| 879 | | class AtomEntry { |
| 880 | | var $links = array(); |
| 881 | | var $categories = array(); |
| 882 | | } |
| 883 | | |
| 884 | | class AtomParser { |
| 885 | | |
| 886 | | var $ATOM_CONTENT_ELEMENTS = array('content','summary','title','subtitle','rights'); |
| 887 | | var $ATOM_SIMPLE_ELEMENTS = array('id','updated','published','draft','author'); |
| 888 | | |
| 889 | | var $depth = 0; |
| 890 | | var $indent = 2; |
| 891 | | var $in_content; |
| 892 | | var $ns_contexts = array(); |
| 893 | | var $ns_decls = array(); |
| 894 | | var $is_xhtml = false; |
| 895 | | var $skipped_div = false; |
| 896 | | |
| 897 | | var $entry; |
| 898 | | |
| 899 | | function AtomParser() { |
| 900 | | |
| 901 | | $this->entry = new AtomEntry(); |
| 902 | | $this->map_attrs_func = create_function('$k,$v', 'return "$k=\"$v\"";'); |
| 903 | | $this->map_xmlns_func = create_function('$p,$n', '$xd = "xmlns"; if(strlen($n[0])>0) $xd .= ":{$n[0]}"; return "{$xd}=\"{$n[1]}\"";'); |
| 904 | | } |
| 905 | | |
| 906 | | function parse($xml) { |
| 907 | | |
| 908 | | global $app_logging; |
| 909 | | array_unshift($this->ns_contexts, array()); |
| 910 | | |
| 911 | | $parser = xml_parser_create_ns(); |
| 912 | | xml_set_object($parser, $this); |
| 913 | | xml_set_element_handler($parser, "start_element", "end_element"); |
| 914 | | xml_parser_set_option($parser,XML_OPTION_CASE_FOLDING,0); |
| 915 | | xml_parser_set_option($parser,XML_OPTION_SKIP_WHITE,0); |
| 916 | | xml_set_character_data_handler($parser, "cdata"); |
| 917 | | xml_set_default_handler($parser, "_default"); |
| 918 | | xml_set_start_namespace_decl_handler($parser, "start_ns"); |
| 919 | | xml_set_end_namespace_decl_handler($parser, "end_ns"); |
| 920 | | |
| 921 | | $contents = ""; |
| 922 | | |
| 923 | | xml_parse($parser, $xml); |
| 924 | | |
| 925 | | xml_parser_free($parser); |
| 926 | | |
| 927 | | return true; |
| 928 | | } |
| 929 | | |
| 930 | | function start_element($parser, $name, $attrs) { |
| 931 | | |
| 932 | | $tag = array_pop(split(":", $name)); |
| 933 | | |
| 934 | | array_unshift($this->ns_contexts, $this->ns_decls); |
| 935 | | |
| 936 | | $this->depth++; |
| 937 | | |
| 938 | | if(!empty($this->in_content)) { |
| 939 | | $attrs_prefix = array(); |
| 940 | | |
| 941 | | // resolve prefixes for attributes |
| 942 | | foreach($attrs as $key => $value) { |
| 943 | | $attrs_prefix[$this->ns_to_prefix($key)] = $this->xml_escape($value); |
| 944 | | } |
| 945 | | $attrs_str = join(' ', array_map($this->map_attrs_func, array_keys($attrs_prefix), array_values($attrs_prefix))); |
| 946 | | if(strlen($attrs_str) > 0) { |
| 947 | | $attrs_str = " " . $attrs_str; |
| 948 | | } |
| 949 | | |
| 950 | | $xmlns_str = join(' ', array_map($this->map_xmlns_func, array_keys($this->ns_contexts[0]), array_values($this->ns_contexts[0]))); |
| 951 | | if(strlen($xmlns_str) > 0) { |
| 952 | | $xmlns_str = " " . $xmlns_str; |
| 953 | | } |
| 954 | | |
| 955 | | // handle self-closing tags (case: a new child found right-away, no text node) |
| 956 | | if(count($this->in_content) == 2) { |
| 957 | | array_push($this->in_content, ">"); |
| 958 | | } |
| 959 | | |
| 960 | | array_push($this->in_content, "<". $this->ns_to_prefix($name) ."{$xmlns_str}{$attrs_str}"); |
| 961 | | } else if(in_array($tag, $this->ATOM_CONTENT_ELEMENTS) || in_array($tag, $this->ATOM_SIMPLE_ELEMENTS)) { |
| 962 | | $this->in_content = array(); |
