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

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