regenerate without feedback + # logger.info("Regenerating description") + # description = self.course_description_agent.generate( + # mode=self.state.mode, + # metadata=self.state.metadata, + # context_text=context_text, + # output_language=self.state.metadata.output_language, + # ) + # self.state.description = description + # elif action.action_type == "feedback": + # # Regenerate with user feedback + # logger.info( + # f"Regenerating description with user feedback: {action.feedback}" + # ) + # description = self.course_description_agent.generate( + # mode=self.state.mode, + # metadata=self.state.metadata, + # context_text=context_text + # + "\n\nUser feedback for regeneration: " + # + action.feedback, + # output_language=self.state.metadata.output_language, + # ) + # self.state.description = description + + # # ======================================== + # # CHECKPOINT 2: Complete Framework + # # ======================================== + # logger.info("Generating complete pedagogical framework") + + # # Generate objectives + # objectives = self.learning_objectives_agent.generate( + # description=self.state.description.text, + # context_text=context_text, + # metadata=self.state.metadata, + # mode=self.state.mode, + # output_language=self.state.metadata.output_language, + # ) + # self.state.objectives = objectives + + # # Integrate sustainability + # query = ( + # self.state.description.text + # + " " + # + " ".join([obj.text for obj in objectives.objectives]) + # ) + # # filters = {"discipline": self.state.metadata.discipline} + # # sdg_resources = rag_retrieve(query, filters, top_k=15) + + # high_relevance_docs = [ + # doc + # for doc in sdg_resources + # if doc.relevance_score and doc.relevance_score > 0.7 + # ] + # if len(high_relevance_docs) < 3: + # logger.warning("Insufficient RAG results, falling back to web search") + # # web_results = web_search(query) + # logger.info(f"Web search returned {len(web_results)} results") + + # self.state.retrieved_resources.extend(sdg_resources) + + # sustainability = self.sustainability_agent.integrate( + # description=self.state.description.text, + # objectives=objectives, + # sdg_resources=sdg_resources, + # metadata=self.state.metadata, + # mode=self.state.mode, + # output_language=self.state.metadata.output_language, + # ) + + # # Handle Mode 2B variants + # if self.state.mode == "mode_2b_augment": + # # Augment mode: Keep original objectives, add suggested ones + # if sustainability.suggested_objectives: + # logger.info( + # f"Mode 2B-Augment: Adding {len(sustainability.suggested_objectives)} suggested objectives" + # ) + # all_objectives = ( + # objectives.objectives + sustainability.suggested_objectives + # ) + # self.state.objectives.objectives = all_objectives + # elif self.state.mode == "mode_2b_transform": + # # Transform mode: Objectives already rewritten with sustainability + # logger.info( + # "Mode 2B-Transform: Using rewritten objectives with sustainability" + # ) + # # No need to merge - objectives are already transformed + + # self.state.sustainability_integration = sustainability + + # # Generate outcomes + # outcomes = self.learning_outcomes_agent.generate( + # objectives=self.state.objectives, + # sustainability_map=sustainability, + # metadata=self.state.metadata, + # output_language=self.state.metadata.output_language, + # ) + # self.state.outcomes = outcomes + + # # Map competencies + # greencomp = { + # "C1": "Valuing sustainability", + # "C2": "Supporting fairness", + # "C3": "Promoting nature", + # "C4": "Systems thinking", + # "C5": "Critical thinking", + # "C6": "Problem framing", + # "C7": "Futures literacy", + # "C8": "Adaptability", + # "C9": "Exploratory thinking", + # "C10": "Political agency", + # "C11": "Collective action", + # "C12": "Individual initiative", + # } + # competencies = self.competency_mapping_agent.map_competencies( + # outcomes=outcomes, + # greencomp_framework=greencomp, + # output_language=self.state.metadata.output_language, + # ) + # self.state.competencies = competencies + + # # Validate framework + # validation = self.pedagogical_engineer_agent.validate( + # description=self.state.description.text, + # objectives=self.state.objectives, + # outcomes=outcomes, + # competencies=competencies, + # sustainability_map=sustainability, + # metadata=self.state.metadata, + # output_language=self.state.metadata.output_language, + # ) + # self.state.validation_history.append(validation) + + # # Auto-apply