Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
Azure AI Projects Python SDK (#38338)
* First version of client with endpoint operations * Make Endpoints sample work * Create EndpointProperties class. Implement client.endpoints.get_default * Support AzureAIClient.from_connection_string() * Make default input argument for .endpoints.get * Re-emit from latest TypeSpec. First attempt to support 3 different endpoints * Fix construction of the three endpoints * Re-emit from TypeSpec. Rename host_name to endpoint * Bug fix * Add methods ai_client.inference.get_chat_completions_client() and ai_client.inference.get_azure_openai_client() * Adding sample and changing client name (#37724) * Add .inference.get_embeddings_client() method * Jhakulin/azure ai client agents (#37748) * generated agents * merge latest * update * add patches * update * new version generated * fixes * fix merge error * update * use assistant_id for now * update * Create seperate inference samples * Move SASTokenCredential into .models * Add more agents samples (#37777) * add more samples * adjust sleep * review updates * Update _patch.py (#37760) * Online Evaluation changes (#37794) * Patch constructor for async AzureAIClient (#37804) * Users/singankit/evaluation add data mapping (#37824) * Update sample_evaluations.py * Update sample_evaluations.py * Fix .inference.get_azure_openai_client() for the async AzureAIClient (#37828) * migrate async, poll functions, and samples from assistant to agent (#37831) * migrate async, poll functions, and samples from assistant to agent * migrate async, poll functions, and samples from assistant to agent * resolved comments * resolve comments * using print instead of logging * moved ai_client to main function * Resolved comment * Fixed samples * [AI Client] adapt Sophia's assistant tests to client repo (#37819) * [AI Client] adapt Sophia's assistant tests to client repo * get tests running and passing with pytest * recordings pass and pushed * review feedback * Took out streaming from create_run, create_and_process_run, and submit_tool_outputs_to_run and declare separate functions for streaming (#37851) * Took out streaming from create_run, create_and_process_run, and submit_tool_outputs_to_run and declare separate functions for streaming * Took out create_and_process_stream and update samples accordingly * fixed samples * Miscellaneous updates (#37852) * Jhakulin/azure ai client pylance (#37887) * Fixes to Pylance error * fix sample * update * use file_upload_and_poll * updates to async samples * remove comment as we are using poll methods * more fixes * Use List instead of Sequence and fixed samples (#37898) * fixed warning for aio and get call function tools for stream within t… (#37913) * fixed warning for aio and get call function tools for stream within the SDK * Fixed another warning * Fixed return type * Evaluation API Review Feedback (#37915) * Evaluation API Review Feedback * add sample for evaluation schedule --------- Co-authored-by: Sai Kothinti <[email protected]> * Fixed poll while loop and sample (#37922) * Jhakulin/azure ai client agents updates2 (#37937) * minor fixes and updates * update * update * create_vector_store_file_batch_and_poll + samples and update aio using the latest code from syn (#37952) * create_vector_store_file_batch_and_poll + samples and update aio using the latest code from sync * Resolved comments * Clean up on the Optional (#37971) * Switch from .endpoints to .connection (re-emit from latest TypeSpec) * updates and minor fixes (#37984) * Clean up (#37985) * Rename connection string environment variables Rename AI_CLIENT_CONNECTION_STRING to PROJECT_CONNECTION_STRING, to match how it's shown in Azure AI Studio. Also rename AZURE_AI_CLIENT_AGENTS_CONNECTION_STRING (used in Agents tests) to PROJECT_CONNECTION_STRING_AGENTS_TESTS * updated codeinterpreter attachment sample (#37990) * Filter parameters, returned by SSE event (#37991) * nirovins/filter_parameters * Fixes * Move utility functions out of class * Evaluation data set id change revert (#38007) * Updates to .connection operators (#38024) - Re-emit from latest TypeSpec - Updates to .connections operators to support the new preview version of REST API - Add "id" on ConnectionProperties * Users/singankit/ai project utils (#38033) * Sample upload file method on AI Client * Updating sample evaluation * Improving Evaluation user experience * Adding pip install link for evaluation * Update sample_evaluations.py * Reverting connection changes * Update _patch.py * First emit of azure-ai-project (_patch.py, samples and test not yet copied) * Copy _patch.py, samples, test, update names * Fix wrong import * M hietala/agent tools updates (#38031) * no need for function name separately in function definitions * more changes related to not needing to specify function names separately * adding capability to remove vector stores from file search tool * adding remove file to code interpreter * review comment related change --------- Co-authored-by: Marko Hietala <[email protected]> * add changes to latest typespec and ai-project changes (#38051) * Move agents test to agents folder (#38055) * Move agents test to agents folder * Copy changes to project * Prep for connection tests. (#38060) * Create azure-ai-projects folder * Add Inference tests (#38069) * re-doing the agent toolu updates after package changes (#38078) Co-authored-by: Marko Hietala <[email protected]> * Add .diagnostics.enable() method to fetch App Insights connection strings (#38109) * Users/singankit/evaluation async samples (#38104) * Adding async sample for evaluation * Update sample_evaluations_async.py * Fix sphinx docstring and cSpell (#38125) * Fix sphinx * Fix spelling * Fix * Fix II * Generate new code * Fix test * Fix * Disable evaluation schedule and remove sampling rate (#38161) * sdk changes for online eval disable schedule and remove sampling rate * change sample * Add tests for Connection operations (#38156) * update to get file content (#38165) * Jhakulin/parse messages (#38167) * parse messages * update * add test file * update * use stream for get file content * add get_file_content_stream * update * fix merge error * review feedback * more review comments * more review comments * Update README.md (#38173) * update to get file content (#38192) * Enable tracing (#38185) * added validation to tools and tools_resources, added update agent (#38195) * Rename `.diagnostics` to `.telemetry` (#38209) * Jhakulin/agents get file content async (#38201) * update to get file content * agents async updates * update * fix await/pylance issue * fix pylance error on AsyncToolSet * return empty set if no tool resources for code interpreter * added async toolset sample * Fixed test (#38222) * Support the new unified models deployed to the AIServices connection * Fixed validation to put AyncFunctionTool into ToolSet and FunctionTool into AsyncToolSet. Introduce BaseToolset. (#38244) * adding ai agents instrumentor (#38200) * adding ai agents instrumentor * streaming tracing sample and fixes to streaming tracing * moved tracing to agents tracing folder * adding support for tracing list_messages and async tracing sample * removing some commented out code * adding separate console and azure monitor tracing samples * removing unused code * removing more unused code * renamed tracing to telemetry. removed throwing from instrument and uninstrument * Update sdk/ai/azure-ai-projects/azure/ai/projects/telemetry/agents/_utils.py Co-authored-by: Liudmila Molkova <[email protected]> * Update sdk/ai/azure-ai-projects/azure/ai/projects/telemetry/agents/_ai_agents_instrumentor.py Co-authored-by: Liudmila Molkova <[email protected]> * Update sdk/ai/azure-ai-projects/azure/ai/projects/telemetry/agents/_ai_agents_instrumentor.py Co-authored-by: Liudmila Molkova <[email protected]> * fix for streaming with eventhandler and toolset * adding previous fix for async version also * adding missing function --------- Co-authored-by: Marko Hietala <[email protected]> Co-authored-by: Liudmila Molkova <[email protected]> * Jhakulin/user functions updates (#38248) * update to get file content * agents async updates * update * fix await/pylance issue * fix pylance error on AsyncToolSet * return empty set if no tool resources for code interpreter * added async toolset sample * updates to user functions * update * fix the array type for parameters * Connections & Inference test updates to support new unified models * Improve tracing samples and add docs on telemetry.enable (#38257) * Improve tracig samples and add docs on telemetry.enable * fix typo * Jhakulin/user functions optional (#38261) * update to get file content * agents async updates * update * fix await/pylance issue * fix pylance error on AsyncToolSet * return empty set if no tool resources for code interpreter * added async toolset sample * updates to user functions * update * fix the array type for parameters * support for Optional * update * one more example with list of dicts * adding content recording control to agents instrumentor api (#38268) Co-authored-by: Marko Hietala <[email protected]> * Howie/save file test (#38306) * Added test for save_file * resolved comment * Mypy fixes for instrumentor (#38302) * Fixes * Regenerate code * Merge again * Linter fixes * Regenerate the code * Fix mypy * More mypy fixes * Fix * Fix handling of SAS token and fix mypy (#38312) * Fix handling of SAS token and fix mypy * typo * Fix also pylance issue * Glharper/ai client new tools (#38202) * [AI Client] Add new search