Commit | Line | Data |
---|---|---|
168e428f PH |
1 | #! /usr/bin/perl |
2 | ||
ddec0cd9 | 3 | # $Cambridge: exim/doc/doc-docbook/Tidytxt,v 1.6 2008/02/04 17:18:18 fanf2 Exp $ |
168e428f | 4 | |
9b371988 PH |
5 | # Script to tidy up the output of w3m when it makes a text file. First we |
6 | # convert sequences of blank lines into a single blank line, to get everything | |
7 | # uniform. Then we go through and insert blank lines before chapter and | |
8 | # sections, also converting chapter titles to uppercase. | |
168e428f | 9 | |
595028e4 PH |
10 | # We also have to do some character translation in the first pass. It seems |
11 | # that xmlto now generates Unicode in its HTML pages. This gives three problems: | |
12 | # (1) It inserts the byte sequence C2 A0 (U+00A0) as a fixed-width space; | |
8473d4ee TF |
13 | # (2) It uses U+25CF as its bullet character. |
14 | # (3) It inserts a whole slew of "box drawing" characters round the heading. | |
595028e4 | 15 | |
9b371988 PH |
16 | @lines = <>; |
17 | ||
18 | $lastwasblank = 0; | |
19 | foreach $line (@lines) | |
168e428f | 20 | { |
8473d4ee | 21 | # (1) non-break space -> normal space |
595028e4 | 22 | $line =~ s/\x{c2}\x{a0}/ /g; |
8473d4ee | 23 | # (2) bullet -> asterisk |
595028e4 | 24 | $line =~ s/\x{e2}\x{97}\x{8f}/*/g; |
8473d4ee TF |
25 | # (3a) horizontal box drawing -> hyphen |
26 | $line =~ s/\x{e2}\x{94}[\x{80}\x{81}\x{84}\x{85}\x{88}\x{89}]/-/g; | |
27 | $line =~ s/\x{e2}\x{95}[\x{8c}\x{8d}\x{90}]/-/g; | |
28 | $line =~ s/\x{e2}\x{95}[\x{b4}\x{b6}\x{b8}\x{ba}\x{bc}\x{be}]/-/g; | |
29 | # (3b) vertical box drawing -> bar | |
30 | $line =~ s/\x{e2}\x{94}[\x{82}\x{83}\x{86}\x{87}\x{8a}\x{8b}]/|/g; | |
ddec0cd9 TF |
31 | $line =~ s/\x{e2}\x{95}[\x{8e}\x{8f}\x{91}]/|/g; |
32 | $line =~ s/\x{e2}\x{95}[\x{b5}\x{b7}\x{b9}\x{bb}\x{bd}\x{bf}]/|/g; | |
8473d4ee TF |
33 | # (3c) corner box drawing -> plus |
34 | $line =~ s/\x{e2}\x{94}[\x{8c}-\x{bf}]/+/g; | |
35 | $line =~ s/\x{e2}\x{95}[\x{80}-\x{8b}\x{92}-\x{b0}]/+/g; | |
36 | # other | |
37 | $line =~ s/\x{e2}\x{95}\x{b1}/\//g; | |
38 | $line =~ s/\x{e2}\x{95}\x{b2}/\\/g; | |
39 | $line =~ s/\x{e2}\x{95}\x{b3}/X/g; | |
595028e4 | 40 | |
9b371988 | 41 | if ($line =~ /^\s*$/) |
168e428f | 42 | { |
9b371988 PH |
43 | $line = "" if $lastwasblank; |
44 | $lastwasblank = 1; | |
168e428f PH |
45 | next; |
46 | } | |
9b371988 PH |
47 | $lastwasblank = 0; |
48 | } | |
49 | ||
50 | # Find start of TOC, uppercasing its title | |
51 | ||
52 | for ($i = 0; $i < scalar @lines; $i++) | |
53 | { | |
54 | $lines[$i] = "TABLE OF CONTENTS\n" if $lines[$i] =~ /^Table of Contents/; | |
595028e4 | 55 | last if $lines[$i] =~ /^1\. /; |
9b371988 PH |
56 | } |
57 | ||
58 | # Find start of first chapter | |
59 | ||
60 | for ($i++; $i < scalar @lines; $i++) | |
595028e4 | 61 | { last if $lines[$i] =~ /^1\. /; } |
9b371988 PH |
62 | |
63 | # Process the body. We can detect the starts of chapters and sections by | |
64 | # looking for preceding and following blank lines, and then matching against | |
65 | # the numbers. | |
66 | ||
67 | $chapter = 0; | |
68 | for (; $i < scalar @lines; $i++) | |
69 | { | |
70 | next if $lines[$i-1] !~ /^$/ || $lines[$i+1] !~ /^$/; | |
71 | ||
72 | # Start of chapter | |
73 | ||
74 | if ($lines[$i] =~ /^(\d+)\. / && $1 == $chapter + 1) | |
75 | { | |
76 | $chapter++; | |
77 | $section = 0; | |
78 | $lines[$i] = "\n\n" . ("=" x 79) . "\n" . uc($lines[$i]); | |
79 | } | |
80 | ||
81 | # Start of next section | |
82 | ||
83 | elsif ($lines[$i] =~ /^(\d+)\.(\d+) / && $1 == $chapter && $2 == $section + 1) | |
84 | { | |
85 | $section++; | |
86 | $lines[$i] = "\n$lines[$i]" . "-" x (length($lines[$i]) - 1) . "\n"; | |
87 | } | |
168e428f PH |
88 | } |
89 | ||
9b371988 PH |
90 | print @lines; |
91 | ||
168e428f | 92 | # End |