proxy:

(de)compression.
library:
cnofig file reading
html entity decoding

git-svn-id: svn://botcastle1b/yplom/proxy@12 05ce6ebb-7522-4a6e-a768-0026ae12be9f

diff --git a/cleararch.1.pl b/cleararch.1.pl
index 63c1b5f58dcc1c1fbcb5dbf23e754100c2be4b0d..25dae8c9ec89f333167782fae3bb852f12b1b672 100644 (file)
--- a/cleararch.1.pl
+++ b/cleararch.1.pl
@@ -11,7 +11,7 @@ use POSIX qw(strftime);
 ###TIMEOUT_ARCH;\r
 \r
 $time = time();\r
-print strftime("%d.%m.%Y %H:%M:%S", gmtime($time))."\n";\r
+print strftime('%d.%m.%Y %H:%M:%S', gmtime($time))."\n";\r
 cleardir(ARCH_PATH);\r
 \r
 sub cleardir {\r
@@ -25,7 +25,7 @@ sub cleardir {
                print "Failed to open: $dirpath\/\n";\r
                return 0;\r
        }\r
-       while ($subpath = readdir $dir) {\r
+       while (defined($subpath = readdir $dir)) {\r
                $subpathfull=$dirpath.$subpath;\r
                if ($subpath =~ /^\.\.?$/) {\r
                        next;\r

diff --git a/proxy.1.pl b/proxy.1.pl
index 0de727fb2d243994aeb8c44f5be8d138c9cdfc87..af0b02c6ddf004fffd6859ba31d89b770d90983f 100755 (executable)
--- a/proxy.1.pl
+++ b/proxy.1.pl
@@ -118,7 +118,7 @@ sub proxy {
        
        my $contentlength;
        
-       my @curl_arg=(CURL_PATH,'-i','-#'); # the command line arguments for curl
+       my @curl_arg=(CURL_PATH,'-i','-#','--compressed'); # the command line arguments for curl
        my @reqhead;                        # the HTTP request header
        my @resphead;
        
@@ -625,7 +625,7 @@ sub unlock {
        # All unlocking access must be logged: time, IP, user. The server owner must
        # be able to know who did what with the proxy (in  case of legal problems).
        open ($logfile, ">>", UNLOCK_LOG) or return unlockpage("Couldn't log your action.","Status: 500 Internal Server Error\n");
-       print $logfile strftime("%d.%m.%Y %H:%M:%S", gmtime($accesstime))."  $IP  $CGI{'username'}\n";
+       print $logfile strftime('%d.%m.%Y %H:%M:%S', gmtime($accesstime))."  $IP  $CGI{'username'}\n";
        close($logfile);
        
