#!/usr/bin/env perl # # CDDL HEADER START # # The contents of this file are subject to the terms of the # Common Development and Distribution License (the "License"). # You may not use this file except in compliance with the License. # # You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE # or http://www.opensolaris.org/os/licensing. # See the License for the specific language governing permissions # and limitations under the License. # # When distributing Covered Code, include this CDDL HEADER in each # file and include the License file at usr/src/OPENSOLARIS.LICENSE. # If applicable, add the following below this CDDL HEADER, with the # fields enclosed by brackets "[]" replaced with your own identifying # information: Portions Copyright [yyyy] [name of copyright owner] # # CDDL HEADER END # # # Copyright 2008 Sun Microsystems, Inc. All rights reserved. # Use is subject to license terms. # # Copyright 2011 Joyent, Inc. All rights reserved. # # jsstyle - check for some common stylistic errors. # # jsstyle is a sort of "lint" for Javascript coding style. This tool is # derived from the cstyle tool, used to check for the style used in the # Solaris kernel, sometimes known as "Bill Joy Normal Form". # # There's a lot this can't check for, like proper indentation of code # blocks. There's also a lot more this could check for. # # A note to the non perl literate: # # perl regular expressions are pretty much like egrep # regular expressions, with the following special symbols # # \s any space character # \S any non-space character # \w any "word" character [a-zA-Z0-9_] # \W any non-word character # \d a digit [0-9] # \D a non-digit # \b word boundary (between \w and \W) # \B non-word boundary # require 5.0; use IO::File; use Getopt::Std; use strict; my $usage = "Usage: jsstyle [-h?vcC] [-t ] [-f ] [-o ] file ... Check your JavaScript file for style. See for details on config options. Report bugs to . Options: -h print this help and exit -v verbose -c check continuation indentation inside functions -t specify tab width for line length calculation -C don't check anything in header block comments -f PATH path to a jsstyle config file -o OPTION1,OPTION2 set config options, e.g. '-o doxygen,indent=2' "; my %opts; if (!getopts("ch?o:t:f:vC", \%opts)) { print $usage; exit 2; } if (defined($opts{'h'}) || defined($opts{'?'})) { print $usage; exit; } my $check_continuation = $opts{'c'}; my $verbose = $opts{'v'}; my $ignore_hdr_comment = $opts{'C'}; my $tab_width = $opts{'t'}; # By default, tabs are 8 characters wide if (! defined($opts{'t'})) { $tab_width = 8; } # Load config my %config = ( indent => "tab", doxygen => 0, # doxygen comments: /** ... */ splint => 0, # splint comments. Needed? "unparenthesized-return" => 1, "literal-string-quote" => "single", # 'single' or 'double' "blank-after-start-comment" => 1, ); sub add_config_var ($$) { my ($scope, $str) = @_; if ($str !~ /^([\w-]+)(?:\s*=\s*(.*?))?$/) { die "$scope: invalid option: '$str'"; } my $name = $1; my $value = ($2 eq '' ? 1 : $2); #print "scope: '$scope', str: '$str', name: '$name', value: '$value'\n"; # Validate config var. if ($name eq "indent") { # A number of spaces or "tab". if ($value !~ /^\d+$/ && $value ne "tab") { die "$scope: invalid '$name': must be a number (of ". "spaces) or 'tab'"; } } elsif ($name eq "doxygen" || # boolean vars $name eq "splint" || $name eq "unparenthesized-return" || $name eq "blank-after-start-comment") { if ($value != 1 && $value != 0) { die "$scope: invalid '$name': don't give a value"; } } elsif ($name eq "literal-string-quote") { if ($value !