1#!/usr/bin/perl
2
3# Copyright (C) 2007, 2008 Simon Josefsson <simon@josefsson.org>
4# Copyright (C) 2007 Luis Mondesi <lemsx1@gmail.com>
5# * calls git directly. To use it just:
6#   cd ~/Project/my_git_repo; git2cl > ChangeLog
7# * implements strptime()
8# * fixes bugs in $comment parsing
9#   - copy input before we remove leading spaces
10#   - skip "merge branch" statements as they don't
11#     have information about files (i.e. we never
12#     go into $state 2)
13#   - behaves like a pipe/filter if input is given from the CLI
14#     else it calls git log by itself
15#
16# The functions mywrap, last_line_len, wrap_log_entry are derived from
17# the cvs2cl tool, see <http://www.red-bean.com/cvs2cl/>:
18# Copyright (C) 2001,2002,2003,2004 Martyn J. Pearce <fluffy@cpan.org>
19# Copyright (C) 1999 Karl Fogel <kfogel@red-bean.com>
20#
21# git2cl is free software; you can redistribute it and/or modify it
22# under the terms of the GNU General Public License as published by
23# the Free Software Foundation; either version 2, or (at your option)
24# any later version.
25#
26# git2cl is distributed in the hope that it will be useful, but
27# WITHOUT ANY WARRANTY; without even the implied warranty of
28# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
29# General Public License for more details.
30#
31# You should have received a copy of the GNU General Public License
32# along with git2cl; see the file COPYING.  If not, write to the Free
33# Software Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA
34# 02111-1307, USA.
35
36use strict;
37use POSIX qw(strftime);
38use Text::Wrap qw(wrap);
39use FileHandle;
40
41use constant EMPTY_LOG_MESSAGE => '*** empty log message ***';
42
43# this is a helper hash for stptime.
44# Assumes you are calling 'git log ...' with LC_ALL=C
45my %month = (
46    'Jan'=>0,
47    'Feb'=>1,
48    'Mar'=>2,
49    'Apr'=>3,
50    'May'=>4,
51    'Jun'=>5,
52    'Jul'=>6,
53    'Aug'=>7,
54    'Sep'=>8,
55    'Oct'=>9,
56    'Nov'=>10,
57    'Dec'=>11,
58);
59
60my $fh = new FileHandle;
61
62sub key_ready
63{
64    my ($rin, $nfd);
65    vec($rin, fileno(STDIN), 1) = 1;
66    return $nfd = select($rin, undef, undef, 0);
67}
68
69sub strptime {
70    my $str = shift;
71    return undef if not defined $str;
72
73    # we are parsing this format
74    # Fri Oct 26 00:42:56 2007 -0400
75    # to these fields
76    # sec, min, hour, mday, mon, year, wday = -1, yday = -1, isdst = -1
77    # Luis Mondesi <lemsx1@gmail.com>
78    my @date;
79    if ($str =~ /([[:alpha:]]{3})\s+([[:alpha:]]{3})\s+([[:digit:]]{1,2})\s+([[:digit:]]{1,2}):([[:digit:]]{1,2}):([[:digit:]]{1,2})\s+([[:digit:]]{4})/){
80        push(@date,$6,$5,$4,$3,$month{$2},($7 - 1900),-1,-1,-1);
81    } else {
82        die ("Cannot parse date '$str'\n'");
83    }
84    return @date;
85}
86
87sub mywrap {
88    my ($indent1, $indent2, @text) = @_;
89    # If incoming text looks preformatted, don't get clever
90    my $text = Text::Wrap::wrap($indent1, $indent2, @text);
91    if ( grep /^\s+/m, @text ) {
92	return $text;
93    }
94    my @lines = split /\n/, $text;
95    $indent2 =~ s!^((?: {8})+)!"\t" x (length($1)/8)!e;
96    $lines[0] =~ s/^$indent1\s+/$indent1/;
97    s/^$indent2\s+/$indent2/
98	for @lines[1..$#lines];
99    my $newtext = join "\n", @lines;
100    $newtext .= "\n"
101	if substr($text, -1) eq "\n";
102    return $newtext;
103}
104
105sub last_line_len {
106    my $files_list = shift;
107    my @lines = split (/\n/, $files_list);
108    my $last_line = pop (@lines);
109    return length ($last_line);
110}
111
112# A custom wrap function, sensitive to some common constructs used in
113# log entries.
114sub wrap_log_entry {
115    my $text = shift;                  # The text to wrap.
