xref: /aoo41x/main/xmerge/workben/jstyle.pl (revision cdf0e10c)
1#!/bin/sh -- # This comment tells perl not to loop!
2#
3#***************************************************************************
4#
5# DO NOT ALTER OR REMOVE COPYRIGHT NOTICES OR THIS FILE HEADER.
6#
7# Copyright 2000, 2010 Oracle and/or its affiliates.
8#
9# OpenOffice.org - a multi-platform office productivity suite
10#
11# This file is part of OpenOffice.org.
12#
13# OpenOffice.org is free software: you can redistribute it and/or modify
14# it under the terms of the GNU Lesser General Public License version 3
15# only, as published by the Free Software Foundation.
16#
17# OpenOffice.org is distributed in the hope that it will be useful,
18# but WITHOUT ANY WARRANTY; without even the implied warranty of
19# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
20# GNU Lesser General Public License version 3 for more details
21# (a copy is included in the LICENSE file that accompanied this code).
22#
23# You should have received a copy of the GNU Lesser General Public License
24# version 3 along with OpenOffice.org.  If not, see
25# <http://www.openoffice.org/license.html>
26# for a copy of the LGPLv3 License.
27#
28#***************************************************************************
29
30
31eval 'exec perl -S $0 "$@"'
32if 0;
33#
34# @(#)jstyle 1.2 98/01/08
35#
36# jstyle - check for some common stylistic errors.
37#
38#	jstyle is a sort of "lint" for Java coding style.
39#
40#	There's a lot this can't check for, like proper
41#	indentation of continuation lines.  There's also
42#	a lot more this could check for.
43#
44#	A note to the non perl literate:
45#
46#		perl regular expressions are pretty much like egrep
47#		regular expressions, with the following special symbols
48#
49#		\s	any space character
50#		\S	any non-space character
51#		\w	any "word" character [a-zA-Z0-9_]
52#		\W	any non-word character
53#		\d	a digit [0-9]
54#		\D	a non-digit
55#		\b	word boundary (between \w and \W)
56#		\B	non-word boundary
57#
58#require "getopts.pl";
59# XXX - because some versions of perl can not find the lib directory,
60# we just include this here.
61;# getopts.pl - a better getopt.pl
62
63;# Usage:
64;#      do Getopts("a:bc");  # -a takes arg. -b & -c not. Sets opt_* as a
65;#                           #  side effect.
66
67sub Getopts {
68    local($argumentative) = @_;
69    local(@args,$_,$first,$rest);
70    local($[) = 0;
71    local($errs) = 0;
72
73    @args = split( / */, $argumentative );
74    while(($_ = $ARGV[0]) =~ /^-(.)(.*)/) {
75	($first,$rest) = ($1,$2);
76	$pos = index($argumentative,$first);
77	if($pos >= $[) {
78	    if($args[$pos+1] eq ":") {
79		shift(@ARGV);
80		if($rest eq "") {
81		    $rest = shift(@ARGV);
82		}
83		eval "\$opt_$first = \$rest;";
84	    }
85	    else {
86		eval "\$opt_$first = 1";
87		if($rest eq "") {
88		    shift(@ARGV);
89		}
90		else {
91		    $ARGV[0] = "-$rest";
92		}
93	    }
94	}
95	else {
96	    print STDERR "Unknown option: $first\n";
97	    ++$errs;
98	    if($rest ne "") {
99		$ARGV[0] = "-$rest";
100	    }
101	    else {
102		shift(@ARGV);
103	    }
104	}
105    }
106    $errs == 0;
107}
108
1091;
110# end of getopts.pl
111
112$usage =
113"usage: jstyle [-c] [-h] [-p] [-s] [-t] [-v] [-C] file ...
114	-c	check continuation line indenting
115	-h	perform heuristic checks that are sometimes wrong
116	-p	perform some of the more picky checks
117	-s	check for spaces vs. tabs
118	-t	insist on indenting by tabs
119	-v	verbose
120	-C	don't check anything in header block comments
121	-S	print out overall statistics
122";
123
124if (!&Getopts("chpstvCS")) {
125	print $usage;
126	exit 1;
127}
128
129$check_continuation = $opt_c;
130$heuristic = $opt_h;
131$picky = $opt_p;
132$spaces = $opt_s;
133$tabs = $opt_t;
134$verbose = $opt_v;
135$ignore_hdr_comment = $opt_C;
136$statistics = $opt_S;
137
138if ($verbose) {
139	$fmt = "%s: %d: %s\n%s\n";
140} else {
141	$fmt = "%s: %d: %s\n";
142}
143
144# Note, following must be in single quotes so that \s and \w work right.