        # Ok, unlocked. Create access file.

diff --git a/proxy_lib.1.pm b/proxy_lib.1.pm
index f3bf29a85d82239df797130ad36051c951774c72..82fc153e730469ba53101d46f97a080eba30530b 100644 (file)
--- a/proxy_lib.1.pm
+++ b/proxy_lib.1.pm
@@ -10,16 +10,271 @@ use strict;
 use Exporter;\r
 use vars qw($VERSION @ISA @EXPORT @EXPORT_OK %EXPORT_TAGS);\r
 \r
+use constant entitycode => {\r
+       'amp' => '&',\r
+       'gt' => '>',\r
+       'lt' => '<',\r
+       'quot' => '"',\r
+       'acute' => '´',\r
+       'cedil' => '¸',\r
+       'circ' => 'ˆ',\r
+       'macr' => '¯',\r
+       'middot' => '·',\r
+       'tilde' => '˜',\r
+       'uml' => '¨',\r
+       'Aacute' => 'Á',\r
+       'aacute' => 'á',\r
+       'Acirc' => 'Â',\r
+       'acirc' => 'â',\r
+       'AElig' => 'Æ',\r
+       'aelig' => 'æ',\r
+       'Agrave' => 'À',\r
+       'agrave' => 'à',\r
+       'Aring' => 'Å',\r
+       'aring' => 'å',\r
+       'Atilde' => 'Ã',\r
+       'atilde' => 'ã',\r
+       'Auml' => 'Ä',\r
+       'auml' => 'ä',\r
+       'Ccedil' => 'Ç',\r
+       'ccedil' => 'ç',\r
+       'Eacute' => 'É',\r
+       'eacute' => 'é',\r
+       'Ecirc' => 'Ê',\r
+       'ecirc' => 'ê',\r
+       'Egrave' => 'È',\r
+       'egrave' => 'è',\r
+       'ETH' => 'Ð',\r
+       'eth' => 'ð',\r
+       'Euml' => 'Ë',\r
+       'euml' => 'ë',\r
+       'Iacute' => 'Í',\r
+       'iacute' => 'í',\r
+       'Icirc' => 'Î',\r
+       'icirc' => 'î',\r
+       'Igrave' => 'Ì',\r
+       'igrave' => 'ì',\r
+       'Iuml' => 'Ï',\r
+       'iuml' => 'ï',\r
+       'Ntilde' => 'Ñ',\r
+       'ntilde' => 'ñ',\r
+       'Oacute' => 'Ó',\r
+       'oacute' => 'ó',\r
+       'Ocirc' => 'Ô',\r
+       'ocirc' => 'ô',\r
+       'OElig' => 'Œ',\r
+       'oelig' => 'œ',\r
+       'Ograve' => 'Ò',\r
+       'ograve' => 'ò',\r
+       'Oslash' => 'Ø',\r
+       'oslash' => 'ø',\r
+       'Otilde' => 'Õ',\r
+       'otilde' => 'õ',\r
+       'Ouml' => 'Ö',\r
+       'ouml' => 'ö',\r
+       'Scaron' => 'Š',\r
+       'scaron' => 'š',\r
+       'szlig' => 'ß',\r
+       'THORN' => 'Þ',\r
+       'thorn' => 'þ',\r
+       'Uacute' => 'Ú',\r
+       'uacute' => 'ú',\r
+       'Ucirc' => 'Û',\r
+       'ucirc' => 'û',\r
+       'Ugrave' => 'Ù',\r
+       'ugrave' => 'ù',\r
+       'Uuml' => 'Ü',\r
+       'uuml' => 'ü',\r
+       'Yacute' => 'Ý',\r
+       'yacute' => 'ý',\r
+       'yuml' => 'ÿ',\r
+       'Yuml' => 'Ÿ',\r
+       'cent' => '¢',\r
+       'curren' => '¤',\r
+       'euro' => '€',\r
+       'pound' => '£',\r
+       'yen' => '¥',\r
+       'brvbar' => '¦',\r
+       'bull' => '•',\r
+       'copy' => '©',\r
+       'dagger' => '†',\r
+       'Dagger' => '‡',\r
+       'frasl' => '⁄',\r
+       'hellip' => '…',\r
+       'iexcl' => '¡',\r
+       'image' => 'ℑ',\r
+       'iquest' => '¿',\r
+       'lrm' => '‎',\r
+       'mdash' => '—',\r
+       'ndash' => '–',\r
+       'not' => '¬',\r
+       'oline' => '‾',\r
+       'ordf' => 'ª',\r
+       'ordm' => 'º',\r
+       'para' => '¶',\r
