drupal/scripts/code-style.pl

#!/usr/bin/perl -w
# $Id: code-style.pl,v 1.11 2006/02/21 18:46:54 dries Exp $

# Author: Alexander Schwartz (alexander.schwartz@gmx.net)
# Licence: GPL
# First version: 2001-10-15

# Originally written for Drupal (http://drupal.org/) to ensure stylish
# code.  This program tries to show as many improvements as possible with
# no false positives.

# $Id: code-style.pl,v 1.11 2006/02/21 18:46:54 dries Exp $

$comment = 0;
$program = 0;
if ($ARGV[0] eq '-debug') {
  $debug=1;
  shift (@ARGV);
}
else {
  $debug=0;
}
while (<>) {
  $org=$_;
  s/\\["']//g;
  # please don't use nested comments for now... thanks!
  # handles comments // style, but don't mess with http://
  s/\/\/[^:].*//;
  # handles comments /**/ on a single line
  s/\/\*.*\*\///g;
  # handles comments /**/ over several lines
  if ($comment == 1) {
    if (s/.*\*\///) {
      $comment = 0;
    }
    else {
      next;
    }
  }
  if (s/\/\*.*//) {
    $comment = 1;
  }
  if (/^\s*#/) {
    next;
  }

  if (s/<\?php//) {
    $program = 1;
  }
  if (/\?>/) {
    $program = 0;
  }

  # enforce "bar". foo() ."bar" syntax
  if (/^("[^"]*"|[^"])*("[^"]*")\.[^ ]/ && $program) {
    $msg = "'\".' -> '\". '";
  }
  elsif (/^("[^"]*"|[^"])*("[^"]*")\s+\./ && $program) {
    $msg = "'\" .' -> '\".'";
  }
  # enforce "bar". foo() ."bar" syntax
  elsif (/^("[^"]*"|[^"])*[^ "]\.("[^"]*")/ && $program) {
    $msg = "'.\"' -> '.\"'";
  }
  elsif (/^("[^"]*"|[^"])*[^ "]\.\s+("[^"]*")/ && $program) {
    $msg = "'. \"' -> '.\"'";
  }
  # XHTML requires closing tag
  elsif (/<br>/i) {
    $msg = "'<br>' -> '<br />'";
  }
  elsif (/\$REQUEST_URI/i) {
    $msg = "the use of REQUEST_URI is prone to XSS exploits and does not work on IIS; use request_uri() instead";
  }
  elsif (/\"REQUEST_URI\"/i) {
    $msg = "the use of REQUEST_URI is prone to XSS exploits and does not work on IIS; use request_uri() instead";
  }

  # XHTML compatibility mode suggests a blank before /
  # i.e. <br />
  elsif (/<[a-z][^>]*[^ >]\/>/i) {
    $msg = "'<foo/".">' -> '<foo />'";
  }
  # we write '{' on the same line, not on the next
  elsif (/^\s*{/ && $program) {
    $msg = "take '{' to previous line";
  }
  elsif (/([a-z])([A-Z])/) {
    $msg = "no mixed case function or variable names, use lower case and _";
  }
  elsif (/<[\/]*[A-Z]+[^>]*>/) {
    $msg = "XHTML demands tags to be lowercase";
  }

  # trying to recognize splitted lines
  # there are only a few valid last characters in programming mode,
  # only sometimes it is ( if you use if/else with a single statement

