8 # For now we can't rely on a perl >= 5.14 on
9 # the build sites, thus we throw away all unicode
10 # awareness and do the matching byte by byte
15 'u|utf8!' => \my $want_utf8, # do not replace unicode characters
16 ) or die "Usage: $0 [-u|--utf8]\n";
18 # Script to tidy up the output of w3m when it makes a text file. First we
19 # convert sequences of blank lines into a single blank line, to get everything
20 # uniform. Then we go through and insert blank lines before chapter and
21 # sections, also converting chapter titles to uppercase.
23 # We also have to do some character translation in the first pass. It seems
24 # that xmlto now generates Unicode in its HTML pages. This gives three problems:
25 # (1) It inserts the byte sequence C2 A0 (U+00A0) as a fixed-width space;
26 # (2) It uses U+25CF as its bullet character.
27 # (3) It inserts a whole slew of "box drawing" characters round the heading.
32 foreach my $line (@lines)
34 # (1) non-break space -> normal space
35 $line =~ s/\x{c2}\x{a0}/ /g;
39 # (2) bullet -> asterisk
40 $line =~ s/\x{e2}\x{97}\x{8f}/*/g;
41 $line =~ s/\x{e2}\x{80}\x{a2}/*/g; # OpenSUSE
42 $line =~ s/\x{e2}\x{96}\x{a1}/*/g; # OpenSUSE
43 # (3a) horizontal box drawing -> hyphen
44 $line =~ s/\x{e2}\x{94}[\x{80}\x{81}\x{84}\x{85}\x{88}\x{89}]/-/g;
45 $line =~ s/\x{e2}\x{95}[\x{8c}\x{8d}\x{90}]/-/g;
46 $line =~ s/\x{e2}\x{95}[\x{b4}\x{b6}\x{b8}\x{ba}\x{bc}\x{be}]/-/g;
47 # (3b) vertical box drawing -> bar
48 $line =~ s/\x{e2}\x{94}[\x{82}\x{83}\x{86}\x{87}\x{8a}\x{8b}]/|/g;
49 $line =~ s/\x{e2}\x{95}[\x{8e}\x{8f}\x{91}]/|/g;
50 $line =~ s/\x{e2}\x{95}[\x{b5}\x{b7}\x{b9}\x{bb}\x{bd}\x{bf}]/|/g;
51 # (3c) corner box drawing -> plus
52 $line =~ s/\x{e2}\x{94}[\x{8c}-\x{bf}]/+/g;
53 $line =~ s/\x{e2}\x{95}[\x{80}-\x{8b}\x{92}-\x{b0}]/+/g;
55 $line =~ s/\x{e2}\x{95}\x{b1}/\//g;
56 $line =~ s/\x{e2}\x{95}\x{b2}/\\/g;
57 $line =~ s/\x{e2}\x{95}\x{b3}/X/g;
60 # w3m rendering issue apparently only seen by pdp
61 # affects section numbers after the ToC, some info on spool-file -lines, etc
62 # always appears to be a spurious extra character, safely just dropped.
67 $line = "" if $lastwasblank;
74 # Find start of TOC, uppercasing its title
77 for ($i = 0; $i < scalar @lines; $i++)
79 $lines[$i] = "TABLE OF CONTENTS\n" if $lines[$i] =~ /^Table of Contents/;
80 last if $lines[$i] =~ /^1\. /;
83 # Find start of first chapter
85 for ($i++; $i < scalar @lines; $i++)
86 { last if $lines[$i] =~ /^1\. /; }
88 # Process the body. We can detect the starts of chapters and sections by
89 # looking for preceding and following blank lines, and then matching against
94 for (; $i < scalar @lines; $i++)
96 next if $lines[$i-1] !~ /^$/ || $lines[$i+1] !~ /^$/;
100 if ($lines[$i] =~ /^(\d+)\. / && $1 == $chapter + 1)
104 $lines[$i] = "\n\n" . ("=" x 79) . "\n" . uc($lines[$i]);
107 # Start of next section
109 elsif ($lines[$i] =~ /^(\d+)\.(\d+) / && $1 == $chapter && $2 == $section + 1)
112 $lines[$i] = "\n$lines[$i]" . "-" x (length($lines[$i]) - 1) . "\n";