feat: anthropic batching code

[jxnl]

I'm pretty sure this works, but I would love some time for someone else to verify some of the loose ends. I don't know how to test this yet.

---

Important

Adds support for managing batch jobs on Anthropic alongside OpenAI, updating CLI and batch processing logic.

• CLI Enhancements:
  • Added --use-anthropic flag in batch.py to switch between OpenAI and Anthropic.
  • Updated create_from_file, cancel, and download_file commands to support Anthropic API.
  • Modified generate_table and get_jobs to handle Anthropic batch jobs.
• Batch Processing:
  • Updated BatchJob class in batch.py to parse and create batch jobs for both OpenAI and Anthropic.
  • Removed openai_models type and generalized model handling.
  • Adjusted RequestBody and BatchModel to accommodate Anthropic requests.
• Documentation:
  • Updated batch.md to include instructions for using Anthropic API.
  • Added examples for creating, listing, and managing batch jobs with Anthropic.

^{This description was created by for fbf81a7. It will automatically update as commits are pushed.}

[ellipsis-dev]

❌ Changes requested. Reviewed everything up to 1bdc2d5 in 39 seconds

More details

• Looked at 692 lines of code in 4 files
• Skipped 0 files when reviewing.
• Skipped posting 4 drafted comments based on config settings.

1. docs/cli/batch.md:1

• Draft comment:
  If this is a new file, ensure it's added to mkdocs.yml for documentation consistency.
• Reason this comment was not posted:
  Confidence changes required: 80%
  The mkdocs.yml file needs to be updated to include the new batch.md file for documentation consistency.

2. instructor/batch.py:114

• Draft comment:
  Add comments to explain the logic in create_from_messages, especially around handling different modes for OpenAI and Anthropic.
• Reason this comment was not posted:
  Confidence changes required: 80%
  The function create_from_messages in batch.py is quite complex and would benefit from comments explaining the logic, especially around the handling of different modes for OpenAI and Anthropic.

3. instructor/cli/batch.py:39

• Draft comment:
  The use of str() around getattr calls is unnecessary and can be removed for cleaner code. This applies to lines 39, 40, and 41.
• Reason this comment was not posted:
  Confidence changes required: 50%
  The generate_table function in batch.py uses getattr to access attributes, which is a good practice to avoid errors if the attribute doesn't exist. However, the use of str() around these calls is unnecessary and can be removed for cleaner code.

4. instructor/cli/batch.py:104

• Draft comment:
  Avoid using eval for parsing JSON data. Use json.loads instead for security reasons.
• Reason this comment was not posted:
  Marked as duplicate.

Workflow ID: wflow_0XQLXZjyvf4ZN6vE

---

Want Ellipsis to fix these issues? Tag @ellipsis-dev in a comment. You can customize Ellipsis with 👍 / 👎 feedback, review rules, user-specific overrides, quiet mode, and more.

[ellipsis-dev]

Using eval to parse JSON is a security risk. Consider using json.loads(line) instead.

[ellipsis-dev]

👍 Looks good to me! Incremental review on fbf81a7 in 20 seconds

More details

• Looked at 56 lines of code in 3 files
• Skipped 0 files when reviewing.
• Skipped posting 3 drafted comments based on config settings.

1. instructor/cli/batch.py:103

• Draft comment:
  Using eval to parse file content is a security risk. Consider using json.loads instead for safer parsing.
• Reason this comment was not posted:
  Comment was on unchanged code.

2. instructor/cli/batch.py:13

• Draft comment:
  list[Any] is not compatible with Python versions below 3.9. Use List[Any] for compatibility.
• Reason this comment was not posted:
  Confidence changes required: 80%
  The generate_table function uses list[Any] which is not compatible with Python versions below 3.9. It should be replaced with List[Any] for compatibility.

3. instructor/cli/batch.py:13

• Draft comment:
  For compatibility with Python versions below 3.9, use List[Any] instead of list[Any]. This also applies to other instances in the codebase.
• Reason this comment was not posted:
  Confidence changes required: 80%
  The generate_table function uses list[Any] which is not compatible with Python versions below 3.9. It should be updated for compatibility.

Workflow ID: wflow_j3F1tpwsSjPIJ9Ne

---

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

[cloudflare-workers-and-pages]

Deploying instructor-py with    Cloudflare Pages

Latest commit:	e2e6b6c
Status:	⚡️  Build in progress...

View logs