  # from here on we need no more strings
  while (s/^([^"]*)"[^"]*"/$1#/) {};
  while (s/^([^']*)'[^']*'/$1#/) {};

  # it should be 'if (' all the time
  if (/(^|[^a-zA-Z])(if|else|elseif|while|foreach|switch|return|for)\(/) {
    $msg = "'(' -> ' ('";
  }
  #elsif (/[^;{}:\s\n]\s*\n*$/ && $program && !/^[\s}]*(if|else)/) {
  #  $msg = "don't split lines";
  #}
  elsif (/\}\s*else/) {
    $msg = "'} else' -> '}\\nelse'";
  }
  elsif (/[^{\s\n]\s*\n*$/ && $program && /^\s*(if|else)/) {
    $msg = "every if/else needs a { at eol";
  }
  elsif (/([\(\[]) / && $program) {
    $msg = "'$1 ' -> '$1'";
  }
  elsif (/\S ([\)\]])/ && $program) {
    $msg = "' $1' -> '$1'";
  }
  # but no brackets
  elsif (/([a-z-A-Z_][a-zA-Z0-9_-]*)\s+\(/ && $program) {
    if ($1 ne "switch" and $1 ne "if" and $1 ne "while" and $1 ne "foreach" and $1 ne "return" and $1 ne "for" and $1 ne "elseif") {
      $msg = "'$1 (' -> '$1('";
    }
  }
  # there should be a space before '{'
  if (/[^ ]{/ && $program) {
    $msg = "missing space before '{'";
  }
  # there should be a space after ','
  elsif (/[,][^ \n\r]/ && $program) {
    $msg = "missing space after ','";
  }
  # spaces before and after, only foreach may use $foo=>bar
  elsif (/[^ =|\-|\+](\+|\-)[^ =>|\-|\+]/ && $program && !/foreach/) {
    $msg = "'$1' -> ' $1 '";
  }
  elsif (/[^ =](\*|==|\.=|=>|=|\|\|)[^ =>]/ && $program && !/foreach/) {
    $msg = "'$1' -> ' $1 '";
  }
  # ensure $bar["foo"] and $bar[$foo] and $bar[0]
  elsif (/\[[^#][^\]]*\]/ && !/\[[0-9\$][^\]]*\]/ && !/\[\]/) {
    $msg = "only [\"foo\"], [\$foo] or [0] is allowed";
  }
  # first try to find missing quotes after = in (X)HTML tags
  elsif (/<[^>]*=[a-zA-Z0-9][^>]*>/) {
    $msg = "=... -> =\"...\"";
  }
  if (defined $msg) {
    if ($debug==0) {
      print $ARGV .":". $. .": $msg : ". $org;
    }
    undef $msg;
  }
  elsif ($debug==1) {
    print $org;
  }
} continue {
  close ARGV if eof;
}
1	dries	1.1	#!/usr/bin/perl -w
2	dries	1.12	# $Id: code-style.pl,v 1.11 2006/02/21 18:46:54 dries Exp $
3	dries	1.1
4			# Author: Alexander Schwartz (alexander.schwartz@gmx.net)
5			# Licence: GPL
6			# First version: 2001-10-15
7
8	dries	1.11	# Originally written for Drupal (http://drupal.org/) to ensure stylish
9	dries	1.2	# code. This program tries to show as many improvements as possible with
10	dries	1.1	# no false positives.
11
12	dries	1.12	# $Id: code-style.pl,v 1.11 2006/02/21 18:46:54 dries Exp $
13	dries	1.1
14			$comment = 0;
15			$program = 0;
16			if ($ARGV[0] eq '-debug') {
17			$debug=1;
18			shift (@ARGV);
19			}
20			else {
21			$debug=0;
22			}
23			while (<>) {
24			$org=$_;
25			s/\\["']//g;
26			# please don't use nested comments for now... thanks!
27			# handles comments // style, but don't mess with http://
28	dries	1.3	s/\/\/[^:].*//;
29	dries	1.1	# handles comments /**/ on a single line
30			s/\/\.\*\///g;
31			# handles comments /**/ over several lines
32			if ($comment == 1) {
33			if (s/.\\///) {
34	dries	1.2	$comment = 0;
35			}
36	dries	1.1	else {
37	dries	1.2	next;
38	dries	1.1	}
39			}
40			if (s/\/\.//) {
41			$comment = 1;
42			}
43			if (/^\s*#/) {
44			next;
45			}
46
47			if (s/<\?php//) {
48			$program = 1;
49			}
50			if (/\?>/) {
51			$program = 0;
52			}
53	dries	1.2
54	dries	1.9	# enforce "bar". foo() ."bar" syntax
55	dries	1.1	if (/^("[^"]"\|[^"])("[^"]*")\.[^ ]/ && $program) {
56			$msg = "'\".' -> '\". '";
57			}
58			elsif (/^("[^"]"\|[^"])("[^"]*")\s+\./ && $program) {
59			$msg = "'\" .' -> '\".'";
60			}
61	dries	1.9	# enforce "bar". foo() ."bar" syntax
62	dries	1.1	elsif (/^("[^"]"\|[^"])[^ "]\.("[^"]*")/ && $program) {
63			$msg = "'.\"' -> '.\"'";
