libxml

  1 /**
  2  * section: Tree
  3  * synopsis: Navigates a tree to print element names
  4  * purpose: Parse a file to a tree, use xmlDocGetRootElement() to
  5  *          get the root element, then walk the document and print
  6  *          all the element name in document order.
  7  * usage: tree1 filename_or_URL
  8  * test: tree1 test2.xml > tree1.tmp && diff tree1.tmp $(srcdir)/tree1.res
  9  * author: Dodji Seketeli
 10  * copy: see Copyright for the status of this software.
 11  */
 12 #include <stdio.h>
 13 #include <string.h>
 14 #include <libxml/parser.h>
 15 #include <libxml/tree.h>
 16 
 17 /*
 18  *To compile this file using gcc you can type
 19  *  gcc `xml2-config --cflags --libs` -o tree1 tree1.c
 20  *Run this program
 21  *  ./tree1 test.xml
 22  */
 23 
 24 static const char* LEVELS[] = {"", " ", "  ", "   ", "    ", "     ", "      ", "       " };
 25 
 26 static void printTree(xmlNode * a_node, int level);
 27 
 28 /**
 29  * print_element_names:
 30  * @a_node: the initial xml node to consider.
 31  *
 32  * 打印所有兄弟节点和子节点的名字.
 33  */
 34 static void print_element_names(xmlNode * a_node, const char* msg);
 35 
 36 
 37 // 根据标签名称获取节点（可以实现更加复杂的逻辑，获取指定节点）
 38 static xmlNode *getNode(xmlNode *rootNode, const char* tag, xmlNode **parentNode);
 39 
 40 // 删除当前节点，但是保留子节点
 41 static void removeNode(xmlNode *parentNode, xmlNode *nodeToDelete);
 42 
 43 // 用一个父节点包装子节点
 44 static void wrapWithNode(xmlNode *parentNode, xmlNode *node, xmlNode *newNode);
 45 
 46 // 增加一个新节点
 47 static void appendNewChildNode(xmlNode *parentNode, xmlNode *newNode);
 48 
 49 /**
 50  * print_element_names:
 51  * @a_node: the initial xml node to consider.
 52  *
 53  * Prints the names of the all the xml elements
 54  * that are siblings or children of a given xml node.
 55  */
 56 static int test_removeNode(const char* filepath);
 57 
 58 /**
 59  * print_element_names:
 60  * @a_node: the initial xml node to consider.
 61  *
 62  * Prints the names of the all the xml elements
 63  * that are siblings or children of a given xml node.
 64  */
 65 static int test_wrapWithNode(const char* filepath);
 66 
 67 int main(int argc, char **argv)
 68 {
 69     if (argc != 2) {
 70         printf("error: invalid arguments");
 71         return -1;
 72     }
 73 
 74    /*
 75     * this initialize the library and check potential ABI mismatches
 76     * between the version it was compiled for and the actual shared
 77     * library used.
 78     */
 79    LIBXML_TEST_VERSION
 80 
 81     printf("test: removeNode:\n");
 82     test_removeNode(argv[1]);
 83 
 84     printf("\n\ntest: wrapWithNode\n");
 85     test_wrapWithNode(argv[1]);
 86 
 87    /*
 88     *Free the global variables that may
 89     *have been allocated by the parser.
 90     */
 91    xmlCleanupParser();
 92 
 93    return 0;
 94 }
 95 
 96 void print_element_names(xmlNode * a_node, const char* msg)
 97 {
 98     xmlNode *cur_node = NULL;
 99 
100     if (msg != NULL && strlen(msg) > 0) {
101         printf("print: %s\n", msg);
102     }
103 
104     for (cur_node = a_node; cur_node; cur_node = cur_node->next) {
105         if (cur_node->type == XML_ELEMENT_NODE) {
106             printf("node type: Element, name: %s\n", cur_node->name);
107         }
108 
109         print_element_names(cur_node->children, "");
110     }
111 }
112 
113 void printTree(xmlNode * a_node, int level)
114 {
115     xmlNode *cur_node = NULL;
116 
117     //printf("%s", LEVELS[level]);
118 
119     for (cur_node = a_node; cur_node; cur_node = cur_node->next) {
120         if (cur_node->type == XML_ELEMENT_NODE) {
121             printf("%s%s <%d>\n", LEVELS[level], cur_node->name, cur_node->type);
122             printTree(cur_node->children, level + 1);
123         } else {
124             printf("%s#%s <%d>\n", LEVELS[level], cur_node->name, cur_node->type);
125         }
126     }
127 }
128 
129 xmlNode *getNode(xmlNode *rootNode, const char* tag, xmlNode **parentNode) {
130     xmlNode *cur = rootNode;
131     if ((cur->type == XML_ELEMENT_NODE) && (!xmlStrcmp(cur->name, (const xmlChar *)tag))){
132         *parentNode = NULL;
133         return cur;
134     }
135 
136     *parentNode = cur;
137     cur = cur->xmlChildrenNode;
138     while (cur != NULL) {
139         if ((cur->type == XML_ELEMENT_NODE) && (!xmlStrcmp(cur->name, (const xmlChar *)tag))){