145$typename = '(int|char|boolean|byte|short|long|float|double)';
146
147if ($#ARGV >= 0) {
148	foreach $arg (@ARGV) {
149		if (!open(STDIN, $arg)) {
150			printf "%s: can not open\n", $arg;
151		} else {
152			&jstyle($arg);
153			close STDIN;
154		}
155	}
156} else {
157	&jstyle("<stdin>");
158}
159
160if ($statistics != 0) {
161	foreach $key (sort(keys %errcount)) {
162		printf "%6d %s\n", $errcount{$key}, $key;
163	}
164	printf " -----\n";
165	printf "%6d Total warnings\n", $tot_errcount;
166	printf "%6d Lines of code\n", $totlines;
167}
168
169sub err {
170	if ($statistics == 0) {
171		printf $fmt, $filename, $., $_[0], $line;
172	} else {
173		$msg = $_[0];
174		$msg =~ s/ \([0-9][0-9]*\)$//;
175		$errcount{$msg} += 1;
176		$tot_errcount += 1;
177	}
178}
179
180sub jstyle {
181
182$in_comment = 0;
183$in_header_comment = 0;
184$in_continuation = 0;
185$in_class = 0;
186$in_declaration = 0;
187$note_level = 0;
188$nextok = 0;
189$nocheck = 0;
190$expect_continuation = 0;
191$prev = '';
192
193$filename = $_[0];
194
195line: while (<STDIN>) {
196	++$totlines;
197	s/\r?\n$//;	# strip return and newline
198
199	# save the original line, then remove all text from within
200	# double or single quotes, we do not want to check such text.
201
202	$line = $_;
203	s/"[^"]*"/\"\"/g;
204	s/'.'/''/g;
205
206	# an /* END JSTYLED */ comment ends a no-check block.
207	if ($nocheck) {
208		if (/\/\* *END *JSTYLED *\*\//) {
209			$nocheck = 0;
210		} else {
211			next line;
212		}
213	}
214
215	# a /*JSTYLED*/ comment indicates that the next line is ok.
216	if ($nextok) {
217		if ($okmsg) {
218			do err($okmsg);
219		}
220		$nextok = 0;
221		$okmsg = 0;
222		if (/\/\* *JSTYLED.*\*\//) {
223			/^.*\/\* *JSTYLED *(.*) *\*\/.*$/;
224			$okmsg = $1;
225			$nextok = 1;
226		}
227		$prev = $line;
228		next line;
229	}
230
231	# check length of line.
232	# first, a quick check to see if there is any chance of being too long.
233	if ($line =~ tr/\t/\t/ * 7 + length($line) > 100) {
234		# yes, there is a chance.
235		# replace tabs with spaces and check again.
236		$eline = $line;
237		1 while $eline =~
238		    s/\t+/' ' x (length($&) * 8 - length($`) % 8)/e;
239		$l = length($eline);
240		if (length($eline) > 100) {
241			do err("line > 100 characters ($l)");
242		}
243	}
244#	this is the fastest way to check line length,
245#	but it doesnt work with perl 3.0.
246#	if ($line =~ tr/\t/\t/ * 7 + length($line) > 80) {
247#		$pos = $oldp = $p = 0;
248#		while (($p = index($line, "\t", $p)) >= 0) {
249#			$pos = ($pos + $p - $oldp + 8) & ~7;
250#			$oldp = ++$p;
251#		}
252#		$pos += length($line) - $oldp;
253#		if ($pos > 80) {
254#			do err("line > 80 characters");
255#		}
256#	}
257
258	# remember whether we expect to be inside a continuation line.
259	$in_continuation = $expect_continuation;
260
261	# check for proper continuation line.  blank lines
262	# in the middle of the
263	# continuation do not count.
264	# XXX - only check within functions.
265	if ($check_continuation && $expect_continuation && $in_class &&
266	    !/^\s*$/) {
267		# continuation line must start with whitespace of
268		# previous line, plus either 4 spaces or a tab, but
269		# do not check lines that start with a string constant
270		# since they are often shifted to the left to make them
271		# fit on the line.
272		if (!/^$continuation_indent    \S/ &&
273		    !/^$continuation_indent\t\S/ && !/^\s*"/) {
274			do err("continuation line improperly indented");
275		}
276		$expect_continuation = 0;
277	}
278
279	# a /* BEGIN JSTYLED */ comment starts a no-check block.
280	if (/\/\* *BEGIN *JSTYLED *\*\//) {
281		$nocheck = 1;
282	}
283
284	# a /*JSTYLED*/ comment indicates that the next line is ok.
