Add ARM files
[dh-make-perl] / dev / arm / libhtml-parser-perl / libhtml-parser-perl-3.56 / eg / htext
diff --git a/dev/arm/libhtml-parser-perl/libhtml-parser-perl-3.56/eg/htext b/dev/arm/libhtml-parser-perl/libhtml-parser-perl-3.56/eg/htext
new file mode 100755 (executable)
index 0000000..e4d276d
--- /dev/null
@@ -0,0 +1,29 @@
+#!/usr/bin/perl -w
+
+# Extract all plain text from an HTML file
+
+use strict;
+use HTML::Parser 3.00 ();
+
+my %inside;
+
+sub tag
+{
+   my($tag, $num) = @_;
+   $inside{$tag} += $num;
+   print " ";  # not for all tags
+}
+
+sub text
+{
+    return if $inside{script} || $inside{style};
+    print $_[0];
+}
+
+HTML::Parser->new(api_version => 3,
+                 handlers    => [start => [\&tag, "tagname, '+1'"],
+                                 end   => [\&tag, "tagname, '-1'"],
+                                 text  => [\&text, "dtext"],
+                                ],
+                 marked_sections => 1,
+       )->parse_file(shift) || die "Can't open file: $!\n";;