License Stars Issues Last Commit

A low-level tool for analyzing and potentially recovering data from broken Firebird database files (.fdb, .gdb). It bypasses the standard Firebird engine to read raw page structures and data.

Warning: This is a development version. It directly manipulates database files. Always create a backup of your database before using this tool. Use at your own risk.

• Database Structure Analysis:
  • Opens and reads Firebird database files.
  • Identifies and counts pages by type (Header, PIP, TIP, Pointer, Data, Index Root/BTree, Blob, Generator, WAL).
  • Displays basic page information (number, type, checksum, relation ID).
• Header Manipulation:
  • Reads and interprets database header flags.
  • Allows setting/clearing specific flags (e.g., Read Only, Force Write).
• Page Content Inspection:
  • Reads specific pages by number.
  • Extracts and displays raw record fragments from Data Pages.
  • Allows changing page type and checksum (Dangerous!).
• Transaction Analysis:
  • Reads Transaction Inventory Pages (TIP).
  • Counts transactions by state (Active, Limbo, Dead, Committed).
• Basic Data Recovery:
  • Extracts raw data fragments from Data Pages.
  • (WIP) Basic MVCC (Multi-Version Concurrency Control) awareness for data visibility (see TODO).
  • (New) Parses basic record headers to identify deleted records and data length.
• Record Fragment Parsing:
  • Introduces uRecordParser module for analyzing individual record fragments.
  • Extracts key information like transaction ID, back pointer, flags, and data length from the record header.

• Advanced Data Recovery:
  • Implement full MVCC logic to reconstruct the correct, visible version of records based on transaction states and TIP page data.
  • Export recovered data to a new, valid Firebird database or other formats (e.g., CSV).
  • Analyze and potentially repair B-Tree index structures.
  • Analyze and potentially repair Blob page chains.
• Deeper Integrity Checks:
  • Verify logical links between pages (e.g., Pointer Pages -> Data Pages, Record Back Pointers).
  • Identify orphaned pages.
• Page Generation/Repair:
  • Implement logic to generate missing TIP or PIP pages if necessary.
  • Attempt to rebuild damaged page structures.
• User Interface Improvements:
  • Add search functionality within page data.
  • Provide more detailed views for different page types (B-Tree keys, Blob data).
  • Add a 'Restore from Backup' feature.
  • Improve error handling and user feedback.

• Database Header Page
• Page Inventory Page (PIP)
• Transaction Inventory Page (TIP)
• Pointer Page
• Data Page
• Index Root Page
• Index B-Tree Page
• Blob Data Page
• Generator Page
• Write Ahead Log Page (WAL) (Note: WAL functionality might be limited)

• Language: Pascal (Delphi).
• Target: Windows.
• Approach: Direct file I/O, parsing raw page structures according to Firebird Internal Format documentation (fbint-page-1.html) and Firebird source code conventions.
• Modules:
  • uStructs: Core data structures (TPag, THdr, etc.).
  • uPageAnalyzer: General page reading and type identification.
  • uDataPage: Data Page header parsing and record fragment extraction.
  • uTipPage: TIP page header parsing and transaction state analysis.
  • uBtreePage: B-Tree page header parsing and key extraction.
  • uRecordParser: (New) Parses headers of individual record fragments.
  • uDatabaseStats: Aggregates statistics using uPageAnalyzer.
  • uFlagManager: Manages database header flags.

MIT

Gordienko Roman