httputil: try to guess content-type from extension if it's not defined

[user/henk/code/ruby/rbot.git] / lib / rbot / core / utils / utils.rb

diff --git a/lib/rbot/core/utils/utils.rb b/lib/rbot/core/utils/utils.rb
index 115263707bdd27676568897ce24fb9103a93ff3a..535ae190179c1bfbfa577a180d42baa24014f079 100644 (file)
--- a/lib/rbot/core/utils/utils.rb
+++ b/lib/rbot/core/utils/utils.rb
@@ -44,254 +44,70 @@ rescue LoadError
     'gt' => '>',
     'hellip' => '…',
     'nbsp' => ' ',
-=begin
-    # extras codes, for future use...
-    'zwnj' => '‌',
-    'aring' => '\xe5',
-    'gt' => '>',
-    'yen' => '\xa5',
-    'ograve' => '\xf2',
-    'Chi' => 'Χ',
-    'bull' => '•',
-    'Egrave' => '\xc8',
-    'Ntilde' => '\xd1',
-    'upsih' => 'ϒ',
-    'Yacute' => '\xdd',
-    'asymp' => '≈',
-    'radic' => '√',
-    'otimes' => '⊗',
-    'nabla' => '∇',
-    'aelig' => '\xe6',
-    'oelig' => 'œ',
-    'equiv' => '≡',
-    'Psi' => 'Ψ',
-    'auml' => '\xe4',
-    'circ' => 'ˆ',
-    'Acirc' => '\xc2',
-    'Epsilon' => 'Ε',
-    'Yuml' => 'Ÿ',
-    'Eta' => 'Η',
-    'Icirc' => '\xce',
-    'Upsilon' => 'Υ',
-    'ndash' => '–',
-    'there4' => '∴',
-    'Prime' => '″',
-    'prime' => '′',
-    'psi' => 'ψ',
-    'Kappa' => 'Κ',
-    'rsaquo' => '›',
-    'Tau' => 'Τ',
-    'darr' => '↓',
-    'ocirc' => '\xf4',
-    'lrm' => '‎',
-    'zwj' => '‍',
-    'cedil' => '\xb8',
-    'Ecirc' => '\xca',
-    'not' => '\xac',
-    'AElig' => '\xc6',
-    'oslash' => '\xf8',
-    'acute' => '\xb4',
-    'lceil' => '⌈',
-    'shy' => '\xad',
-    'rdquo' => '”',
-    'ge' => '≥',
-    'Igrave' => '\xcc',
-    'Ograve' => '\xd2',
-    'euro' => '€',
-    'dArr' => '⇓',
-    'sdot' => '⋅',
-    'nbsp' => '\xa0',
-    'lfloor' => '⌊',
-    'lArr' => '⇐',
-    'Auml' => '\xc4',
-    'larr' => '←',
-    'Atilde' => '\xc3',
-    'Otilde' => '\xd5',
-    'szlig' => '\xdf',
-    'clubs' => '♣',
-    'diams' => '♦',
-    'agrave' => '\xe0',
-    'Ocirc' => '\xd4',
-    'Iota' => 'Ι',
-    'Theta' => 'Θ',
-    'Pi' => 'Π',
-    'OElig' => 'Œ',
-    'Scaron' => 'Š',
-    'frac14' => '\xbc',
-    'egrave' => '\xe8',
-    'sub' => '⊂',
-    'iexcl' => '\xa1',
-    'frac12' => '\xbd',
-    'sbquo' => '‚',
-    'ordf' => '\xaa',
-    'sum' => '∑',
-    'prop' => '∝',
-    'Uuml' => '\xdc',
-    'ntilde' => '\xf1',
-    'sup' => '⊃',
-    'theta' => 'θ',
-    'prod' => '∏',
-    'nsub' => '⊄',
-    'hArr' => '⇔',
-    'rlm' => '‏',
-    'THORN' => '\xde',
-    'infin' => '∞',
-    'yuml' => '\xff',
-    'Mu' => 'Μ',
-    'le' => '≤',
-    'Eacute' => '\xc9',
-    'thinsp' => ' ',
-    'ecirc' => '\xea',
-    'bdquo' => '„',
-    'Sigma' => 'Σ',
-    'fnof' => 'ƒ',
-    'Aring' => '\xc5',
-    'tilde' => '˜',
-    'frac34' => '\xbe',