285	if (/\/\* *JSTYLED.*\*\//) {
286		/^.*\/\* *JSTYLED *(.*) *\*\/.*$/;
287		$okmsg = $1;
288		$nextok = 1;
289	}
290	if (/\/\/ *JSTYLED/) {
291		/^.*\/\/ *JSTYLED *(.*)$/;
292		$okmsg = $1;
293		$nextok = 1;
294	}
295
296	# is this the beginning or ending of a class?
297	if (/^(public\s+)*\w(class|interface)\s/) {
298		$in_class = 1;
299		$in_declaration = 1;
300		$prev = $line;
301		next line;
302	}
303	if (/^}\s*(\/\*.*\*\/\s*)*$/) {
304		$in_class = 0;
305		$prev = $line;
306		next line;
307	}
308
309	if (!$spaces) {
310		# strip trailing spaces
311		s/\s*$//;
312	}
313
314	# does this looks like the start of a block comment?
315	if (/^\s*\/\*(\*|)$/) {
316		if (!/^(\t|    )*\/\*(\*|)$/) {
317			do err("block comment not indented properly");
318		}
319		$in_comment = 1;
320		s/\/\*(\*|)/ /;
321		$comment_prefix = $_;
322		if ($comment_prefix eq " ") {
323			$in_header_comment = 1;
324		}
325		$prev = $line;
326		next line;
327	}
328	if (/^\s*\/\*./ && !/^\s*\/\*\*$/ && !/^\s*\/\*.*\*\//) {
329		do err("improper first line of block comment");
330		# it's a bad one, but it still is one.
331		# avoid ripple effect of not recognizing this.
332		if (!/^(\t|    )*\/\*(\*|)/) {
333			do err("block comment not indented properly");
334		}
335		$in_comment = 1;
336		s/\/\*.*/ /;
337		$comment_prefix = $_;
338		if ($comment_prefix eq " ") {
339			$in_header_comment = 1;
340		}
341		$prev = $line;
342		next line;
343	}
344	# are we still in the block comment?
345	if ($in_comment && !/^$comment_prefix\*/) {
346		# assume out of comment
347		$in_comment = 0;
348		$in_header_comment = 0;
349	}
350
351	if ($in_header_comment && $ignore_hdr_comment) {
352		$prev = $line;
353		next line;
354	}
355
356	# check for errors that might occur in comments and in code.
357
358	# allow spaces to be used to draw pictures in header comments.
359	if ($spaces && /[^ ]     / && !/".*     .*"/ && !$in_header_comment) {
360		do err("spaces instead of tabs");
361	}
362	if ($tabs && /^ / && !/^ \*[ \t\/]/ && !/^ \*$/ &&
363	    (!/^    \w/ || $in_class != 0)) {
364		do err("indent by spaces instead of tabs");
365	}
366	if (!$in_comment && (/^(\t    )* {1,3}\S/ || /^(\t    )* {5,7}\S/) &&
367	    !(/^\s*[-+|&\/?:=]/ || ($prev =~ /,\s*$/))) {
368		do err("indent not a multiple of 4");
369	}
370	if ($spaces && /\s$/) {
371		do err("space or tab at end of line");
372	}
373if (0) {
374	if (/^[\t]+ [^ \t\*]/ || /^[\t]+  \S/ || /^[\t]+   \S/) {
375		do err("continuation line not indented by 4 spaces");
376	}
377}
378	if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) {
379		do err("comment preceded by non-blank");
380	}
381	if ($spaces && /\t[ ]+\t/) {
382		do err("spaces between tabs");
383	}
384	if ($spaces && / [\t]+ /) {
385		do err("tabs between spaces");
386	}
387
388	if ($in_comment) {	# still in comment
389		$prev = $line;
390		next line;
391	}
392
393	if ((/\/\*\S/ && !/\/\*\*/) || /\/\*\*\S/) {
394		do err("missing blank after open comment");
395	}
396	if (/\S\*\//) {
397		do err("missing blank before close comment");
398	}
399	# allow // at beginnging of line, often used to comment out code
400	if (/.\/\/\S/) {		# C++ comments
401		do err("missing blank after start comment");
402	}
403	# check for unterminated single line comments.
404	if (/\S.*\/\*/ && !/\S.*\/\*.*\*\//) {
405		do err("unterminated single line comment");
406	}
407
408	# delete any comments and check everything else.
409	s/\/\*.*\*\///g;
410	s/\/\/.*$//;		# C++ comments
411
412	# delete any trailing whitespace; we have already checked for that.
413	s/\s*$//;
414
415	# following checks do not apply to text in comments.
