In order to be able to search fast and efficient, doxysearch does not search the generated documentation directly. Instead, it uses an index file, that should be generated with doxytag. The index file is extracted from the generated HTML files and contains all words and substrings of words present in the HTML files, in a compact form, together with their frequencies and links. Although I tried to store all information as compactly as possible, the size of the index is still quite large. Usually it is about the same size as the original HTML files.
I have tried to make the search engine highly portable, because it must run on the target system. As a result doxysearch does not require the Qt library. All that is required to build doxysearch is a C++ compiler. If you are using g++
for example, you can build the search engine manually, by typing:
g++ doxysearch.cpp -o doxysearch
To include a search engine in the documentation generated by doxygen follow these steps:
-g
option, if you haven't done this already.
SEARCHENGINE
tag is set to YES
and that all paths are correct.
CGI_NAME
tag in the configuration file. The script is a small wrapper that calls doxysearch
with the correct parameters. Using this script allows multiple search engines for different projects to be present in one directory. search.cfg
: this file is a small configuration file for the search engine. It contains two lines of text. The first line should be the absolute URL to the documentation. The second line should be the absolute URL to the CGI script. This information is taken from the configuration file. search.gif
: this is the image that is used for the search button. .cgi
files that were renamed executables (so DOS batch files do not seem to work either). Therefore, on Windows a small C program will be generated by doxygen. You should compile and link the program with your favourite compiler and change the extension of the executable from .exe
to .cgi
.
doxytag -s search.idxThis will create a search index with the name
search.idx
. Currently the index file must be called like this.
Using doxytag
and doxysearch
it is possible to create a search engine for the Qt documentation, without needing the sources! This can be done by carefully following these steps:
cd $QTDIR/html
doxytag -s search.idxin the directory where the HTML files are located. This will parse all files and build a search index. Apart from the file
search.idx
two other files will be generated: search.gif
and search.cgi
search.cgi
with a text editor.
Fill in the absolute path to the doxysearch
binary after the DOXYSEARCH=
tag. On my system this becomes:
DOXYSEARCH=/usr/local/bin/doxysearch
Fill in the absolute path to the qt documentation after the DOXYPATH=
tag. On my system this becomes:
DOXYPATH=/usr/local/qt/html
search.cgi
script to this directory. If needed you may change the name of the script. On my system, this becomes: cp search.cgi /usr/local/lib/httpd/cgi-bin/
search.cfg
. On the first line, you must put the absolute URL to the Qt documentation. Since, I only use the search engine on my own standalone system, I use the file:
protocol. On the second line, you must put the absolute URL to the cgi script. On my system the resulting file looks like this: file:///usr/local/qt/html http://blizzard/cgi-bin/search.cgi
<li><a href="http://blizzard/cgi-bin/search.cgi">Search the documentation<a>
index.html
file.