| 963 | | $this->is_xhtml = $attrs['type'] == 'xhtml'; |
| 964 | | array_push($this->in_content, array($tag,$this->depth)); |
| 965 | | } else if($tag == 'link') { |
| 966 | | array_push($this->entry->links, $attrs); |
| 967 | | } else if($tag == 'category') { |
| 968 | | array_push($this->entry->categories, $attrs['term']); |
| 969 | | } |
| 970 | | |
| 971 | | $this->ns_decls = array(); |
| 972 | | } |
| 973 | | |
| 974 | | function end_element($parser, $name) { |
| 975 | | |
| 976 | | $tag = array_pop(split(":", $name)); |
| 977 | | |
| 978 | | if(!empty($this->in_content)) { |
| 979 | | if($this->in_content[0][0] == $tag && |
| 980 | | $this->in_content[0][1] == $this->depth) { |
| 981 | | array_shift($this->in_content); |
| 982 | | if($this->is_xhtml) { |
| 983 | | $this->in_content = array_slice($this->in_content, 2, count($this->in_content)-3); |
| 984 | | } |
| 985 | | $this->entry->$tag = join('',$this->in_content); |
| 986 | | $this->in_content = array(); |
| 987 | | } else { |
| 988 | | $endtag = $this->ns_to_prefix($name); |
| 989 | | if (strpos($this->in_content[count($this->in_content)-1], '<' . $endtag) !== false) { |
| 990 | | array_push($this->in_content, "/>"); |
| 991 | | } else { |
| 992 | | array_push($this->in_content, "</$endtag>"); |
| 993 | | } |
| 994 | | } |
| 995 | | } |
| 996 | | |
| 997 | | array_shift($this->ns_contexts); |
| 998 | | |
| 999 | | #print str_repeat(" ", $this->depth * $this->indent) . "end_element('$name')" ."\n"; |
| 1000 | | |
| 1001 | | $this->depth--; |
| 1002 | | } |
| 1003 | | |
| 1004 | | function start_ns($parser, $prefix, $uri) { |
| 1005 | | #print str_repeat(" ", $this->depth * $this->indent) . "starting: " . $prefix . ":" . $uri . "\n"; |
| 1006 | | array_push($this->ns_decls, array($prefix,$uri)); |
| 1007 | | } |
| 1008 | | |
| 1009 | | function end_ns($parser, $prefix) { |
| 1010 | | #print str_repeat(" ", $this->depth * $this->indent) . "ending: #" . $prefix . "#\n"; |
| 1011 | | } |
| 1012 | | |
| 1013 | | function cdata($parser, $data) { |
| 1014 | | #print str_repeat(" ", $this->depth * $this->indent) . "data: #" . $data . "#\n"; |
| 1015 | | if(!empty($this->in_content)) { |
| 1016 | | // handle self-closing tags (case: text node found, need to close element started) |
| 1017 | | if (strpos($this->in_content[count($this->in_content)-1], '<') !== false) { |
| 1018 | | array_push($this->in_content, ">"); |
| 1019 | | } |
| 1020 | | array_push($this->in_content, $this->xml_escape($data)); |
| 1021 | | } |
| 1022 | | } |
| 1023 | | |
| 1024 | | function _default($parser, $data) { |
| 1025 | | # when does this gets called? |
| 1026 | | } |
| 1027 | | |
| 1028 | | |
| 1029 | | function ns_to_prefix($qname) { |
| 1030 | | $components = split(":", $qname); |
| 1031 | | $name = array_pop($components); |
| 1032 | | |
| 1033 | | if(!empty($components)) { |
| 1034 | | $ns = join(":",$components); |
| 1035 | | foreach($this->ns_contexts as $context) { |
| 1036 | | foreach($context as $mapping) { |
| 1037 | | if($mapping[1] == $ns && strlen($mapping[0]) > 0) { |
| 1038 | | return "$mapping[0]:$name"; |
| 1039 | | } |
| 1040 | | } |
| 1041 | | } |
| 1042 | | } |
| 1043 | | return $name; |
| 1044 | | } |
| 1045 | | |
| 1046 | | function xml_escape($string) |
| 1047 | | { |
| 1048 | | return str_replace(array('&','"',"'",'<','>'), |
| 1049 | | array('&','"',''','<','>'), |
| 1050 | | $string ); |
| 1051 | | } |
| 1052 | | } |
| 1053 | | |