NAME

af - Amberfish text retrieval software

SYNOPSIS

af [-i] [options] [file ...]
af [-s] [options]
af [-L] [options]
af [-l] [options]
af [--fetch] [file] [begin] [end]
af [--version]

DESCRIPTION

The program af is a text-based interface to Amberfish functions for indexing and searching documents. A simple indexing example would look something like:

af -iCv -d mydb *.txt

This creates a new database, mydb, containing an index to the set of files, *.txt. To enable faster searching, an optional "linearize" step can be done (this can take a long time to run):

af -L -d mydb

Here is a typical search command:

af -s -d mydb -Q 'cat & (dog | mouse)'

COMMAND OPTIONS

Only one of these options can be used at a time.

-i, --index: Index documents (either file ... or specified via standard input if -F is used).
-s, --search: Search an indexed database.
-L, --linearize: Linearize an indexed database. Linearizing is an optional step that can be done after indexing. The advantages of linearizing are that it reduces searching time and slightly reduces the size of the index files. The disadvantages are that the linearizing process is very slow, especially when used on large databases, and it prevents any additional documents from being added to the database.
-l, --list: List the documents contained in a database.
--fetch: Output a portion of a file. This command takes no other options. The file name file, starting offset begin, and ending offset end are specified at the end of the line.
--version: Print the af version number.

GENERAL OPTIONS

These options are generally available with all command options.

-d, --db dbname: Use dbname as the database name. With some command options such as -s, this option can be supplied multiple times to specify multiple databases.
-v, --verbose: Show verbose output. This option can be supplied multiple times to increase verbosity.
-D, --debug: Show extremely verbose (debugging) output. Using this option once is equivalent to -vvvvv, and it can be supplied multiple times to increase verbosity further.

INDEX OPTIONS

The following options can only be used together with the indexing (-i) command.

-C, --create

Create a new database, overwriting any existing one with the same name.

-m, --memory maximum

Set the maximum amount of memory in megabytes to use for indexing. More memory speeds up indexing.

--phrase

Enable phrase searching. This can only be used together with -C.

--split delimiter

Parse input files into multiple documents at points where the specified delimiter string is found.

-t, --doctype=text, --doctype=xml

Set the document type. The default is text. Specifying xml enables functions related to searching and retrieving within nested tags in XML documents.

--dlevel level

The maximum resolution (levels of descent) for retrieval of nested documents. The default value is 1; increasing it lengthens indexing time significantly. Use this for XML instead of --split to subdivide documents. Note that this only affects resolution of elements returned from searches and is unrelated to nested queries which have much higher (fixed) resolution.

--no-stem

Do not perform stemming. This can only be used together with -C. Normally, stemming is automatically enabled if Amberfish was compiled with the stemming function. This option disables stemming even if it is available. Note that the stemming function is not distributed with this package and must be installed manually.

-F

Read list of documents to be indexed from standard input, rather than from the end of the command line.

SEARCH OPTIONS

The following options can only be used together with the searching (-s) command.

-Q, --query-boolean query_string: Search for the specified Boolean query string.
-n, --numhits x: Output a maximum of x results.
--skiphits x: Do not output the first x results.
--totalhits: Output the total number of results.
--style=list, --style=lineage, --style=trec: Set style of printed result sets. The default is list. Use the lineage style with XML to see hierarchical results. For the trec style, it is assumed that the indexed file names are the document numbers and that --skiphits is not used (because rank always starts at 1).
--trec-tag run_tag: Output TREC results with the specified run tag. (This is to be used with --style=trec.)
--trec-topic topic_number: Output TREC results with the specified topic number. (This is to be used with --style=trec.)

LINEARIZE OPTIONS

The following options can only be used together with the linearize (-L) command.

-m, --memory maximum: Set the maximum amount of memory in megabytes to use for linearizing. More memory speeds up linearizing.
--no-linear-buffer: Do not use a memory buffer to speed up linearizing. This option will be removed once the linearization buffer code proves to be reliable.

AUTHOR

Nassib Nassar; see http://www.etymon.com/ for updates.