tools * add hooks to update connection lists for bing search * use updated typespec * use ConnectionType and ConnectionsOperations * endpointOperations -> ConnectionsOperations * add AzureAISearch tool def * adapt project with new tool changes * Add new tool defs for bing and azure ai search to projects.models * use service schema for BingGroundingTool * add samples for new tools * add sharepoint tool * update prompt to use bing browsing * add sharepoint_grounding discriminator * code gen from merged typespec * update to latest typespec * review feedback * add preview header to bing grounding and azure AI serach samples * Use 'Union' insted of '|' to make it run on Pyton 3.8 and up (#38344) * Fix mypy issues and re generate the code. (#38340) * Add more mypy fixes * Fix imports * Fix * Add safe conversion method * Fix mypy * Merge to main * Fix mypy * Fix unit test * Sphinx fix * Fix cspell * Use DefaultAzureCredential from azure.identity.aio for async samples (#38362) * Rename AAD to ENTRA_ID (#38363) * added agents example section in readme and added tags to samples (#38351) * added agents example section in readme and added tags to samples * Resolved comments * Added more detail to readme * Fix mypy and do not scan examples (#38375) * Changes in a package * Fix * Fix mypy * fix mypy * Fix * Connections APIs throw instead of returning None if a connection name/type is not found (#38383) * Make some APIs throw instead of returning None. Re-try to get connection string * Update README.md. Work in progress * Fixes * Add code interpreter file attachment sample and link to readme. Add alot more detail to readme also. (#38384) * Fix broken async .telemetry.enable() method, when you don't pass in "destination" (#38399) * Fix README. Fix Optional[str] * Fix async telemetry * Sample for thread with resource and document it in readme (#38393) * Users/singankit/evaluation test (#38402) * Test file * Adding fixture * Adding evaluation e2e test * Update azure_ai_projects_tests.env * Update how the inference URL is constructed (#38409) * First * Revert back this change * Fix some cspell errors, and link errors (#38420) * cspell lint --config .vscode/cspell.json --no-summary "sdk/ai/azure-ai-projects/**/*" * Resolved comments from Krista (#38405) * Address comments * Run black * Add support of Azure asset IDs for the vector store. (#38066) * Generate code * Generate new code * Expose file API and add the async test. * Remove unnecessary change * Add async samples * Do not create ML Client in tests * Better file retrieval * Regenerate code * Re generate the code * Re generate code * Fix samle name and make iterface more clear * Add samples for attachment enterprise search * Add unit tests for message attachment * Finzlize sync tests. * Add unit tests and use constants as a URI types * Regenerate code * Fix * Generate new code * Draft commit * Add another unit test * Fix and add test data * Regenerate files * Add more unit tests * Re generate code * Regenerate files * Rename classes and regenerate the code * Fixes * Fix parameter name * Fix file names * Script to check sample names inside file comments. * Fix PR comment by Krista on async credential in ConnectionProperties class (#38424) * Black * User/singankit/evaluation spellcheck (#38427) * Spell check errors evaluation * Removing test environment from evaluation schedule test * Fix doc string errors * Fix typing in samples (#38354) * Fix samples * Re enable type checks and remove extra code * Remove non needed code * Remove non needed changes * Remove non needed changes * Fixes * Return non optional client * Remove extra line * Fix file names * Fixes * Fix async file * Fix mypy issues and suppress the ones related to typespec * Fix cSpell * Re generate code and fix unit tests * Delete sdk/ai/azure-ai-projects/samples/config.json * Fix typo * Fix unit test * Run "black" (again) * Fix a few pyright errors, and address a comment in the Python PR (#38435) * Address some of Krista's PR comments (#38444) * sdk changes for name change for provisioning state and isEnabled online evaluation (#38298) * sdk changes for name change for provisioning state and isEnabled online evaluation * add correct tsp-location.yaml * azure-ai-projects: configure otel logging in telemetry.enable() helper (#38434) * Configure otel events in projects.telemetry.enable() * [ai] fix(pylint): Resolve `W1309`, `W0107`, `W0707`, `R1705`, `W0401`, `R1714`, `W0611`, `W0404`, `W0201`, `W1201` and `W1203` (#38447) * style: Run isort * fix(pylint): Resolve W1309(f-string-without-interpolation) From sdk/ai/azure-ai-projects, run: ruff check --select=F541 azure --fix * fix(pylint): Resolve W0107(unnecessary-pass) From sdk/ai/azure-ai-projects, run ruff check --select=PIE790 --fix azure * fix(pylint): Resolve