git.ozlabs.org Git - ccan/blob - ccan/ttxml/ttxml.c

   1 /* Licensed under GPL - see LICENSE file for details */\r
   2 \r
   3 #include <stdlib.h>\r
   4 #include <string.h>\r
   5 #include <stdio.h>\r
   6 \r
   7 #include "ttxml.h"\r
   8 \r
   9 \r
  10 #define BUFFER 3264\r
  11 \r
  12 #define XML_LETTER      1\r
  13 #define XML_NUMBER      2\r
  14 #define XML_SPACE       4\r
  15 #define XML_SLASH       8\r
  16 #define XML_OPEN        16\r
  17 #define XML_EQUALS      32\r
  18 #define XML_CLOSE       64\r
  19 #define XML_QUOTE       128\r
  20 #define XML_OTHER       256\r
  21 \r
  22 #define XML_ALL 0xFFFFFFFF\r
  23 \r
  24 \r
  25 typedef struct XMLBUF\r
  26 {\r
  27         FILE * fptr;\r
  28         char * buf;\r
  29         int len;\r
  30         int read_index;\r
  31         int eof;\r
  32 } XMLBUF;\r
  33 \r
  34 \r
  35 /* Allocate a new XmlNode */\r
  36 XmlNode* xml_new(char * name)\r
  37 {\r
  38         XmlNode * ret = malloc(sizeof(XmlNode));\r
  39         if(!ret)return NULL;\r
  40 \r
  41         ret->attrib = NULL;\r
  42         ret->nattrib = 0;\r
  43         ret->child = ret->next = NULL;\r
  44 \r
  45         ret->name = name;\r
  46         return ret;\r
  47 }\r
  48 \r
  49 /* free a previously allocated XmlNode */\r
  50 void xml_free(XmlNode *target)\r
  51 {\r
  52         int i;\r
  53         for(i=0; i<target->nattrib*2; i++)\r
  54                 if(target->attrib[i])\r
  55                         free(target->attrib[i]);\r
  56 \r
  57         if(target->attrib)free(target->attrib);\r
  58         if(target->child)xml_free(target->child);\r
  59         if(target->next)xml_free(target->next);\r
  60         free(target->name);\r
  61         free(target);\r
  62 }\r
  63 \r
  64 /* Raise flags if we have a character of special meaning.\r
  65  * This is where I've hidden the switch statements :-p\r
  66  */\r
  67 static int is_special(char item)\r
  68 {\r
  69         if((item >= 'a' && item <= 'z') || (item >= 'A' && item <='Z'))\r
  70                 return XML_LETTER;\r
  71         if( item >= '0' && item <='9' )\r
  72                 return XML_NUMBER;\r
  73         if( item == 0x20 || item == '\t' ||     item == 0x0D || item == 0x0A )\r
  74                 return XML_SPACE;\r
  75         if( item == '/' )\r
  76                 return XML_SLASH;\r
  77         if( item == '<' )\r
  78                 return XML_OPEN;\r
  79         if( item == '=' )\r
  80                 return XML_EQUALS;\r
  81         if( item == '>' )\r
  82                 return XML_CLOSE;\r
  83         if( item == '"' || item == '\'' )\r
  84                 return XML_QUOTE;\r
  85         return 128;\r
  86 }\r
  87 \r
  88 /* Refresh the buffer, if possible */\r
  89 static void xml_read_file(XMLBUF *xml)\r
  90 {\r
  91         int size;\r
  92         \r
  93         if(xml->eof)return;\r
  94           \r
  95         size = fread( xml->buf, 1, xml->len, xml->fptr);\r
  96         if( size != xml->len )\r
  97         {\r
  98                 printf("Buffer reduction\n");\r
  99                 xml->len = size;\r
 100                 xml->buf[size]=0;\r
 101                 xml->eof = 1;\r
 102         }\r
 103 }\r
 104 \r
 105 \r
 106 /* All reading of the XML buffer done through these two functions */\r
 107 /*** read a byte without advancing the offset */\r
 108 static char xml_peek(XMLBUF *xml)\r
 109 {\r
 110         return xml->buf[xml->read_index];\r
 111 }\r
 112 \r
 113 /*** read a byte and advance the offset */\r
 114 static char xml_read_byte(XMLBUF *xml)\r
 115 {\r
 116         char ret = xml_peek(xml);\r
 117         xml->read_index++;\r
 118         if(xml->read_index >= xml->len)\r
 119         {\r
 120                 if(xml->eof)\r
 121                 {\r
 122                   xml->read_index = xml->len;\r
 123                   return ret;\r
 124                 }\r
 125                 xml->read_index = 0 ;\r
 126                 xml_read_file(xml);\r
 127         }\r
 128         return ret;\r
