stream/url.c

   1 /*
   2  * URL Helper
   3  *
   4  * Copyright (C) 2001 Bertrand Baudet <bertrand_baudet@yahoo.com>
   5  *
   6  * This file is part of MPlayer.
   7  *
   8  * MPlayer is free software; you can redistribute it and/or modify
   9  * it under the terms of the GNU General Public License as published by
  10  * the Free Software Foundation; either version 2 of the License, or
  11  * (at your option) any later version.
  12  *
  13  * MPlayer is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  16  * GNU General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU General Public License along
  19  * with MPlayer; if not, write to the Free Software Foundation, Inc.,
  20  * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
  21  */
  22
  23 #include <string.h>
  24 #include <stdlib.h>
  25 #include <stdio.h>
  26 #include <ctype.h>
  27 #include <inttypes.h>
  28
  29 #include "url.h"
  30 #include "mp_msg.h"
  31
  32 #ifndef SIZE_MAX
  33 #define SIZE_MAX ((size_t)-1)
  34 #endif
  35
  36 URL_t *url_redirect(URL_t **url, const char *redir) {
  37   URL_t *u = *url;
  38   URL_t *res;
  39   if (!strchr(redir, '/') || *redir == '/') {
  40     char *tmp;
  41     char *newurl = malloc(strlen(u->url) + strlen(redir) + 1);
  42     strcpy(newurl, u->url);
  43     if (*redir == '/') {
  44       redir++;
  45       tmp = strstr(newurl, "://");
  46       if (tmp) tmp = strchr(tmp + 3, '/');
  47     } else
  48       tmp = strrchr(newurl, '/');
  49     if (tmp) tmp[1] = 0;
  50     strcat(newurl, redir);
  51     res = url_new(newurl);
  52     free(newurl);
  53   } else
  54     res = url_new(redir);
  55   url_free(u);
  56   *url = res;
  57   return res;
  58 }
  59
  60 static int make_noauth_url(URL_t *url, char *dst, int dst_size)
  61 {
  62     if (url->port)
  63         return snprintf(dst, dst_size, "%s://%s:%d%s", url->protocol,
  64                         url->hostname, url->port, url->file);
  65     else
  66         return snprintf(dst, dst_size, "%s://%s%s", url->protocol,
  67                         url->hostname, url->file);
  68 }
  69
  70 URL_t*
  71 url_new(const char* url) {
  72         int pos1, pos2,v6addr = 0, noauth_len;
  73         URL_t* Curl = NULL;
  74         char *escfilename=NULL;
  75         char *ptr1=NULL, *ptr2=NULL, *ptr3=NULL, *ptr4=NULL;
  76         int jumpSize = 3;
  77
  78         if( url==NULL ) return NULL;
  79
  80         if (strlen(url) > (SIZE_MAX / 3 - 1)) {
  81                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
  82                 goto err_out;
  83         }
  84         escfilename=malloc(strlen(url)*3+1);
  85         if (!escfilename ) {
  86                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
  87                 goto err_out;
  88         }
  89
  90         // Create the URL container
  91         Curl = malloc(sizeof(URL_t));
  92         if( Curl==NULL ) {
  93                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
  94                 goto err_out;
  95         }
  96
  97         // Initialisation of the URL container members
  98         memset( Curl, 0, sizeof(URL_t) );
  99
 100         url_escape_string(escfilename,url);
 101
 102         // Copy the url in the URL container
 103         Curl->url = strdup(escfilename);
 104         if( Curl->url==NULL ) {
 105                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 106                 goto err_out;
 107         }
 108         mp_msg(MSGT_OPEN,MSGL_V,"Filename for url is now %s\n",escfilename);
 109
 110         // extract the protocol
 111         ptr1 = strstr(escfilename, "://");
 112         if( ptr1==NULL ) {
 113                 // Check for a special case: "sip:" (without "//"):
 114                 if (strstr(escfilename, "sip:") == escfilename) {
 115                         ptr1 = (char *)&url[3]; // points to ':'
 116                         jumpSize = 1;
 117                 } else {
 118                         mp_msg(MSGT_NETWORK,MSGL_V,"Not an URL!\n");
 119                         goto err_out;
 120                 }
 121         }
 122         pos1 = ptr1-escfilename;
 123         Curl->protocol = malloc(pos1+1);
 124         if( Curl->protocol==NULL ) {
 125                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 126                 goto err_out;
 127         }
 128         strncpy(Curl->protocol, escfilename, pos1);
 129         Curl->protocol[pos1] = '\0';
 130
 131         // jump the "://"
 132         ptr1 += jumpSize;
 133         pos1 += jumpSize;
 134
 135         // check if a username:password is given
 136         ptr2 = strstr(ptr1, "@");
 137         ptr3 = strstr(ptr1, "/");
 138         if( ptr3!=NULL && ptr3<ptr2 ) {
 139                 // it isn't really a username but rather a part of the path
 140                 ptr2 = NULL;
 141         }
 142         if( ptr2!=NULL ) {
 143                 // We got something, at least a username...
