Awesome Open Source
Awesome Open Source


A tool for extracting information from IDA databases. idbtool knows how to handle databases from all IDA versions since v2.0, both i64 and idb files. You can also use idbtool to recover information from unclosed databases.

idbtool works without change with IDA v7.0.

Much faster than loading a file in IDA

With idbtool you can search thousands of .idb files in seconds.

More precisely: on my laptop it takes:

  • 1.5 seonds to extract 143 idc scripts from 119 idb and i64 files.
  • 3.8 seonds to print idb info for 441 files.
  • 5.6 seconds to extract 281 enums containing 4726 members from 35 files.
  • 67.8 seconds to extract 5942 structs containing 33672 members from 265 files.

Loading an approximately 5 Gbyte idb file in IDA, takes about 45 minutes. While idb3.h takes basically no time at all, no more than a few milliseconds.


Two versions of this tool exist:

One written in python

One written in C++

Both repositories contain a library which can be used for reading .idb or .i64 files.

An IDA Pro plugin making use of idb3.h can be found here:

This is a plugin making it easy to copy scripts, structs or enums from recent ida databases.



idbtool [options] [database file(s)] [-- address-list]
  • -n or --names will list all named values in the database.

  • -s or --scripts will list all scripts stored in the database.

  • -u or --structs will list all structs stored in the database.

  • -e or --enums will list all enums stored in the database.

  • -i or --info will print some general info about the database.

  • -a list all names, including ..todo..

  • -d dump btree page tree contents.

  • --inc, --dec list all records in ascending / descending order.

  • -q or --query search specific records in the database.

  • -m or --limit limit the number of results returned by -q.

All addresses after -- will be printed as symbol+offset.


Queries need to be specified last on the command line.


idbtool [database file(s)]  --query  "Root Node;V"

Will list the source binary for all the databases specified on the command line.

A query is a string with the following format:

  • [==,<=,>=,<,>] - optional relation, default: ==
  • a base node key:
    • a DOT followed by the numeric value of the nodeid.
    • a HASH followed by the numeric value of the system-nodeid.
    • a QUESTION followed by the name of the node. -> a 'N'ame node
    • the name of the node. -> the name is resolved, results in a '.'Dot node
  • an optional tag ( A for Alt, S for Supval, etc )
  • an optional index value

Example queries:

  • Root Node;V -> prints record containing the source binary name
  • ?Root Node -> prints the Name record pointing to the root
  • >Root Node -> prints the first 10 records starting with the root node id.
  • <Root Node -> prints the 10 records startng with the records before the root node.
  • .0xff000001;N -> prints the root node name entry.
  • #1;N -> prints the root node name entry.

List the highest node and following record in the database in two different ways, the first: starting at the first record below ffc00000, and listing the next. The second: starting at the first record after ffc00000, and listing the previous:

  • --query "<#0xc00000" --limit 2 --inc -v
  • --query ">#0xc00000" --limit 2 --dec -v

Note that this should be the nodeid in the $ MAX NODE record.

List the last two records:

  • --limit 2 --dec -v

List the first two records, the $ MAX LINK and $ MAX NODE records:

  • --limit 2 --inc -v

A full database dump

Several methods exist for printing all records in the database. This may be useful if you want to investigate more of IDA''s internals. But can also be useful in recovering data from corrupted databases.

  • --inc, --dec can be used to enumerate all b-tree records in either forward, or backward direction.
  • --id0 walks the page tree, instead of the b-tree, printing the contents of each page


The header file idb3.h contains a library for reading from IDA Pro databases.


Class for accessing sections of an .idb or .i64 file.

Constructor Parameters:

  • std::shared_ptr<std::istream> ( typedefed to stream_ptr )


  • stream_ptr getsection(int)

ID0File, ID1File, NAMFile

Constructor Parameters:

  • IDBFile& idb
  • stream_ptr


  • INDEX - the argument for idb.getsection



  • Cursor find(relation_t, nodeid, ...)

    • ... can be:
      • tag, index
      • tag, hash
      • tag
  • Cursor find(relation_t, std::string key)

  • std::string blob(nodeid, tag, ...)

  • uint64_t node(std::string name)

  • bool is64bit()

    • true for .i64 files.
  • uint64_t nodebase()

    • return 0xFF000000(00000000) for 32/64 bit databases.
  • void enumlist(uint64_t nodeid, char tag, CB cb)

    • call cb for each value in the list.

Convenience Methods

  • std::string getdata(ARGS...args)
  • std::string getstr(ARGS...args)
  • uint64_t getuint(ARGS...args)
  • uint64_t getuint(BtreeBase::Cursor& c)
  • std::string getname(uint64_t node)



  • uint32_t GetFlags(uint64_t ea)



  • uint64_t findname(uint64_t ea)



  • void next()
    • move cursor to the next btree record
  • void prev()
    • move cursor to the previous btree record
  • bool eof()
    • did we reach the start/end of the btree?
  • std::stringgetkey()`
    • return the key pointed to by the cursor
  • std::stringgetval()`
    • return the value pointed to by the cursor


  • add option to list all comments stored in the database
  • support compressed sections
  • add option to list flags for a list of addresses.


Willem Hengeveld [email protected]

Get A Weekly Email With Trending Projects For These Topics
No Spam. Unsubscribe easily at any time.
c-plus-plus (18,586
reverse-engineering (472