From: b <rowerynaksiezycu@gmail.com>
Date: Sun, 25 Sep 2022 10:27:48 +0000 (+0000)
Subject: added html entity encoding,
X-Git-Tag: v1.0.0~1
X-Git-Url: http://bicyclesonthemoon.info/git-projects/?a=commitdiff_plain;h=c9d714f574b7b0d68c8eb7f86f445f893a56514d;p=botm%2Fconfig

added html entity encoding,
fixed whitespace handling for empty pattern
---

diff --git a/configure.pl b/configure.pl
index 3a13a61..d575160 100755
--- a/configure.pl
+++ b/configure.pl
@@ -271,6 +271,10 @@
 #     $0 - the text to be encoded
 #     $1 - the regexp which specifies which characters should be encoded
 #          if not specified then a default regexp is used
+#   @_HT_ENCODE is a function which will perform HTML entity encoding
+#     $0 - the text to be encoded
+#     $1 - the regexp which specifies which characters should be encoded
+#          if not specified then a default regexp is used
 #   @_SPRINTF is a call to the Perl sprintf() function
 #     $0 - the format string
 #     $1 and continuing - the parameters to sprintf()
@@ -372,6 +376,7 @@ use constant PATH_SEPARATOR  => '_PATH_SEPARATOR';
 
 use constant ESCAPE     => '_ESCAPE';
 use constant URL_ENCODE => '_URL_ENCODE';
+use constant HT_ENCODE  => '_HT_ENCODE';
 use constant PATH       => '_PATH';
 use constant SPRINTF    => '_SPRINTF';
 
@@ -562,6 +567,7 @@ sub escape {
 	return $outcome;
 }
 
+# NOT UNICODE AWARE
 sub urlencode {
 	(my $text, my $match) = @_;
 	unless (defined $match) {
@@ -580,6 +586,26 @@ sub urlencode {
 	return $outcome;
 }
 
+# NOT UNICODE AWARE
+# (but there no real need to escaping non-ascii characters)
+sub entityencode {
+	(my $text, my $match) = @_;
+	unless (defined $match) {
+		$match = '[\"=><\&]'
+	}
+	
+	my $outcome = '';
+	foreach my $ch (split('', $text)) {
+		if ($ch =~ $match) {
+			$outcome .= sprintf('&#%02hu;',ord($ch));
+		}
+		else {
+			$outcome .= $ch;
+		}
+	}
+	return $outcome;
+}
+
 sub join_path {
 	(my $joiner, my @segments) = @_;
 	
@@ -790,6 +816,9 @@ sub parse_pattern {
 	elsif ($name eq URL_ENCODE) {
 		$return = urlencode(@parameter_list);
 	}
+	elsif ($name eq HT_ENCODE) {
+		$return = entityencode(@parameter_list);
+	}
 	elsif ($name eq PATH) {
 		$return = join_path($cfg{PATH_SEPARATOR()},@parameter_list);
 	}
@@ -834,7 +863,7 @@ sub parse_value {
 	if ($to_parse =~ /^((([^#\\])|(\\.))*)#/) {
 		$to_parse = $1;
 	}
-	if ($to_parse =~ /^[ \t]*([^ \t](.*[^ \t])?)[ \t]*$/) {
+	if ($to_parse =~ /^[ \t]*(([^ \t](.*[^ \t])?)?)[ \t]*$/) {
 		$to_parse = $1;
 	}