Parse::DebControl

Section: User Contributed Perl Documentation (3pm)
Updated: 2022-10-14
Index Return to Main Contents
 

NAME

Parse::DebControl - Easy OO parsing of debian control-like files  

SYNOPSIS

        use Parse::DebControl

        $parser = new Parse::DebControl;

        $data = $parser->parse_mem($control_data, $options);
        $data = $parser->parse_file('./debian/control', $options);
        $data = $parser->parse_web($url, $options);

        $writer = new Parse::DebControl;

        $string = $writer->write_mem($singlestanza);
        $string = $writer->write_mem([$stanza1, $stanza2]);

        $writer->write_file($filename, $singlestanza, $options);
        $writer->write_file($filename, [$stanza1, $stanza2], $options);

        $writer->write_file($handle, $singlestanza, $options);
        $writer->write_file($handle, [$stanza1, $stanza2], $options);

        $parser->DEBUG();

 

DESCRIPTION

        Parse::DebControl is an easy OO way to parse debian control files and 
        other colon separated key-value pairs. It's specifically designed
        to handle the format used in Debian control files, template files, and
        the cache files used by dpkg.

        For basic format information see:
        http://www.debian.org/doc/debian-policy/ch-controlfields.html#s-controlsyntax

        This module does not actually do any intelligence with the file content
        (because there are a lot of files in this format), but merely handles
        the format. It can handle simple control files, or files hundreds of lines 
        long efficiently and easily.

 

Class Methods

"new()"
"new($debug)"

Returns a new Parse::DebControl object. If a true parameter $debug is passed in, it turns on debugging, similar to a call to "DEBUG()" (see below);

"parse_file($control_filename,$options)"

Takes a filename as a scalar and an optional hashref of options (see below). Will parse as much as it can, warning (if "DEBUG"ing is turned on) on parsing errors.

Returns an array of hashrefs, containing the data in the control file, split up by stanza. Stanzas are deliniated by newlines, and multi-line fields are expressed as such post-parsing. Single periods are treated as special extra newline deliniators, per convention. Whitespace is also stripped off of lines as to make it less-easy to make mistakes with hand-written conf files).

The options hashref can take parameters as follows. Setting the string to true enables the option.

        useTieIxHash - Instead of an array of regular hashrefs, uses Tie::IxHash-
                based hashrefs

        discardCase  - Remove all case items from keys (not values)             

        stripComments - Remove all commented lines in standard #comment format.
                Literal #'s are represented by ##. For instance

                Hello there #this is a comment
                Hello there, I like ##CCCCCC as a grey.

                The first is a comment, the second is a literal "#".

        verbMultiLine - Keep the description AS IS, and no not collapse leading
                spaces or dots as newlines. This also keeps whitespace from being
                stripped off the end of lines.

        tryGzip - Attempt to expand the data chunk with gzip first. If the text is
                already expanded (ie: plain text), parsing will continue normally. 
                This could optionally be turned on for all items in the future, but
                it is off by default so we don't have to scrub over all the text for
                performance reasons.

    singleBlock - Only parse the first block of data and return it. This is
        useful when you have possible "junk" data after the metadata.

    strict - Tries to parse obeying the strict rules for real debian control files.
        This will force comment stripping for debian/control (must start line) and for
        other files will check if a field may span multiple lines.

    allowUnknownFields - In strict mode, allow unknown fields.

    type - If the strict option is chosen, then this parameter defines what format
        we have. Available formats is:
            - debian/control
            - DEBIAN/control
            - .dsc
            - .changes

"parse_mem($control_data, $options)"

Similar to "parse_file", except takes data as a scalar. Returns the same array of hashrefs as "parse_file". The options hashref is the same as "parse_file" as well; see above.

"parse_web($url, $options)"

Similar to the other parse_* functions, this pulls down a control file from the web and attempts to parse it. For options and return values, see "parse_file", above

"write_file($filename, $data, $options)"
"write_file($handle, $data)"
"write_file($filename, [$data1, $data2, $data3], $options)"
"write_file($handle, [$data, $data2, $data3])"

This function takes a filename or a handle and writes the data out. The data can be given as a single hashref or as an arrayref of hashrefs. It will then write it out in a format that it can parse. The order is dependent on your hash sorting order. If you care, use Tie::IxHash. Remember for reading back in, the module doesn't care.

The $options hashref can contain one of the following two items:

        addNewline - At the end of the last stanza, add an additional newline.
        appendFile  - (default) Write to the end of the file
        clobberFile - Overwrite the file given.
        gzip - Compress the data with gzip before writing

Since you determine the mode of your filehandle, passing it along with an options hashref obviously won't do anything; rather, it is ignored.