 129 }\r
 130 \r
 131 \r
 132 /* skip over bytes matching the is_special mask */\r
 133 static void xml_skip( XMLBUF *xml, int mask)\r
 134 {\r
 135         while( is_special(xml_peek(xml)) & mask && !(xml->eof && xml->read_index >= xml->len) )\r
 136                 xml_read_byte(xml);\r
 137 }\r
 138 \r
 139 \r
 140 /* character matching tests for the feed functions */\r
 141 static char quotechar = 0;\r
 142 static int test_quote(const char x)\r
 143 {\r
 144         static int escaped=0;\r
 145         if( escaped || '\\' == x )\r
 146         {\r
 147                 escaped = !escaped;\r
 148                 return 1;\r
 149         }\r
 150         if( x != quotechar )\r
 151                 return 1;\r
 152         return 0;\r
 153 }\r
 154 \r
 155 static int feed_mask = 0;\r
 156 static int test_mask(const char x)\r
 157 {\r
 158         return !(is_special(x) & feed_mask);\r
 159 }\r
 160 \r
 161 /*\r
 162  * char* xml_feed(x, test)\r
 163  *\r
 164  * Reads as many contiguous chars that pass test() into a newly allocated\r
 165  * string.\r
 166  *\r
 167  * Instead of calling xml_read_byte and flogging realloc() for each byte,\r
 168  * it checks the buffer itself.\r
 169 */\r
 170 static char* xml_feed( XMLBUF *xml, int (*test)(char) )\r
 171 {\r
 172         int offset = xml->read_index;\r
 173         int delta;\r
 174         char *ret = NULL;\r
 175         int size = 0;\r
 176 \r
 177         /* perform first and N middle realloc()'s */\r
 178         while( test(xml->buf[offset]) )\r
 179         {\r
 180                 offset ++;\r
 181 \r
 182                 if(offset >= xml->len)\r
 183                 {\r
 184                         delta = offset - xml->read_index;\r
 185                         ret = realloc(ret, size + delta + 1);\r
 186                         memcpy(ret+size, xml->buf + xml->read_index, delta);\r
 187                         size += delta;\r
 188                         ret[size]=0;\r
 189                         if(xml->eof)return ret;\r
 190                         xml_read_file(xml);\r
 191                         xml->read_index = 0;\r
 192                         offset = 0;\r
 193                 }\r
 194         }\r
 195         /* perform final realloc() if needed */\r
 196         if(offset > xml->read_index)\r
 197         {\r
 198                 delta = offset - xml->read_index;\r
 199                 ret = realloc(ret, size + delta + 1);\r
 200                 memcpy(ret+size, xml->buf + xml->read_index, delta);\r
 201                 xml->read_index = offset;\r
 202                 size += delta;\r
 203                 ret[size]=0;\r
 204         }\r
 205         return ret;\r
 206 }\r
 207 \r
 208 /* this reads attributes from tags, of the form...\r
 209  *\r
 210  * <tag attr1="some arguments" attr2=argument>\r
 211  *\r
 212  * It is aware of quotes, and will allow anything inside quoted arguments\r
 213  */\r
 214 static void xml_read_attr(struct XMLBUF *xml, XmlNode *node)\r
 215 {\r
 216         int n=0;\r
 217 \r
 218         // how does this tag finish?\r
 219         while(xml->len)\r
 220         {\r
 221                 if( is_special(xml_peek(xml)) & (XML_CLOSE | XML_SLASH) )\r
 222                         return;\r
 223 \r
 224                 n = ++node->nattrib;\r
 225                 node->attrib = realloc(node->attrib, n * 2 * sizeof(char*) );\r
 226                 node->attrib[--n*2+1] = 0;\r
 227                 \r
 228                 feed_mask = XML_EQUALS | XML_SPACE | XML_CLOSE | XML_SLASH;\r
 229                 node->attrib[n*2] = xml_feed(xml, test_mask );\r
 230                 if( xml_peek(xml) == '=' )\r
 231                 {\r
 232                         xml_read_byte(xml);\r
 233                         if( is_special(xml_peek(xml)) & XML_QUOTE )\r
 234                         {\r
 235                                 quotechar = xml_read_byte(xml);\r
 236                                 node->attrib[n*2+1] = xml_feed(xml, test_quote);\r
