rundiff revision 404:9d95995db57c
1#! /usr/bin/env perl
2
3# Copyright (c) 2003 The Regents of The University of Michigan
4# All rights reserved.
5#
6# Redistribution and use in source and binary forms, with or without
7# modification, are permitted provided that the following conditions are
8# met: redistributions of source code must retain the above copyright
9# notice, this list of conditions and the following disclaimer;
10# redistributions in binary form must reproduce the above copyright
11# notice, this list of conditions and the following disclaimer in the
12# documentation and/or other materials provided with the distribution;
13# neither the name of the copyright holders nor the names of its
14# contributors may be used to endorse or promote products derived from
15# this software without specific prior written permission.
16#
17# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
18# "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
19# LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
20# A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
21# OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
22# SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
23# LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
24# DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
25# THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
26# (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
27# OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
28
29# Diff two streams.
30#
31# Unlike regular diff, this script does not read in the entire input
32# before doing a diff, so it can be used on lengthy outputs piped from
33# other programs (e.g., M5 traces).  The best way to do this is to
34# take advantage of the power of Perl's open function, which will
35# automatically fork a subprocess if the last character in the
36# "filename" is a pipe (|).  Thus to compare the instruction traces
37# from two versions of m5 (m5a and m5b), you can do this:
38#
39# rundiff 'm5a --trace:flags=InstExec |' 'm5b --trace:flags=InstExec |'
40#
41
42use strict;
43
44use Getopt::Std;
45
46#
47# Options:
48#  -c <n> : print n lines of context before & after changes
49#  -l <n> : use n lines of lookahead
50#  -x     : use "complex" diff from Algorithm::Diff (see below)
51#
52our ($opt_c, $opt_l, $opt_x);
53getopts('c:l:x');
54
55#
56# For the highest-quality (minimal) diffs, we can use the
57# Algorithm::Diff package.  By default, a built-in, simple, and
58# generally quite adequate algorithm will be used.  If you have
59# Algorithm::Diff installed on your system, and don't mind having the
60# script go slower (like 3-4x slower, based on informal observation),
61# then specify '-x' on the command line to use it.
62my $use_complexdiff = defined($opt_x);
63
64if ($use_complexdiff) {
65    # Don't use 'use', as that's a compile-time option and will fail
66    # on systems that don't have Algorithm::Diff installed even if
67    # $use_complexdiff is false.  'require' is evaluated at runtime,
68    # so it's OK.
69    require Algorithm::Diff;
70    import Algorithm::Diff qw(traverse_sequences);
71};
72
73my $lookahead_lines = $opt_l || 200;
74
75# in theory you could have different amounts of context before and
76# after a diff, but until someone needs that there's only one arg to
77# set both.
78my $precontext_lines = $opt_c || 3;
79my $postcontext_lines = $precontext_lines;
80
81my $file1 = $ARGV[0];
82my $file2 = $ARGV[1];
83
84die "Need two args." if (!(defined($file1) && defined($file2)));
85
86my ($fh1, $fh2);
87open($fh1, $file1) or die "Can't open $file1";
88open($fh2, $file2) or die "Can't open $file2";
89
90# buffer of matching lines for pre-diff context
91my @precontext = ();
92# number of post-diff matching lines remaining to print
93my $postcontext = 0;
94
95# lookahead buffers for $file1 and $file2 respectively
96my @lines1 = ();
97my @lines2 = ();
98
99# Next line number available to print from each file.  Generally this
100# corresponds to the oldest line in @precontext, or the oldest line in
101# @lines1 and @lines2 if @precontext is empty.
102my $lineno1 = 1;
103my $lineno2 = 1;
104
105# Fill a lookahead buffer to $lookahead_lines lines (or until EOF).
106sub fill
107{
108    my ($fh, $array) = @_;
109
110    while (@$array < $lookahead_lines) {
111	my $line = <$fh>;
112	last if (!defined($line));
113	push @$array, $line;
114    }
115}
116
117# Print and delete n lines from front of given array with given prefix.
118sub printlines
119{
120    my ($array, $n, $prefix) = @_;
121
122    while ($n--) {
123	my $line = shift @$array;
124	last if (!defined($line));
125	print $prefix, $line;
126    }
127}
128
129# Print a difference region where n1 lines of file1 were replaced by
130# n2 lines of file2 (where either n1 or n2 could be zero).
131sub printdiff
132{
133    my ($n1, $n2)= @_;
134
135    # If the precontext buffer is full or we're at the beginning of a
136    # file, then this is a new diff region, so we should print a
137    # header indicating the current line numbers.  If we're past the
138    # beginning and the precontext buffer isn't full, then whatever
139    # we're about to print is contiguous with the end of the last
140    # region we printed, so we just concatenate them on the output.
141    if (@precontext == $precontext_lines || ($lineno1 == 0 && $lineno2 == 0)) {
142	print "@@ -$lineno1 +$lineno2 @@\n";
143    }
144
145    # Print and clear the precontext buffer.