+       'permil' => '‰',\r
+       'prime' => '′',\r
+       'Prime' => '″',\r
+       'real' => 'ℜ',\r
+       'reg' => '®',\r
+       'rlm' => '‏',\r
+       'sect' => '§',\r
+       'shy' => '­',\r
+       'sup1' => '¹',\r
+       'trade' => '™',\r
+       'weierp' => '℘',\r
+       'bdquo' => '„',\r
+       'laquo' => '«',\r
+       'ldquo' => '“',\r
+       'lsaquo' => '‹',\r
+       'lsquo' => '‘',\r
+       'raquo' => '»',\r
+       'rdquo' => '”',\r
+       'rsaquo' => '›',\r
+       'rsquo' => '’',\r
+       'sbquo' => '‚',\r
+       'emsp' => ' ',\r
+       'ensp' => ' ',\r
+       'nbsp' => ' ',\r
+       'thinsp' => ' ',\r
+       'zwj' => '‍',\r
+       'zwnj' => '‌',\r
+       'deg' => '°',\r
+       'divide' => '÷',\r
+       'frac12' => '½',\r
+       'frac14' => '¼',\r
+       'frac34' => '¾',\r
+       'ge' => '≥',\r
+       'le' => '≤',\r
+       'minus' => '−',\r
+       'sup2' => '²',\r
+       'sup3' => '³',\r
+       'times' => '×',\r
+       'alefsym' => 'ℵ',\r
+       'and' => '∧',\r
+       'ang' => '∠',\r
+       'asymp' => '≈',\r
+       'cap' => '∩',\r
+       'cong' => '≅',\r
+       'cup' => '∪',\r
+       'empty' => '∅',\r
+       'equiv' => '≡',\r
+       'exist' => '∃',\r
+       'fnof' => 'ƒ',\r
+       'forall' => '∀',\r
+       'infin' => '∞',\r
+       'int' => '∫',\r
+       'isin' => '∈',\r
+       'lang' => '⟨',\r
+       'lceil' => '⌈',\r
+       'lfloor' => '⌊',\r
+       'lowast' => '∗',\r
+       'micro' => 'µ',\r
+       'nabla' => '∇',\r
+       'ne' => '≠',\r
+       'ni' => '∋',\r
+       'notin' => '∉',\r
+       'nsub' => '⊄',\r
+       'oplus' => '⊕',\r
+       'or' => '∨',\r
+       'otimes' => '⊗',\r
+       'part' => '∂',\r
+       'perp' => '⊥',\r
+       'plusmn' => '±',\r
+       'prod' => '∏',\r
+       'prop' => '∝',\r
+       'radic' => '√',\r
+       'rang' => '⟩',\r
+       'rceil' => '⌉',\r
+       'rfloor' => '⌋',\r
+       'sdot' => '⋅',\r
+       'sim' => '∼',\r
+       'sub' => '⊂',\r
+       'sube' => '⊆',\r
+       'sum' => '∑',\r
+       'sup' => '⊃',\r
+       'supe' => '⊇',\r
+       'there4' => '∴',\r
+       'Alpha' => 'Α',\r
+       'alpha' => 'α',\r
+       'Beta' => 'Β',\r
+       'beta' => 'β',\r
+       'Chi' => 'Χ',\r
+       'chi' => 'χ',\r
+       'Delta' => 'Δ',\r
+       'delta' => 'δ',\r
+       'Epsilon' => 'Ε',\r
+       'epsilon' => 'ε',\r
+       'Eta' => 'Η',\r
+       'eta' => 'η',\r
+       'Gamma' => 'Γ',\r
+       'gamma' => 'γ',\r
+       'Iota' => 'Ι',\r
+       'iota' => 'ι',\r
+       'Kappa' => 'Κ',\r
+       'kappa' => 'κ',\r
+       'Lambda' => 'Λ',\r
+       'lambda' => 'λ',\r
+       'Mu' => 'Μ',\r
+       'mu' => 'μ',\r
+       'Nu' => 'Ν',\r
+       'nu' => 'ν',\r
+       'Omega' => 'Ω',\r
+       'omega' => 'ω',\r
+       'Omicron' => 'Ο',\r
+       'omicron' => 'ο',\r
+       'Phi' => 'Φ',\r
+       'phi' => 'φ',\r
+       'Pi' => 'Π',\r
+       'pi' => 'π',\r
+       'piv' => 'ϖ',\r
+       'Psi' => 'Ψ',\r
+       'psi' => 'ψ',\r
+       'Rho' => 'Ρ',\r