~ /single|double/) { die "$scope: invalid '$name': must be 'single' ". "or 'double'"; } } else { die "$scope: unknown config var: $name"; } $config{$name} = $value; } if (defined($opts{'f'})) { my $path = $opts{'f'}; my $fh = new IO::File $path, "r"; if (!defined($fh)) { die "cannot open config path '$path'"; } my $line = 0; while (<$fh>) { $line++; s/^\s*//; # drop leading space s/\s*$//; # drop trailing space next if ! $_; # skip empty line next if /^#/; # skip comments add_config_var "$path:$line", $_; } } if (defined($opts{'o'})) { for my $x (split /,/, $opts{'o'}) { add_config_var "'-o' option", $x; } } my ($filename, $line, $prev); # shared globals my $fmt; my $hdr_comment_start; if ($verbose) { $fmt = "%s: %d: %s\n%s\n"; } else { $fmt = "%s: %d: %s\n"; } if ($config{"doxygen"}) { # doxygen comments look like "/*!" or "/**"; allow them. $hdr_comment_start = qr/^\s*\/\*[\!\*]?$/; } else { $hdr_comment_start = qr/^\s*\/\*$/; } # Note, following must be in single quotes so that \s and \w work right. my $lint_re = qr/\/\*(?: jsl:\w+?|ARGSUSED[0-9]*|NOTREACHED|LINTLIBRARY|VARARGS[0-9]*| CONSTCOND|CONSTANTCOND|CONSTANTCONDITION|EMPTY| FALLTHRU|FALLTHROUGH|LINTED.*?|PRINTFLIKE[0-9]*| PROTOLIB[0-9]*|SCANFLIKE[0-9]*|JSSTYLED.*? )\*\//x; my $splint_re = qr/\/\*@.*?@\*\//x; my $err_stat = 0; # exit status if ($#ARGV >= 0) { foreach my $arg (@ARGV) { my $fh = new IO::File $arg, "r"; if (!defined($fh)) { printf "%s: cannot open\n", $arg; } else { &jsstyle($arg, $fh); close $fh; } } } else { &jsstyle("", *STDIN); } exit $err_stat; my $no_errs = 0; # set for JSSTYLED-protected lines sub err($) { my ($error) = @_; unless ($no_errs) { printf $fmt, $filename, $., $error, $line; $err_stat = 1; } } sub err_prefix($$) { my ($prevline, $error) = @_; my $out = $prevline."\n".$line; unless ($no_errs) { printf $fmt, $filename, $., $error, $out; $err_stat = 1; } } sub err_prev($) { my ($error) = @_; unless ($no_errs) { printf $fmt, $filename, $. - 1, $error, $prev; $err_stat = 1; } } sub jsstyle($$) { my ($fn, $filehandle) = @_; $filename = $fn; # share it globally my $in_cpp = 0; my $next_in_cpp = 0; my $in_comment = 0; my $in_header_comment = 0; my $comment_done = 0; my $in_function = 0; my $in_function_header = 0; my $in_declaration = 0; my $note_level = 0; my $nextok = 0; my $nocheck = 0; my $in_string = 0; my ($okmsg, $comment_prefix); $line = ''; $prev = ''; reset_indent(); line: while (<$filehandle>) { s/\r?\n$//; # strip return and newline # save the original line, then remove all text from within # double or single quotes, we do not want to check such text. $line = $_; # # C allows strings to be continued with a backslash at the end of # the line. We translate that into a quoted string on the previous # line followed by an initial quote on the next line. # # (we assume that no-one will use backslash-continuation with character # constants) # $_ = '"' . $_ if ($in_string && !$nocheck && !$in_comment); # # normal strings and characters # s/'([^\\']|\\.)*'/\'\'/g; s/"([^\\"]|\\.)*"/\"\"/g; # # detect string continuation # if ($nocheck || $in_comment) { $in_string = 0; } else { # # Now that all full strings are replaced with "", we check # for unfinished strings continuing onto the next line. # $in_string = (s/([^"](?:"")*)"([^\\"]|\\.)*\\$/$1""/ || s/^("")*"([^\\"]|\\.)