validation recommendations + # if not validation.passed: + # if validation.severity == "minor": + # logger.info("Auto-applying minor validation fixes") + # for suggestion in validation.suggestions: + # logger.info(f"Applied: {suggestion}") + # elif validation.severity == "major": + # logger.warning("Major validation issues - showing to user") + # for issue in validation.issues: + # logger.warning(f"Issue: {issue}") + + # # Yield checkpoint 2 + # checkpoint = CheckpointData( + # phase="framework", + # data={ + # "description": self.state.description.text, + # "objectives": self.state.objectives, + # "sustainability": sustainability, + # "outcomes": outcomes, + # "competencies": competencies, + # "validation": validation, + # }, + # actions=["accept", "edit"], + # ) + # action = yield checkpoint + + # # Handle user action + # if action.action_type == "edit" and action.edited_data: + # # User edited framework components + # if "objectives" in action.edited_data: + # self.state.objectives = action.edited_data["objectives"] + # if "outcomes" in action.edited_data: + # self.state.outcomes = action.edited_data["outcomes"] + # logger.info("User edited framework") + # elif action.action_type == "regenerate": + # # Regenerate framework components + # logger.info("Regenerating framework") + # # Re-run the framework generation steps + # objectives = self.learning_objectives_agent.generate( + # description=self.state.description.text, + # context_text=context_text, + # metadata=self.state.metadata, + # mode=self.state.mode, + # output_language=self.state.metadata.output_language, + # ) + # self.state.objectives = objectives + # # Re-integrate sustainability and regenerate outcomes/competencies + # # (simplified - could be more granular) + # elif action.action_type == "feedback": + # # Regenerate framework with user feedback + # logger.info(f"Regenerating framework with user feedback: {action.feedback}") + # feedback_context = ( + # context_text + "\n\nUser feedback for improvement: " + action.feedback + # ) + + # objectives = self.learning_objectives_agent.generate( + # description=self.state.description.text, + # context_text=feedback_context, + # metadata=self.state.metadata, + # mode=self.state.mode, + # output_language=self.state.metadata.output_language, + # ) + # self.state.objectives = objectives + + # # Re-generate sustainability integration + # query = ( + # self.state.description.text + # + " " + # + " ".join([obj.text for obj in objectives.objectives]) + # ) + # sdg_resources = rag_retrieve( + # query, {"discipline": self.state.metadata.discipline}, top_k=15 + # ) + + # sustainability = self.sustainability_agent.integrate( + # description=self.state.description.text, + # objectives=objectives, + # sdg_resources=sdg_resources, + # metadata=self.state.metadata, + # mode=self.state.mode, + # output_language=self.state.metadata.output_language, + # ) + # self.state.sustainability_integration = sustainability + + # # Re-generate outcomes + # outcomes = self.learning_outcomes_agent.generate( + # objectives=objectives, + # sustainability_map=sustainability, + # metadata=self.state.metadata, + # output_language=self.state.metadata.output_language, + # ) + # self.state.outcomes = outcomes + + # # Re-map competencies + # greencomp = { + # "C1": "Valuing sustainability", + # "C2": "Supporting fairness", + # "C3": "Promoting nature", + # "C4": "Systems thinking", + # "C5": "Critical thinking", + # "C6": "Problem framing", + # "C7": "Futures literacy", + # "C8": "Adaptability", + # "C9": "Exploratory thinking", + # "C10": "Political agency", + # "C11": "Collective action", + # "C12": "Individual initiative", + # } + # competencies = self.competency_mapping_agent.map_competencies( + # outcomes=outcomes, + # greencomp_framework=greencomp, + # output_language=self.state.metadata.output_language, + # ) + # self.state.competencies = competencies + + # # ======================================== + # # CHECKPOINT 3: Session Planning + # # ======================================== + # logger.info("Phase 2: Distributing sessions") + # self.state.current_phase = Phase.SESSIONS + + # sessions = distribute_objectives( + # objectives=self.state.objectives.objectives, metadata=self.state.metadata + # ) + # self.state.sessions = sessions + + # # Yield checkpoint 3 + # checkpoint = CheckpointData( + # phase="sessions", + # data={"sessions": sessions, "objectives": self.state.objectives}, + # actions=["accept", "edit"], + # ) + # action = yield checkpoint + + # if action.action_type == "edit" and action.edited_data: + # if "sessions" in action.edited_data: + # self.state.sessions = action.edited_data["sessions"] + # logger.info("User edited session plan") + # elif action.action_type == "regenerate": + # # Regenerate session distribution + # logger.info("Regenerating session plan") + # sessions = distribute_objectives( + # objectives=self.state.objectives.objectives, + # metadata=self.state.metadata, + # ) + # self.state.sessions = sessions + # elif action.action_type == "feedback": + # # For sessions, feedback is noted but distribution is algorithmic + # # User can edit objectives if they want different distribution + # logger.info(f"Session plan feedback noted: {action.feedback}") + # # Optionally: could implement custom distribution based on feedback + # logger.warning( + # "Session feedback handling: currently sessions are algorithmically distributed" + # ) + + # # ======================================== + # # CHECKPOINT 4+: Activities per Session + # # ======================================== + # logger.info("Phase 3: Designing activities") + # self.state.current_phase = Phase.ACTIVITIES + + # activities = {} + + # for session in self.state.sessions: + # logger.info(f"Designing activities for session {session.session_number}") + + # # Filter compatible activities + # filtered = filter_activities( + # session_type=session.type, + # session_mode=session.mode, + # duration=session.duration, + # class_size=session.class_size, + # ) + + # if not filtered: + # logger.warning( + # f"No compatible activities for session {session.session_number}" + # ) + # activities[session.session_number] = [] + # continue + + # # Recommend activities based on learning outcomes + # recommendations = self.activity_recommendation_agent.recommend( + # session=session, + # learning_outcomes=self.state.outcomes, + # filtered_activities=filtered, + # output_language=self.state.metadata.output_language, + # ) + + # # Generate guides + # guides = [] + # for template in recommendations.recommended_activities: + # guide = self.activity_guide_generator_agent.generate( + # activity_template=template, + # session=session, + # metadata=self.state.metadata, + # sustainability_map=self.state.sustainability_integration, + # output_language=self.state.metadata.output_language, + # ) + # guides.append(guide) + + # activities[session.session_number] = guides + + # # Yield checkpoint for this session + # checkpoint = CheckpointData( + # phase=f"session_{session.session_number}", + # data={ + # "session": session, + # "activities": guides, + # "session_number": session.session_number, + # "total_sessions": len(self.state.sessions), + # }, + # actions=["accept", "edit"], + # ) + # action = yield checkpoint + + # if action.action_type == "edit" and action.edited_data: + # if "activities" in action.edited_data: + # activities[session.session_number] = action.edited_data[ + # "activities" + # ] + # logger.info( + # f"User edited activities for session {session.session_number}" + # ) + # elif action.action_type == "regenerate": + # # Regenerate activities for this session + # logger.info( + # f"Regenerating activities for session {session.session_number}" + # ) + # recommendations = self.activity_recommendation_agent.recommend( + # session=session, + # learning_outcomes=self.state.outcomes, + # filtered_activities=filtered, + # output_language=self.state.metadata.output_language, + # ) + # guides = [] + # for template in recommendations.recommended_activities: + # guide = self.activity_guide_generator_agent.generate( + # activity_template=template, + # session=session, + # metadata=self.state.metadata, + # sustainability_map=self.state.sustainability_integration, + # output_language=self.state.metadata.output_language, + # ) + # guides.append(guide) + # activities[session.session_number] = guides + # elif action.action_type == "feedback": + # # Regenerate activities with user feedback + # logger.info(f"Regenerating activities with feedback: {action.feedback}") + # # Could pass feedback to agent prompts - for now, simple regeneration + # # Future enhancement: pass