146    if (@precontext) {
147	print ' ', join(' ', @precontext);
148	$lineno1 += scalar(@precontext);
149	$lineno2 += scalar(@precontext);
150	@precontext = ();
151    }
152
153    # Print the differing lines.
154    printlines(\@lines1, $n1, '-');
155    printlines(\@lines2, $n2, '+');
156    $lineno1 += $n1;
157    $lineno2 += $n2;
158
159    # Set $postcontext to print the next $postcontext_lines matching lines.
160    $postcontext = $postcontext_lines;
161}
162
163
164########################
165#
166# Complex diff algorithm
167#
168########################
169
170{
171    my $match_found;
172    my $discard_lines1;
173    my $discard_lines2;
174
175    sub match { $match_found = 1; }
176    sub discard1 { $discard_lines1++ unless $match_found; }
177    sub discard2 { $discard_lines2++ unless $match_found; }
178
179    sub complex_diff
180    {
181	$match_found = 0;
182	$discard_lines1 = 0;
183	$discard_lines2 = 0;
184
185	# See Diff.pm.  Note that even though this call generates a
186	# complete diff of both lookahead buffers, all we use it for
187	# is to figure out how many lines to discard off the front of
188	# each buffer to resync the streams.
189	traverse_sequences( \@lines1, \@lines2,
190			    { MATCH => \&match,
191			      DISCARD_A => \&discard1,
192			      DISCARD_B => \&discard2 });
193
194	if (!$match_found) {
195	    printdiff(scalar(@lines1), scalar(@lines2));
196	    die "Lost sync!";
197	}
198
199	# Since we shouldn't get here unless the first lines of the
200	# buffers are different, then we must discard some lines off
201	# at least one of the buffers.
202	die if ($discard_lines1 == 0 && $discard_lines2 == 0);
203
204	printdiff($discard_lines1, $discard_lines2);
205    }
206}
207
208#######################
209#
210# Simple diff algorithm
211#
212#######################
213
214# Check for a pair of matching lines; if found, generate appropriate
215# diff output.
216sub checkmatch
217{
218    my ($n1, $n2) = @_;
219
220    # Check if two adjacent lines match, to reduce false resyncs
221    # (particularly on unrelated blank lines).  This generates
222    # larger-than-necessary diffs when a single line really should be
223    # treated as common; if that bugs you, use Algorithm::Diff.
224    if ($lines1[$n1] eq $lines2[$n2] && $lines1[$n1+1] eq $lines2[$n2+1]) {
225	printdiff($n1, $n2);
226	return 1;
227    }
228
229    return 0;
230}
231
232sub simple_diff
233{
234    # Look for differences of $cnt lines to resync,
235    # increasing $cnt from 1 to $lookahead_lines until we find
236    # something.
237    for (my $cnt = 1; $cnt < $lookahead_lines-1; ++$cnt) {
238	# Check for n lines in one file being replaced by
239	# n lines in the other.
240	return if checkmatch($cnt, $cnt);
241	# Find differences where n lines in one file were
242	# replaced by m lines in the other.  We let m = $cnt
243	# and iterate for n = 0 to $cnt-1.
244	for (my $n = 0; $n < $cnt; ++$n) {
245	    return if checkmatch($n, $cnt);
246	    return if checkmatch($cnt, $n);
247	}
248    }
249
250    printdiff(scalar(@lines1), scalar(@lines2));
251    die "Lost sync!";
252}
253
254# Set the pointer to the appropriate diff function.
255#
256# Note that in either case the function determines how many lines to
257# discard from the front of each lookahead buffer to resync the
258# streams, then prints the appropriate diff output and discards them.
259# After the function returns, it should always be the case that
260# $lines1[0] eq $lines2[0].
261my $find_diff = $use_complexdiff ? \&complex_diff : \&simple_diff;
262
263# The main loop.
264while (1) {
265    # keep lookahead buffers topped up
266    fill($fh1, \@lines1);
267    fill($fh2, \@lines2);
268
269    # peek at first line in each buffer
270    my $l1 = $lines1[0];
271    my $l2 = $lines2[0];
272
273    if (!defined($l1) && !defined($l2)) {
274	# reached EOF on both streams: exit
275	exit(1);
276    }
277
278    if ($l1 eq $l2) {
279	# matching lines: delete from lookahead buffer
280	shift @lines1;
281	shift @lines2;
282	# figure out what to do with this line
283	if ($postcontext > 0) {
284	    # we're in the post-context of a diff: print it
285	    $postcontext--;
286	    print ' ', $l1;
287	    $lineno1++;
288	    $lineno2++;
289	}
290	else {
291	    # we're in the middle of a matching region... save this
292	    # line for precontext in case we run into a difference.
293	    push @precontext, $l1;
294	    # don't let precontext buffer get bigger than needed
295	    while (@precontext > $precontext_lines) {
296		shift @precontext;
297		$lineno1++;
298		$lineno2++;
299	    }
300	}
301    }
302    else {
303	# Mismatch.  Deal with it.
304	&$find_diff();
305    }
306}
307