#! /usr/bin/perl -w # $Cambridge: exim/doc/doc-scripts/fc2k,v 1.1 2004/10/07 15:04:35 ph10 Exp $ # Script to read the HTML table of contents for the Exim FAQ and create an # HTML KWIC index out of it. ######################################################################## # List of words to ignore - kept alphabetically for reference, but they # don't have to be in order. $ignore_list = " a ability able about address addresses addressed affect affected after against all allow allowed allows already also although always am an and and/or any anybody anyone anything anywhere are aren't arrange arrive as at back bad based basically be because been behave behaviour being best between bob both bug build builds built busy but by call called calls can can't cannot causes causing central certain code comes coming command commands complain complaining complains configure configured conjunction contact contain contains contained correct correctly could currently customer day days defined deliver delivers delivered delivery deliveries did do does doesn't doing don't down during e-mail e-mails each easy else email emails entirely entries entry especially etc even ever every example exim exim's experiencing far few file files find fine fly following for form found from fully get gets getting given gives giving go goes going got handle handles handled handling happen happens has have haven't having helpful him host hosts how however i i'd i'm i've if in indeed instead into is issue issues isn't it it's its jim just keep keeps know knows like line lines look looked looking lot machine machines machine's mail mails main make me mean means message messages might more must my myself near need neither no nor not now occur of off often ok on one only or other our out over own part parts particular per place possibility possible present problem problems put puts quite raised rather really reason rid right round run runs same say saying see seeing seem seems seen sees set setting she should so some somehow something sometimes stand state statement still strange such supposed system systems take takes than that the their them then there these they things think this those to try though to/for told too tried tries trying under until up use uses used using usually valid value values via want wanted wanting was way we we've well what what's when where whereabouts whenever whether which while who whose why will with within without wish won't wondered work worked working works would wrong xxx yet yyy "; ######################################################################## # The regular expression fragment that defines the separator between words $wordgap = "(?:[]().?,;:\"']|(?><[^>]*>))*(?:\\s+|\$)(?:[[(\"'`]|(?><[^>]*>))*"; ######################################################################## # Function to add to a length to accommodate HTML stuff sub setlen{ my($len, $s) = @_; $len += length($1) while ($s =~ /(<\/?[a-z]+>)/ig); $len += 1 while ($s =~ /&#\d+;/g); return $len; } ######################################################################## # Function to write out the list of initials with references sub write_initials { my($this_initial) = "$_[0]"; print OUT "

\n  "; foreach $initial (sort keys %initials) { if ($initial eq $this_initial) { print OUT "   $initial "; } else { print OUT "  $initial"; } } print OUT " "x4 . "FAQ Contents\n

\n"; } ######################################################################## # The main program. We can pick out the contents lines because they lie # between
  • and
  • in the file, sometimes on more than one physical # line. # Turn the list of ignorable words into a hash for quick lookup. Add the # empty word to the list. @words = split /\s+/, $ignore_list; foreach $word (@words) { $ignore{$word} = 1; } $ignore{""} = 1; # Open the file and do the job open(IN, "html/FAQ.html") || die "Can't open html/FAQ.html\n"; while () { next unless /^
  • /; $_ .= while !/<\/li>$/; chomp; s/\n\s*/ /g; # Extract the operative text into $text, with the beginning in $pre. my($pre,$text,$post) = /^
  • (.*<\/a>:(?: )*)(.*)

    <\/li>$/; # Now split into words. As well as punctuation, there may be HTML thingies # between words. Absorb them into the separators. my(@words) = split /$wordgap/, $text; # Lower case all the words, and remove those that we don't want. # Then keep a list of all the used initials. REMOVE_IGNORE: for ($i = 0; $i < scalar @words; $i++) { my($word) = $words[$i] = "\L$words[$i]\E"; # Remove certain forms of word and those on the ignore list if (defined $ignore{$word} || # word on ignore list $word =~ /^-+$/ || # word consists entirely of hyphens $word =~ /^-[^a-z]/ || # follows leading hyphen with non-letter $word =~ /^[^a-z-]/ || # starts with a non-letter or hyphen $word =~ /[@^.]/ # contains @ or ^ or . ) { splice(@words, $i, 1); redo REMOVE_IGNORE if $i < scalar @words; } # Otherwise, build up a list of initials else { my($inword) = $word; $inword =~ s/^-//; $initial = substr($inword, 0, 1); $initials{"\U$initial\E"} = 1; } } # Create the lines for the KWIC index, and store them in associative # arrays, with the keyword as the key. That will get them sorted # automatically. while (scalar @words > 0) { my($word) = shift @words; my($pretext, $casedword, $posttext) = $text =~ /(.*?)(? $leftlen) { my($cutoff) = $leftlen; $cutoff++ while ($cutoff < $prelen && substr($pretext, -$cutoff, 1) ne " "); $pretext = "... " . substr($pretext, -$cutoff); } if ($postlen > $rightlen) { my($cutoff) = $rightlen; $cutoff++ while ($cutoff < $postlen && substr($posttext, $cutoff, 1) ne " "); $posttext = substr($posttext, 0, $cutoff) . "..."; } # If the pre text has a font-ending not preceded by a font beginning # (i.e. we've chopped the beginning off), we must insert a beginning. while ($pretext =~ /^(.*?)<\/(small|tt|b|i)>/ && $1 !~ /<$2>/) { $pretext = "<$2>" . $pretext; } # If the pre text ends in a special font, we have to terminate that, # and reset it at the start of the post text. my($poststart) = ""; while ($pretext =~ /<(small|tt|b|i)>(?!.*?<\/\1>)/) { $pretext .= ""; $poststart .= "<$1>"; } # If the post text changes font but doesn't close it, we must add # the closure. while ($posttext =~ /<(small|tt|b|i)>(?!.*?<\/\1>)/) { $posttext .= ""; } # Remove any unnecessary changes in either of them $pretext =~ s/<(small|tt|b|i)>\s*<\/\1>//g; $posttext =~ s/<(small|tt|b|i)>\s*<\/\1>//g; # Save the texts in associative arrays. Add the question number to # the end of the word to make the key. $pre =~ /(Q\d\d\d\d)/; my($key) = "$word-$1"; $tableft{$key} = $pre . $pretext; $tabright{$key} = $poststart . "$casedword" . $posttext; } } close(IN); # Now write out the files. Each letter in the index goes in a different file $current_initial = ""; foreach $key (sort keys %tableft) { my($initial) = $key =~ /^(.)/; $initial = "\U$initial\E"; if ($initial ne $current_initial) { if ($current_initial ne "") { print OUT "\n"; &write_initials($current_initial); print OUT "\n\n"; close OUT; } open (OUT, ">html/FAQ-KWIC_$initial.html") || die "Can't open html/FAQ-KWIC_$initial.html\n"; print OUT "\n" . "\n" . "Exim FAQ: KWIC index section $initial\n" . "\n" . "\n" . "

    Exim FAQ: Keyword-in-context index

    \n"; write_initials($initial); if ($initial eq "A") { print OUT < This Keyword-in-context index for the Exim FAQ is generated automatically from the FAQ source. Browsers may not display the data very prettily, but it is hoped that it may provide a useful aid for finding things in the FAQ.

    End } print OUT "\n"; $current_initial = $initial; } print OUT "\n"; print OUT "\n"; print OUT "\n"; print OUT "\n"; } # Close the final file if ($current_initial ne "") { print OUT "
    $tableft{$key}$tabright{$key}
    \n"; &write_initials($current_initial); print OUT "\n\n"; close OUT; } # End