stream/url.c

   1 /*
   2  * URL Helper
   3  * by Bertrand Baudet <bertrand_baudet@yahoo.com>
   4  * (C) 2001, MPlayer team.
   5  *
   6  */
   7
   8 #include <string.h>
   9 #include <stdlib.h>
  10 #include <stdio.h>
  11 #include <ctype.h>
  12 #include <inttypes.h>
  13
  14 #include "url.h"
  15 #include "mp_msg.h"
  16 #include "help_mp.h"
  17
  18 #ifndef SIZE_MAX
  19 #define SIZE_MAX ((size_t)-1)
  20 #endif
  21
  22 URL_t *url_redirect(URL_t **url, const char *redir) {
  23   URL_t *u = *url;
  24   URL_t *res;
  25   if (!strchr(redir, '/') || *redir == '/') {
  26     char *tmp;
  27     char *newurl = malloc(strlen(u->url) + strlen(redir) + 1);
  28     strcpy(newurl, u->url);
  29     if (*redir == '/') {
  30       redir++;
  31       tmp = strstr(newurl, "://");
  32       if (tmp) tmp = strchr(tmp + 3, '/');
  33     } else
  34       tmp = strrchr(newurl, '/');
  35     if (tmp) tmp[1] = 0;
  36     strcat(newurl, redir);
  37     res = url_new(newurl);
  38     free(newurl);
  39   } else
  40     res = url_new(redir);
  41   url_free(u);
  42   *url = res;
  43   return res;
  44 }
  45
  46 URL_t*
  47 url_new(const char* url) {
  48         int pos1, pos2,v6addr = 0;
  49         URL_t* Curl = NULL;
  50         char *escfilename=NULL;
  51         char *ptr1=NULL, *ptr2=NULL, *ptr3=NULL, *ptr4=NULL;
  52         int jumpSize = 3;
  53
  54         if( url==NULL ) return NULL;
  55
  56         if (strlen(url) > (SIZE_MAX / 3 - 1)) {
  57                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
  58                 goto err_out;
  59         }
  60         escfilename=malloc(strlen(url)*3+1);
  61         if (!escfilename ) {
  62                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
  63                 goto err_out;
  64         }
  65
  66         // Create the URL container
  67         Curl = malloc(sizeof(URL_t));
  68         if( Curl==NULL ) {
  69                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
  70                 goto err_out;
  71         }
  72
  73         // Initialisation of the URL container members
  74         memset( Curl, 0, sizeof(URL_t) );
  75
  76         url_escape_string(escfilename,url);
  77
  78         // Copy the url in the URL container
  79         Curl->url = strdup(escfilename);
  80         if( Curl->url==NULL ) {
  81                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
  82                 goto err_out;
  83         }
  84         mp_msg(MSGT_OPEN,MSGL_V,"Filename for url is now %s\n",escfilename);
  85
  86         // extract the protocol
  87         ptr1 = strstr(escfilename, "://");
  88         if( ptr1==NULL ) {
  89                 // Check for a special case: "sip:" (without "//"):
  90                 if (strstr(escfilename, "sip:") == escfilename) {
  91                         ptr1 = (char *)&url[3]; // points to ':'
  92                         jumpSize = 1;
  93                 } else {
  94                         mp_msg(MSGT_NETWORK,MSGL_V,"Not an URL!\n");
  95                         goto err_out;
  96                 }
  97         }
  98         pos1 = ptr1-escfilename;
  99         Curl->protocol = malloc(pos1+1);
 100         if( Curl->protocol==NULL ) {
 101                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
 102                 goto err_out;
 103         }
 104         strncpy(Curl->protocol, escfilename, pos1);
 105         Curl->protocol[pos1] = '\0';
 106
 107         // jump the "://"
 108         ptr1 += jumpSize;
 109         pos1 += jumpSize;
 110
 111         // check if a username:password is given
 112         ptr2 = strstr(ptr1, "@");
 113         ptr3 = strstr(ptr1, "/");
 114         if( ptr3!=NULL && ptr3<ptr2 ) {
 115                 // it isn't really a username but rather a part of the path
 116                 ptr2 = NULL;
 117         }
 118         if( ptr2!=NULL ) {
 119                 // We got something, at least a username...