416
417	# if it looks like an operator at the end of the line, and it is
418	# not really the end of a comment (...*/), and it is not really
419	# a label (done:), and it is not a case label (case FOO:),
420	# or we are not in a function definition (ANSI C style) and the
421	# operator is a "," (to avoid hitting "int\nfoo(\n\tint i,\n\tint j)"),
422	# or we are in a function and the operator is a
423	# "*" (to avoid hitting on "char*\nfunc()").
424	if ((/[-+|&\/?:=]$/ && !/\*\/$/ && !/^\s*\w*:$/ &&
425	    !/^\s\s*case\s\s*\w*:$/) ||
426	    /,$/ ||
427	    ($in_class && /\*$/)) {
428		$expect_continuation = 1;
429		if (!$in_continuation) {
430			/^(\s*)\S/;
431			$continuation_indent = $1;
432		}
433	}
434	if (/[^<>\s][!<>=]=/ || /[^<>][!<>=]=\S/ ||
435	    (/[^->]>[^=>\s]/ && !/[^->]>$/) || (/[^<]<[^=<\s]/ && !/[^<]<$/) ||
436	    /[^<\s]<[^<]/ || /[^->\s]>[^>]/) {
437		do err("missing space around relational operator");
438	}
439	if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ ||
440	    (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) ||
441	    (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) {
442		do err("missing space around assignment operator");
443	}
444	if (/[,;]\S/ && !/\bfor \(;;\)/) {
445		do err("comma or semicolon followed by non-blank");
446	}
447	# allow "for" statements to have empty "while" clauses
448	if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) {
449		do err("comma or semicolon preceded by blank");
450	}
451if (0) {
452	if (/^\s*(&&|\|\|)/) {
453		do err("improper boolean continuation");
454	}
455}
456	if ($picky && /\S   *(&&|\|\|)/ || /(&&|\|\|)   *\S/) {
457		do err("more than one space around boolean operator");
458	}
459	if (/\b(for|if|while|switch|return|case|catch|synchronized)\(/) {
460		do err("missing space between keyword and paren");
461	}
462	if (/(\b(for|if|while|switch|return|catch|synchronized)\b.*){2,}/) {
463		# multiple "case" allowed
464		do err("more than one keyword on line");
465	}
466	if (/\b(for|if|while|switch|return|case|catch|synchronized)\s\s+\(/ &&
467	    !/^#if\s+\(/) {
468		do err("extra space between keyword and paren");
469	}
470	# try to detect "func (x)" but not "if (x)" or
471	# "int (*func)();"
472	if (/\w\s\(/) {
473		$s = $_;
474		# strip off all keywords on the line
475		s/\b(for|if|while|switch|return|case|catch|synchronized)\s\(/XXX(/g;
476		#s/\b($typename|void)\s+\(+/XXX(/og;
477		if (/\w\s\(/) {
478			do err("extra space between function name and left paren");
479		}
480		$_ = $s;
481	}
482	if (/\(\s/) {
483		do err("whitespace after left paren");
484	}
485	# allow "for" statements to have empty "continue" clauses
486	if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/) {
487		do err("whitespace before right paren");
488	}
489	if (/^\s*\(void\)[^ ]/) {
490		do err("missing space after (void) cast");
491	}
492	if (/\S{/ && !/{{/) {
493		do err("missing space before left brace");
494	}
495	if ($in_class && /^\s+{/ && ($prev =~ /\)\s*$/)) {
496		do err("left brace starting a line");
497	}
498	if (/}(else|while)/) {
499		do err("missing space after right brace");
500	}
501	if (/}\s\s+(else|while)/) {
502		do err("extra space after right brace");
503	}
504	if (/\b$typename\*/o) {
505		do err("missing space between type name and *");
506	}
507	if ($heuristic) {
508		# cannot check this everywhere due to "struct {\n...\n} foo;"
509		if ($in_class && !$in_declaration &&
510		    /}./ && !/}\s+=/ && !/{.*}[;,]$/ && !/}(\s|)*$/ &&
511		    !/} (else|while)/ && !/}}/) {
512			do err("possible bad text following right brace");
513		}
514		# cannot check this because sub-blocks in
515		# the middle of code are ok
516		if ($in_class && /^\s+{/) {
517			do err("possible left brace starting a line");
518		}
519	}
520	if (/^\s*else\W/) {
521		if ($prev =~ /^\s*}$/) {
522			$str = "else and right brace should be on same line";
523			if ($statistics == 0) {
524				printf $fmt, $filename, $., $str, $prev;
525				if ($verbose) {
526					printf "%s\n", $line;
527				}
528			} else {
529				$errcount{$str} += 1;
530				$tot_errcount += 1;
531			}
532		}
533	}
534	$prev = $line;
535}
536
537if ($picky && $prev eq "") {
538	do err("last line in file is blank");
539}
540
541}
542