*\\$/""/); } # # figure out if we are in a cpp directive # $in_cpp = $next_in_cpp || /^\s*#/; # continued or started $next_in_cpp = $in_cpp && /\\$/; # only if continued # strip off trailing backslashes, which appear in long macros s/\s*\\$//; # an /* END JSSTYLED */ comment ends a no-check block. if ($nocheck) { if (/\/\* *END *JSSTYLED *\*\//) { $nocheck = 0; } else { reset_indent(); next line; } } # a /*JSSTYLED*/ comment indicates that the next line is ok. if ($nextok) { if ($okmsg) { err($okmsg); } $nextok = 0; $okmsg = 0; if (/\/\* *JSSTYLED.*\*\//) { /^.*\/\* *JSSTYLED *(.*) *\*\/.*$/; $okmsg = $1; $nextok = 1; } $no_errs = 1; } elsif ($no_errs) { $no_errs = 0; } # check length of line. # first, a quick check to see if there is any chance of being too long. if ((($line =~ tr/\t/\t/) * ($tab_width - 1)) + length($line) > 80) { # yes, there is a chance. # replace tabs with spaces and check again. my $eline = $line; 1 while $eline =~ s/\t+/' ' x (length($&) * $tab_width - length($`) % $tab_width)/e; if (length($eline) > 80) { err("line > 80 characters"); } } # ignore NOTE(...) annotations (assumes NOTE is on lines by itself). if ($note_level || /\b_?NOTE\s*\(/) { # if in NOTE or this is NOTE s/[^()]//g; # eliminate all non-parens $note_level += s/\(//g - length; # update paren nest level next; } # a /* BEGIN JSSTYLED */ comment starts a no-check block. if (/\/\* *BEGIN *JSSTYLED *\*\//) { $nocheck = 1; } # a /*JSSTYLED*/ comment indicates that the next line is ok. if (/\/\* *JSSTYLED.*\*\//) { /^.*\/\* *JSSTYLED *(.*) *\*\/.*$/; $okmsg = $1; $nextok = 1; } if (/\/\/ *JSSTYLED/) { /^.*\/\/ *JSSTYLED *(.*)$/; $okmsg = $1; $nextok = 1; } # universal checks; apply to everything if (/\t +\t/) { err("spaces between tabs"); } if (/ \t+ /) { err("tabs between spaces"); } if (/\s$/) { err("space or tab at end of line"); } if (/[^ \t(]\/\*/ && !/\w\(\/\*.*\*\/\);/) { err("comment preceded by non-blank"); } # is this the beginning or ending of a function? # (not if "struct foo\n{\n") if (/^{$/ && $prev =~ /\)\s*(const\s*)?(\/\*.*\*\/\s*)?\\?$/) { $in_function = 1; $in_declaration = 1; $in_function_header = 0; $prev = $line; next line; } if (/^}\s*(\/\*.*\*\/\s*)*$/) { if ($prev =~ /^\s*return\s*;/) { err_prev("unneeded return at end of function"); } $in_function = 0; reset_indent(); # we don't check between functions $prev = $line; next line; } if (/^\w*\($/) { $in_function_header = 1; } # a blank line terminates the declarations within a function. # XXX - but still a problem in sub-blocks. if ($in_declaration && /^$/) { $in_declaration = 0; } if ($comment_done) { $in_comment = 0; $in_header_comment = 0; $comment_done = 0; } # does this looks like the start of a block comment? if (/$hdr_comment_start/) { if ($config{"indent"} eq "tab") { if (!/^\t*\/\*/) { err("block comment not indented by tabs"); } } elsif (!/^ *\/\*/) { err("block comment not indented by spaces"); } $in_comment = 1; /^(\s*)\//; $comment_prefix = $1; if ($comment_prefix eq "") { $in_header_comment = 1; } $prev = $line; next line; } # are we still in the block comment? if ($in_comment) { if (/^$comment_prefix \*\/$/) { $comment_done = 1; } elsif (/\*\//) { $comment_done = 1; err("improper block comment close") unless ($ignore_hdr_comment && $in_header_comment); } elsif (!/^$comment_prefix \*[ \t]/ && !/^$comment_prefix \*$/) { err("improper block comment") unless ($ignore_hdr_comment && $in_header_comment); } } if ($in_header_comment && $ignore_hdr_comment) { $prev = $line; next line; } # check for errors that might occur in comments and in code. # allow spaces to be used to draw pictures in header comments. #if (/[^ ] / && !/".* .*"/ && !