Contents of /code/trunk/CleanTxt

#! /usr/bin/perl -w

# Script to take the output of nroff -man and remove all the backspacing and
# the page footers and the screen commands etc so that it is more usefully
# readable online. In fact, in the latest nroff, intermediate footers don't
# seem to be generated any more.

$blankcount = 0;
$lastwascut = 0;
$firstheader = 1;

# Input on STDIN; output to STDOUT.

while (<STDIN>)
  {
  s/\x1b\[\d+m//g;   # Remove screen controls "ESC [ number m"
  s/.\x8//g;         # Remove "char, backspace"

  # Handle header lines. Retain only the first one we encounter, but remove
  # the blank line that follows. Any others (e.g. at end of document) and the
  # following blank line are dropped.

  if (/^PCRE(\w*)\(([13])\)\s+PCRE\1\(\2\)$/)
    {
    if ($firstheader)
      {
      $firstheader = 0;
      print;
      $lastprinted = $_;
      $lastwascut = 0;
      }
    $_=<STDIN>;       # Remove a blank that follows
    next;
    }

  # Count runs of empty lines

  if (/^\s*$/)
    {
    $blankcount++;
    $lastwascut = 0;
    next;
    }

  # If a chunk of lines has been cut out (page footer) and the next line
  # has a different indentation, put back one blank line.

  if ($lastwascut && $blankcount < 1 && defined($lastprinted))
    {
    ($a) = $lastprinted =~ /^(\s*)/;
    ($b) = $_ =~ /^(\s*)/;
    $blankcount++ if ($a ne $b);
    }

  # We get here only when we have a non-blank line in hand. If it was preceded
  # by 3 or more blank lines, read the next 3 lines and see if they are blank.
  # If so, remove all 7 lines, and remember that we have just done a cut.

  if ($blankcount >= 3)
    {
    for ($i = 0; $i < 3; $i++)
      {
      $next[$i] = <STDIN>;
      $next[$i] = "" if !defined $next[$i];
      $next[$i] =~ s/\x1b\[\d+m//g;   # Remove screen controls "ESC [ number m"
      $next[$i] =~ s/.\x8//g;         # Remove "char, backspace"
      }

    # Cut out chunks of the form <3 blanks><non-blank><3 blanks>

    if ($next[0] =~ /^\s*$/ &&
        $next[1] =~ /^\s*$/ &&
        $next[2] =~ /^\s*$/)
      {
      $blankcount -= 3;
      $lastwascut = 1;
      }

    # Otherwise output the saved blanks, the current, and the next three
    # lines. Remember the last printed line.

    else
      {
      for ($i = 0; $i < $blankcount; $i++) { print "\n"; }
      print;
      for ($i = 0; $i < 3; $i++)
        {
        $next[$i] =~ s/.\x8//g;
        print $next[$i];
        $lastprinted = $_;
        }
      $lastwascut = 0;
      $blankcount = 0;
      }
    }

  # This non-blank line is not preceded by 3 or more blank lines. Output
  # any blanks there are, and the line. Remember it. Force two blank lines
  # before headings.

  else
    {
    $blankcount = 2 if /^\S/ && !/^Last updated/ && !/^Copyright/ &&
      defined($lastprinted);
    for ($i = 0; $i < $blankcount; $i++) { print "\n"; }
    print;
    $lastprinted = $_;
    $lastwascut = 0;
    $blankcount = 0;
    }
  }

# End
1	ph10	97	#! /usr/bin/perl -w
2
3			# Script to take the output of nroff -man and remove all the backspacing and
4			# the page footers and the screen commands etc so that it is more usefully
5			# readable online. In fact, in the latest nroff, intermediate footers don't
6			# seem to be generated any more.
7
8			$blankcount = 0;
9			$lastwascut = 0;
10			$firstheader = 1;
11
12			# Input on STDIN; output to STDOUT.
13
14			while (<STDIN>)
15			{
16			s/\x1b\[\d+m//g; # Remove screen controls "ESC [ number m"
17			s/.\x8//g; # Remove "char, backspace"
18
19			# Handle header lines. Retain only the first one we encounter, but remove
20			# the blank line that follows. Any others (e.g. at end of document) and the
21			# following blank line are dropped.
22
23			if (/^PCRE(\w*)\(([13])\)\s+PCRE\1\(\2\)$/)
24			{
25			if ($firstheader)
26			{
27			$firstheader = 0;
28			print;
29			$lastprinted = $_;
30			$lastwascut = 0;
31			}
32			$_=<STDIN>; # Remove a blank that follows
33			next;
34			}
35
36			# Count runs of empty lines
37
38			if (/^\s*$/)
39			{
40			$blankcount++;
41			$lastwascut = 0;
42			next;
43			}
44
45			# If a chunk of lines has been cut out (page footer) and the next line
46			# has a different indentation, put back one blank line.
47
48			if ($lastwascut && $blankcount < 1 && defined($lastprinted))
49			{
50			($a) = $lastprinted =~ /^(\s*)/;
51			($b) = $_ =~ /^(\s*)/;
52			$blankcount++ if ($a ne $b);
53			}
54
55			# We get here only when we have a non-blank line in hand. If it was preceded
56			# by 3 or more blank lines, read the next 3 lines and see if they are blank.
57			# If so, remove all 7 lines, and remember that we have just done a cut.
58
59			if ($blankcount >= 3)
60			{
61			for ($i = 0; $i < 3; $i++)
62			{
63			$next[$i] = <STDIN>;
64			$next[$i] = "" if !defined $next[$i];
65			$next[$i] =~ s/\x1b\[\d+m//g; # Remove screen controls "ESC [ number m"
66			$next[$i] =~ s/.\x8//g; # Remove "char, backspace"
67			}
68
69			# Cut out chunks of the form <3 blanks><non-blank><3 blanks>
70
71			if ($next[0] =~ /^\s*$/ &&
72			$next[1] =~ /^\s*$/ &&
73			$next[2] =~ /^\s*$/)
74			{
75			$blankcount -= 3;
76			$lastwascut = 1;
77			}
78
79			# Otherwise output the saved blanks, the current, and the next three
80			# lines. Remember the last printed line.
81
82			else
83			{
84			for ($i = 0; $i < $blankcount; $i++) { print "\n"; }
85			print;
86			for ($i = 0; $i < 3; $i++)
87			{
88			$next[$i] =~ s/.\x8//g;
89			print $next[$i];
90			$lastprinted = $_;
91			}
92			$lastwascut = 0;
93			$blankcount = 0;
94			}
95			}
96
97			# This non-blank line is not preceded by 3 or more blank lines. Output
98			# any blanks there are, and the line. Remember it. Force two blank lines
99			# before headings.
100
101			else
102			{
103			$blankcount = 2 if /^\S/ && !/^Last updated/ && !/^Copyright/ &&
104			defined($lastprinted);
105			for ($i = 0; $i < $blankcount; $i++) { print "\n"; }
106			print;
107			$lastprinted = $_;
108			$lastwascut = 0;
109			$blankcount = 0;
110			}
111			}
112
113			# End