116    my $left_pad_str = shift;          # String to pad with on the left.
117
118    # These do NOT take left_pad_str into account:
119    my $length_remaining = shift;      # Amount left on current line.
120    my $max_line_length  = shift;      # Amount left for a blank line.
121
122    my $wrapped_text = '';             # The accumulating wrapped entry.
123    my $user_indent = '';              # Inherited user_indent from prev line.
124
125    my $first_time = 1;                # First iteration of the loop?
126    my $suppress_line_start_match = 0; # Set to disable line start checks.
127
128    my @lines = split (/\n/, $text);
129    while (@lines)   # Don't use `foreach' here, it won't work.
130    {
131	my $this_line = shift (@lines);
132	chomp $this_line;
133
134	if ($this_line =~ /^(\s+)/) {
135	    $user_indent = $1;
136	}
137	else {
138	    $user_indent = '';
139	}
140
141	# If it matches any of the line-start regexps, print a newline now...
142	if ($suppress_line_start_match)
143	{
144	    $suppress_line_start_match = 0;
145	}
146	elsif (($this_line =~ /^(\s*)\*\s+[a-zA-Z0-9]/)
147	       || ($this_line =~ /^(\s*)\* [a-zA-Z0-9_\.\/\+-]+/)
148	       || ($this_line =~ /^(\s*)\([a-zA-Z0-9_\.\/\+-]+(\)|,\s*)/)
149	       || ($this_line =~ /^(\s+)(\S+)/)
150	       || ($this_line =~ /^(\s*)- +/)
151	       || ($this_line =~ /^()\s*$/)
152	       || ($this_line =~ /^(\s*)\*\) +/)
153	       || ($this_line =~ /^(\s*)[a-zA-Z0-9](\)|\.|\:) +/))
154	{
155	    $length_remaining = $max_line_length - (length ($user_indent));
156	}
157
158	# Now that any user_indent has been preserved, strip off leading
159	# whitespace, so up-folding has no ugly side-effects.
160	$this_line =~ s/^\s*//;
161
162	# Accumulate the line, and adjust parameters for next line.
163	my $this_len = length ($this_line);
164	if ($this_len == 0)
165	{
166	    # Blank lines should cancel any user_indent level.
167	    $user_indent = '';
168	    $length_remaining = $max_line_length;
169	}
170	elsif ($this_len >= $length_remaining) # Line too long, try breaking it.
171	{
172	    # Walk backwards from the end.  At first acceptable spot, break
173	    # a new line.
174	    my $idx = $length_remaining - 1;
175	    if ($idx < 0) { $idx = 0 };
176	    while ($idx > 0)
177	    {
178		if (substr ($this_line, $idx, 1) =~ /\s/)
179		{
180		    my $line_now = substr ($this_line, 0, $idx);
181		    my $next_line = substr ($this_line, $idx);
182		    $this_line = $line_now;
183
184		    # Clean whitespace off the end.
185		    chomp $this_line;
186
187		    # The current line is ready to be printed.
188		    $this_line .= "\n${left_pad_str}";
189
190		    # Make sure the next line is allowed full room.
191		    $length_remaining = $max_line_length - (length ($user_indent));
192
193		    # Strip next_line, but then preserve any user_indent.
194		    $next_line =~ s/^\s*//;
195
196		    # Sneak a peek at the user_indent of the upcoming line, so
197		    # $next_line (which will now precede it) can inherit that
198		    # indent level.  Otherwise, use whatever user_indent level
199		    # we currently have, which might be none.
200		    my $next_next_line = shift (@lines);
201		    if ((defined ($next_next_line)) && ($next_next_line =~ /^(\s+)/)) {
202			$next_line = $1 . $next_line if (defined ($1));
203			# $length_remaining = $max_line_length - (length ($1));
204			$next_next_line =~ s/^\s*//;
205		    }
206		    else {
207			$next_line = $user_indent . $next_line;
208		    }
209		    if (defined ($next_next_line)) {
210			unshift (@lines, $next_next_line);
211		    }
212		    unshift (@lines, $next_line);
213
214		    # Our new next line might, coincidentally, begin with one of
215		    # the line-start regexps, so we temporarily turn off
216		    # sensitivity to that until we're past the line.
217		    $suppress_line_start_match = 1;
218
219		    last;
220		}
221		else
222		{
223		    $idx--;
224		}
225	    }
226
227	    if ($idx == 0)
228	    {
229		# We bottomed out because the line is longer than the
230		# available space.  But that could be because the space is
231		# small, or because the line is longer than even the maximum
232		# possible space.  Handle both cases below.