$in_header_comment) { # err("spaces instead of tabs"); #} #if (/^ / && !/^ \*[ \t\/]/ && !/^ \*$/ && # (!/^ \w/ || $in_function != 0)) { # err("indent by spaces instead of tabs"); #} if ($config{"indent"} eq "tab") { if (/^ {2,}/ && !/^ [^ ]/) { err("indent by spaces instead of tabs"); } } elsif (/^\t/) { err("indent by tabs instead of spaces") } elsif (/^( +)/ && !$in_comment) { my $indent = $1; if (length($indent) < $config{"indent"}) { err("indent of " . length($indent) . " space(s) instead of " . $config{'indent'}); } } if (/^\t+ [^ \t\*]/ || /^\t+ \S/ || /^\t+ \S/) { err("continuation line not indented by 4 spaces"); } # A multi-line block comment must not have content on the first line. if (/^\s*\/\*./ && !/^\s*\/\*.*\*\// && !/$hdr_comment_start/) { err("improper first line of block comment"); } if ($in_comment) { # still in comment, don't do further checks $prev = $line; next line; } if ((/[^(]\/\*\S/ || /^\/\*\S/) && !(/$lint_re/ || ($config{"splint"} && /$splint_re/))) { err("missing blank after open comment"); } if (/\S\*\/[^)]|\S\*\/$/ && !(/$lint_re/ || ($config{"splint"} && /$splint_re/))) { err("missing blank before close comment"); } if ($config{"blank-after-start-comment"} && /(?\s][!<>=]=/ || /[^<>!=][!<>=]==?[^\s,=]/ || (/[^->]>[^,=>\s]/ && !/[^->]>$/) || (/[^<]<[^,=<\s]/ && !/[^<]<$/) || /[^<\s]<[^<]/ || /[^->\s]>[^>]/) { err("missing space around relational operator"); } if (/\S>>=/ || /\S<<=/ || />>=\S/ || /<<=\S/ || /\S[-+*\/&|^%]=/ || (/[^-+*\/&|^%!<>=\s]=[^=]/ && !/[^-+*\/&|^%!<>=\s]=$/) || (/[^!<>=]=[^=\s]/ && !/[^!<>=]=$/)) { # XXX - should only check this for C++ code # XXX - there are probably other forms that should be allowed if (!/\soperator=/) { err("missing space around assignment operator"); } } if (/[,;]\S/ && !/\bfor \(;;\)/ && # Allow a comma in a regex quantifier. !/\/.*?\{\d+,?\d*\}.*?\//) { err("comma or semicolon followed by non-blank"); } # allow "for" statements to have empty "while" clauses if (/\s[,;]/ && !/^[\t]+;$/ && !/^\s*for \([^;]*; ;[^;]*\)/) { err("comma or semicolon preceded by blank"); } if (/^\s*(&&|\|\|)/) { err("improper boolean continuation"); } if (/\S *(&&|\|\|)/ || /(&&|\|\|) *\S/) { err("more than one space around boolean operator"); } if (/\b(delete|typeof|instanceOf|throw|with|catch|new|function|in|for|if|while|switch|return|case)\(/) { err("missing space between keyword and paren"); } if (/(\b(catch|for|if|with|while|switch|return)\b.*){2,}/) { # multiple "case" and "sizeof" allowed err("more than one keyword on line"); } if (/\b(delete|typeof|instanceOf|with|throw|catch|new|function|in|for|if|while|switch|return|case)\s\s+\(/ && !/^#if\s+\(/) { err("extra space between keyword and paren"); } # try to detect "func (x)" but not "if (x)" or # "#define foo (x)" or "int (*func)();" if (/\w\s\(/) { my $s = $_; # strip off all keywords on the line s/\b(delete|typeof|instanceOf|throw|with|catch|new|function|in|for|if|while|switch|return|case)\s\(/XXX(/g; s/#elif\s\(/XXX(/g; s/^#define\s+\w+\s+\(/XXX(/; # do not match things like "void (*f)();" # or "typedef void (func_t)();" s/\w\s\(+\*/XXX(*/g; s/\b(void)\s+\(+/XXX(/og; if (/\w\s\(/) { err("extra space between function name and left paren"); } $_ = $s; } if ($config{"unparenthesized-return"} && /^\s*return\W[^;]*;/ && !/^\s*return\s*\(.*\);/) { err("unparenthesized return expression"); } if (/\btypeof\b/ && !/\btypeof\s*\(.*\)/) { err("unparenthesized typeof expression"); } if (/\(\s/) { err("whitespace after left paren"); } # allow "for" statements to have empty "continue" clauses if (/\s\)/ && !