feat: don't print relevance score by default

[Miyou]

For post-processing RAG output with an LLM we don't want to influence the LLM by including the relevance score.

See this PR for the post-processing: ErikBjare/gptme#435

---

Important

Adds --print-relevance flag to search command in gptme_rag/cli.py to control printing of relevance scores, defaulting to not printing them.

• Behavior:
  • Adds --print-relevance flag to search command in gptme_rag/cli.py to control printing of relevance scores.
  • By default, relevance scores are not printed unless --print-relevance is specified.
• Functions:
  • Updates search() function to include print_relevance parameter and conditionally print relevance scores in get_expanded_content() and summary view.

^{This description was created by for 1aa5797. It will automatically update as commits are pushed.}

[ellipsis-dev]

👍 Looks good to me! Reviewed everything up to 1aa5797 in 54 seconds

More details

• Looked at 43 lines of code in 1 files
• Skipped 0 files when reviewing.
• Skipped posting 5 drafted comments based on config settings.

1. gptme_rag/cli.py:477

• Draft comment:
  Include print_relevance condition here so that relevance is printed only when requested. Consider adding a brief comment explaining that relevance is intentionally hidden by default for LLM processing.
• Reason this comment was not posted:
  Comment was not on a location in the diff, so it can't be submitted as a review comment.

2. gptme_rag/cli.py:538

• Draft comment:
  Guard the relevance printing in summary view with the print_relevance flag. This ensures consistency with the full view update.
• Reason this comment was not posted:
  Comment looked like it was already resolved.

3. gptme_rag/cli.py:358

• Draft comment:
  Good addition of the '--print-relevance' flag. Consider updating the search command's docstring to mention that relevance scores are hidden by default unless this flag is set.
• Reason this comment was not posted:
  Confidence changes required: 0%
  None

4. gptme_rag/cli.py:483

• Draft comment:
  The condition now checks both 'distances' and 'print_relevance' before printing the relevance score in full format. This meets the intended behavior.
• Reason this comment was not posted:
  Confidence changes required: 0%
  None

5. gptme_rag/cli.py:535

• Draft comment:
  Similarly, in the summary view the check is updated to ensure relevance is printed only when explicitly requested. This change is consistent with the new flag.
• Reason this comment was not posted:
  Confidence changes required: 0%
  None

Workflow ID: wflow_DNii9swN1saNOhBc

---

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

[ErikBjare]

Looks good, just a nit about parameter naming

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 0% with 4 lines in your changes missing coverage. Please review.

Project coverage is 48.16%. Comparing base (6c1900f) to head (eba8be5).

Files with missing lines	Patch %	Lines
gptme_rag/cli.py	0.00%	4 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master      #11      +/-   ##
==========================================
- Coverage   48.23%   48.16%   -0.08%     
==========================================
  Files          10       10              
  Lines        1362     1364       +2     
==========================================
  Hits          657      657              
- Misses        705      707       +2     

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