Koha-community/Koha
13
7
Fork 0
Code Issues Releases Wiki Activity
Koha/misc/translator/text-extract2.pl

354 lines
12 KiB
Prolog
Executable file
Raw Blame History

#!/usr/bin/perl
# Test filter partially based on Ambrose's hideous subst.pl code
# The idea is that the .tmpl files are not valid HTML, and as a result
# HTML::Parse would be completely confused by these templates.
# This is just a simple scanner (not a parser) & should give better results.
# This script is meant to be a drop-in replacement of text-extract.pl
# A grander plan: Code could be written to detect template variables and
# construct gettext-c-format-string-like meta-strings (e.g., "Results %s
# through %s of %s records" that will be more likely to be translatable
# to languages where word order is very unlike English word order.
# --> This will be relatively major rework, and requires corresponding
# rework in tmpl_process.pl
use Getopt::Long;
use strict;
use vars qw( $input );
use vars qw( $debug_dump_only_p );
use vars qw( $pedantic_p $pedantic_tag );
use vars qw( $pedantic_attribute_error_in_nonpedantic_mode_p );
use vars qw( $pedantic_tmpl_var_use_in_nonpedantic_mode_p );
use vars qw( $fatal_p );
###############################################################################
# Hideous stuff
use vars qw( $re_directive $re_tmpl_var $re_tmpl_var_escaped $re_tmpl_include );
use vars qw( $re_directive_control $re_tmpl_endif_endloop );
BEGIN {
# $re_directive must not do any backreferences
$re_directive = q{<(?:(?i)(?:!--\s*)?\/?TMPL_(?:VAR|LOOP|INCLUDE|IF|ELSE|UNLESS)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
# TMPL_VAR or TMPL_INCLUDE
$re_tmpl_var = q{<(?:(?i)(?:!--\s*)?TMPL_(?:VAR)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
$re_tmpl_include = q{<(?:(?i)(?:!--\s*)?TMPL_(?:INCLUDE)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
# TMPL_VAR ESCAPE=1/HTML/URL
$re_tmpl_var_escaped = q{<(?:(?i)(?:!--\s*)?TMPL_(?:VAR|INCLUDE)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))\s+ESCAPE=(?:1|HTML|URL)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
# Any control flow directive
$re_directive_control = q{<(?:(?i)(?:!--\s*)?\/?TMPL_(?:LOOP|IF|ELSE|UNLESS)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
# /LOOP or /IF or /UNLESS
$re_tmpl_endif_endloop = q{<(?:(?i)(?:!--\s*)?\/TMPL_(?:LOOP|IF|UNLESS)(?:\s+(?:[a-zA-Z][-a-zA-Z0-9]*=)?(?:'[^']*'|"[^"]*"|[^\s<>]+))*\s*(?:--)?)>};
}
# Hideous stuff from subst.pl, slightly modified to use the above hideous stuff
# Note: The $re_tag's set $1 (<tag), $2 (>), and $3 (rest of string)
use vars qw( $re_comment $re_entity_name $re_end_entity $re_etag );
use vars qw( $re_tag_strict $re_tag_compat @re_tag );
sub re_tag ($) {
my($compat) = @_;
my $etag = $compat? '>': '<>\/';
# This is no longer similar to the original regexp in subst.pl :-(
# Note that we don't want <> in compat mode; Mozilla knows about <
q{(<\/?(?:|(?:"(?:} . $re_directive . q{|[^"])*"|'(?:} . $re_directive . q{|[^'])*'|--(?:[^-]|-[^-])*--|(?:}
. $re_directive
. q{|(?!--)[^"'<>} . $etag . q{]))+))([} . $etag . q{]|(?=<))(.*)};
}
BEGIN {
$re_comment = '(?:--(?:[^-]|-[^-])*--)';
$re_entity_name = '(?:[^&%#;<>\s]+)'; # NOTE: not really correct SGML
$re_end_entity = '(?:;|$|(?=\s))'; # semicolon or before-whitespace
$re_etag = q{(?:<\/?(?:"[^"]*"|'[^']*'|[^"'>\/])*[>\/])}; # end-tag
@re_tag = ($re_tag_strict, $re_tag_compat) = (re_tag(0), re_tag(1));
}
# End of the hideous stuff
sub KIND_TEXT () { 'TEXT' }
sub KIND_CDATA () { 'CDATA' }
sub KIND_TAG () { 'TAG' }
sub KIND_DECL () { 'DECL' }
sub KIND_PI () { 'PI' }
sub KIND_DIRECTIVE () { 'HTML::Template' }
sub KIND_COMMENT () { 'COMMENT' } # empty DECL with exactly one SGML comment
sub KIND_UNKNOWN () { 'ERROR' }
use vars qw( $readahead $lc_0 $lc $syntaxerror_p );
use vars qw( $cdata_mode_p $cdata_close );
###############################################################################
sub warn_pedantic ($$) {
my($flag, $msg) = @_;