W0707(raise-missing-from) * fix(pylint): Resolve R1705(no-else-return) From azure/ai/azure-ai-projects, run ruff check --select=RET505 --fix azure * fix(pylint): Resolve W0401(wildcard-import) * fix(pylint): Resolve R1714(consider-using-in) * fix(pylint): Resolve W0611(unused-import) * fix(pylint): Resolve W0404(reimported) * fix(pylint): Resolve W0201(attribute-defined-outside-init) * fix(pylint): Resolve W1201(logging-not-lazy) and W1203(logging-fstring-interpolation) * Add readme for get_message, save_file, get_file_content, tracing, Bing grounding, enterprise (#38426) * git Add readme for get_message, save_file, get_file_content, tracing * Readme for enterprise search, code interpreter, attachment * More documentation for tools * Fixed create_thread documentation * resolved comments * Resolved comments * Resolved comments from Krista * More fixes to mypy (#38446) * Fix unit tests and some type issues. * Fixes * Fix mypy * agents telemetry pyright fixes (#38457) Co-authored-by: Marko Hietala <[email protected]> * Fix pyright for agents. (#38464) * Fix as if python version is 3.8 * Fixed * Add custom config * Move line, ignoring import * fix pip install instruction (#38480) * [ai] fix(pylint): Resolve `C0301`, `C4739`, `C4740`, `C4741`, `C4742`, `C4743` for `azure-ai-projects` (#38459) * fix(pylint): Resolve C0301(line-too-long) * fix(pylint): Resolve C4743(docstring-should-be-keyword) * fix(pylint): Resolve C4739(docstring-missing-param) * fix(pylint): Resolve C4740(docstring-missing-type) * fix(pylint): Resolve C4741(docstring-missing-return) * fix(pylint): Resolve C4742(docstring-missing-rtype) * style: Run black * Enable test recordings for tests\inference, tests\telemetry and tests\connection (#38475) * Update cspell.json * Removed enterprise and add installation for tracing (#38483) * Removed enterprise and add installation for tracing * clean up * resolved comments * Fix mypy errors (#38487) * Run black --config ../../../eng/black-pyproject.toml * agents tracing pylint fixes (#38496) Co-authored-by: Marko Hietala <[email protected]> * Remove premature adding of asserts.json file * [ai] fix(pylint): Resolving remaining pylint errors (#38500) * fix(pylint): Resolve R1705(no-else-return) * fix(pylint): Resolve C0207(use-maxsplit-arg) * fix(pylint): Resolve C0412(ungrouped-imports) * fix(pylint): Resolve W1401(anomalous-backslash-in-string) * fix(pylint): Resolve W0105(pointless-string-statement) * fix(pylint): Ignore W0221(arguments-differ) pylint has an open issue tracking false positives for arguments-differ with overloads in subclasses * fix(pylint): Ignore W0718(broad-exception-caught) * fix(pylint): Ignore E0401(import-error) and E0611(no-name-in-module) * fix(pylint): Ignore C4748(client-accepts-api-verison-keyword) Currently, neither client classes supports overriding the api version * fix(pylint): Ignore W0212(protected-access) * fix(pylint): resolve W0707(raise-missing-from) * fix(pylint): Ignore E1102(not-callable) pylint appears to fail to correctly infer that settings.tracing_implementation actually is callable * fix(pylint): Ignore W0231(super-init-not-called) * fix(pylint): Ignore W0613(unused-argument) * fix(pylint): Ignore W0236(invalid-overriden-method) * fix(pylint): Ignore R0914(too-many-locals) * fix(pylint): Ignore R0902(too-many-instance-attributes) * fix(pylint): Ignore R0915(too-many-statements) * fix(pylint): Ignore R0911(too-many-return-statements) * Disable all agents tests. Push recordings for all other tests to the assents folder (#38481) * run "black --config ../../../eng/black-pyproject.toml ." * Fix pylint and new mypy issues (#38503) * Ignore most of pylint errors * Add generated paths to pylint.rc * Use backslash for consistency in pylintrc files --------- Co-authored-by: Darren Cohen <[email protected]> * Remove async from .telemetry.enable (#38514) * Fix pylint * Rename `with_credentials` to `include_credentials` per apiview.dev review comment (#38517) * Try to fix test issues on Python 3.13. Run black * Try again to fix Python 3.13 test failure * Attempt to fix azure-ai-projects not showing up in the release pipeline * Fix release date in CHANGELOG.md * Fix CHANGELOG.md --------- Co-authored-by: Ankit Singhal <[email protected]> Co-authored-by: Jarno Hakulinen <[email protected]> Co-authored-by: howieleung <[email protected]> Co-authored-by: Glenn Harper <[email protected]> Co-authored-by: Sai Kothinti <[email protected]> Co-authored-by: nick863 <[email protected]> Co-authored-by: M-Hietala <[email protected]> Co-authored-by: Marko Hietala <[email protected]> Co-authored-by: Liudmila Molkova <[email protected]> Co-authored-by: kdestin <[email protected]>
- Loading branch information
11 people
authored
Nov 14, 2024
1 parent
9fd4095
commit 9ebd895