64			}
65			elsif (/^("[^"]"\|[^"])[^ "]\.\s+("[^"]*")/ && $program) {
66			$msg = "'. \"' -> '.\"'";
67			}
68			# XHTML requires closing tag
69	dries	1.3	elsif (/<br>/i) {
70			$msg = "'<br>' -> '<br />'";
71	dries	1.1	}
72	dries	1.8	elsif (/\$REQUEST_URI/i) {
73			$msg = "the use of REQUEST_URI is prone to XSS exploits and does not work on IIS; use request_uri() instead";
74	dries	1.6	}
75	dries	1.8	elsif (/\"REQUEST_URI\"/i) {
76			$msg = "the use of REQUEST_URI is prone to XSS exploits and does not work on IIS; use request_uri() instead";
77	dries	1.6	}
78
79	dries	1.2	# XHTML compatibility mode suggests a blank before /
80	dries	1.1	# i.e. <br />
81			elsif (/<[a-z][^>]*[^ >]\/>/i) {
82			$msg = "'<foo/".">' -> '<foo />'";
83			}
84			# we write '{' on the same line, not on the next
85			elsif (/^\s*{/ && $program) {
86	dries	1.2	$msg = "take '{' to previous line";
87	dries	1.1	}
88	dries	1.7	elsif (/([a-z])([A-Z])/) {
89			$msg = "no mixed case function or variable names, use lower case and _";
90	dries	1.1	}
91	dries	1.7	elsif (/<[\/][A-Z]+[^>]>/) {
92			$msg = "XHTML demands tags to be lowercase";
93			}
94
95	dries	1.1	# trying to recognize splitted lines
96			# there are only a few valid last characters in programming mode,
97			# only sometimes it is ( if you use if/else with a single statement
98	dries	1.2
99	dries	1.1	# from here on we need no more strings
100			while (s/^([^"])"[^"]"/$1#/) {};
101			while (s/^([^'])'[^']'/$1#/) {};
102
103			# it should be 'if (' all the time
104			if (/(^\|[^a-zA-Z])(if\|else\|elseif\|while\|foreach\|switch\|return\|for)\(/) {
105	dries	1.3	$msg = "'(' -> ' ('";
106	dries	1.1	}
107	dries	1.8	#elsif (/[^;{}:\s\n]\s\n$/ && $program && !/^[\s}]*(if\|else)/) {
108			# $msg = "don't split lines";
109			#}
110	dries	1.1	elsif (/\}\s*else/) {
111			$msg = "'} else' -> '}\\nelse'";
112			}
113			elsif (/[^{\s\n]\s\n$/ && $program && /^\s*(if\|else)/) {
114			$msg = "every if/else needs a { at eol";
115			}
116			elsif (/([\(\[]) / && $program) {
117			$msg = "'$1 ' -> '$1'";
118			}
119	dries	1.12	elsif (/\S ([\)\]])/ && $program) {
120	dries	1.1	$msg = "' $1' -> '$1'";
121			}
122	dries	1.2	# but no brackets
123	dries	1.1	elsif (/([a-z-A-Z_][a-zA-Z0-9_-]*)\s+\(/ && $program) {
124			if ($1 ne "switch" and $1 ne "if" and $1 ne "while" and $1 ne "foreach" and $1 ne "return" and $1 ne "for" and $1 ne "elseif") {
125			$msg = "'$1 (' -> '$1('";
126			}
127			}
128			# there should be a space before '{'
129	dries	1.3	if (/[^ ]{/ && $program) {
130	dries	1.1	$msg = "missing space before '{'";
131			}
132			# there should be a space after ','
133	kjartan	1.4	elsif (/[,][^ \n\r]/ && $program) {
134	dries	1.1	$msg = "missing space after ','";
135			}
136			# spaces before and after, only foreach may use $foo=>bar
137	dries	1.10	elsif (/[^ =\|\-\|\+](\+\|\-)[^ =>\|\-\|\+]/ && $program && !/foreach/) {
138	kjartan	1.4	$msg = "'$1' -> ' $1 '";
139			}
140			elsif (/[^ =](\*\|==\|\.=\|=>\|=\|\\|\\|)[^ =>]/ && $program && !/foreach/) {
141	dries	1.3	$msg = "'$1' -> ' $1 '";
142			}
143			# ensure $bar["foo"] and $bar[$foo] and $bar[0]
144			elsif (/\[[^#][^\]]\]/ && !/\[[0-9\$][^\]]\]/ && !/\[\]/) {
145			$msg = "only [\"foo\"], [\$foo] or [0] is allowed";
146			}
147			# first try to find missing quotes after = in (X)HTML tags
148			elsif (/<[^>]=[a-zA-Z0-9][^>]>/) {
149			$msg = "=... -> =\"...\"";
150	dries	1.1	}
151			if (defined $msg) {
152			if ($debug==0) {
153			print $ARGV .":". $. .": $msg : ". $org;
154			}
155			undef $msg;
156	dries	1.2	}
157	dries	1.1	elsif ($debug==1) {
158			print $org;
159			}
160			} continue {
161			close ARGV if eof;
162			}