 237                                 xml_read_byte(xml);\r
 238                         }\r
 239                         else\r
 240                         {\r
 241                                 feed_mask = XML_SPACE | XML_CLOSE | XML_SLASH;\r
 242                                 node->attrib[n*2+1] = xml_feed(xml, test_mask);\r
 243                         }\r
 244                 }\r
 245                 xml_skip(xml, XML_SPACE);\r
 246         }\r
 247 }\r
 248 \r
 249 /* The big decision maker, is it a regular node, or a text node.\r
 250  * If it's a node, it will check if it should have children, and if so\r
 251  * will recurse over them.\r
 252  * Text nodes don't have children, so no recursing.\r
 253  */\r
 254 static XmlNode* xml_parse(struct XMLBUF *xml)\r
 255 {\r
 256         int offset;\r
 257         int toff;\r
 258         char *tmp;\r
 259         XmlNode **this, *ret = NULL;\r
 260         \r
 261         this = &ret;\r
 262 \r
 263         xml_skip(xml, XML_SPACE);       // skip whitespace\r
 264         offset=0;\r
 265         while( (xml->read_index < xml->len) || !xml->eof )\r
 266         {\r
 267                 switch(is_special(xml_peek(xml)))\r
 268                 {\r
 269                         case XML_OPEN:\r
 270                                 xml_read_byte(xml);\r
 271                                 if(xml_peek(xml) == '/')\r
 272                                         return ret;             // parents close tag\r
 273                                 // read the tag name\r
 274                                 feed_mask = XML_SPACE | XML_SLASH | XML_CLOSE;\r
 275                                 *this = xml_new( xml_feed(xml, test_mask));\r
 276                                 xml_skip(xml, XML_SPACE);       // skip any whitespace\r
 277 \r
 278                                 xml_read_attr(xml, *this);      // read attributes\r
 279 \r
 280                                 // how does this tag finish?\r
 281                                 switch(is_special(xml_peek(xml)))\r
 282                                 {\r
 283                                         case XML_CLOSE:         // child-nodes ahead\r
 284                                                 xml_read_byte(xml);\r
 285                                                 (*this)->child = xml_parse(xml);\r
 286                                                 xml_skip(xml, XML_ALL ^ XML_CLOSE);\r
 287                                                 xml_read_byte(xml);\r
 288                                                 break;\r
 289                                         case XML_SLASH:         // self closing tag\r
 290                                                 xml_read_byte(xml);\r
 291                                                 xml_read_byte(xml);\r
 292                                                 break;\r
 293                                 }\r
 294                                 break;\r
 295 \r
 296                         default:        // text node\r
 297                                 *this = xml_new(0);\r
 298                                 xml_skip(xml, XML_SPACE);       // skip any whitespace\r
 299                                 feed_mask = XML_OPEN;\r
 300                                 (*this)->nattrib=1;\r
 301                                 (*this)->attrib = malloc(sizeof(char*)*2);\r
 302                                 (*this)->attrib[1] = NULL;\r
 303                                 tmp = (*this)->attrib[0] = xml_feed(xml, test_mask);\r
 304 \r
 305                                 /* trim the whitespace off the end of text nodes,\r
 306                                  * by overwriting the spaces will null termination. */\r
 307                                 toff = strlen(tmp)-1;\r
 308                                 while( ( is_special(tmp[toff]) & XML_SPACE ) )\r
 309                                 {\r
 310                                         tmp[toff] = 0;\r
 311                                         toff --;\r
