As far as the necessity to configure some Frontend options has emerged, we arrived to the decision to abandon use of template-file and introduced the configuration file optimized for fast load. More information about new format of the configuration file may be found here
It is possible to configure search logic with the help of the option SearchType in the Frontend configuration file. The following logical operators are supported:
Now results can be sorted by relevance (by default) and date. Sorting can by configured by changing key symbols %A and %L in the Frontend configuration file.
It is possible to keep frequently used words (prepositions, articles, pronouns, etc.) from being added to the search index. More information about this topic may be found here
It is possible now to set type of the documents returned by the Fronend (default is text/html). For example:
::CONFIG content-type = text/html; charset=windows-1251or
::CONFIG content-type = text/xml
When the index is buil each task is assigned its number (startin with 0). This number may be handled to the frontend by means of d parameter. So, it is possible to make search at selected sites at once. Example:
<form action=%F method=get> <input type=text name=q size=40 maxlength=64 value="%Q"> <input type=submit value="Search"> <input type=checkbox name=d value=0>First site <input type=checkbox name=d value=1>Second site <input type=checkbox name=d value=2>Third site </form>
If the Indexer is started several times, one and the same number will be assigned to 2 and more tasks.
Now it is possible to display information about found words using key symbol %W.
It is possible now to change position of the navigation links on the page (<< < 1 2 3 4 5 > >>). It can be done by adjusting the key symbol %B.
To distinguish between morphological forms you need to create file 'lang.cns' and save it in the directory, where index files are stored (or will be created). We do not include file 'lang.cns' in this distribution, because of its size - 16 Mb.
If file 'lang.cns' is not found, the search and indexing process will be performed without taking morphology into account.
We have developed a special utility allowing building 'lang.cns' from ispell dictionaries. You may find necessary dictionaries at http://fmg-www.cs.ucla.edu/geoff/ispell-dictionaries.html.
ispell dictionary comprises of two files - a list of words (lang.dict) and a set of word formation rules (lang.aff). These files may have some other names in downloaded archives. You will have to rename them to 'lang.dict' and 'lang.aff'.
ATTENTION!!! If you have built the index taking into consideration morphology, you will have to search also taking into consideration morpholgy and using the same dictionary.
To make search process faster a search index defragmenter has been added - file idefrag(.exe) in directory 'indexer'. Defragmenter allows placement of index files in the way which provides their consecutive access. On some servers, defragmentation gives an increase of performance up to 60%.
In all versions of CNSearch and CNSearch Pro 1.0 the search was perfomed during every jump between pages with results. In CNSearch Pro 1.2 the search is performed only once, the search results being saved to a temporary file. When jumping from one page to another the search is not performed and saved results are used instead.
In UNIX versions temporary files are stored in /tmp directory.
In Windows versions these files are stored in the directory defined in TEMP or TMP environment. If none of these variables is defined, the temporary files are stored in %WINDIR%\temp. If WINDIR variable is not set either, the temporary files are stored to the root directory on disk C: .
A bug causing appearance of hexadecimal numbers (25F, 3FE etc.) in search results has been fixed.
Because of this error there was no possibility to add the same pages to the search index.
Now it is possible to leave blank lines and use Perl-style commentaries in the configuration file:
[Job localhost] [Index] URL http://127.0.0.1/ # this is a test site Statistic Append CharSet ByHTTPHeader MaxFiles 10000 Exclude search/,mail/,.zip,.gif,.jpg
The search engine can index .xls files as well as .htm, .rtf, and .doc files.
Modules 'runner' and 'searchctl' have been removed. Now the Indexer is started as follows:
For Unix/Linux:
./indexer JobName
For Windows:
indexer.exe JobName