233
234		if ($length_remaining == ($max_line_length - (length ($user_indent))))
235		{
236		    # The line is simply too long -- there is no hope of ever
237		    # breaking it nicely, so just insert it verbatim, with
238		    # appropriate padding.
239		    $this_line = "\n${left_pad_str}${this_line}";
240		}
241		else
242		{
243		    # Can't break it here, but may be able to on the next round...
244		    unshift (@lines, $this_line);
245		    $length_remaining = $max_line_length - (length ($user_indent));
246		    $this_line = "\n${left_pad_str}";
247		}
248	    }
249	}
250	else  # $this_len < $length_remaining, so tack on what we can.
251	{
252	    # Leave a note for the next iteration.
253	    $length_remaining = $length_remaining - $this_len;
254
255	    if ($this_line =~ /\.$/)
256	    {
257		$this_line .= "  ";
258		$length_remaining -= 2;
259	    }
260	    else  # not a sentence end
261	    {
262		$this_line .= " ";
263		$length_remaining -= 1;
264	    }
265	}
266
267	# Unconditionally indicate that loop has run at least once.
268	$first_time = 0;
269
270	$wrapped_text .= "${user_indent}${this_line}";
271    }
272
273    # One last bit of padding.
274    $wrapped_text .= "\n";
275
276    return $wrapped_text;
277}
278
279# main
280
281my @date;
282my $author;
283my @files;
284my $comment;
285
286my $state; # 0-header 1-comment 2-files
287my $done = 0;
288
289$state = 0;
290
291# if reading from STDIN, we assume that we are
292# getting git log as input
293if (key_ready())
294{
295
296    #my $dummyfh; # don't care about writing
297    #($fh,$dummyfh) = FileHandle::pipe;
298    $fh->fdopen(*STDIN, 'r');
299}
300else
301{
302    $fh->open("LC_ALL=C git log --pretty --numstat --summary|")
303	or die("Cannot execute git log...$!\n");
304}
305
306while (my $_l = <$fh>) {
307    #print STDERR "debug ($state, " . (@date ? (strftime "%Y-%m-%d", @date) : "") . "): `$_'\n";
308    if ($state == 0) {
309	if ($_l =~ m,^Author: (.*),) {
310	    $author = $1;
311	}
312	if ($_l =~ m,^Date: (.*),) {
313	    @date = strptime($1);
314	}
315	$state = 1 if ($_l =~ m,^$, and $author and (@date+0>0));
316    } elsif ($state == 1) {
317        # * modifying our input text is a bad choice
318        #   let's make a copy of it first, then we remove spaces
319        # * if we meet a "merge branch" statement, we need to start
320        #   over and find a real entry
321        # Luis Mondesi <lemsx1@gmail.com>
322        my $_s = $_l;
323	$_s =~ s/^    //g;
324        if ($_s =~ m/^Merge branch|^Merge remote branch/)
325        {
326            $state=0;
327            $author=0;
328            next;
329        }
330	$comment = $comment . $_s;
331	$state = 2 if ($_l =~ m,^$,);
332    } elsif ($state == 2) {
333	if ($_l =~ m,^([0-9]+)\t([0-9]+)\t(.*)$,) {
334	    push @files, $3;
335	}
336	$done = 1 if ($_l =~ m,^$,);
337    }
338
339    if ($done) {
340	print (strftime "%Y-%m-%d  $author\n\n", @date);
341
342	my $files = join (", ", @files);
343	$files = mywrap ("\t", "\t", "* $files"), ": ";
344
345	if (index($comment, EMPTY_LOG_MESSAGE) > -1 ) {
346	    $comment = "[no log message]\n";
347	}
348
349	my $files_last_line_len = 0;
350	$files_last_line_len = last_line_len($files) + 1;
351	my $msg = wrap_log_entry($comment, "\t", 69-$files_last_line_len, 69);
352
353	$msg =~ s/[ \t]+\n/\n/g;
354
355	print "$files: $msg\n";
356
357	@date = ();
358	$author = "";
359	@files = ();
360	$comment = "";
361
362	$state = 0;
363	$done = 0;
364    }
365}
366
367if (@date + 0)
368{
369    print (strftime "%Y-%m-%d  $author\n\n", @date);
370    my $msg = wrap_log_entry($comment, "\t", 69, 69);
371    $msg =~ s/[ \t]+\n/\n/g;
372    print "\t* $msg\n";
373}
374