 120                 int len = ptr2-ptr1;
 121                 Curl->username = malloc(len+1);
 122                 if( Curl->username==NULL ) {
 123                         mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
 124                         goto err_out;
 125                 }
 126                 strncpy(Curl->username, ptr1, len);
 127                 Curl->username[len] = '\0';
 128
 129                 ptr3 = strstr(ptr1, ":");
 130                 if( ptr3!=NULL && ptr3<ptr2 ) {
 131                         // We also have a password
 132                         int len2 = ptr2-ptr3-1;
 133                         Curl->username[ptr3-ptr1]='\0';
 134                         Curl->password = malloc(len2+1);
 135                         if( Curl->password==NULL ) {
 136                                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
 137                                 goto err_out;
 138                         }
 139                         strncpy( Curl->password, ptr3+1, len2);
 140                         Curl->password[len2]='\0';
 141                 }
 142                 ptr1 = ptr2+1;
 143                 pos1 = ptr1-escfilename;
 144         }
 145
 146         // before looking for a port number check if we have an IPv6 type numeric address
 147         // in IPv6 URL the numeric address should be inside square braces.
 148         ptr2 = strstr(ptr1, "[");
 149         ptr3 = strstr(ptr1, "]");
 150         ptr4 = strstr(ptr1, "/");
 151         if( ptr2!=NULL && ptr3!=NULL && ptr2 < ptr3 && (!ptr4 || ptr4 > ptr3)) {
 152                 // we have an IPv6 numeric address
 153                 ptr1++;
 154                 pos1++;
 155                 ptr2 = ptr3;
 156                 v6addr = 1;
 157         } else {
 158                 ptr2 = ptr1;
 159
 160         }
 161
 162         // look if the port is given
 163         ptr2 = strstr(ptr2, ":");
 164         // If the : is after the first / it isn't the port
 165         ptr3 = strstr(ptr1, "/");
 166         if(ptr3 && ptr3 - ptr2 < 0) ptr2 = NULL;
 167         if( ptr2==NULL ) {
 168                 // No port is given
 169                 // Look if a path is given
 170                 if( ptr3==NULL ) {
 171                         // No path/filename
 172                         // So we have an URL like http://www.hostname.com
 173                         pos2 = strlen(escfilename);
 174                 } else {
 175                         // We have an URL like http://www.hostname.com/file.txt
 176                         pos2 = ptr3-escfilename;
 177                 }
 178         } else {
 179                 // We have an URL beginning like http://www.hostname.com:1212
 180                 // Get the port number
 181                 Curl->port = atoi(ptr2+1);
 182                 pos2 = ptr2-escfilename;
 183         }
 184         if( v6addr ) pos2--;
 185         // copy the hostname in the URL container
 186         Curl->hostname = malloc(pos2-pos1+1);
 187         if( Curl->hostname==NULL ) {
 188                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
 189                 goto err_out;
 190         }
 191         strncpy(Curl->hostname, ptr1, pos2-pos1);
 192         Curl->hostname[pos2-pos1] = '\0';
 193
 194         // Look if a path is given
 195         ptr2 = strstr(ptr1, "/");
 196         if( ptr2!=NULL ) {
 197                 // A path/filename is given
 198                 // check if it's not a trailing '/'
 199                 if( strlen(ptr2)>1 ) {
 200                         // copy the path/filename in the URL container
 201                         Curl->file = strdup(ptr2);
 202                         if( Curl->file==NULL ) {
 203                                 mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
 204                                 goto err_out;
 205                         }
 206                 }
 207         }
 208         // Check if a filename was given or set, else set it with '/'
 209         if( Curl->file==NULL ) {
 210                 Curl->file = malloc(2);
 211                 if( Curl->file==NULL ) {
 212                         mp_msg(MSGT_NETWORK,MSGL_FATAL,MSGTR_MemAllocFailed);
 213                         goto err_out;
 214                 }
 215                 strcpy(Curl->file, "/");
 216         }
 217
 218         free(escfilename);
 219         return Curl;
 220 err_out:
 221         if (escfilename) free(escfilename);
 222         if (Curl) url_free(Curl);
 223         return NULL;
 224 }
 225
 226 void
 227 url_free(URL_t* url) {
 228         if(!url) return;
 229         if(url->url) free(url->url);
 230         if(url->protocol) free(url->protocol);
 231         if(url->hostname) free(url->hostname);
 232         if(url->file) free(url->file);
 233         if(url->username) free(url->username);
 234         if(url->password) free(url->password);
 235         free(url);
 236 }
 237
 238
 239 /* Replace escape sequences in an URL (or a part of an URL) */
 240 /* works like strcpy(), but without return argument */
 241 void
 242 url_unescape_string(char *outbuf, const char *inbuf)
 243 {
 244         unsigned char c,c1,c2;
 245         int i,len=strlen(inbuf);
 246         for (i=0;i<len;i++){
 247                 c = inbuf[i];
 248                 if (c == '%' && i<len-2) { //must have 2 more chars
 249                         c1 = toupper(inbuf[i+1]); // we need uppercase characters
 250                         c2 = toupper(inbuf[i+2]);
 251                         if (    ((c1>='0' && c1<='9') || (c1>='A' && c1<='F')) &&
 252                                 ((c2>='0' && c2<='9') || (c2>='A' && c2<='F')) ) {
 253                                 if (c1>='0' && c1<='9') c1-='0';
 254                                 else c1-='A'-10;
 255                                 if (c2>='0' && c2<='9') c2-='0';
 256                                 else c2-='A'-10;
 257                                 c = (c1<<4) + c2;
 258                                 i=i+2; //only skip next 2 chars if valid esc
 259                         }
 260                 }
 261                 *outbuf++ = c;
 262         }
 263         *outbuf++='\0'; //add nullterm to string
 264 }
 265
 266 static void
 267 url_escape_string_part(char *outbuf, const char *inbuf) {
 268         unsigned char c,c1,c2;
 269         int i,len=strlen(inbuf);
 270
 271         for  (i=0;i<len;i++) {
 272                 c = inbuf[i];
 273                 if ((c=='%') && i<len-2 ) { //need 2 more characters
 274                     c1=toupper(inbuf[i+1]); c2=toupper(inbuf[i+2]); // need uppercase chars
 275                    } else {
 276                     c1=129; c2=129; //not escape chars
 277                    }
 278
 279                 if(     (c >= 'A' && c <= 'Z') ||
 280                         (c >= 'a' && c <= 'z') ||
 281                         (c >= '0' && c <= '9') ||
 282                         (c >= 0x7f)) {
 283                         *outbuf++ = c;
 284                 } else if ( c=='%' && ((c1 >= '0' && c1 <= '9') || (c1 >= 'A' && c1 <= 'F')) &&
 285                            ((c2 >= '0' && c2 <= '9') || (c2 >= 'A' && c2 <= 'F'))) {
 286                                                               // check if part of an escape sequence
 287                             *outbuf++=c;                      // already
 288
 289                                                               // dont escape again
 290                             mp_msg(MSGT_NETWORK,MSGL_ERR,MSGTR_MPDEMUX_URL_StringAlreadyEscaped,c,c1,c2);
 291                                                               // error as this should not happen against RFC 2396
 292                                                               // to escape a string twice
 293                 } else {
 294                         /* all others will be escaped */
 295                         c1 = ((c & 0xf0) >> 4);
 296                         c2 = (c & 0x0f);
 297                         if (c1 < 10) c1+='0';
 298                         else c1+='A'-10;
 299                         if (c2 < 10) c2+='0';
 300                         else c2+='A'-10;
 301                         *outbuf++ = '%';
 302                         *outbuf++ = c1;
 303                         *outbuf++ = c2;
 304                 }
 305         }
 306         *outbuf++='\0';
 307 }
 308
 309 /* Replace specific characters in the URL string by an escape sequence */
 310 /* works like strcpy(), but without return argument */
 311 void
 312 url_escape_string(char *outbuf, const char *inbuf) {
 313         unsigned char c;
 314         int i = 0,j,len = strlen(inbuf);
 315         char* tmp,*unesc = NULL, *in;
 316
 317         // Look if we have an ip6 address, if so skip it there is
 318         // no need to escape anything in there.
