cvs2cl1

Usage

Overview

cvs2cl produces a GNU-style ChangeLog for CVS-controlled sources by running "cvs log" and parsing the output. Duplicate log messages get unified in the Right Way.

The default output of cvs2cl is designed to be compact, formally unambiguous, but still easy for humans to read. It should be largely self-explanatory; the one abbreviation that might not be obvious is "utags". That stands for "universal tags" -- a universal tag is one held by all the files in a given change entry.

If you need output that's easy for a program to parse, use the --xml option. Note that with XML output, just about all available information is included with each change entry, whether you asked for it or not, on the theory that your parser can ignore anything it's not looking for.

If filenames are given as arguments cvs2cl only shows log information for the named files.

Options

-h, -help, --help, -?

Show a short help and exit.

--version

Show version and exit.

-r, --revisions

Show revision numbers in output.

-b, --branches

Show branch names in revisions when possible.

-t, --tags

Show tags (symbolic names) in output.

-T, --tagdates

Show tags in output on their first occurance.

--show-dead

Show dead files.

--stdin

Read from stdin, don't run cvs log.

--stdout

Output to stdout not to ChangeLog.

-d, --distributed

Put ChangeLogs in subdirs.

-f FILE, --file FILE

Write to FILE instead of ChangeLog.

--fsf

Use this if log data is in FSF ChangeLog style.

--FSF

Attempt strict FSF-standard compatible output (incompatible with --accum).

-W SECS, --window SECS

Window of time within which log entries unify.

-U UFILE, --usermap UFILE

Expand usernames to email addresses from UFILE.

--passwd PASSWORDFILE

Use system passwd file for user name expansion. If no mail domain is provided (via --domain), it tries to read one from /etc/mailname, output of hostname -d, dnsdomainname, or domain-name. cvs2cl exits with an error if none of those options is successful. Use a domain of '' to prevent the addition of a mail domain.

--domain DOMAIN

Domain to build email addresses from.

--gecos

Get user information from GECOS data.

-R REGEXP, --regexp REGEXP

Include only entries that match REGEXP. This option may be used multiple times.

-I REGEXP, --ignore REGEXP

Ignore files whose names match REGEXP. This option may be used multiple times. The regexp is a perl regular expression. It is matched as is; you may want to prefix with a ^ or suffix with a $ to anchor the match.

-C, --case-insensitive

Any regexp matching is done case-insensitively.

-F BRANCH, --follow BRANCH

Show only revisions on or ancestral to BRANCH.

--follow-only BRANCH

Like --follow, but sub-branches are not followed.

--no-ancestors

When using -F, only track changes since the BRANCH started.

--no-hide-branch-additions

By default, entries generated by cvs for a file added on a branch (a dead 1.1 entry) are not shown. This flag reverses that action.

-S, --separate-header

Blank line between each header and log message.

--group-within-date

Group ChangeLog entries on the same date together, instead of having a separate entry for each commit on that date.

--summary

Add CVS change summary information.

--no-wrap

Don't auto-wrap log message (recommend -S also).

--no-indent

Don't indent log message

--gmt, --utc

Show times in GMT/UTC instead of local time.

--accum

Add to an existing ChangeLog (incompatible with --xml and --FSF).

-w, --day-of-week

Show day of week.

--no-times

Don't show times in output.

--chrono

Output log in chronological order (default is reverse chronological order).

--header FILE

Get ChangeLog header from FILE ("-" means stdin).

--xml

Output XML instead of ChangeLog format (incompatible with --accum).

--xml-encoding ENCODING

Insert encoding clause in XML header.

--xml-stylesheet FILE

Insert xml-stylesheet processing instruction with FILE formatting stylesheet file path in XML header.

--noxmlns

Don't include xmlns= attribute in root element.

--hide-filenames

Don't show filenames (ignored for XML output).

--no-common-dir

Don't shorten directory names from filenames.

--rcs CVSROOT

Handle filenames from raw RCS, for instance those produced by "cvs rlog" output, stripping the prefix CVSROOT.

-P, --prune

Don't show empty log messages.

--lines-modified

Output the number of lines added and the number of lines removed for each checkin (if applicable). At the moment, this only affects the XML output mode.

--ignore-tag TAG

Ignore individual changes that are associated with a given tag. May be repeated, if so, changes that are associated with any of the given tags are ignored.

--show-tag TAG

Log only individual changes that are associated with a given tag. May be repeated, if so, changes that are associated with any of the given tags are logged.

--delta FROM_TAG:TO_TAG

Attempt a delta between two tags (since FROM_TAG up to and including TO_TAG). The algorithm is a simple date-based one (this is a hard problem) so results are imperfect.

-g OPTS, --global-opts OPTS

Pass OPTS to cvs like in "cvs OPTS log ...".

-l OPTS, --log-opts OPTS

Pass OPTS to cvs log like in "cvs ... log OPTS".

Notes about the options and arguments:

•

The -I and -F options may appear multiple times.

•

To follow trunk revisions, use "-F trunk" ("-F TRUNK" also works). This is okay because no would ever, ever be crazy enough to name a branch "trunk", right? Right.

•

For the -U option, the UFILE should be formatted like CVSROOT/users. That is, each line of UFILE looks like this:

       jrandom:[email protected]

or maybe even like this

       jrandom:'Jesse Q. Random <[email protected]>'

Don't forget to quote the portion after the colon if necessary.

•

Many people want to filter by date. To do so, invoke cvs2cl like this:

       cvs2cl -l "-d'DATESPEC'"

where DATESPEC is any date specification valid for "cvs log -d". (Note that CVS 1.10.7 and below requires there be no space between -d and its argument).

•

Dates/times are interpreted in the local time zone.

•

Remember to quote the argument to `-l' so that your shell doesn't interpret spaces as argument separators.

•

See the 'Common Options' section of the cvs manual ('info cvs' on UNIX-like systems) for more information.

•

Note that the rules for quoting under windows shells are different.

•

To run in an automated environment such as CGI or PHP, suidperl may be needed in order to execute as the correct user to enable /cvsroot read lock files to be written for the 'cvs log' command. This is likely just a case of changing the /usr/bin/perl command to /usr/bin/suidperl, and explicitly declaring the PATH variable.

Examples

Some examples (working on UNIX shells):

      # logs after 6th March, 2003 (inclusive)
      cvs2cl -l "-d'>2003-03-06'"
      # logs after 4:34PM 6th March, 2003 (inclusive)
      cvs2cl -l "-d'>2003-03-06 16:34'"
      # logs between 4:46PM 6th March, 2003 (exclusive) and
      # 4:34PM 6th March, 2003 (inclusive)
      cvs2cl -l "-d'2003-03-06 16:46>2003-03-06 16:34'"

Some examples (on non-UNIX shells):

      # Reported to work on windows xp/2000
      cvs2cl -l  "-d"">2003-10-18;today<"""

Prerequisites

This script requires "Text::Wrap", "Time::Local", and "File::Basename". It also seems to require "Perl 5.004_04" or higher.

Operating system compatibility

Should work on any OS.

Script categories

Version_Control/CVS