1#! /usr/bin/perl -w
2
3# Script to turn PCRE2 man pages into HTML
4
5
6# Subroutine to handle font changes and other escapes
7
8sub do_line {
9my($s) = $_[0];
10
11$s =~ s/</&#60;/g;                   # Deal with < and >
12$s =~ s/>/&#62;/g;
13$s =~ s"\\fI(.*?)\\f[RP]"<i>$1</i>"g;
14$s =~ s"\\fB(.*?)\\f[RP]"<b>$1</b>"g;
15$s =~ s"\\e"\\"g;
16$s =~ s/(?<=Copyright )\(c\)/&copy;/g;
17$s;
18}
19
20# Subroutine to ensure not in a paragraph
21
22sub end_para {
23if ($inpara)
24  {
25  print TEMP "</PRE>\n" if ($inpre);
26  print TEMP "</P>\n";
27  }
28$inpara = $inpre = 0;
29$wrotetext = 0;
30}
31
32# Subroutine to start a new paragraph
33
34sub new_para {
35&end_para();
36print TEMP "<P>\n";
37$inpara = 1;
38}
39
40
41# Main program
42
43$innf = 0;
44$inpara = 0;
45$inpre = 0;
46$wrotetext = 0;
47$toc = 0;
48$ref = 1;
49
50while ($#ARGV >= 0 && $ARGV[0] =~ /^-/)
51  {
52  $toc = 1 if $ARGV[0] eq "-toc";
53  shift;
54  }
55
56# Initial output to STDOUT
57
58print <<End ;
59<html>
60<head>
61<title>$ARGV[0] specification</title>
62</head>
63<body bgcolor="#FFFFFF" text="#00005A" link="#0066FF" alink="#3399FF" vlink="#2222BB">
64<h1>$ARGV[0] man page</h1>
65<p>
66Return to the <a href="index.html">PCRE2 index page</a>.
67</p>
68<p>
69This page is part of the PCRE2 HTML documentation. It was generated
70automatically from the original man page. If there is any nonsense in it,
71please consult the man page, in case the conversion went wrong.
72<br>
73End
74
75print "<ul>\n" if ($toc);
76
77open(TEMP, ">/tmp/$$") || die "Can't open /tmp/$$ for output\n";
78
79while (<STDIN>)
80  {
81  # Handle lines beginning with a dot
82
83  if (/^\./)
84    {
85    # Some of the PCRE2 man pages used to contain instances of .br. However,
86    # they should have all been removed because they cause trouble in some
87    # (other) automated systems that translate man pages to HTML. Complain if
88    # we find .br or .in (another macro that is deprecated).
89
90    if (/^\.br/ || /^\.in/)
91      {
92      print STDERR "\n*** Deprecated macro encountered - rewrite needed\n";
93      print STDERR "*** $_\n";
94      die "*** Processing abandoned\n";
95      }
96
97    # Instead of .br, relevent "literal" sections are enclosed in .nf/.fi.
98
99    elsif (/^\.nf/)
100      {
101      $innf = 1;
102      }
103
104    elsif (/^\.fi/)
105      {
106      $innf = 0;
107      }
108
109    # Handling .sp is subtle. If it is inside a literal section, do nothing if
110    # the next line is a non literal text line; similarly, if not inside a
111    # literal section, do nothing if a literal follows, unless we are inside
112    # a .nf/.fi section or about to enter one. The point being that the <pre>
113    # and </pre> that delimit literal sections will do the spacing. Always skip
114    # if no previous output.
115
116    elsif (/^\.sp/)
117      {
118      if ($wrotetext)
119        {
120        $_ = <STDIN>;
121        if ($inpre)
122          {
123          print TEMP "\n" if (/^[\s.]/);
124          }
125        else
126          {
127          print TEMP "<br>\n<br>\n" if ($innf || /^\.nf/ || !/^[\s.]/);
128          }
129        redo;    # Now process the lookahead line we just read
130        }
131      }
132    elsif (/^\.TP/ || /^\.PP/ || /^\.P/)
133      {
134      &new_para();
135      }
136    elsif (/^\.SH\s*("?)(.*)\1/)
137      {
138      # Ignore the NAME section
139      if ($2 =~ /^NAME\b/)
140        {
141        <STDIN>;
142        next;
143        }
144
145      &end_para();
146      my($title) = &do_line($2);
147      if ($toc)
148        {
149        printf("<li><a name=\"TOC%d\" href=\"#SEC%d\">$title</a>\n",
150          $ref, $ref);
151        printf TEMP ("<br><a name=\"SEC%d\" href=\"#TOC1\">$title</a><br>\n",
152          $ref);
153        $ref++;
154        }
155      else
156        {
157        print TEMP "<br><b>\n$title\n</b><br>\n";
158        }
159      }
160    elsif (/^\.SS\s*("?)(.*)\1/)
161      {
162      &end_para();
163      my($title) = &do_line($2);
164      print TEMP "<br><b>\n$title\n</b><br>\n";
165      }
166    elsif (/^\.B\s*(.*)/)
167      {
168      &new_para() if (!$inpara);
169      $_ = &do_line($1);
170      s/"(.*?)"/$1/g;
171      print TEMP "<b>$_</b>\n";
172      $wrotetext = 1;
173      }
174    elsif (/^\.I\s*(.*)/)
175      {
176      &new_para() if (!$inpara);
177      $_ = &do_line($1);
178      s/"(.*?)"/$1/g;
179      print TEMP "<i>$_</i>\n";
180      $wrotetext = 1;
181      }
182
183    # A comment that starts "HREF" takes the next line as a name that
184    # is turned into a hyperlink, using the text given, which might be
185    # in a special font. If it ends in () or (digits) or punctuation, they
186    # aren't part of the link.
