The indexer indexes the content of the databses published in a Phraseanet application and allows searching into the captions of documents.
The indexer runs in background, it monitors databases to index (upcoming or shutdown of databases) and maintains “full-text”, “value” and “thesaurus” indexes.
The indexer is a command-line program, “phraseanet-indexer”.
It can be ran in command-line, scheduled (cron, scheduled task...) or via the Phraseanet “Task-Manager”.
The indexer needs those libraries :
After building from sources (see Windows specifics), copy the program in a directory allowing excution.
–host : adress of the database
–port : database port (usualy 3306 for MySQL)
–base : name of the “application-box” database
–user : SQL account for connection
–password : password of connection account
–default-character-set : charset of the connection
Those options selects the “application-box” (SQL database of the Phraseanet program) where databases (“data-boxes”) to index are published.
When running, the indexer can be stopped with Ctrl-C (or sending sigint signal under linux or OSX). The “socket” option allows to stop the indexer via telnet.
To optimize speed, the indexer works on batches of records (default 50). If the amount of ram consumed is too high, this number can be decreased with the “flush” option.
When indexing fields linked the a thesaurus, new terms are set as “candidates”. The option “clng” allows to define the primary language to set for those terms.
–debug : Type of generated logs.
It can be useful to not display some options from the command-line (for example the password). The indexer can read options from a file (which must be located into the same directory as the program file).
The indexer is supposed to run continuously. This option allows to index changed records and then quit.
–help : print help
The “help” option will print information and default values for every option.
To be checked, the indexer can log various kind of operations (xml ops, sql...). 7 “classes” of messages can be filtered with 7 bits (mask) of the “debug” value.
Those logs are usualy sent to the system (syslog for linux or OSX, event-log for Windows). The “nolog” option allows to send messages on screen.
Note
A debug mask set to –debug=64 (flush ops.) allows control of indexer without overloading the logs.
The indexer can be checked with “-?” option which must print the help :
phraseanet_indexer -?
phraseanet_indexer version 3.9.0.4
Usage : phraseanet_indexer <options>
[-? | --help] : this help
[-v | --version : display version and quit
[-h | --host]=<addr> : host addr. of applicationBox (default '127.0.0.1')
[-P | --port]=<port> : port of applicationBox (default '3306')
[-b | --base]=<base> : database of applicationBox (default 'phrasea')
[-u | --user]=<user> : user account for connection to applicationBox
(default 'root')
[-p | --password]=<pwd> : password for connection to applicationBox
(default '')
[-s | --socket]=<port> : port for telnet control (default none)
[-f | --flush]=<n> : flush every n records (default 50)
[-o | --old] : use old 'sbas' table instead of 'xbas'
[ --quit] : index once and quit
[-c | --clng]=<lng> : default language for new candidates terms
(default 'fr')
[-n | --nolog] : do not log, but out to console
[-d | --debug]=<mask> : debug mask (to console)
1 : xml parsing
2 : sql errors
4 : sql ok
8 : memory alloc.
16 : record ops.
32 : structure ops.
64 : flush ops.
[-@ | --optfile]=<file> : read (more) arguments from text file
(see 'sample_args.txt')
[--default-character-set]=<charset> : charset of applicationBox AND dataBoxes
(default none)
Example of reading options from a file “indexerargs.txt”
phraseanet_indexer --debug=64 --nolog --optfile=indexerargs.txt
# connection to application-box...
# ...host, port, base, user, password
--host=127.0.0.1
--port=3306
--base=phrasea
--user=phraseanet
--password=xxxxxxxxx
# socket to talk (telnet) to indexer
--socket=2055
# use 'sbas' table (mandatory)
-o
# candidates default language
--clng=fr
# sql connections encoding
--default-character-set=utf8
The indexder can be executed by the “Indexer” task of the Phraseanet Task-Manager. Options can then be set via the gui of the task.
After build, “sudo make install” will copy the program file into the “bin” directory (for ex. /usr/local/bin).
The indexer is pre-builded for Windows.
Download the last version “https://github.com/alchemy-fr/Phraseanet-Indexer/blob/master/WIN32/Release_win32/”, and copy the program file into a separate directory (for ex. into “Program Files\Phraseanet-Indexer\phraseanet_indexer.exe”).
Note
Some files (fixing bugs for older versions of Phraseanet) are available for download.
For compatibility reasons and matching versions, it’s recommended to copy the dll files of requested libraries into the directory of the program file (beside phraseanet_indexer.exe)
Those libraries can be downloaded from :
Usualy the indexer is executed by the Phraseanet task “Indexer”. Under Windows, the indexer can also be installed as a service via 3 specific options.
--install : install the service "Phraseanet-Indexer"
--remove : uninstall the service
--run : run the indexer as a program (mandatory to run via the task)
ex :
C:\Phraseanet-Indexer\phraseanet_indexer.exe -h=localhost -P=3306 -b=ab_test
-u=phuser -p=**** --socket=25200 --default-character-set=utf8 -o -n
-d=0 --install
If the indexer is executed as a Windows service, the “indexer task” can be deleted (or not activated).