+       'rho' => 'ρ',\r
+       'Sigma' => 'Σ',\r
+       'sigma' => 'σ',\r
+       'sigmaf' => 'ς',\r
+       'Tau' => 'Τ',\r
+       'tau' => 'τ',\r
+       'Theta' => 'Θ',\r
+       'theta' => 'θ',\r
+       'thetasym' => 'ϑ',\r
+       'upsih' => 'ϒ',\r
+       'Upsilon' => 'Υ',\r
+       'upsilon' => 'υ',\r
+       'Xi' => 'Ξ',\r
+       'xi' => 'ξ',\r
+       'Zeta' => 'Ζ',\r
+       'zeta' => 'ζ',\r
+       'crarr' => '↵',\r
+       'darr' => '↓',\r
+       'dArr' => '⇓',\r
+       'harr' => '↔',\r
+       'hArr' => '⇔',\r
+       'larr' => '←',\r
+       'lArr' => '⇐',\r
+       'rarr' => '→',\r
+       'rArr' => '⇒',\r
+       'uarr' => '↑',\r
+       'uArr' => '⇑',\r
+       'clubs' => '♣',\r
+       'diams' => '♦',\r
+       'hearts' => '♥',\r
+       'spades' => '♠',\r
+       'loz' => '◊',\r
+};\r
+\r
 ###ARCH_PATH;\r
 ###ACCESS_PATH;\r
 ###TIMEOUT_UNLOCK;\r
 ###TIMEOUT_INACT;\r
 ###UNLOCK_PROXY_HOST;\r
 \r
-$VERSION     = 0.000002;\r
+$VERSION     = 0.000003;\r
 @ISA         = qw(Exporter);\r
 @EXPORT      = ();\r
-@EXPORT_OK   = qw(access divideurl formatheader getcgi joinurl path2url url2path path2urldiv urldecode urldiv2path);\r
+@EXPORT_OK   = qw(access divideurl entitydecode formatheader getcgi joinurl path2url url2path path2urldiv readconfigfile urldecode urldiv2path);\r
 %EXPORT_TAGS = ();\r
 \r
 # This function checks if the user has unlocked the proxy. Nonzero means yes.\r
@@ -192,6 +447,28 @@ sub urldecode {
        return $t;\r
 }\r
 \r
+sub entitydecode {\r
+       my $t = $_[0];\r
+       $t =~ s/&(#?[a-zA-Z0-9]+);/entitydecode1ch($1)/eg;\r
+       return $t;\r
+}\r
+\r
+sub entitydecode1ch {\r
+       my $t = $_[0];\r
+       if ($t =~ /^#([0-9]+)$/) {\r
+               return chr($1);\r
+       }\r
+       elsif ($t =~ /^#x([0-9A-Fa-f]+)$/) {\r
+               return chr(hex($1));\r
+       }\r
+       elsif ($t =~ /^([A-Za-z0-9]+)$/) {\r
+               return entitycode->{$1};\r
+       }\r
+       else {\r
+               return '';\r
+       }\r
+}\r
+\r
 # The function to get CGI parameters from string.\r
 # Format is: name=url_encoded_value&name=url_encoded_value& ... &name=url_encoded_value\r
 sub getcgi {\r
@@ -534,4 +811,25 @@ sub path2url {
        return joinurl(path2urldiv($_[0]));\r
 }\r
 \r
+sub readconfigfile {\r
+       (my $configpath) = @_;\r
+       my $configfile;\r
+       my %config;\r
+       \r
+       if (open ($configfile, "<", $configpath)) {\r
+               while (defined(my $line = <$configfile>)) {\r
+                       $line =~ s/[\r\n]//g;\r
+                       $line =~ s/#.*$//;\r
+                       if ($line =~ /^ *([a-zA-Z0-9_\-\.]+) *= *(.*)$/){\r
+                               my $name=$1;\r
+                               my $value=$2;\r
+                               $value =~ s/ *$//;\r
+                               $config{$name}=$value;\r
+                       }\r
+               }\r
+               close ($configfile);\r
+       }\r
+       return %config;\r
+}\r
+\r
 1;\r