187
188    elsif (/^\.\\"\s*HREF/)
189      {
190      $_=<STDIN>;
191      chomp;
192      $_ = &do_line($_);
193      $_ =~ s/\s+$//;
194      $_ =~ /^(?:<.>)?([^<(]+)(?:\(\))?(?:<\/.>)?(?:\(\d+\))?[.,;:]?$/;
195      print TEMP "<a href=\"$1.html\">$_</a>\n";
196      }
197
198    # A comment that starts "HTML" inserts literal HTML
199
200    elsif (/^\.\\"\s*HTML\s*(.*)/)
201      {
202      print TEMP $1;
203      }
204
205    # A comment that starts < inserts that HTML at the end of the
206    # *next* input line - so as not to get a newline between them.
207
208    elsif (/^\.\\"\s*(<.*>)/)
209      {
210      my($markup) = $1;
211      $_=<STDIN>;
212      chomp;
213      $_ = &do_line($_);
214      $_ =~ s/\s+$//;
215      print TEMP "$_$markup\n";
216      }
217
218    # A comment that starts JOIN joins the next two lines together, with one
219    # space between them. Then that line is processed. This is used in some
220    # displays where two lines are needed for the "man" version. JOINSH works
221    # the same, except that it assumes this is a shell command, so removes
222    # continuation backslashes.
223
224    elsif (/^\.\\"\s*JOIN(SH)?/)
225      {
226      my($one,$two);
227      $one = <STDIN>;
228      $two = <STDIN>;
229      $one =~ s/\s*\\e\s*$// if (defined($1));
230      chomp($one);
231      $two =~ s/^\s+//;
232      $_ = "$one $two";
233      redo;            # Process the joined lines
234      }
235
236    # .EX/.EE are used in the pcre2demo page to bracket the entire program,
237    # which is unmodified except for turning backslash into "\e".
238
239    elsif (/^\.EX\s*$/)
240      {
241      print TEMP "<PRE>\n";
242      while (<STDIN>)
243        {
244        last if /^\.EE\s*$/;
245        s/\\e/\\/g;
246        s/&/&amp;/g;
247        s/</&lt;/g;
248        s/>/&gt;/g;
249        print TEMP;
250        }
251      }
252
253    # Ignore anything not recognized
254
255    next;
256    }
257
258  # Line does not begin with a dot. Replace blank lines with new paragraphs
259
260  if (/^\s*$/)
261    {
262    &end_para() if ($wrotetext);
263    next;
264    }
265
266  # Convert fonts changes and output an ordinary line. Ensure that indented
267  # lines are marked as literal.
268
269  $_ = &do_line($_);
270  &new_para() if (!$inpara);
271
272  if (/^\s/)
273    {
274    if (!$inpre)
275      {
276      print TEMP "<pre>\n";
277      $inpre = 1;
278      }
279    }
280  elsif ($inpre)
281    {
282    print TEMP "</pre>\n";
283    $inpre = 0;
284    }
285
286  # Add <br> to the end of a non-literal line if we are within .nf/.fi
287
288  $_ .= "<br>\n" if (!$inpre && $innf);
289
290  print TEMP;
291  $wrotetext = 1;
292  }
293
294# The TOC, if present, will have been written - terminate it
295
296print "</ul>\n" if ($toc);
297
298# Copy the remainder to the standard output
299
300close(TEMP);
301open(TEMP, "/tmp/$$") || die "Can't open /tmp/$$ for input\n";
302
303print while (<TEMP>);
304
305print <<End ;
306<p>
307Return to the <a href="index.html">PCRE2 index page</a>.
308</p>
309End
310
311close(TEMP);
312unlink("/tmp/$$");
313
314# End
315