1 /* Licensed under GPL - see LICENSE file for details */
\r
14 #define XML_LETTER 1
\r
15 #define XML_NUMBER 2
\r
19 #define XML_EQUALS 32
\r
20 #define XML_CLOSE 64
\r
21 #define XML_QUOTE 128
\r
22 #define XML_OTHER 256
\r
24 #define XML_ALL 0xFFFFFFFF
\r
27 typedef struct XMLBUF
\r
37 /* Allocate a new XmlNode */
\r
38 static XmlNode* xml_new(char * name)
\r
40 XmlNode * ret = malloc(sizeof(XmlNode));
\r
41 if(!ret)return NULL;
\r
45 ret->child = ret->next = NULL;
\r
51 /* free a previously allocated XmlNode */
\r
52 void xml_free(XmlNode *target)
\r
55 for(i=0; i<target->nattrib*2; i++)
\r
56 if(target->attrib[i])
\r
57 free(target->attrib[i]);
\r
59 if(target->attrib)free(target->attrib);
\r
60 if(target->child)xml_free(target->child);
\r
61 if(target->next)xml_free(target->next);
\r
66 /* Raise flags if we have a character of special meaning.
\r
67 * This is where I've hidden the switch statements :-p
\r
69 static int is_special(char item)
\r
71 if((item >= 'a' && item <= 'z') || (item >= 'A' && item <='Z'))
\r
73 if( item >= '0' && item <='9' )
\r
75 if( item == 0x20 || item == '\t' || item == 0x0D || item == 0x0A )
\r
85 if( item == '"' || item == '\'' )
\r
90 /* Refresh the buffer, if possible */
\r
91 static void xml_read_file(XMLBUF *xml)
\r
97 size = fread( xml->buf, 1, xml->len, xml->fptr);
\r
98 if( size != xml->len )
\r
107 /* All reading of the XML buffer done through these two functions */
\r
108 /*** read a byte without advancing the offset */
\r
109 static char xml_peek(XMLBUF *xml)
\r
111 return xml->buf[xml->read_index];
\r
114 /*** read a byte and advance the offset */
\r
115 static char xml_read_byte(XMLBUF *xml)
\r
117 char ret = xml_peek(xml);
\r
119 if(xml->read_index >= xml->len)
\r
123 xml->read_index = xml->len;
\r
126 xml->read_index = 0 ;
\r
127 xml_read_file(xml);
\r
133 /* skip over bytes matching the is_special mask */
\r
134 static void xml_skip( XMLBUF *xml, int mask)
\r
136 while( is_special(xml_peek(xml)) & mask && !(xml->eof && xml->read_index >= xml->len) )
\r
137 xml_read_byte(xml);
\r
141 /* character matching tests for the feed functions */
\r
142 static char quotechar = 0;
\r
143 static int test_quote(const char x)
\r
145 static int escaped=0;
\r
146 if( escaped || '\\' == x )
\r
148 escaped = !escaped;
\r
151 if( x != quotechar )
\r
156 static int feed_mask = 0;
\r
157 static int test_mask(const char x)
\r
159 return !(is_special(x) & feed_mask);
\r
163 * char* xml_feed(x, test)
\r
165 * Reads as many contiguous chars that pass test() into a newly allocated
\r
168 * Instead of calling xml_read_byte and flogging realloc() for each byte,
\r
169 * it checks the buffer itself.
\r
171 static char* xml_feed( XMLBUF *xml, int (*test)(char) )
\r
173 int offset = xml->read_index;
\r
178 /* perform first and N middle realloc()'s */
\r
179 while( test(xml->buf[offset]) )
\r
183 if(offset >= xml->len)
\r
185 delta = offset - xml->read_index;
\r
186 ret = realloc(ret, size + delta + 1);
\r
187 memcpy(ret+size, xml->buf + xml->read_index, delta);
\r
190 if(xml->eof)return ret;
\r
191 xml_read_file(xml);
\r
192 xml->read_index = 0;
\r
196 /* perform final realloc() if needed */
\r
197 if(offset > xml->read_index)
\r
199 delta = offset - xml->read_index;
\r
200 ret = realloc(ret, size + delta + 1);
\r
201 memcpy(ret+size, xml->buf + xml->read_index, delta);
\r
202 xml->read_index = offset;
\r
209 /* this reads attributes from tags, of the form...
\r
211 * <tag attr1="some arguments" attr2=argument>
\r
213 * It is aware of quotes, and will allow anything inside quoted arguments
\r
215 static void xml_read_attr(struct XMLBUF *xml, XmlNode *node)
\r
219 // how does this tag finish?