 319         tmp = strstr(inbuf,"://[");
 320         if(tmp) {
 321                 tmp = strchr(tmp+4,']');
 322                 if(tmp && (tmp[1] == '/' || tmp[1] == ':' ||
 323                            tmp[1] == '\0')) {
 324                         i = tmp+1-inbuf;
 325                         strncpy(outbuf,inbuf,i);
 326                         outbuf += i;
 327                         tmp = NULL;
 328                 }
 329         }
 330
 331         tmp = NULL;
 332         while(i < len) {
 333                 // look for the next char that must be kept
 334                 for  (j=i;j<len;j++) {
 335                         c = inbuf[j];
 336                         if(c=='-' || c=='_' || c=='.' || c=='!' || c=='~' ||    /* mark characters */
 337                            c=='*' || c=='\'' || c=='(' || c==')' ||             /* do not touch escape character */
 338                            c==';' || c=='/' || c=='?' || c==':' || c=='@' ||    /* reserved characters */
 339                            c=='&' || c=='=' || c=='+' || c=='$' || c==',')      /* see RFC 2396 */
 340                                 break;
 341                 }
 342                 // we are on a reserved char, write it out
 343                 if(j == i) {
 344                         *outbuf++ = c;
 345                         i++;
 346                         continue;
 347                 }
 348                 // we found one, take that part of the string
 349                 if(j < len) {
 350                         if(!tmp) tmp = malloc(len+1);
 351                         strncpy(tmp,inbuf+i,j-i);
 352                         tmp[j-i] = '\0';
 353                         in = tmp;
 354                 } else // take the rest of the string
 355                         in = (char*)inbuf+i;
 356
 357                 if(!unesc) unesc = malloc(len+1);
 358                 // unescape first to avoid escaping escape
 359                 url_unescape_string(unesc,in);
 360                 // then escape, including mark and other reserved chars
 361                 // that can come from escape sequences
 362                 url_escape_string_part(outbuf,unesc);
 363                 outbuf += strlen(outbuf);
 364                 i += strlen(in);
 365         }
 366         *outbuf = '\0';
 367         if(tmp) free(tmp);
 368         if(unesc) free(unesc);
 369 }
 370
 371 #ifdef URL_DEBUG
 372 void
 373 url_debug(const URL_t *url) {
 374         if( url==NULL ) {
 375                 printf("URL pointer NULL\n");
 376                 return;
 377         }
 378         if( url->url!=NULL ) {
 379                 printf("url=%s\n", url->url );
 380         }
 381         if( url->protocol!=NULL ) {
 382                 printf("protocol=%s\n", url->protocol );
 383         }
 384         if( url->hostname!=NULL ) {
 385                 printf("hostname=%s\n", url->hostname );
 386         }
 387         printf("port=%d\n", url->port );
 388         if( url->file!=NULL ) {
 389                 printf("file=%s\n", url->file );
 390         }
 391         if( url->username!=NULL ) {
 392                 printf("username=%s\n", url->username );
 393         }
 394         if( url->password!=NULL ) {
 395                 printf("password=%s\n", url->password );
 396         }
 397 }
 398 #endif /* URL_DEBUG */