The addNewline option solves a situation where if you are writing stanzas to a file in a loop (such as logging with this module), then the data will be streamed together, and won't parse back in correctly. It is possible that this is the behavior that you want (if you wanted to write one key at a time), so it is optional.

This function returns the number of bytes written to the file, undef otherwise.

"write_mem($data)"
"write_mem([$data1,$data2,$data3])";

This function works similarly to the "write_file" method, except it returns the control structure as a scalar, instead of writing it to a file. There is no %options for this file (yet);

"DEBUG()"

Turns on debugging. Calling it with no parameter or a true parameter turns on verbose "warn()"ings. Calling it with a false parameter turns it off. It is useful for nailing down any format or internal problems.

 

CHANGES

Version 2.005 - January 13th, 2004
More generic test suite fix for earlier versions of Test::More
Updated copyright statement

Version 2.004 - January 12th, 2004

More documentation formatting and typo fixes
CHANGES file now generated automatically
Fixes for potential test suite failure in Pod::Coverage run
Adds the ``addNewline'' option to write_file to solve the streaming stanza problem.
Adds tests for the addNewline option

Version 2.003 - January 6th, 2004

Added optional Test::Pod test
Skips potential Win32 test failure in the module where it wants to write to /tmp.
Added optional Pod::Coverage test

Version 2.002 - October 7th, 2003

No code changes. Fixes to test suite

Version 2.001 - September 11th, 2003

Cleaned up more POD errors
Added tests for file writing
Fixed bug where write_file ignored the gzip parameter

Version 2.0 - September 5th, 2003

Version increase.
Added gzip support (with the tryGzip option), so that compresses control files can be parsed on the fly
Added gzip support for writing of control files
Added parse_web to snag files right off the web. Useful for things such as apt's Sources.gz and Packages.gz

Version 1.10b - September 2nd, 2003

Documentation fix for ## vs # in stripComments

Version 1.10 - September 2nd, 2003

Documentation fixes, as pointed out by pudge
Adds a feature to stripComments where ## will get interpolated as a literal pound sign, as suggested by pudge.

Version 1.9 - July 24th, 2003

Fix for warning for edge case (uninitialized value in chomp)
Tests for CRLF

Version 1.8 - July 11th, 2003

By default, we now strip off whitespace unless verbMultiLine is in place. This makes sense for things like conf files where trailing whitespace has no meaning. Thanks to pudge for reporting this.

Version 1.7 - June 25th, 2003

POD documentation error noticed again by Frank Lichtenheld
Also by Frank, applied a patch to add a ``verbMultiLine'' option so that we can hand multiline fields back unparsed.
Slightly expanded test suite to cover new features

Version 1.6.1 - June 9th, 2003

POD cleanups noticed by Frank Lichtenheld. Thank you, Frank.

Version 1.6 - June 2nd, 2003

Cleaned up some warnings when you pass in empty hashrefs or arrayrefs
Added stripComments setting
Cleaned up POD errors

Version 1.5 - May 8th, 2003

Added a line to quash errors with undef hashkeys and writing
Fixed the Makefile.PL to straighten up DebControl.pm being in the wrong dir

Version 1.4 - April 30th, 2003

Removed exports as they were unnecessary. Many thanks to pudge, who pointed this out.

Version 1.3 - April 28th, 2003

Fixed a bug where writing blank stanzas would throw a warning. Fix found and supplied by Nate Oostendorp.

Version 1.2b - April 25th, 2003

Fixed:

A bug in the test suite where IxHash was not disabled in 40write.t. Thanks to Jeroen Latour from cpan-testers for the report.

Version 1.2 - April 24th, 2003

Fixed:

A bug in IxHash support where multiple stanzas might be out of order

Version 1.1 - April 23rd, 2003

Added:

Writing support
Tie::IxHash support
Case insensitive reading support

Version 1.0 - April 23rd, 2003

This is the initial public release for CPAN, so everything is new.
 

BUGS

The module will let you parse otherwise illegal key-value pairs and pairs with spaces. Badly formed stanzas will do things like overwrite duplicate keys, etc. This is your problem.

As of 1.10, the module uses advanced regexp's to figure out about comments. If the tests fail, then stripComments won't work on your earlier perl version (should be fine on 5.6.0+)  

TODO

Change the name over to the Debian:: namespace, probably as Debian::ControlFormat. This will happen as soon as the project that uses this module reaches stability, and we can do some minor tweaks.  

COPYRIGHT

Parse::DebControl is copyright 2003,2004 Jay Bonci <jaybonci@cpan.org>. This program is free software; you can redistribute it and/or modify it under the same terms as Perl itself.


 

Index

NAME
SYNOPSIS
DESCRIPTION
Class Methods
CHANGES
BUGS
TODO
COPYRIGHT

This document was created by man2html, using the manual pages.
Time: 05:43:02 GMT, March 29, 2024