warn "Warning$pedantic_tag: $msg\n" if $pedantic_p || !$$flag;
if (!$pedantic_p) {
warn "Warning$pedantic_tag: Further similar negligible warnings will not be reported, use --pedantic for details\n" unless $$flag;
$$flag = 1;
}
}
###############################################################################
sub extract_attributes ($;$) {
my($s, $lc) = @_;
my %attr;
$s = $1 if $s =~ /^<\S+(.*)\/\S$/s # XML-style self-closing tags
|| $s =~ /^<\S+(.*)\S$/s; # SGML-style tags
for (my $i = 0; $s =~ /^(?:$re_directive_control)?\s+(?:$re_directive_control)?(?:([a-zA-Z][-a-zA-Z0-9]*)\s*=\s*)?('((?:$re_directive|[^'])*)'|"((?:$re_directive|[^"])*)"|((?:$re_directive|[^\s<>])+))/os;) {
my($key, $val, $val_orig, $rest)
= ($1, (defined $3? $3: defined $4? $4: $5), $2, $');
$i += 1;
$attr{+lc($key)} = [$key, $val, $val_orig, $i];
$s = $rest;
if ($val =~ /$re_tmpl_include/os) {
warn "Warning: TMPL_INCLUDE in attribute"
. (defined $lc? " near line $lc": '') . ": $val_orig\n";
} elsif ($val =~ /$re_tmpl_var/os && $val !~ /$re_tmpl_var_escaped/os) {
warn_pedantic \$pedantic_tmpl_var_use_in_nonpedantic_mode_p,
"Unescaped TMPL_VAR in attribute"
. (defined $lc? " near line $lc": '') . ": $val_orig"
if $pedantic_p || !$pedantic_tmpl_var_use_in_nonpedantic_mode_p;
} elsif ($val_orig !~ /^['"]/) {
warn_pedantic \$pedantic_attribute_error_in_nonpedantic_mode_p,
"Unquoted attribute contains character(s) that should be quoted"
. (defined $lc? " near line $lc": '') . ": $val_orig"
if $val =~ /[^-\.A-Za-z0-9]/s;
}
}
my $s2 = $s; $s2 =~ s/$re_tmpl_endif_endloop//g; # for the next check
if ($s2 =~ /\S/s) { # should never happen
if ($s =~ /^([^\n]*)\n/s) { # this is even worse
warn "Error: Completely confused while extracting attributes"
. (defined $lc? " near line $lc": '') . ": $1\n";
warn "Error: " . (scalar split(/\n/, $s) - 1) . " more line(s) not shown.\n";
$fatal_p = 1;
} else {
warn "Warning: Strange attribute syntax"
. (defined $lc? " near line $lc": '') . ": $s\n";
}
}
return \%attr;
}
sub next_token_internal (*) {
my($h) = @_;
my($it, $kind);
my $eof_p = 0;
if (!defined $readahead || !length $readahead) {
my $next = scalar <$h>;
$eof_p = !defined $next;
if (!$eof_p) {
$lc += 1;
$readahead .= $next;
}
}
$lc_0 = $lc; # remember line number of first line
if ($eof_p && !length $readahead) { # nothing left to do
;
} elsif ($readahead =~ /^\s+/s) { # whitespace
($kind, $it, $readahead) = (KIND_TEXT, $&, $');
# FIXME the following (the [<\s] part) is an unreliable HACK :-(
} elsif ($readahead =~ /^(?:[^<]|<[<\s])+/s) { # non-space normal text
($kind, $it, $readahead) = (KIND_TEXT, $&, $');
warn "Warning: Unescaped < near line $lc_0: $it\n"
if !$cdata_mode_p && $it =~ /</s;
} else { # tag/declaration/processing instruction
my $ok_p = 0;
for (;;) {
if ($cdata_mode_p) {
if ($readahead =~ /^$cdata_close/) {
($kind, $it, $readahead) = (KIND_TAG, $&, $');
$ok_p = 1;
} else {
($kind, $it, $readahead) = (KIND_TEXT, $readahead, undef);
$ok_p = 1;
}
} elsif ($readahead =~ /^$re_tag_compat/os) {
($kind, $it, $readahead) = (KIND_TAG, "$1>", $3);
$ok_p = 1;
warn "Warning: SGML \"closed start tag\" notation near line $lc_0: $1<\n" if $2 eq '';
} elsif ($readahead =~ /^<!--(?:(?!-->).)*-->/s) {
($kind, $it, $readahead) = (KIND_COMMENT, $&, $');
$ok_p = 1;
warn "Warning: Syntax error in comment at line $lc_0: $&\n";
$syntaxerror_p = 1;
}
last if $ok_p;
my $next = scalar <$h>;
$eof_p = !defined $next;
last if $eof_p;
$lc += 1;
$readahead .= $next;
}
if ($kind ne KIND_TAG) {
;
} elsif ($it =~ /^<!/) {
$kind = KIND_DECL;
$kind = KIND_COMMENT if $it =~ /^<!--(?:(?!-->).)*-->/;
} elsif ($it =~ /^<\?/) {
$kind = KIND_PI;
}
if ($it =~ /^$re_directive/ios && !$cdata_mode_p) {
$kind = KIND_DIRECTIVE;
}
if (!$ok_p && $eof_p) {
($kind, $it, $readahead) = (KIND_UNKNOWN, $readahead, undef);
$syntaxerror_p = 1;
}
}
warn "Warning: Unrecognizable token found near line $lc_0: $it\n"
if $kind eq KIND_UNKNOWN;
return defined $it? (wantarray? ($kind, $it):