feedback to activity guide generator + # recommendations = self.activity_recommendation_agent.recommend( + # session=session, + # learning_outcomes=self.state.outcomes, + # filtered_activities=filtered, + # output_language=self.state.metadata.output_language, + # ) + # guides = [] + # for template in recommendations.recommended_activities: + # guide = self.activity_guide_generator_agent.generate( + # activity_template=template, + # session=session, + # metadata=self.state.metadata, + # sustainability_map=self.state.sustainability_integration, + # output_language=self.state.metadata.output_language, + # ) + # guides.append(guide) + # activities[session.session_number] = guides + + # self.state.activities = activities + + # # ======================================== + # # COMPLETE + # # ======================================== + # self.state.current_phase = Phase.COMPLETE + # logger.info("========== CHECKPOINT-BASED ORCHESTRATION COMPLETE ==========") + + # # Final checkpoint with complete output + # yield CheckpointData( + # phase="complete", data=self.get_output().dict(), actions=[] + # ) diff --git a/src/app/tutor/tools/__init__.py b/src/app/tutor/tools/__init__.py new file mode 100644 index 00000000..94f22149 --- /dev/null +++ b/src/app/tutor/tools/__init__.py @@ -0,0 +1,28 @@ +"""Tools module - Deterministic utilities for agents""" + +from .distribute_objectives import distribute_objectives +from .filter_activities import filter_activities, load_activity_templates +from .greencomp_data import ( + get_all_methods, + get_greencomp_competency_details, + get_method_description, + get_suitable_methods_for_competencies, + get_suitable_methods_for_outcome_text, + load_greencomp_methods, + load_greencomp_outcomes, +) +from .validate_format import validate_format + +__all__ = [ + "filter_activities", + "load_activity_templates", + "validate_format", + "distribute_objectives", + "load_greencomp_outcomes", + "load_greencomp_methods", + "get_suitable_methods_for_competencies", + "get_suitable_methods_for_outcome_text", + "get_greencomp_competency_details", + "get_all_methods", + "get_method_description", +] diff --git a/src/app/tutor/tools/distribute_objectives.py b/src/app/tutor/tools/distribute_objectives.py new file mode 100644 index 00000000..6d768f95 --- /dev/null +++ b/src/app/tutor/tools/distribute_objectives.py @@ -0,0 +1,73 @@ +""" +Objective distribution tool - Deterministic distribution of objectives across sessions (Frugal AI) +""" + +import logging +from typing import List + +from src.app.baml_client.async_client import types + +logger = logging.getLogger(__name__) + + +def distribute_objectives( + objectives: List[types.LearningObjective], metadata: types.CourseMetadata +) -> List[types.SessionPlan]: + """ + Distribute learning objectives evenly across sessions + + Simple algorithm for prototype: + - Divide objectives evenly + - Distribute remainder to first sessions + - In future: pedagogical sequencing (foundational concepts first) + + Args: + objectives: List of learning objectives to distribute + metadata: Course metadata with session info + + Returns: + List of SessionPlan objects with assigned objectives + """ + num_sessions = metadata.num_sessions + num_objectives = len(objectives) + + logger.info( + f"Distributing {num_objectives} objectives across {num_sessions} sessions" + ) + + # Calculate distribution + objectives_per_session = num_objectives // num_sessions + remainder = num_objectives % num_sessions + + sessions = [] + current_objective_idx = 0 + + for session_num in range(1, num_sessions + 1): + # Calculate how many objectives for this session + count = objectives_per_session + (1 if session_num <= remainder else 0) + + # Assign next 'count' objectives + session_objectives = [ + objectives[current_objective_idx + i].number for i in range(count) + ] + current_objective_idx += count + + # Create session plan + session = types.SessionPlan( + session_number=session_num, + objectives=session_objectives, + type=metadata.session_type, + mode=metadata.session_mode, + duration=metadata.session_duration, + class_size=metadata.class_size, + ) + + sessions.append(session) + + logger.debug( + f"Session {session_num}: {len(session_objectives)} objectives " + f"(numbers: {session_objectives})" + ) + + logger.info(f"Created {len(sessions)} session plans") + return sessions diff --git