feat: Add optional cross-page table merging functionality

[Libres-coder]

Summary

Implements optional table merging for tables spanning multiple pages. Currently disabled by default to ensure backward compatibility.

Related Issye

#131

Solution

• Added post-processing to merge Markdown and HTML tables from consecutive pages
• Merges based on column count and header similarity (≥80% for Markdown)
• Graceful fallback if merging fails

Changes

Modified:

• config.py - Added ENABLE_TABLE_MERGE, ENABLE_HTML_TABLE_MERGE, TABLE_MERGE_LOG flags
• run_dpsk_ocr_pdf.py - Integrated optional merge post-processing

Added:

• process/table_merger.py - Table extraction, parsing, and merge logic

Features

• Backward compatible (disabled by default)
• No new dependencies
• Supports Markdown and HTML tables
• Error handling with fallback

Usage

Enable in config.py:

ENABLE_TABLE_MERGE = True

[minhnguyent546]

Could you please support merging HTML tables also?

[Libres-coder]

Could you please support merging HTML tables also?

get it

[Libres-coder]

@ImadSaddik @Ucas-HaoranWei ptal,thx

[athrael-soju]

This is a much needed feature. Any progress on this PR?

[Libres-coder]

@athrael-soju This feature has been implemented.The cross-page table merging is now available. It's disabled by default to ensure backward compatibility.The feature supports both Markdown and HTML tables with automatic fallback on errors.
To enable it, simply set in config.py:

ENABLE_TABLE_MERGE = True

[Libres-coder]

@athrael-soju I've tested it with both Markdown and HTML tables. Here's my local demo:

[athrael-soju]

@Libres-coder this is excellent, thanks for sharing!