/^\s*for \([^;]*;[^;]*; \)/) { err("whitespace before right paren"); } if (/^\s*\(void\)[^ ]/) { err("missing space after (void) cast"); } if (/\S{/ && !/({|\(){/ && # Allow a brace in a regex quantifier. !/\/.*?\{\d+,?\d*\}.*?\//) { err("missing space before left brace"); } if ($in_function && /^\s+{/ && ($prev =~ /\)\s*$/ || $prev =~ /\bstruct\s+\w+$/)) { err("left brace starting a line"); } if (/}(else|while)/) { err("missing space after right brace"); } if (/}\s\s+(else|while)/) { err("extra space after right brace"); } if (/^\s+#/) { err("preprocessor statement not in column 1"); } if (/^#\s/) { err("blank after preprocessor #"); } # # We completely ignore, for purposes of indentation: # * lines outside of functions # * preprocessor lines # if ($check_continuation && $in_function && !$in_cpp) { process_indent($_); } if (/^\s*else\W/) { if ($prev =~ /^\s*}$/) { err_prefix($prev, "else and right brace should be on same line"); } } $prev = $line; } if ($prev eq "") { err("last line in file is blank"); } } # # Continuation-line checking # # The rest of this file contains the code for the continuation checking # engine. It's a pretty simple state machine which tracks the expression # depth (unmatched '('s and '['s). # # Keep in mind that the argument to process_indent() has already been heavily # processed; all comments have been replaced by control-A, and the contents of # strings and character constants have been elided. # my $cont_in; # currently inside of a continuation my $cont_off; # skipping an initializer or definition my $cont_noerr; # suppress cascading errors my $cont_start; # the line being continued my $cont_base; # the base indentation my $cont_first; # this is the first line of a statement my $cont_multiseg; # this continuation has multiple segments my $cont_special; # this is a C statement (if, for, etc.) my $cont_macro; # this is a macro my $cont_case; # this is a multi-line case my @cont_paren; # the stack of unmatched ( and [s we've seen sub reset_indent() { $cont_in = 0; $cont_off = 0; } sub delabel($) { # # replace labels with tabs. Note that there may be multiple # labels on a line. # local $_ = $_[0]; while (/^(\t*)( *(?:(?:\w+\s*)|(?:case\b[^:]*)): *)(.*)$/) { my ($pre_tabs, $label, $rest) = ($1, $2, $3); $_ = $pre_tabs; while ($label =~ s/^([^\t]*)(\t+)//) { $_ .= "\t" x (length($2) + length($1) / 8); } $_ .= ("\t" x (length($label) / 8)).$rest; } return ($_); } sub process_indent($) { require strict; local $_ = $_[0]; # preserve the global $_ s///g; # No comments s/\s+$//; # Strip trailing whitespace return if (/^$/); # skip empty lines # regexps used below; keywords taking (), macros, and continued cases my $special = '(?:(?:\}\s*)?else\s+)?(?:if|for|while|switch)\b'; my $macro = '[A-Z_][A-Z_0-9]*\('; my $case = 'case\b[^:]*$'; # skip over enumerations, array definitions, initializers, etc. if ($cont_off <= 0 && !/^\s*$special/ && (/(?:(?:\b(?:enum|struct|union)\s*[^\{]*)|(?:\s+=\s*)){/ || (/^\s*{/ && $prev =~ /=\s*(?:\/\*.*\*\/\s*)*$/))) { $cont_in = 0; $cont_off = tr/{/{/ - tr/}/}/; return; } if ($cont_off) { $cont_off += tr/{/{/ - tr/}/}/; return; } if (!$cont_in) { $cont_start = $line; if (/^\t* /) { err("non-continuation indented 4 spaces"); $cont_noerr = 1; # stop reporting } $_ = delabel($_); # replace labels with tabs # check if the statement is complete return if (/^\s*\}?$/); return if (/^\s*\}?\s*else\s*\{?$/); return if (/^\s*do\s*\{?$/); return if (/{$/); return if (/}[,;]?