a/src/app/tutor/tools/filter_activities.py b/src/app/tutor/tools/filter_activities.py new file mode 100644 index 00000000..bffd43b2 --- /dev/null +++ b/src/app/tutor/tools/filter_activities.py @@ -0,0 +1,176 @@ +""" +Activity filtering tool - Deterministic filtering of activity templates (Frugal AI) +""" + +import json +import logging +from pathlib import Path +from typing import List + +from src.app.baml_client.async_client import types + +logger = logging.getLogger(__name__) + +# Cached activity templates (CAG not RAG) +_activity_templates: List[types.ActivityTemplate] = [] + + +def load_activity_templates(filepath: str = None) -> List[types.ActivityTemplate]: + """ + Load activity templates from JSON file + + Args: + filepath: Path to activity templates JSON file + + Returns: + List of ActivityTemplate objects + """ + global _activity_templates + + if filepath is None: + # Use default path + filepath = Path(__file__).parent.parent / "data" / "activity_templates.json" + + if not Path(filepath).exists(): + logger.warning( + f"Activity templates file not found: {filepath}. Using mock data." + ) + _activity_templates = _get_mock_templates() + return _activity_templates + + with open(filepath, "r", encoding="utf-8") as f: + data = json.load(f) + + _activity_templates = [types.ActivityTemplate(**template) for template in data] + logger.info(f"Loaded {len(_activity_templates)} activity templates") + + return _activity_templates + + +def _get_mock_templates() -> List[types.ActivityTemplate]: + """Get mock activity templates for testing""" + return [ + types.ActivityTemplate( + id="act_001", + name="Débat structuré", + description="Débat en groupes sur des questions controversées", + compatible_types=["CM", "TD", "Séminaire"], + compatible_modes=[types.SessionMode.PRESENTIEL, types.SessionMode.HYBRIDE], + min_duration=1.5, + max_duration=3.0, + min_size=15, + max_size=100, + estimated_duration=2.0, + ), + types.ActivityTemplate( + id="act_002", + name="Étude de cas collaborative", + description="Analyse de cas en petits groupes", + compatible_types=["TD", "TP"], + compatible_modes=[ + types.SessionMode.PRESENTIEL, + types.SessionMode.DISTANCIEL, + types.SessionMode.HYBRIDE, + ], + min_duration=1.0, + max_duration=2.5, + min_size=10, + max_size=40, + estimated_duration=1.5, + ), + types.ActivityTemplate( + id="act_003", + name="Présentation magistrale interactive", + description="Cours avec moments d'interaction et questions", + compatible_types=["CM"], + compatible_modes=[ + types.SessionMode.PRESENTIEL, + types.SessionMode.DISTANCIEL, + types.SessionMode.HYBRIDE, + ], + min_duration=1.0, + max_duration=4.0, + min_size=20, + max_size=300, + estimated_duration=2.0, + ), + types.ActivityTemplate( + id="act_004", + name="Atelier pratique", + description="Exercices pratiques individuels ou en binôme", + compatible_types=["TP", "TD"], + compatible_modes=[types.SessionMode.PRESENTIEL], + min_duration=1.5, + max_duration=3.0, + min_size=8, + max_size=30, + estimated_duration=2.0, + ), + types.ActivityTemplate( + id="act_005", + name="Quiz formatif", + description="Quiz interactif pour vérifier la compréhension", + compatible_types=["CM", "TD", "TP"], + compatible_modes=[ + types.SessionMode.PRESENTIEL, + types.SessionMode.DISTANCIEL, + types.SessionMode.HYBRIDE, + ], + min_duration=0.5, + max_duration=1.5, + min_size=5, + max_size=200, + estimated_duration=0.5, + ), + ] + + +def filter_activities( + session_type: str, session_mode: types.SessionMode, duration: float, class_size: int +) -> List[types.ActivityTemplate]: + """ + Filter activity templates based on session constraints (deterministic, no LLM) + + Args: + session_type: Type of session (CM, TD, TP, Séminaire, etc.) + session_mode: Mode of session (Présentiel, Distanciel, Hybride) + duration: Session duration in hours + class_size: Number of students + + Returns: + List of compatible ActivityTemplate objects + """ + global _activity_templates + + # Load templates if not already loaded + if not _activity_templates: + load_activity_templates() + + logger.info( + f"Filtering activities: type={session_type}, mode={session_mode}, " + f"duration={duration}h, size={class_size}" + ) + + filtered = [] + + for activity in _activity_templates: + # Check session type compatibility + if session_type not in activity.compatible_types: + continue + + # Check mode compatibility + if session_mode not in activity.compatible_modes: + continue + + # Check duration constraints + if not (activity.min_duration <= duration <= activity.max_duration): + continue + + # Check class size constraints + if not (activity.min_size <= class_size <= activity.max_size): + continue + + filtered.append(activity) + + logger.info(f"Filtered to {len(filtered)} compatible activities") + return filtered diff --git a/src/app/tutor/tools/greencomp_data.py b/src/app/tutor/tools/greencomp_data.py new file mode 100644 index 00000000..9f53ba40 --- /dev/null +++ b/src/app/tutor/tools/greencomp_data.py @@ -0,0 +1,238 @@ +""" +GreenComp Data Loader +Loads and provides access to GreenComp learning outcomes and methods CSV files +""" + +import logging +from functools import lru_cache +from pathlib import Path +from typing import Dict, List, Optional + +import pandas as pd + +logger = logging.getLogger(__name__) + +# Get project root directory (parent of welearn_mas) +PROJECT_ROOT = Path(__file__).parent.parent.parent + + +@lru_cache(maxsize=1) +def load_greencomp_outcomes() -> pd.DataFrame: + """ + Load GreenComp learning outcomes CSV + + Returns: + DataFrame with columns: Competence area, Competence, Learning outcome, + KSA coverage, Observable, Measurable, Evaluation Methods + """ + csv_path = PROJECT_ROOT / "GreenComp-LearningOutcomes.csv" + + if not csv_path.exists(): + logger.warning(f"GreenComp outcomes CSV not found at {csv_path}") + return pd.DataFrame() + + df = pd.read_csv(csv_path) + logger.info(f"Loaded {len(df)} GreenComp learning outcomes") + return df + + +@lru_cache(maxsize=1) +def load_greencomp_methods() -> pd.DataFrame: + """ + Load GreenComp learning methods CSV + + Returns: + DataFrame with columns: Method_Name, Description, Validation_Source, + Best_For_LO_Types + """ + csv_path = PROJECT_ROOT / "GreenComp-LearningMethods.csv" + + if not csv_path.exists(): + logger.warning(f"GreenComp methods CSV not found at {csv_path}") + return pd.DataFrame() + + df = pd.read_csv(csv_path) + logger.info(f"Loaded {len(df)} GreenComp learning methods") + return df + + +def get_suitable_methods_for_competencies(competencies: List[str]) -> List[Dict]: + """ + Get suitable learning methods for given GreenComp competencies + + Args: + competencies: List of GreenComp competency codes (e.g., ["C4", "C5"]) + + Returns: + List of dicts with method info: {name, description, best_for} + """ + methods_df = load_greencomp_methods() + outcomes_df = load_greencomp_outcomes() + + if methods_df.empty or outcomes_df.empty: + return [] + + # Map competencies to learning outcome types + # Filter outcomes by competencies and get their types + competency_names = [] + for comp_code in competencies: + # Extract competency names from outcomes based on competency area + matching_outcomes = outcomes_df[ + outcomes_df["Competence"].str.contains(comp_code, case=False, na=False) + ] + if not matching_outcomes.empty: + competency_names.extend(matching_outcomes["Competence"].unique().tolist()) + + # Find methods that match these competency types + suitable_methods = [] + for _, method in methods_df.iterrows(): + best_for = str(method.get("Best_For_LO_Types", "")) + + # Check if any competency name appears in the best_for field + if any( + comp_name.lower() in best_for.lower() + for comp_name in competency_names + if comp_name + ): + suitable_methods.append( + { + "name": method.get("Method_Name", ""), + "description": method.get("Description", ""), + "best_for": best_for, + } + ) + + return suitable_methods + + +def get_suitable_methods_for_outcome_text(outcome_text: str) -> List[Dict]: + """ + Get suitable learning methods for a given learning outcome by matching keywords + + Args: + outcome_text: Text of the learning outcome + + Returns: + List of dicts with method info, sorted by relevance + """ + methods_df = load_greencomp_methods() + + if methods_df.empty: + return [] + + # Keywords to match in learning outcome text + keyword_mapping = { + "analyze": [ + "Case Study Analysis", + "Critical Analysis / Deconstruction", + "Systems Mapping", + ], + "evaluate": [ + "Critical Analysis / Deconstruction", + "Policy & Governance Analysis", + ], + "design": ["Futures Literacy Workshop", "Living Lab / Co-creation"], + "demonstrate": [ + "Community Project (Participatory Action)", + "Reflective Practice / Portfolio", + ], + "stakeholder": [ + "Stakeholder Simulation / Role-Play", + "Living Lab / Co-creation", + ], + "systems": ["Systems Mapping", "Interdisciplinary Research Project"], + "policy": ["Policy & Governance Analysis", "Advocacy Campaign"], + "future": ["Futures Literacy Workshop", "Socio-Ecological Transition Planning"], + "personal": ["Personal Action Plan", "Reflective Practice / Portfolio"], + "community": [ + "Community Project (Participatory Action)", + "Living Lab / Co-creation", + ], + "sustainability": ["Case Study Analysis", "Interdisciplinary Research Project"], + } + + outcome_lower = outcome_text.lower() + method_scores = {} + + # Score each method based on keyword matches + for keyword, method_names in keyword_mapping.items(): + if keyword in outcome_lower: + for method_name in method_names: + method_scores[method_name] = method_scores.get(method_name, 0) + 1 + + # Get method details and sort by score + suitable_methods = [] + for method_name, score in sorted( + method_scores.items(), key=lambda x: x[1], reverse=True + ): + method_row = methods_df[methods_df["Method_Name"] == method_name] + if not method_row.empty: + method = method_row.iloc[0] + suitable_methods.append( + { + "name": method.get("Method_Name", ""), + "description": method.get("Description", ""), + "best_for": method.get("Best_For_LO_Types", ""), + "relevance_score": score, + } + ) + + return suitable_methods[:5] # Return top 5 + + +def get_greencomp_competency_details(competency_code: str) -> Optional[Dict]: + """ + Get detailed information about a GreenComp competency + + Args: + competency_code: Competency code (e.g., "C4", "Systems thinking") + + Returns: + Dict with competency details or None if not found + """ + outcomes_df = load_greencomp_outcomes() + + if outcomes_df.empty: + return None + + # Try to find by code or name + matching = outcomes_df[ + (outcomes_df["Competence"].str.contains(competency_code, case=False, na=False)) + | ( + outcomes_df["Competence area"].str.contains( + competency_code, case=False, na=False + ) + ) + ] + + if matching.empty: + return None + + first_match = matching.iloc[0] + return { + "competence_area": first_match.get("Competence area", ""), + "competence": first_match.get("Competence", ""), + "sample_outcome": first_match.get("Learning outcome", ""), + "evaluation_methods": first_match.get("Evaluation Methods", ""), + } + + +def get_all_methods() -> List[str]: + """Get list of all available learning method names""" + methods_df = load_greencomp_methods() + if methods_df.empty: + return [] + return methods_df["Method_Name"].tolist() + + +def get_method_description(method_name: str) -> Optional[str]: + """Get description of a specific learning method""" + methods_df = load_greencomp_methods() + if methods_df.empty: + return None + + method = methods_df[methods_df["Method_Name"] == method_name] + if method.empty: + return None + + return method.iloc[0].get("Description", "") diff --git a/src/app/tutor/tools/validate_format.py b/src/app/tutor/tools/validate_format.py new file mode 100644 index 00000000..900bf008 --- /dev/null +++ b/src/app/tutor/tools/validate_format.py @@ -0,0 +1,48 @@ +""" +Format validation tool - Check output structure using Pydantic +""" + +import logging +from typing import Any, Dict, Type + +from pydantic import BaseModel, ValidationError + +logger = logging.getLogger(__name__) + + +class ValidationResult(BaseModel): + """Result of format validation""" + + valid: bool + errors: list[str] = [] + + +def validate_format( + content: Dict[str, Any], schema: Type[BaseModel] +) -> ValidationResult: + """ + Validate content against Pydantic schema + + Args: + content: Content to validate (as dict) + schema: Pydantic model class to validate against + + Returns: + ValidationResult with valid status and any errors + """ + logger.info(f"Validating content against schema: {schema.__name__}") + + try: + # Attempt to instantiate the schema with content + schema(**content) + logger.info("Validation passed") + return ValidationResult(valid=True, errors=[]) + + except ValidationError as e: + errors = [f"{err['loc']}: {err['msg']}" for err in e.errors()] + logger.warning(f"Validation failed: {errors}") + return ValidationResult(valid=False, errors=errors) + + except Exception as e: + logger.error(f"Unexpected validation error: {str(e)}") + return ValidationResult(valid=False, errors=[str(e)])