e7a86aee47
Repaired problem in commit 577fd9d3a634 which broke case-sensitive indexing term prefix wrapping
Jean-Francois Dockes2014-07-24 15:29:19 +02:00
368ebf2f86
All docs get containerfilename field, indexed exclusively with prefix (only a field search containerfilename:someterm can match).
Jean-Francois Dockes2014-07-23 19:30:41 +02:00
19da7020d6
Add new kind of aliases for field names, to be used only at query time
Jean-Francois Dockes2014-07-23 16:52:08 +02:00
578511c3e2
Add flag qualifying field to be indexed exclusively with prefix
Jean-Francois Dockes2014-07-23 15:28:16 +02:00
c7a9aced05
tests: make it easier to move to other machine by sorting on url and other small changes
Jean-Francois Dockes2014-07-23 10:23:03 +02:00
846079184a
mimeconf: create entries for application/javascript and application/x-php (for using xdg-mime)
Jean-Francois Dockes2014-07-23 10:21:37 +02:00
46e924f3ea
Allow configuration of the last-resort external command used to identify mime types
Jean-Francois Dockes2014-07-19 16:15:43 +02:00
c11453fc27
Treat the "open parent" link click more like the right-click menu entry for open parent. The latter is still different because it opens the enclosing document, not necessarily a folder if the doc result is embedded. But they now behave the same if the parent is a folder
Jean-Francois Dockes2014-07-19 15:31:03 +02:00
73f15d134b
Processs xdg.tags and xdg.comment freedesktop extended attribute fields, as complementing keywords and description
Jean-Francois Dockes2014-07-19 15:03:40 +02:00
ad2576c8f7
Gui result list: add %P substitution for parent folder and F link target for opening the parent folder
Jean-Francois Dockes2014-07-17 10:46:15 +02:00
088f1f9a9f
try out modification to use container file name for display, but not search, purposes when the subdocument file name is not set
Jean-Francois Dockes2014-07-16 17:05:26 +02:00
8c239bef6f
python module: query_iter function must incref the query (problem surfaced because the query ref count is now correct following the previous incref fix
Jean-Francois Dockes2014-07-16 14:29:18 +02:00
b6eb3589ba
do not unaccent Bengali characters (process like the Hindi ones)
Jean-Francois Dockes2014-07-16 12:47:30 +02:00
39b46dc55f
Consequences of 3656:ee3c7f3d964f. Deal with the fact that some subdocs may not have a filename field at all
Jean-Francois Dockes2014-07-07 18:58:00 +02:00
0a7f5637eb
Python module: stray INCREF on db object prevented descriptor and memory freeing. Fixed prototypes for db_close and query_close methods (fixes possible core dump)
Jean-Francois Dockes2014-06-26 12:43:18 +02:00
25271db690
msword docs: avoid generating an error for files containing only a picture (empty antiword output)
Jean-Francois Dockes2014-07-06 16:24:11 +02:00
4a783beadc
Query language: it used to be that an entry lacking whitespace but splittable, like [term1,term2] was transformed into a phrase search, which made sense in some cases, but no so many. This commit changes this: [term1,term2] now means term1 AND term2. [term1/term2] means term1 OR term2. This is especially useful for field searches where you would previously be forced to repeat the field name for every term. [somefield:term1 somefield:term2] can now be expressed as somefield:term1,term2
Jean-Francois Dockes2014-06-12 17:12:08 +02:00
5f05f5d588
forgot to return false when no app found -> bad alloc/core dump
Jean-Francois Dockes2014-06-12 16:55:24 +02:00
3814da2a20
execmd: limit the cases where the child process could call the LOGXX functions to really exceptional cases. Previously, the child process could be called with a non-existant command to exec, which would trigger a LOGERR call, which in turn could block because the log mutex had been taken in the father process (and there is nobody in the child to free it). This would manifest itself by 20mn of "selectloop returned 1" messages as the father was waiting for a blocked child until the filter timeout. Other threads would go on, and the timeout would finally trigger, so this did not end up as a failed indexing as long as someone was patient...
Jean-Francois Dockes2014-06-07 18:48:35 +02:00
50ee2838dc
Use a single Xapian Database object for update and query operations while indexing instead of separately open read/write ones. Separate objects seem to trigger Xapian issues
Jean-Francois Dockes2014-06-07 11:44:04 +02:00
12acdc4faf
Allow access to Xapian docid for Recoll document in recollq and Python API. Add sample Python program to find document duplicates, using MD5 terms
Jean-Francois Dockes2014-05-19 12:00:15 +02:00
4ac34cb134
Off by one error in maximum embedding depth test caused overflow of FileInterner m_tmpflgs temp flags array and possibly bus error depending on arch (only seen on 32 bits arch)
Jean-Francois Dockes2014-05-15 15:15:01 +02:00
e867f855ad
get rid of numerous probably inocuous valgrind/helgrind messages by ensuring that actual string copies are passed between threads, without refcount/shared data magic
Jean-Francois Dockes2014-05-05 19:01:58 +02:00
deb4cc8d12
mt indexing: do not pass copies of Xapian::Document between threads: the reference counting is not mt-safe. Replace with pointers
Jean-Francois Dockes2014-05-05 10:59:50 +02:00
d9fba74dbe
add /media to the default skippedPaths list mostly as a reminder that blindly processing these with the general indexer is a bad idea (use separate indexes instead).
Jean-Francois Dockes2014-04-30 10:50:58 +02:00
0ded457258
command line query: add option -N to print field names between values when -F is used. Interpret [-F ""] as print all fields
Jean-Francois Dockes2014-04-30 10:01:03 +02:00
077aed3018
fix term byte offsets produced by new textsplit: for highlighting
Jean-Francois Dockes2014-04-24 12:42:10 +02:00
ece15318ab
New text splitter with word accumulator and full partial span generation. Search/Index seem ok. Still a pb with use for highlighting (preview)
Jean-Francois Dockes2014-04-24 10:13:19 +02:00
Jean-Francois Dockes