|
| 1 | +/**@file |
| 2 | + * @brief Swift-HTML-Parser |
| 3 | + * @author _tid_ |
| 4 | + */ |
| 5 | + |
| 6 | +/** |
| 7 | + * HTMLNode |
| 8 | + */ |
| 9 | +class HTMLNode { |
| 10 | + enum HTMLNodeType : String { |
| 11 | + case HTMLUnkownNode = "" |
| 12 | + case HTMLHrefNode = "href" |
| 13 | + case HTMLTextNode = "text" |
| 14 | + case HTMLCodeNode = "code" |
| 15 | + case HTMLSpanNode = "span" |
| 16 | + case HTMLPNode = "p" |
| 17 | + case HTMLLiNode = "li" |
| 18 | + case HTMLUiNode = "ui" |
| 19 | + case HTMLImageNode = "image" |
| 20 | + case HTMLOlNode = "ol" |
| 21 | + case HTMLStrongNode = "strong" |
| 22 | + case HTMLPreNode = "pre" |
| 23 | + case HTMLBlockQuoteNode = "blockquote" |
| 24 | + } |
| 25 | + |
| 26 | + var doc : htmlDocPtr |
| 27 | + var node : xmlNode? |
| 28 | + let nodeType : HTMLNodeType = HTMLNodeType.HTMLUnkownNode |
| 29 | + |
| 30 | + /** |
| 31 | + * 親ノード |
| 32 | + */ |
| 33 | + var parent : HTMLNode? { |
| 34 | + if let p = self.node?.parent { |
| 35 | + return HTMLNode(doc: self.doc, node: p) |
| 36 | + } |
| 37 | + return nil |
| 38 | + } |
| 39 | + |
| 40 | + /** |
| 41 | + * 次ノード |
| 42 | + */ |
| 43 | + var next : HTMLNode? { |
| 44 | + if let n : UnsafePointer<xmlNode> = node?.next { |
| 45 | + if n { |
| 46 | + return HTMLNode(doc: doc, node: n) |
| 47 | + } |
| 48 | + } |
| 49 | + return nil |
| 50 | + } |
| 51 | + |
| 52 | + /** |
| 53 | + * 子ノード |
| 54 | + */ |
| 55 | + var child : HTMLNode? { |
| 56 | + if let c = node?.children { |
| 57 | + if c { |
| 58 | + return HTMLNode(doc: doc, node: c) |
| 59 | + } |
| 60 | + } |
| 61 | + return nil |
| 62 | + } |
| 63 | + |
| 64 | + /** |
| 65 | + * クラス名 |
| 66 | + */ |
| 67 | + var className : String { |
| 68 | + return getAttributeNamed("class") |
| 69 | + } |
| 70 | + |
| 71 | + /** |
| 72 | + * タグ名 |
| 73 | + */ |
| 74 | + var tagName : String { |
| 75 | + if node { |
| 76 | + return ConvXmlCharToString(self.node!.name) |
| 77 | + } |
| 78 | + return "" |
| 79 | + } |
| 80 | + |
| 81 | + /** |
| 82 | + * コンテンツ |
| 83 | + */ |
| 84 | + var contents : String { |
| 85 | + if node { |
| 86 | + var n = self.node!.children |
| 87 | + if n { |
| 88 | + return ConvXmlCharToString(n.memory.content) |
| 89 | + } |
| 90 | + } |
| 91 | + return "" |
| 92 | + } |
| 93 | + |
| 94 | + /** |
| 95 | + * Initializer |
| 96 | + * @param[in] doc xmlDoc |
| 97 | + */ |
| 98 | + init(doc: htmlDocPtr = nil) { |
| 99 | + self.doc = doc |
| 100 | + var node = xmlDocGetRootElement(doc) |
| 101 | + if node { |
| 102 | + self.node = node.memory |
| 103 | + } |
| 104 | + } |
| 105 | + |
| 106 | + init(doc: htmlDocPtr, node: UnsafePointer<xmlNode>) { |
| 107 | + self.doc = doc |
| 108 | + self.node = node.memory |
| 109 | + |
| 110 | + if let type = HTMLNodeType.fromRaw(tagName) { |
