Page 25 of 34

Re: Developer Builds (for testing)

PostPosted: Wed Apr 23, 2014 3:38 pm
by Abs62
goldendict-1.5.0-RC-314-g948db01(EXE only).7z - 1.19 MB (based on last EA build 1.5.0-RC-293-g75ed194)
Version with full-text search (via menu or Ctrl+Shift+F).

Search modes:
"Whole words" - strict words match. For example "rain" matches "rain" but don't match "rained" or "train". Fastest mode.
"Plain text" - partial words match. In this mode "rain" matches "rain", "rained", "train", etc.
"Wildcards" - Unix-like wildcards are allowed.
"RegExp" - standard regular expressions.

To increase search speed I earnestly recommend to wait for end of indexing of the target dictionaries. Dictionaries will be indexed for full-text search in background, at first small ones up to 150000 headwords, then last ones.

For search modes "Whole words" and "Plain text" the presence at least one word with length 4 or more symbols in the search string is necessary, for other modes it is recommended to increase of search speed.

If you don't want to use full-text search you can just turn it off in preferences. It also will block dictionaries indexation for such search.

Re: Developer Builds (for testing)

PostPosted: Fri Apr 25, 2014 4:24 am
by hanyl05
Thanks Abs62, a very great version with significant breakthrough.

It seems that stardict and mdx dictionary has a much faster full text search speed than that for dsl and bgl, especially for the bilingual dictionaries.

Is it possible to further accelerate the FTS speed for dsl and bgl dictionaries?

Re: Developer Builds (for testing)

PostPosted: Fri Apr 25, 2014 4:47 am
by hanyl05
To Abs62
As to the FTS speed, I found an interesting problem for dsl and bgl dictionaries.

Take dsl format for example, the head for a dsl dictionary is as follows,

#INDEX_LANGUAGE "English"
#CONTENTS_LANGUAGE "English"

In the "#CONTENTS_LANGUAGE" section, if it is not defined as "English" (such as "Chinese" or other languages),
the FTS speed is pretty slow.

Re: Developer Builds (for testing)

PostPosted: Fri Apr 25, 2014 7:42 am
by Abs62
Yes, for languages with writing without spaces (Chinese and Japan now) GD don't make FTS index and all searches going on in all dictionary articles. I don't know how to split solid string without spaces to separate words.

Re: Developer Builds (for testing)

PostPosted: Fri Apr 25, 2014 2:20 pm
by bradm
Abs62 wrote:goldendict-1.5.0-RC-314-g948db01(EXE only).7z - 1.19 MB (based on last EA build 1.5.0-RC-293-g75ed194)
Version with full-text search (via menu or Ctrl+Shift+F).


Hi, abs62,
Thank you so much for this wonderful FTS addition to an already flawlessly running program.

As new tools are coming up with new ideas, I wonder if you plan to implemenent a SAVE FTS list option.

All my best

Re: Developer Builds (for testing)

PostPosted: Sat Apr 26, 2014 5:30 am
by hanyl05
Abs62 wrote:Yes, for languages with writing without spaces (Chinese and Japan now) GD don't make FTS index and all searches going on in all dictionary articles. I don't know how to split solid string without spaces to separate words.

Abs62,
Thank you very much for your reply.
Is it possible to simply index the bilingual Chinese or Japanese dictionaries, ignoring the separation of Chinese and Japanese words? :)
Since the index process makes the FTS make quite fast. :)

Re: Developer Builds (for testing)

PostPosted: Sat Apr 26, 2014 7:53 am
by Abs62
hanyl05 wrote:Is it possible to simply index the bilingual Chinese or Japanese dictionaries, ignoring the separation of Chinese and Japanese words? :)

Now I haven't ideas how to make it. May be later.

Re: Developer Builds (for testing)

PostPosted: Mon Apr 28, 2014 12:51 pm
by yahaira
Abs62 wrote:goldendict-1.5.0-RC-314-g948db01(EXE only).7z - 1.19 MB (based on last EA build 1.5.0-RC-293-g75ed194)
Version with full-text search (via menu or Ctrl+Shift+F).

Search modes:
"Whole words" - strict words match. For example "rain" matches "rain" but don't match "rained" or "train". Fastest mode.
"Plain text" - partial words match. In this mode "rain" matches "rain", "rained", "train", etc.
"Wildcards" - Unix-like wildcards are allowed.
"RegExp" - standard regular expressions.

To increase search speed I earnestly recommend to wait for end of indexing of the target dictionaries. Dictionaries will be indexed for full-text search in background, at first small ones up to 150000 headwords, then last ones.

For search modes "Whole words" and "Plain text" the presence at least one word with length 4 or more symbols in the search string is necessary, for other modes it is recommended to increase of search speed.

If you don't want to use full-text search you can just turn it off in preferences. It also will block dictionaries indexation for such search.






very great version with significant breakthrough.

Re: Developer Builds (for testing)

PostPosted: Tue Apr 29, 2014 7:06 pm
by Abs62
goldendict-1.5.0-RC-340-gfb19fb1(EXE only).7z - 1.20 MB - some optimizations for speed of full-text search.

Re: Developer Builds (for testing)

PostPosted: Thu May 01, 2014 7:41 am
by veole
Abs62 wrote:goldendict-1.5.0-RC-340-gfb19fb1(EXE only).7z - 1.20 MB - some optimizations for speed of full-text search.


It works fine for me.

A small bug: The next and previous buttons are always activated, even if there is only one result / or no previous result (after going foward once).

A request: Could it be possible to full text search only one dictionary like in the "dictionary headwords" option?