19.10.1 Understanding
how OmniHelp FTS works
OmniHelp supports single-term and Boolean
(AND, OR, NOT) full-text search. A search on a phrase is implemented
by successively ANDing the search terms: topics found include all terms
in the phrase, except for stop words (see §19.10.6
Excluding search terms), whether or not those terms occur
together.
There are some limitations:
- Search
does not find terms that start with non-alphanumeric characters. For
example, to find $$_currbase,
you would have to search for currbase;
and to find -progid, you
would have to search for progid.
- Search
does not find partial terms; for example, a search for curr finds
<$Curr>,
but not $$_currbase.
- Search
reports every instance of a hit, even if several instances are in the
same topic. To remove extra instances of a term from the search index,
you can delete duplicate entries from the JavaScript array in myproj_ohs.js, either
by hand or with a UNIX-style utility such as uniq, from
Cygwin.
Because OmniHelp is Open Source, anyone
can modify or replace the search function to overcome these limitations.
You can contribute to the OmniHelp project any tool you make for this
purpose, at Sourceforge:
https://sourceforge.net/projects/omnihelp/
See §19.1 Understanding
how OmniHelp works.
Previous Topic:
19.10 Configuring
full-text search for OmniHelp
Next Topic:
19.10.2 Generating
search data
Parent Topic:
19.10 Configuring
full-text search for OmniHelp
Sibling Topics:
19.10.2 Generating
search data
19.10.3 Making
compound terms searchable
19.10.4 Supporting
search for non-ANSI text
19.10.5 Specifying
length of search terms
19.10.6 Excluding
search terms
19.10.7 Excluding
content from being searched
19.10.8 Using
regular expressions in search
19.10.9 Highlighting
search terms found in topics
Table of Contents • Index