| 111 | + nodeType = type |
| 112 | + } |
| 113 | + } |
| 114 | + |
| 115 | + /** |
| 116 | + * 属性名を取得する |
| 117 | + * @param[in] name 属性 |
| 118 | + * @return 属性名 |
| 119 | + */ |
| 120 | + func getAttributeNamed(name: String) -> String { |
| 121 | + for var attr : xmlAttrPtr = node!.properties; attr; attr = attr.memory.next { |
| 122 | + if attr { |
| 123 | + var mem = attr.memory |
| 124 | + |
| 125 | + if name == ConvXmlCharToString(mem.name) { |
| 126 | + return ConvXmlCharToString(mem.children.memory.content) |
| 127 | + } |
| 128 | + } |
| 129 | + } |
| 130 | + |
| 131 | + return "" |
| 132 | + } |
| 133 | + |
| 134 | + /** |
| 135 | + * タグ名に一致する全ての子ノードを探す |
| 136 | + * @param[in] tagName タグ名 |
| 137 | + * @return 子ノードの配列 |
| 138 | + */ |
| 139 | + func findChildTags(tagName: String) -> HTMLNode[] { |
| 140 | + var nodes : HTMLNode[] = [] |
| 141 | + |
| 142 | + return findChildTags(tagName, node: self.child, retAry: &nodes) |
| 143 | + } |
| 144 | + |
| 145 | + func findChildTags(tagName: String, node: HTMLNode?, inout retAry: HTMLNode[] ) -> HTMLNode[] { |
| 146 | + if node { |
| 147 | + for n in node! { |
| 148 | + if n.tagName == tagName { |
| 149 | + retAry.append(n) |
| 150 | + } |
| 151 | + |
| 152 | + findChildTags(tagName, node: n.child, retAry: &retAry) |
| 153 | + } |
| 154 | + } |
| 155 | + return retAry |
| 156 | + } |
| 157 | + |
| 158 | + /** |
| 159 | + * タグ名で子ノードを探す |
| 160 | + * @param[in] tagName タグ名 |
| 161 | + * @return 子ノード。見つからなければnil |
| 162 | + */ |
| 163 | + func findChildTag(tagName: String) -> HTMLNode? { |
| 164 | + return findChildTag(tagName, node: self) |
| 165 | + } |
| 166 | + |
| 167 | + func findChildTag(tagName: String, node: HTMLNode?) -> HTMLNode? { |
| 168 | + if node == nil { |
| 169 | + return nil |
| 170 | + } |
| 171 | + |
| 172 | + for curNode in node! { |
| 173 | + if tagName == curNode.tagName { |
| 174 | + return curNode |
| 175 | + } |
| 176 | + |
| 177 | + if let c = curNode.child { |
| 178 | + if let n = findChildTag(tagName, node: c) { |
| 179 | + return n |
| 180 | + } |
| 181 | + } |
| 182 | + } |
| 183 | + |
| 184 | + return nil |
| 185 | + } |
| 186 | +} |
| 187 | + |
| 188 | +extension HTMLNode : Sequence { |
| 189 | + func generate() -> HTMLNodeGenerator { |
| 190 | + return HTMLNodeGenerator(node: self) |
| 191 | + } |
| 192 | +} |
| 193 | + |
| 194 | +/** |
| 195 | +* HTMLNodeGenerator |
| 196 | +*/ |
| 197 | +class HTMLNodeGenerator : Generator { |
| 198 | + var node : HTMLNode? |
| 199 | + |
| 200 | + init(node: HTMLNode?) { |
| 201 | + self.node = node |
| 202 | + } |
| 203 | + |
| 204 | + func next() -> HTMLNode? { |
| 205 | + var temp = node? |
| 206 | + node = node?.next |
| 207 | + return temp |
| 208 | + } |
| 209 | +} |
| 210 | + |
0 commit comments