dan
|
b4fc827c0f
Merge latest trunk into this branch.
|
1 year ago |
dan
|
a3e8822c4b
Fix an fts5 problem where a transaction consisting of (a) a DELETE on rowid X, (b) a prefix query, and (c) an INSERT on rowid X, could corrupt the index.
|
1 year ago |
dan
|
a35ae44150
Changes so that fts5 can handle tokens with embedded '\0' bytes.
|
1 year ago |
dan
|
24730de8d1
Add the fts5 'delete-automerge' integer option. A level is eligible for auto-merging if it has a greater than or equal percentage of its entries deleted by tombstones than the 'delete-automerge' option. Default value is 10.
|
1 year ago |
dan
|
2159292ce0
Integrate contentless delete with auto-merge.
|
1 year ago |
dan
|
f7ff755651
Fix an assert() in fts5 that could fail with a corrupt database.
|
3 years ago |
dan
|
f6b0464e36
Fix some fts5 problems with signed integer overflow causing segfaults in -ftrapv builds.
|
4 years ago |
dan
|
f2e151aeab
Fix a problem in fts5 causing the in-memory hash table to be flushed more often than necessary.
|
4 years ago |
mistachkin
|
065f3bf4f2
Fix various harmless compiler warnings seen with MSVC.
|
5 years ago |
dan
|
d406d07262
Fix a buffer overread that could occur when running fts5 prefix queries inside a transaction.
|
5 years ago |
dan
|
b15f19c75e
Fix an fts5 problem with interleaving reads and writes in a single transaction.
|
5 years ago |
drh
|
2d77d80a65
Use 64-bit math to compute the sizes of memory allocations in extensions.
|
6 years ago |
dan
|
d9d7d253fc
Fix another potential buffer overread in fts5.
|
6 years ago |
mistachkin
|
831265f962
Fix harmless compiler warnings in FTS5.
|
7 years ago |
dan
|
22025ef198
Fix some problems in fts5 code detected by -fsanitize=undefined.
|
7 years ago |
mistachkin
|
77fac879d2
More harmless compiler warning fixes.
|
8 years ago |
dan
|
f2d328fa25
Fix a fairly obscure buffer overread in fts5.
|
8 years ago |
dan
|
d11e397b2c
Updates to fts5 to support detail=none mode. As of this commit, many cases are still broken.
|
9 years ago |
dan
|
9f44deed93
Change the name of the offsets=0 option to "detail=column". Have the xInst, xPhraseFirst and other API functions work by parsing the original text for detail=column tables.
|
9 years ago |
dan
|
b12dc84fbb
Add the "offsets=0" option to fts5, to create a smaller index without term offset information. A few things are currently broken on this branch.
|
9 years ago |
mistachkin
|
b9becaa268
Fix even more harmless compiler warnings.
|
9 years ago |
mistachkin
|
cdabd7bd50
Fix harmless compiler warnings.
|
9 years ago |
dan
|
d0dc8cb5a3
Remove a potentially undefined behaviour involving signed integers and bitshift operations from fts5_hash.c.
|
9 years ago |
dan
|
8694d6049f
Fix compiler warnings in fts5 code.
|
9 years ago |
dan
|
3f09beda45
Remove "#ifdef SQLITE_ENABLE_FTS5" from individual fts5 source files. Add a single "#if !defined(SQLITE_CORE) || defined(SQLITE_ENABLE_FTS5)" to fts5.c.
|
9 years ago |
dan
|
3f3074e0c1
Remove the "#include sqlite3Int.h" from fts5Int.h.
|
9 years ago |
dan
|
4591334dd4
Change to storing all keys in a single merge-tree structure instead of one main structure and a separate one for each prefix index. This is a file-format change. Also introduce a mechanism for managing file-format changes.
|
9 years ago |
dan
|
cdbd150170
Improve coverage of fts5 tests.
|
9 years ago |
dan
|
f1e58b7e8d
Begin changing fts5 to use a delete flag so that delete markers may be annihilated more quickly.
|
9 years ago |
dan
|
56e2497aaa
Improve fts5 integrity-check so that it checks that DESC queries return the same as ASC. Change the poslist format slightly to make room for a delete-flag.
|
9 years ago |