 312                                 }\r
 313 \r
 314                                 break;\r
 315                 }\r
 316                 this = &(*this)->next; \r
 317                 xml_skip(xml, XML_SPACE);       // skip whitespace\r
 318         }       \r
 319 \r
 320         return ret;\r
 321 }\r
 322 \r
 323 \r
 324 /* bootstrap the structures for xml_parse() to be able to get started */\r
 325 XmlNode* xml_load(const char * filename)\r
 326 {\r
 327         struct XMLBUF xml;\r
 328         XmlNode *ret = NULL;\r
 329 \r
 330 //      printf("xml_load(\"%s\");\n", filename);\r
 331 \r
 332         xml.eof = 0;\r
 333         xml.read_index = 0;\r
 334         xml.fptr = fopen(filename, "rb");\r
 335         if(!xml.fptr)\r
 336         {\r
 337                 printf("Opening file failed\n");\r
 338                 return NULL;\r
 339         }\r
 340 \r
 341         xml.buf = malloc(BUFFER+1);\r
 342         xml.buf[BUFFER]=0;\r
 343         xml.len = BUFFER;\r
 344         if(!xml.buf)\r
 345                 goto xml_load_fail_malloc_buf;\r
 346         \r
 347         xml_read_file(&xml);\r
 348 \r
 349         ret = xml_parse(&xml);\r
 350 \r
 351         free(xml.buf);\r
 352 xml_load_fail_malloc_buf:\r
 353         fclose(xml.fptr);\r
 354         return ret;\r
 355 }\r
 356 \r
 357 /* very basic function that will get you the first node with a given name */\r
 358 XmlNode * xml_find(XmlNode *xml, const char *name)\r
 359 {\r
 360         XmlNode * ret;\r
 361         if(xml->name)if(!strcmp(xml->name, name))return xml;\r
 362         if(xml->child)\r
 363         {\r
 364                 ret = xml_find(xml->child, name);\r
 365                 if(ret)return ret;\r
 366         }\r
 367         if(xml->next)\r
 368         {\r
 369                 ret = xml_find(xml->next, name);\r
 370                 if(ret)return ret;\r
 371         }\r
 372         return NULL;\r
 373 }\r
 374 \r
 375 /* very basic attribute lookup function */\r
 376 char* xml_attr(XmlNode *x, const char *name)\r
 377 {\r
 378         int i;\r
 379         for(i=0; i<x->nattrib; i++)\r
 380                 if(x->attrib[i*2])\r
 381                         if(!strcmp(x->attrib[i*2], name))\r
 382                                 return x->attrib[i*2+1];\r
 383         return 0;\r
 384 }\r
 385 \r
 386 \r
 387 #ifdef TEST\r
 388 /* print out the heirarchy of an XML file, useful for debugging */\r
 389 void xp(XmlNode *x, int level, int max)\r
 390 {\r
 391         int i;\r
 392         char text[] = "text";\r
 393         char *name = text;\r
 394         if(level > max)return;\r
 395         if(!x)return;\r
 396         if(x->name)name = x->name;\r
 397         for(i=0; i<level; i++)printf("    ");\r
 398         printf("%s:", name);\r
 399         if(x->name)\r
 400         for(i=0; i<x->nattrib; i++)\r
 401                 printf("%s=\"%s\",", x->attrib[i*2], x->attrib[i*2+1]);\r
 402         else printf("%s", x->attrib[0]);\r
 403         printf("\n");\r
 404         if(x->child)xp(x->child, level+1, max);\r
 405         if(x->next)xp(x->next, level, max);\r
 406 }\r
 407 \r
 408 \r
 409 int main(int argc, char *argv[])\r
 410 {\r
 411         XmlNode *x, *tmp;\r
 412         \r
 413         if(!argv[1])\r
 414         {\r
 415                 printf("USAGE: %s name\n\t reads name where name is an XML file.\n",\r
 416                                 argv[0]);\r
 417                 return 1;\r
 418         }\r
 419 \r
 420 #ifdef PROFILE\r
 421         for(int i=0; i<1000; i++)\r
 422         {\r
 423 #endif\r
 424                 x = xml_load(argv[1]);\r
 425 \r
 426                 if(!x)\r
 427                 {\r
 428                         printf("Failed to load.\n");\r
 429                         return 2;\r
 430                 }\r
 431 #ifndef PROFILE\r
 432                 xp(x, 1, 20);\r
 433 #endif\r
 434                 xml_free(x);\r
 435 #ifdef PROFILE\r
 436         }\r
 437 #endif\r
 438 \r
 439         return 0;\r
 440 }\r
 441 #endif\r
 442 \r