140             return cur;
141         }
142 
143         if (cur->type == XML_ELEMENT_NODE) {
144             *parentNode = cur;
145         }
146         cur = cur->next;
147     }
148 
149     return NULL;
150 }
151 
152 // 删除当前节点，但是保留子节点
153 void removeNode(xmlNode *parentNode, xmlNode *nodeToDelete) {
154     if (nodeToDelete == NULL) {
155         printf("error: nodeToDelete is null");
156         return;
157     }
158 
159     xmlNodePtr siblingNode = nodeToDelete->next;
160 
161     while (siblingNode != NULL) {
162         if (siblingNode->type == XML_ELEMENT_NODE) {
163             printf("debug: found sibling: %s\n", siblingNode->name);
164             break;
165         }
166 
167         siblingNode = siblingNode->next;
168     }
169 
170     printf("debug: parentNode: %s, nodeToDelete: %s\n", parentNode->name, nodeToDelete->name);
171     printTree(parentNode, 0);
172 
173     xmlNode *childrenNode = nodeToDelete->children;
174     if (childrenNode == NULL) {
175         printf("warn: childrenNode is null\n");
176     }
177     //xmlUnlinkNode(nodeToDelete->children);
178 
179     xmlNodePtr nextChildNode = NULL;
180 
181     while (childrenNode != NULL) {
182         printf("debug: childrenNode: %s\n", childrenNode->name);
183         nextChildNode = childrenNode->next;
184         xmlUnlinkNode(childrenNode);
185 
186         if (siblingNode != NULL) {
187             printf("debug: addPreSibling: %s, sibling is %s\n", childrenNode->name, siblingNode->name);
188             xmlAddPrevSibling(siblingNode, nextChildNode);
189         } else {
190             printf("debug: addChild: %s, parent is %s\n", childrenNode->name, parentNode->name);
191             printTree(childrenNode, 0);
192             xmlAddChild(parentNode, childrenNode);
193         }
194 
195         childrenNode = nextChildNode;
196     }
197 
198     xmlUnlinkNode(nodeToDelete);    
199     xmlFreeNode(nodeToDelete);
200 }
201 
202 // 用一个父节点包装子节点
203 void wrapWithNode(xmlNode *parentNode, xmlNode *node, xmlNode *newNode) {
204     xmlUnlinkNode(node);
205     xmlAddChild(newNode, node);
206     xmlAddChild(parentNode, newNode);
207 }
208 
209 // 增加一个新节点
210 void appendNewChildNode(xmlNode *parentNode, xmlNode *newNode) {
211     xmlAddChild(parentNode, newNode);
212 }
213 
214 int test_removeNode(const char* filepath) {
215     xmlDoc *doc = NULL;
216     xmlNode *root_element = NULL;
217     xmlNode *parentNode = NULL;
218     xmlNode *curNode = NULL;
219 
220     /*parse the file and get the DOM */
221     doc = xmlReadFile(filepath, NULL, 0);
222 
223     if (doc == NULL) {
224         printf("error: could not parse file %s\n", filepath);
225     }
226 
227     /*Get the root element node */
228     root_element = xmlDocGetRootElement(doc);
229 
230     // 删除节点，但是保留子节点
231     curNode = getNode(root_element, "p", &parentNode);
232     if (curNode == NULL) {
233         printf("error: p node is not found");
234         return -1;
235     }
236     if (parentNode == NULL) {
237         // 根节点只能有一个子节点，这里就不处理了
238         printf("error: This is root node, should treat specially. root node should have only one node");
239         return -1;
240     }
241     removeNode(parentNode, curNode);
242 
243 
244     // 重新获取跟节点，应该是main了
245     root_element = xmlDocGetRootElement(doc);
246 
247     print_element_names(root_element, "after delete");
248 
249     /*free the document */
250     xmlFreeDoc(doc);
251     return 0;
252 }
253 
254 int test_wrapWithNode(const char* filepath) {
255     xmlDoc *doc = NULL;
256     xmlNode *root_element = NULL;
257     xmlNode *newNode = NULL;
258 
259     /*parse the file and get the DOM */
260     doc = xmlReadFile(filepath, NULL, 0);
261 
262     if (doc == NULL) {
263         printf("error: could not parse file %s\n", filepath);
264     }
265 
266     /*Get the root element node */
267     root_element = xmlDocGetRootElement(doc);
268 
269     // 增加一个父节点,根节点需要特殊处理
270     xmlUnlinkNode(root_element);
271     newNode = xmlNewNode(NULL, BAD_CAST "main");
272     xmlAddChild(newNode, root_element);
273     xmlDocSetRootElement(doc, newNode);
274     // 重新获取跟节点，应该是main了
275     root_element = xmlDocGetRootElement(doc);
276 
277     print_element_names(root_element, "after wrap");
278 
279     /*free the document */
280     xmlFreeDoc(doc);
281 
282     return 0;
283 }