$/); # Allow macros on their own lines return if (/^\s*[A-Z_][A-Z_0-9]*$/); # cases we don't deal with, generally non-kosher if (/{/) { err("stuff after {"); return; } # Get the base line, and set up the state machine /^(\t*)/; $cont_base = $1; $cont_in = 1; @cont_paren = (); $cont_first = 1; $cont_multiseg = 0; # certain things need special processing $cont_special = /^\s*$special/? 1 : 0; $cont_macro = /^\s*$macro/? 1 : 0; $cont_case = /^\s*$case/? 1 : 0; } else { $cont_first = 0; # Strings may be pulled back to an earlier (half-)tabstop unless ($cont_noerr || /^$cont_base / || (/^\t*(?: )?(?:gettext\()?\"/ && !/^$cont_base\t/)) { err_prefix($cont_start, "continuation should be indented 4 spaces"); } } my $rest = $_; # keeps the remainder of the line # # The split matches 0 characters, so that each 'special' character # is processed separately. Parens and brackets are pushed and # popped off the @cont_paren stack. For normal processing, we wait # until a ; or { terminates the statement. "special" processing # (if/for/while/switch) is allowed to stop when the stack empties, # as is macro processing. Case statements are terminated with a : # and an empty paren stack. # foreach $_ (split /[^\(\)\[\]\{\}\;\:]*/) { next if (length($_) == 0); # rest contains the remainder of the line my $rxp = "[^\Q$_\E]*\Q$_\E"; $rest =~ s/^$rxp//; if (/\(/ || /\[/) { push @cont_paren, $_; } elsif (/\)/ || /\]/) { my $cur = $_; tr/\)\]/\(\[/; my $old = (pop @cont_paren); if (!defined($old)) { err("unexpected '$cur'"); $cont_in = 0; last; } elsif ($old ne $_) { err("'$cur' mismatched with '$old'"); $cont_in = 0; last; } # # If the stack is now empty, do special processing # for if/for/while/switch and macro statements. # next if (@cont_paren != 0); if ($cont_special) { if ($rest =~ /^\s*{?$/) { $cont_in = 0; last; } if ($rest =~ /^\s*;$/) { err("empty if/for/while body ". "not on its own line"); $cont_in = 0; last; } if (!$cont_first && $cont_multiseg == 1) { err_prefix($cont_start, "multiple statements continued ". "over multiple lines"); $cont_multiseg = 2; } elsif ($cont_multiseg == 0) { $cont_multiseg = 1; } # We've finished this section, start # processing the next. goto section_ended; } if ($cont_macro) { if ($rest =~ /^$/) { $cont_in = 0; last; } } } elsif (/\;/) { if ($cont_case) { err("unexpected ;"); } elsif (!$cont_special) { err("unexpected ;") if (@cont_paren != 0); if (!$cont_first && $cont_multiseg == 1) { err_prefix($cont_start, "multiple statements continued ". "over multiple lines"); $cont_multiseg = 2; } elsif ($cont_multiseg == 0) { $cont_multiseg = 1; } if ($rest =~ /^$/) { $cont_in = 0; last; } if ($rest =~ /^\s*special/) { err("if/for/while/switch not started ". "on its own line"); } goto section_ended; } } elsif (/\{/) { err("{ while in parens/brackets") if (@cont_paren != 0); err("stuff after {") if ($rest =~ /[^\s}]/); $cont_in = 0; last; } elsif (/\}/) { err("} while in parens/brackets") if (@cont_paren != 0); if (!$cont_special && $rest !~ /^\s*(while|else)\b/) { if ($rest =~ /^$/) { err("unexpected }"); } else { err("stuff after }"); } $cont_in = 0; last; } } elsif (/\:/ && $cont_case && @cont_paren == 0) { err("stuff after multi-line case") if ($rest !~ /$^/); $cont_in = 0; last; } next; section_ended: # End of a statement or if/while/for loop. Reset # cont_special and cont_macro based on the rest of the # line. $cont_special = ($rest =~ /^\s*$special/)? 1 : 0; $cont_macro = ($rest =~ /^\s*$macro/)? 1 : 0; $cont_case = 0; next; } $cont_noerr = 0 if (!$cont_in); }