X-Git-Url: http://gitweb.fperrin.net/?p=Dictionary.git;a=blobdiff_plain;f=dictionary-format.txt;h=97182373faa7900a0a86cc2c975e299a4ccbbebf;hp=55fd1c6cc4835b4d671a11ae21bb17f08a016ae7;hb=HEAD;hpb=9c9fb13dce110f98b7a7e23927513f91032bc4c4 diff --git a/dictionary-format.txt b/dictionary-format.txt index 55fd1c6..9718237 100644 --- a/dictionary-format.txt +++ b/dictionary-format.txt @@ -1,26 +1,27 @@ This is a quick write-up of the dictionary file format, v7. -v6 is troublesome as it relies on Java serialization and thus -I won't even attempt to document it. -This is hasn't been checked for correctness and likely has some bugs. +v6 is troublesome as it relies on Java serialization but is now +also documented in dictionary-format-v6.txt. +This hasn't been checked much for correctness and likely has some bugs. Also, I really should have used some standard format for writing this... =========================================== Some basic types: -[String] -[Short]: string length -n bytes: string, modified UTF-8, n is value from previous element - note: no zero termination - [Short] -2 bytes: big-endian, signed value (note: negative values generally not used here) + 2 bytes: big-endian, signed value (note: negative values generally not used here) [Int] -4 bytes: big-endian, signed value (note: negative values generally not used here) + 4 bytes: big-endian, signed value (note: negative values generally not used here) [Long] -8 bytes: big-endian, signed value (note: negative values generally not used here) + 8 bytes: big-endian, signed value (note: negative values generally not used here) + + +[String] + [Short]: string length + n bytes: string, modified UTF-8, n is value from previous element + note: no zero termination ====================================================== @@ -62,7 +63,7 @@ which can take on of these forms: For decoding, the number of leading 1s in the first byte is the overall length - 1. -Note that this scheme would allow storing an even larger range values +Note that this scheme would allow storing an even larger range of values in the 5-byte variant and can be extended to arbitrary length, however that is not currently implemented. @@ -75,24 +76,28 @@ To reduce the cost of this table and enable more efficient compression, multiple entries can be stored in a block that gets one single index entry. I.e. it is only possible to do random-access to the start of a block, seeking to elements further inside the block must be done via reading. -Caching should be used to reduce the impact of this. +Caching should be used to reduce the performance impact of this (so +that when entries 5, 4, 3 etc. of a block are read sequentially, +parsing and decompression is done only once). These lists have the following base format: -[varInt]: number of entries in the list (must be >= 1) () +[varInt]: number of entries in the list (must be >= 0) () [varInt]: compression block size (in entries) (must be >= 1) () [varInt]: flags. Currently only bit 0 used, indicating compression is used -=/*4 + 4 bytes: -(note division with rounding up if not divisible) -table-of-contents. [Int] offset value for each block of entries. -Followed by a final [Int] offset value to the end of the list data (). -Each offset is relative to the start of this block. -Note that currently for simplicity Java int type is used -to process these values, even though negative values make no sense. -This limits the maximum amount of data to around 2GB. +=(/)*4 + 4 bytes: + (note division with rounding up if not divisible) + table-of-contents. + [Int] offset value for each block of entries. + Followed by a final [Int] offset value to the end of the list data (). + Each offset is relative to the start of this block. + Note that currently for simplicity Java int type is used + to process these values, even though negative values make no sense. + This limits the maximum amount of data to around 2GB. -- bytes: data +- bytes: + entry data If compression is enabled, the data for each block is deflate compressed. @@ -103,7 +108,7 @@ the start of the next block must be used to find the end of the compressed data and signal the end of input to the inflater, and whatever data the inflater produces is the decompressed input (note: this might be an implementation -detail, you might need the decompressed size to decompress +detail, you might not need the decompressed size to decompress and parse the contained data, but it prevents things like the inflater overreading and producing spurious errors).