Add a MD5 checksum to the BDIC header....

third_party/hunspell/google/bdict.h

 // Copyright 2008 Google Inc. All Rights Reserved.
 
 #ifndef CHROME_THIRD_PARTY_HUNSPELL_GOOGLE_BDICT_H__
 #define CHROME_THIRD_PARTY_HUNSPELL_GOOGLE_BDICT_H__
 
 #include "base/basictypes.h"
+#include "base/md5.h"
 
 // BDict (binary dictionary) format. All offsets are little endian.
 //
-// Header (12 bytes).
+// Header (28 bytes).
 //   "BDic" Signature (4 bytes)
 //   Version (little endian 4 bytes)
 //   Absolute offset in file of the aff info. (4 bytes)
 //   Absolute offset in file of the dic table. (4 bytes)
+//   (Added by v2.0) MD5 checksum of the aff info and the dic table. (16 bytes)
 //
 // Aff information:
 //   Absolute offset in file of the affix group table (4 bytes)
 //   Absolute offset in file of the affix rules table (4 bytes)
 //   Absolute offset in file of the replacements table (4 bytes)
 //   Absolute offset in file of the "other rules" table (4 bytes)
 //
 //   The data between the aff header and the affix rules table is the comment
 //   from the beginning of the .aff file which often contains copyrights, etc.
-// 
+//
 //   Affix group table:
 //     Array of NULL terminated strings. It will end in a double-NULL.
-//   
+//
 //   Affix rules table:
 //     List of LF termianted lines. NULL terminated.
 //
 //   Replacements table:
 //     List of pairs of NULL teminated words. The end is indicated by a
 //     double-NULL. The first word in the pair is the replacement source, the
 //     second is what to replace it with. Example:
 //       foo\0bar\0a\0b\0\0
 //     for replacing ("foo" with "bar") and ("a" with "b").
 //
 //   Other rules table:
 //     List of LF termianted lines. NULL terminated.
-//     
+//
 //
 // Dic table. This stores the .dic file which contains the words in the
 // dictionary, and indices for each one that indicate a set of suffixes or
 // prefixes that can be applied. We store it in a trie to save space. It
 // replaces Hunspell's hash manager.
 //
 //   0abxxxxx xxxxxxxx (in binary) Leaf node:
 //     The number stored in the bits represented by x is the affix index.
 //
 //     If bit <a> is set, the leaf node has an additional string. Following the
@@ skipping 41 matching lines @@
 
 namespace hunspell {
 
 #pragma pack(push, 1)
 
 class BDict {
  public:
   // File header.
   enum { SIGNATURE = 0x63694442 };
   enum {
-    MAJOR_VERSION = 1,
-    MINOR_VERSION = 1
+    MAJOR_VERSION = 2,
+    MINOR_VERSION = 0
   };
   struct Header {
     uint32 signature;
 
     // Major versions are incompatible with other major versions. Minor versions
     // should be readable by older programs expecting the same major version.
     uint16 major_version;
     uint16 minor_version;
 
     uint32 aff_offset;  // Offset of the aff data.
     uint32 dic_offset;  // Offset of the dic data.
+
+    // Added by version 2.0.
+    MD5Digest digest;  // MD5 digest of the aff data and the dic data.
   };
 
   // AFF section ===============================================================
 
   struct AffHeader {
     uint32 affix_group_offset;
     uint32 affix_rule_offset;
     uint32 rep_offset;  // Replacements table.
     uint32 other_offset;
   };
@@ skipping 59 matching lines @@
   enum { LIST_NODE_TYPE_MASK = 0xE0 };   // 11100000
   enum { LIST_NODE_TYPE_VALUE = 0xE0 };  // 11100000
 
   // The 4th from highest bit indicates a 16 bit (as opposed to 8 bit) list.
   // This mask/value also includes the list ID in the high 3 bits.
   enum { LIST_NODE_16BIT_MASK = 0xF0 };   // 11110000
   enum { LIST_NODE_16BIT_VALUE = 0xF0 };  // 11110000
 
   // The low 4 bits of the list ID byte are the count.
   enum { LIST_NODE_COUNT_MASK = 0xF };  // 00001111
+
+  // Verifies the specified BDICT is sane. This function checks the BDICT header
+  // and compares the MD5 digest of the data with the one in the header.
+  static bool Verify(const char* bdict_data, size_t bdict_length);
 };
 
 #pragma pack(pop)
 
 }  // namespace hunspell
 
 #endif  // CHROME_THIRD_PARTY_HUNSPELL_GOOGLE_BDICT_H__