fix: better anthropic cost logging

[ErikBjare]

Also set a bunch of loglevels for libs higher, so now -v output is actually readable.

Important

Enhance logging in llm_anthropic.py to capture usage data and chunk types, and adjust logging levels in init.py.

• Logging Enhancements in llm_anthropic.py:
  • Add logger.debug(response.usage) in chat() to log usage data.
  • Add logger.debug(chunk.message.usage) in stream() for message_start case.
  • Add logger.debug(chunk.usage) in stream() for message_delta case.
• Logging Configuration in init.py:
  • Set anthropic logger level to INFO to reduce debug log spam.
  • Set httpcore logger level to WARNING.

^{This description was created by for f074706. It will automatically update as commits are pushed.}

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to d0c1c8b in 13 seconds

More details

• Looked at 48 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 2 drafted comments based on config settings.

1. gptme/llm/llm_anthropic.py:97

• Draft comment:
  Combine logging of chunk.type and chunk.usage into a single log entry for better readability and performance.

logger.info(f"Chunk type: {chunk.type}, Usage: {chunk.usage}")

• Reason this comment was not posted:
  Confidence changes required: 50%
  The logging of chunk.type and chunk.usage should be combined into a single log entry for better readability and performance.

2. gptme/llm/llm_anthropic.py:75

• Draft comment:
  The import statement for anthropic.types.beta.prompt_caching is unnecessary here since it's already imported at the module level under TYPE_CHECKING.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The import statement for anthropic.types.beta.prompt_caching is unnecessary in the stream function since it is already imported at the module level under TYPE_CHECKING.

Workflow ID: wflow_Qev6fcPd4ztv9D4m

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.

[codecov-commenter]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 72.57%. Comparing base (65efc3e) to head (f074706).
Report is 1 commits behind head on master.

✅ All tests successful. No failed tests found.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #321      +/-   ##
==========================================
+ Coverage   72.49%   72.57%   +0.07%     
==========================================
  Files          67       67              
  Lines        4912     4922      +10     
==========================================
+ Hits         3561     3572      +11     
+ Misses       1351     1350       -1     

Flag	Coverage Δ
anthropic/claude-3-haiku-20240307	70.62% <100.00%> (+0.08%)	⬆️
openai/gpt-4o-mini	69.36% <25.00%> (-0.04%)	⬇️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[ellipsis-dev]

👍 Looks good to me! Incremental review on f074706 in 15 seconds

More details

• Looked at 64 lines of code in 2 files
• Skipped 0 files when reviewing.
• Skipped posting 2 drafted comments based on config settings.

1. gptme/llm/llm_anthropic.py:68

• Draft comment:
  The logging level for response.usage should be info instead of debug to match the PR description.
• Reason this comment was not posted:
  Comment did not seem useful.

2. gptme/llm/llm_anthropic.py:127

• Draft comment:
  The logging level for chunk.message.usage should be info instead of debug to match the PR description. This also applies to chunk.usage on line 130.
• Reason this comment was not posted:
  Marked as duplicate.

Workflow ID: wflow_Cr8cOkrTD24ehuxo

---

You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.