2005-12-02 Debajyoti Bera * beagled/DumpIndex.cs: term_enumerator returns terms that are already sorted. No need to scan all the terms. * beagled/Server.cs: Pre-initialize the serializer in connection handler. Removes the slight lag(?) in the first response. 2005-11-29 Debajyoti Bera * beagled/KonqHistoryQueryable, Filters/FilterKonqHistory.cs, Util/KonqHistoryUtil.cs, Util/StringFu.cs: Initial support for indexing konqueror webhistory. 2005-11-29 Veerapuram Varadhan * beagled/Filter.cs (AppendText): Don't "Trim()" the string, as lucene takes care of it. This saves good amount of memory-allocation when we have text with line breaks. * Filters/FilterSource.cs (AppendToTokens) (ExtractTokens): Replace un-necessary calls to StringBuilder.ToString() to reduce memory-allocations. 2005-11-27 Daniel Drake * beagled/GaimLogQueryable/GaimLogQueryable.cs: Misc cleanups, don't index arbitrary files, merge watching and crawling into a single task. * beagled/GaimLogQueryable/GaimLogCrawler.cs: Removed, now unused 2005-11-26 Veerapuram Varadhan * beagled/Filter.cs: (AppendText): Check for "\n" before splitting the argument in to array of lines. This saves around 600K of memory allocation for a 130K "C" source code file. 2005-11-25 Veerapuram Varadhan * Filters/FilterShellscript.cs: Shell script filter. * Filters/FilterSource.cs: New style added for Shellscripts. Token generation is moved to a private method. Actually, tokens in shell-scripts are not separated *ONLY* by operators but also by "SPACES". * Filters/Makefile.am: Include Shellscript filter for compilation. 2005-11-22 Dan Winship * libbeagle/beagle/beagle-property.c (prop_to_xml): encode the "Type" attribute correctly. 322140. 2005-11-21 Debajyoti Bera * beagled/LuceneFileQueryable: Fix a bug in LFQ where it wasnt setting the attribute of a file shared across multiple indexables. * BlamQueryable.cs, LifereaQueryable.cs, AkregatorQueryable.cs : Use fixed LFQ. 2005-11-20 Lukas Lipka * Filters/Makefile.am: Disable FilterChm.cs build since it was broken with the recent HTML filter updates. 2005-11-19 Daniel Drake * configure.in, glue/Makefile.am, glue/gecko-utils.cpp, Tiles/TileCanvas.cs: Firefox 1.5 support 2005-11-15 Daniel Drake * Use Mono.Unix.Native. This also includes a switch to Mono's extended attribute bindings which I wrote ages ago. * Tiles/TileBlog.cs, Tiles/template-blog.html: Update with recent property name changes 2005-11-15 Debajyoti Bera * beagled/ExtractContent.cs: Add option to specify output file. * HtmlFilter: Filter non-english URLs correctly. 2005-11-14 Debajyoti Bera * Best: Fix not storing settings when ran with --no-tray 2005-11-13 Debajyoti Bera * Filters/FilterHtml.cs, HtmlAgilityPack: - Instead of building the whole DOM tree, make it event driven. Whenever an HtmlNode is found, a callback function is called and the node is discarded. - HtmlAgilityPack used to load the whole file content in a string using readtoend() and operate on that. In the patched filter, there is class which emulates a readtoend() string but is actually two buffers which are filled on demand. - During encoding detection mode, if the parser sees any non html-head tag, it assumes that tag cannot occur anywhere after. - The filter code undergoes no major change - so the operational behaviour of the filter should be "exactly the same" as before.