-    'emsp' => ' ',
-    'mdash' => '—',
-    'uarr' => '↑',
-    'permil' => '‰',
-    'Ugrave' => '\xd9',
-    'rarr' => '→',
-    'Agrave' => '\xc0',
-    'chi' => 'χ',
-    'forall' => '∀',
-    'eth' => '\xf0',
-    'rceil' => '⌉',
-    'iuml' => '\xef',
-    'gamma' => 'γ',
-    'lambda' => 'λ',
-    'harr' => '↔',
-    'rang' => '〉',
-    'xi' => 'ξ',
-    'dagger' => '†',
-    'divide' => '\xf7',
-    'Ouml' => '\xd6',
-    'image' => 'ℑ',
-    'alefsym' => 'ℵ',
-    'igrave' => '\xec',
-    'otilde' => '\xf5',
-    'Oacute' => '\xd3',
-    'sube' => '⊆',
-    'alpha' => 'α',
-    'frasl' => '⁄',
-    'ETH' => '\xd0',
-    'lowast' => '∗',
-    'Nu' => 'Ν',
-    'plusmn' => '\xb1',
-    'Euml' => '\xcb',
-    'real' => 'ℜ',
-    'sup1' => '\xb9',
-    'sup2' => '\xb2',
-    'sup3' => '\xb3',
-    'Oslash' => '\xd8',
-    'Aacute' => '\xc1',
-    'cent' => '\xa2',
-    'oline' => '‾',
-    'Beta' => 'Β',
-    'perp' => '⊥',
-    'Delta' => 'Δ',
-    'loz' => '◊',
-    'pi' => 'π',
-    'iota' => 'ι',
-    'empty' => '∅',
-    'euml' => '\xeb',
-    'brvbar' => '\xa6',
-    'iacute' => '\xed',
-    'para' => '\xb6',
-    'micro' => '\xb5',
-    'cup' => '∪',
-    'weierp' => '℘',
-    'uuml' => '\xfc',
-    'part' => '∂',
-    'icirc' => '\xee',
-    'delta' => 'δ',
-    'omicron' => 'ο',
-    'upsilon' => 'υ',
-    'Iuml' => '\xcf',
-    'Lambda' => 'Λ',
-    'Xi' => 'Ξ',
-    'kappa' => 'κ',
-    'ccedil' => '\xe7',
-    'Ucirc' => '\xdb',
-    'cap' => '∩',
-    'mu' => 'μ',
-    'scaron' => 'š',
-    'lsquo' => '‘',
-    'isin' => '∈',
-    'Zeta' => 'Ζ',
-    'supe' => '⊇',
-    'deg' => '\xb0',
-    'and' => '∧',
-    'tau' => 'τ',
-    'pound' => '\xa3',
-    'hellip' => '…',
-    'curren' => '\xa4',
-    'int' => '∫',
-    'ucirc' => '\xfb',
-    'rfloor' => '⌋',
-    'ensp' => ' ',
-    'crarr' => '↵',
-    'ugrave' => '\xf9',
-    'notin' => '∉',
-    'exist' => '∃',
-    'uArr' => '⇑',
-    'cong' => '≅',
-    'Dagger' => '‡',
-    'oplus' => '⊕',
-    'times' => '\xd7',
-    'atilde' => '\xe3',
-    'piv' => 'ϖ',
-    'ni' => '∋',
-    'Phi' => 'Φ',
-    'lsaquo' => '‹',
-    'Uacute' => '\xda',
-    'Omicron' => 'Ο',
-    'ang' => '∠',
-    'ne' => '≠',
-    'iquest' => '\xbf',
-    'eta' => 'η',
-    'yacute' => '\xfd',
-    'Rho' => 'Ρ',
-    'uacute' => '\xfa',
-    'Alpha' => 'Α',
-    'zeta' => 'ζ',
-    'Omega' => 'Ω',
-    'nu' => 'ν',
-    'sim' => '∼',
-    'sect' => '\xa7',
-    'phi' => 'φ',
-    'sigmaf' => 'ς',
-    'macr' => '\xaf',
-    'minus' => '−',
-    'Ccedil' => '\xc7',
-    'ordm' => '\xba',
-    'epsilon' => 'ε',
-    'beta' => 'β',
-    'rArr' => '⇒',
-    'rho' => 'ρ',
-    'aacute' => '\xe1',
-    'eacute' => '\xe9',
-    'omega' => 'ω',
