Add two-phase timeout for PDF job processing

[Copilot]

Remove glob wildcard changes, keep only PDF timeout features

Per user request, removed all changes except the PDF timeout functionality:

Removed (glob wildcard security fixes):

• ❌ import glob statement
• ❌ Symlink removal using glob.glob() and os.remove()
• ❌ Wildcard expansion in index_collections method

Kept (PDF timeout features):

• ✅ Two-phase timeout system (queued 5min, running 30min)
• ✅ CLI arguments (--pdf-timeout-queued, --pdf-timeout-running)
• ✅ Enhanced logging with status transitions
• ✅ Context-aware error messages
• ✅ Smart timeout reset on status change

Changes:

• Reverted symlink removal to original shell command: subprocess.run('rm {pattern}', shell=True)
• Reverted index_collections to original: cmd.append(xml_file_path)
• Removed glob import
• All PDF timeout logic remains intact

Result:

Branch now contains ONLY the PDF timeout improvements needed for smooth merge into index_creators.

Original prompt

Problem

Currently, the code uses subprocess.run() with file paths containing wildcards (*), relying on shell=True for expansion. This is problematic because:

1. Using shell=True is a security risk
2. The wildcard expansion happens in the shell, not in Python
3. When the shell command fails, the actual asterisk character is passed literally

Solution

Replace shell-based wildcard expansion with Python's glob.glob() module:

Changes Required

1. Add import: Add import glob at the top of arcflow/main.py (around line 1-12 with other imports)

2. Fix symlink removal (lines ~473-486):

   • Replace the shell command rm {xml_file_path} with Python's os.remove()
   • Use glob.glob() to expand the wildcard pattern before removing files
   • Remove shell=True parameter

   Current code:

   xml_file_path = f'{xml_dir}/{repo_id}_*_batch_{batch_num}.xml'
   result = subprocess.run(
       f'rm {xml_file_path}',
       shell=True,
       cwd=self.arclight_dir,
       stderr=subprocess.PIPE,)

   Should become:

   xml_file_pattern = f'{xml_dir}/{repo_id}_*_batch_{batch_num}.xml'
   xml_files = glob.glob(xml_file_pattern)
   
   for xml_file_path in xml_files:
       try:
           os.remove(xml_file_path)
           self.log.info(f'{" " * indent_size}Removed pending symlink {xml_file_path}')
       except FileNotFoundError:
           self.log.warning(f'{" " * indent_size}File not found: {xml_file_path}')
       except Exception as e:
           self.log.error(f'{" " * indent_size}Error removing pending symlink {xml_file_path}: {e}')

3. Fix index_collections method (line ~576):

   • Use glob.glob() to expand wildcards in xml_file_path before passing to traject command
   • Add a check to warn if no files match the pattern

   Current code:

   cmd.append(xml_file_path)

   Should become:

   # Expand wildcards with glob
   xml_files = glob.glob(xml_file_path)
   
   if not xml_files:
       self.log.warning(f'{indent}No files found matching pattern: {xml_file_path}')
       return
   
   # Add all matching files to the command
   cmd.extend(xml_files)

Benefits

• Security: Eliminates use of shell=True in subprocess calls
• Reliability: Explicit wildcard expansion makes the code behavior clearer
• Portability: Using Python built-ins (os.remove, glob.glob) instead of shell commands
• Error handling: Better control over what happens when files don't exist

Testing

After making these changes, test with the same command that exposed the issue:

python -m arcflow.main \
  --arclight-dir /Users/adryden3/Projects/ASpace/development_env/arcuit \
  --aspace-dir /Users/adryden3/Projects/ASpace/development_env/my-app \
  --solr-url "http://localhost:8983/solr/blacklight-core" \
  --agents-only \
  --force-update

Verify that:

1. Files matching the wildcard pattern are correctly identified
2. No error messages about files not found
3. The indexing completes successfully

This pull request was created from Copilot chat.

---

💡 You can make Copilot smarter by setting up custom instructions, customizing its development environment and configuring Model Context Protocol (MCP) servers. Learn more Copilot coding agent tips in the docs.

[alexdryden]

@copilot the only thing I want to keep from this is the timeout for waiting for pdf generation, can you remove everything else so that I can smoothly merge this into index_creators?

[Copilot]

@copilot the only thing I want to keep from this is the timeout for waiting for pdf generation, can you remove everything else so that I can smoothly merge this into index_creators?

Done! Removed all glob wildcard changes and kept only the PDF timeout features (commit 6674a21).

The branch now contains only:

• Two-phase timeout system (queued 5min, running 30min)
• CLI arguments for timeout configuration
• Enhanced logging and status tracking
• Context-aware error messages

Removed:

• glob import and wildcard expansion
• os.remove() file removal
• All shell security fixes

Ready for smooth merge into index_creators.