Index Server -- Technical Documentation

Technical Documentation about Index Search Server

Index search uses regular CERN server script interface. Index search script (search) is placed in the htbin directory of the server.

The Search Script

The search script search determines the directory of the searchable document, and makes a search down the directory tree starting from that directory. The search only applies to files ending in .html and not beginning with a comma (backup files).

The search is implemented as a WAIS inverted index search. There is a directory hierarchy under the Index Home directory (the variable INDEX_HOME) that corresponds that of the real tree of documents. The document tree begins at WWW home (the variable WWW_HOME). This variable must be set to allow the script to suppress the absolute path leading to WWW home in its output.

In the subdirectories of in the Index Home directory there are files WAIS-index.* which are WAIS index files. They are used to search the corresponding document directories. This search is done by waissearch (standard WAIS program).

If the inverted index file (or even that part of the index directory tree) is missing it is created on the fly by waisindex program (also a standard WAIS program).

When a new index is generated an entry to file IndexMap in Index Home directory is made, in order to notify the automatic index updater about the new index. This file contains tuples:

        indexfile	directory
separated by a tab. indexfile is the full pathname of the index file without the suffix, and directory is the starting point of the directory from which the index is made.

AL 24 November 1993