-    'middot' => '\xb7',
-    'Gamma' => 'Γ',
-    'Iacute' => '\xcd',
-    'lang' => '〈',
-    'spades' => '♠',
-    'rsquo' => '’',
-    'uml' => '\xa8',
-    'thorn' => '\xfe',
-    'ouml' => '\xf6',
-    'thetasym' => 'ϑ',
-    'or' => '∨',
-    'raquo' => '\xbb',
-    'acirc' => '\xe2',
-    'ldquo' => '“',
-    'hearts' => '♥',
-    'sigma' => 'σ',
-    'oacute' => '\xf3',
-=end
+    'Agrave' => 'À',
+    'Aacute' => 'Á',
+    'Acirc' => 'Â',
+    'Atilde' => 'Ã',
+    'Auml' => 'Ä',
+    'Aring' => 'Å',
+    'AElig' => 'Æ',
+    'OElig' => 'Œ',
+    'Ccedil' => 'Ç',
+    'Egrave' => 'È',
+    'Eacute' => 'É',
+    'Ecirc' => 'Ê',
+    'Euml' => 'Ë',
+    'Igrave' => 'Ì',
+    'Iacute' => 'Í',
+    'Icirc' => 'Î',
+    'Iuml' => 'Ï',
+    'ETH' => 'Ð',
+    'Ntilde' => 'Ñ',
+    'Ograve' => 'Ò',
+    'Oacute' => 'Ó',
+    'Ocirc' => 'Ô',
+    'Otilde' => 'Õ',
+    'Ouml' => 'Ö',
+    'Oslash' => 'Ø',
+    'Ugrave' => 'Ù',
+    'Uacute' => 'Ú',
+    'Ucirc' => 'Û',
+    'Uuml' => 'Ü',
+    'Yacute' => 'Ý',
+    'THORN' => 'Þ',
+    'szlig' => 'ß',
+    'agrave' => 'à',
+    'aacute' => 'á',
+    'acirc' => 'â',
+    'atilde' => 'ã',
+    'auml' => 'ä',
+    'aring' => 'å',
+    'aelig' => 'æ',
+    'oelig' => 'œ',
+    'ccedil' => 'ç',
+    'egrave' => 'è',
+    'eacute' => 'é',
+    'ecirc' => 'ê',
+    'euml' => 'ë',
+    'igrave' => 'ì',
+    'iacute' => 'í',
+    'icirc' => 'î',
+    'iuml' => 'ï',
+    'eth' => 'ð',
+    'ntilde' => 'ñ',
+    'ograve' => 'ò',
+    'oacute' => 'ó',
+    'ocirc' => 'ô',
+    'otilde' => 'õ',
+    'ouml' => 'ö',
+    'oslash' => 'ø',
+    'ugrave' => 'ù',
+    'uacute' => 'ú',
+    'ucirc' => 'û',
+    'uuml' => 'ü',
+    'yacute' => 'ý',
+    'thorn' => 'þ',
+    'yuml' => 'ÿ'
         }
       end
     end
@@ -689,9 +505,6 @@ module ::Irc
       when Net::HTTPResponse
         Utils.get_resp_html_info(doc, opts)
       when URI
-        if doc.fragment and not doc.fragment.empty?
-          opts[:uri_fragment] ||= doc.fragment
-        end
         ret = Hash.new
         @@bot.httputil.get_response(doc) { |resp|
           ret = Utils.get_resp_html_info(resp, opts)
@@ -725,8 +538,12 @@ module ::Irc
       when Net::HTTPSuccess
         ret[:headers] = resp.to_hash
 
+        partial = resp.partial_body(@@bot.config['http.info_bytes'])
         if resp['content-type'] =~ /^text\/|(?:x|ht)ml/
-          partial = resp.partial_body(@@bot.config['http.info_bytes'])
+          loc = URI.parse(resp['x-rbot-location'] || resp['location']) rescue nil
+          if loc and loc.fragment and not loc.fragment.empty?
+            opts[:uri_fragment] ||= loc.fragment
+          end
           ret.merge!(Utils.get_string_html_info(partial, opts))
         end
         return ret