\r
222 if( is_special(xml_peek(xml)) & (XML_CLOSE | XML_SLASH) )
\r
225 n = ++node->nattrib;
\r
226 node->attrib = realloc(node->attrib, n * 2 * sizeof(char*) );
\r
227 node->attrib[--n*2+1] = 0;
\r
229 feed_mask = XML_EQUALS | XML_SPACE | XML_CLOSE | XML_SLASH;
\r
230 node->attrib[n*2] = xml_feed(xml, test_mask );
\r
231 if( xml_peek(xml) == '=' )
\r
233 xml_read_byte(xml);
\r
234 if( is_special(xml_peek(xml)) & XML_QUOTE )
\r
236 quotechar = xml_read_byte(xml);
\r
237 node->attrib[n*2+1] = xml_feed(xml, test_quote);
\r
238 xml_read_byte(xml);
\r
242 feed_mask = XML_SPACE | XML_CLOSE | XML_SLASH;
\r
243 node->attrib[n*2+1] = xml_feed(xml, test_mask);
\r
246 xml_skip(xml, XML_SPACE);
\r
250 /* The big decision maker, is it a regular node, or a text node.
\r
251 * If it's a node, it will check if it should have children, and if so
\r
252 * will recurse over them.
\r
253 * Text nodes don't have children, so no recursing.
\r
255 static XmlNode* xml_parse(struct XMLBUF *xml)
\r
260 XmlNode **this, *ret = NULL;
\r
264 xml_skip(xml, XML_SPACE); // skip whitespace
\r
266 while( (xml->read_index < xml->len) || !xml->eof )
\r
268 switch(is_special(xml_peek(xml)))
\r
271 xml_read_byte(xml);
\r
272 if(xml_peek(xml) == '/')
\r
273 return ret; // parents close tag
\r
274 // read the tag name
\r
275 feed_mask = XML_SPACE | XML_SLASH | XML_CLOSE;
\r
276 *this = xml_new( xml_feed(xml, test_mask));
\r
277 xml_skip(xml, XML_SPACE); // skip any whitespace
\r
279 xml_read_attr(xml, *this); // read attributes
\r
281 // how does this tag finish?
\r
282 switch(is_special(xml_peek(xml)))
\r
284 case XML_CLOSE: // child-nodes ahead
\r
285 xml_read_byte(xml);
\r
286 (*this)->child = xml_parse(xml);
\r
287 xml_skip(xml, XML_ALL ^ XML_CLOSE);
\r
288 xml_read_byte(xml);
\r
290 case XML_SLASH: // self closing tag
\r
291 xml_read_byte(xml);
\r
292 xml_read_byte(xml);
\r
297 default: // text node
\r
298 *this = xml_new(0);
\r
299 xml_skip(xml, XML_SPACE); // skip any whitespace
\r
300 feed_mask = XML_OPEN;
\r
301 (*this)->nattrib=1;
\r
302 (*this)->attrib = malloc(sizeof(char*)*2);
\r
303 (*this)->attrib[1] = NULL;
\r
304 tmp = (*this)->attrib[0] = xml_feed(xml, test_mask);
\r
306 /* trim the whitespace off the end of text nodes,
\r
307 * by overwriting the spaces will null termination. */
\r
308 toff = strlen(tmp)-1;
\r
309 while( ( is_special(tmp[toff]) & XML_SPACE ) )
\r
317 this = &(*this)->next;
\r
318 xml_skip(xml, XML_SPACE); // skip whitespace
\r
325 /* bootstrap the structures for xml_parse() to be able to get started */
\r
326 XmlNode* xml_load(const char * filename)
\r
329 XmlNode *ret = NULL;
\r
331 // printf("xml_load(\"%s\");\n", filename);
\r
334 xml.read_index = 0;
\r
335 xml.fptr = fopen(filename, "rb");
\r
339 xml.buf = malloc(BUFFER+1);
\r
343 goto xml_load_fail_malloc_buf;
\r
345 xml_read_file(&xml);
\r
347 ret = xml_parse(&xml);
\r
350 xml_load_fail_malloc_buf:
\r
355 /* very basic function that will get you the first node with a given name */
\r
356 XmlNode * xml_find(XmlNode *xml, const char *name)
\r
359 if(xml->name)if(!strcmp(xml->name, name))return xml;
\r
362 ret = xml_find(xml->child, name);
\r
367 ret = xml_find(xml->next, name);
\r
373 /* very basic attribute lookup function */
\r
374 char* xml_attr(XmlNode *x, const char *name)
\r
377 for(i=0; i<x->nattrib; i++)
\r
379 if(!strcmp(x->attrib[i*2], name))
\r
380 return x->attrib[i*2+1];
\r