[$kind, $it]): undef;
}
sub next_token (*) {
my($h) = @_;
my $it;
if (!$cdata_mode_p) {
$it = next_token_internal($h);
if (defined $it && $it->[0] eq KIND_TAG) { # FIXME
($cdata_mode_p, $cdata_close) = (1, "</$1\\s*>")
if $it->[1] =~ /^<(script|style|textarea)\b/i; #FIXME
push @$it, extract_attributes($it->[1], $lc_0); #FIXME
}
} else {
for ($it = '';;) {
my $lc_prev = $lc;
my $next = next_token_internal($h);
last if !defined $next;
if (defined $next && $next->[1] =~ /$cdata_close/i) { #FIXME
($lc, $readahead) = ($lc_prev, $next->[1] . $readahead); #FIXME
$cdata_mode_p = 0;
}
last unless $cdata_mode_p;
$it .= $next->[1]; #FIXME
}
$it = [KIND_CDATA, $it]; #FIXME
$cdata_close = undef;
}
return defined $it? (wantarray? @$it: $it): undef;
}
###############################################################################
sub debug_dump (*) { # for testing only
my($h) = @_;
print "re_tag_compat is /$re_tag_compat/\n";
for (;;) {
my $s = next_token $h;
last unless defined $s;
printf "%s\n", ('-' x 79);
my($kind, $t, $attr) = @$s; # FIXME
printf "%s:\n", $kind;
printf "%4dH%s\n", length($t),
join('', map {/[\0-\37]/? $_: "$_\b$_"} split(//, $t));
if ($kind eq KIND_TAG && %$attr) {
printf "Attributes:\n";
for my $a (keys %$attr) {
my($key, $val, $val_orig, $order) = @{$attr->{$a}};
printf "%s = %dH%s -- %s\n", $a, length $val,
join('', map {/[\0-\37]/? $_: "$_\b$_"} split(//, $val)),
$val_orig;
}
}
}
}
###############################################################################
sub trim ($) {
my($s) = @_;
$s =~ s/^(?:\s|\&nbsp$re_end_entity)+//os;
$s =~ s/(?:\s|\&nbsp$re_end_entity)+$//os;
return $s;
}
###############################################################################
sub text_extract (*) {
my($h) = @_;
my %text = ();
for (;;) {
my $s = next_token $h;
last unless defined $s;
my($kind, $t, $attr) = @$s; # FIXME
if ($kind eq KIND_TEXT) {
$t = trim $t;
$text{$t} = 1 if $t =~ /\S/s;
} elsif ($kind eq KIND_TAG && %$attr) {
# value [tag=input], meta
my $tag = lc($1) if $t =~ /^<(\S+)/s;
for my $a ('alt', 'content', 'title', 'value') {
if ($attr->{$a}) {
next if $a eq 'content' && $tag ne 'meta';
next if $a eq 'value' && ($tag ne 'input'
|| (ref $attr->{'type'} && $attr->{'type'}->[1] eq 'hidden')); # FIXME
my($key, $val, $val_orig, $order) = @{$attr->{$a}}; #FIXME
$val = trim $val;
$text{$val} = 1 if $val =~ /\S/s;
}
}
}
}
# Emit all extracted strings.
# Don't emit pure whitespace, pure numbers, or TMPL_VAR's.
for my $t (keys %text) {
printf "%s\n", $t
unless $t =~ /^(?:\s|\&nbsp$re_end_entity|$re_tmpl_var)*$/os || $t =~ /^\d+$/;
}
}
###############################################################################
sub usage ($) {
my($exitcode) = @_;
my $h = $exitcode? *STDERR: *STDOUT;
print $h <<EOF;
Usage: $0 [OPTIONS]
Extract strings from HTML file.
--debug-dump-only Do not extract strings; but display scanned tokens
-f, --file=FILE Extract from the specified FILE
--pedantic-warnings Issue warnings even for detected problems which
are likely to be harmless
--help Display this help and exit
EOF
exit($exitcode);
}
###############################################################################
sub usage_error (;$) {
print STDERR "$_[0]\n" if @_;
print STDERR "Try `$0 --help' for more information.\n";
exit(-1);
}
###############################################################################
GetOptions(
'f|file=s' => \$input,
'debug-dump-only' => \$debug_dump_only_p,
'pedantic-warnings' => sub { $pedantic_p = 1 },
'help' => sub { usage(0) },
) || usage_error;
$pedantic_tag = $pedantic_p? '': ' (negligible)';
usage_error('Missing mandatory option -f') unless defined $input;
open(INPUT, "<$input") || die "$0: $input: $!\n";
if ($debug_dump_only_p) {
debug_dump(*INPUT);
} else {
text_extract(*INPUT);
}
warn "Warning: This input will not work with Mozilla standards-compliant mode\n"
if $syntaxerror_p;
close INPUT;
exit(-1) if $fatal_p;
View git blame Copy permalink