 144                 int len = ptr2-ptr1;
 145                 Curl->username = malloc(len+1);
 146                 if( Curl->username==NULL ) {
 147                         mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 148                         goto err_out;
 149                 }
 150                 strncpy(Curl->username, ptr1, len);
 151                 Curl->username[len] = '\0';
 152
 153                 ptr3 = strstr(ptr1, ":");
 154                 if( ptr3!=NULL && ptr3<ptr2 ) {
 155                         // We also have a password
 156                         int len2 = ptr2-ptr3-1;
 157                         Curl->username[ptr3-ptr1]='\0';
 158                         Curl->password = malloc(len2+1);
 159                         if( Curl->password==NULL ) {
 160                                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 161                                 goto err_out;
 162                         }
 163                         strncpy( Curl->password, ptr3+1, len2);
 164                         Curl->password[len2]='\0';
 165                         url_unescape_string(Curl->password, Curl->password);
 166                 }
 167                 url_unescape_string(Curl->username, Curl->username);
 168                 ptr1 = ptr2+1;
 169                 pos1 = ptr1-escfilename;
 170         }
 171
 172         // before looking for a port number check if we have an IPv6 type numeric address
 173         // in IPv6 URL the numeric address should be inside square braces.
 174         ptr2 = strstr(ptr1, "[");
 175         ptr3 = strstr(ptr1, "]");
 176         ptr4 = strstr(ptr1, "/");
 177         if( ptr2!=NULL && ptr3!=NULL && ptr2 < ptr3 && (!ptr4 || ptr4 > ptr3)) {
 178                 // we have an IPv6 numeric address
 179                 ptr1++;
 180                 pos1++;
 181                 ptr2 = ptr3;
 182                 v6addr = 1;
 183         } else {
 184                 ptr2 = ptr1;
 185
 186         }
 187
 188         // look if the port is given
 189         ptr2 = strstr(ptr2, ":");
 190         // If the : is after the first / it isn't the port
 191         ptr3 = strstr(ptr1, "/");
 192         if(ptr3 && ptr3 - ptr2 < 0) ptr2 = NULL;
 193         if( ptr2==NULL ) {
 194                 // No port is given
 195                 // Look if a path is given
 196                 if( ptr3==NULL ) {
 197                         // No path/filename
 198                         // So we have an URL like http://www.hostname.com
 199                         pos2 = strlen(escfilename);
 200                 } else {
 201                         // We have an URL like http://www.hostname.com/file.txt
 202                         pos2 = ptr3-escfilename;
 203                 }
 204         } else {
 205                 // We have an URL beginning like http://www.hostname.com:1212
 206                 // Get the port number
 207                 Curl->port = atoi(ptr2+1);
 208                 pos2 = ptr2-escfilename;
 209         }
 210         if( v6addr ) pos2--;
 211         // copy the hostname in the URL container
 212         Curl->hostname = malloc(pos2-pos1+1);
 213         if( Curl->hostname==NULL ) {
 214                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 215                 goto err_out;
 216         }
 217         strncpy(Curl->hostname, ptr1, pos2-pos1);
 218         Curl->hostname[pos2-pos1] = '\0';
 219
 220         // Look if a path is given
 221         ptr2 = strstr(ptr1, "/");
 222         if( ptr2!=NULL ) {
 223                 // A path/filename is given
 224                 // check if it's not a trailing '/'
 225                 if( strlen(ptr2)>1 ) {
 226                         // copy the path/filename in the URL container
 227                         Curl->file = strdup(ptr2);
 228                         if( Curl->file==NULL ) {
 229                                 mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 230                                 goto err_out;
 231                         }
 232                 }
 233         }
 234         // Check if a filename was given or set, else set it with '/'
 235         if( Curl->file==NULL ) {
 236                 Curl->file = malloc(2);
 237                 if( Curl->file==NULL ) {
 238                         mp_tmsg(MSGT_NETWORK,MSGL_FATAL,"Memory allocation failed.\n");
 239                         goto err_out;
 240                 }
 241                 strcpy(Curl->file, "/");
 242         }
 243
 244         noauth_len = make_noauth_url(Curl, NULL, 0);
 245         if (noauth_len > 0) {
 246                 noauth_len++;
 247                 Curl->noauth_url = malloc(noauth_len);
 248                 if (!Curl->noauth_url) {
 249                         mp_msg(MSGT_NETWORK, MSGL_FATAL, "Memory allocation failed.\n");
 250                         goto err_out;
 251                 }
 252                 make_noauth_url(Curl, Curl->noauth_url, noauth_len);
 253         }
 254
 255         free(escfilename);
 256         return Curl;
 257 err_out:
 258         free(escfilename);
 259         if (Curl) url_free(Curl);
 260         return NULL;
 261 }
 262
 263 void
 264 url_free(URL_t* url) {
 265         if(!url) return;
 266         free(url->url);
 267         free(url->protocol);
 268         free(url->hostname);
 269         free(url->file);
 270         free(url->username);
 271         free(url->password);
 272         free(url);
 273 }
 274
 275
 276 /* Replace escape sequences in an URL (or a part of an URL) */
 277 /* works like strcpy(), but without return argument,
 278    except that outbuf == inbuf is allowed */
 279 void
 280 url_unescape_string(char *outbuf, const char *inbuf)
 281 {
 282         unsigned char c,c1,c2;
 283         int i,len=strlen(inbuf);
 284         for (i=0;i<len;i++){
 285                 c = inbuf[i];
 286                 if (c == '%' && i<len-2) { //must have 2 more chars
 287                         c1 = toupper(inbuf[i+1]); // we need uppercase characters
 288                         c2 = toupper(inbuf[i+2]);
 289                         if (    ((c1>='0' && c1<='9') || (c1>='A' && c1<='F')) &&
 290                                 ((c2>='0' && c2<='9') || (c2>='A' && c2<='F')) ) {
 291                                 if (c1>='0' && c1<='9') c1-='0';
 292                                 else c1-='A'-10;
 293                                 if (c2>='0' && c2<='9') c2-='0';
 294                                 else c2-='A'-10;
 295                                 c = (c1<<4) + c2;
 296                                 i=i+2; //only skip next 2 chars if valid esc
 297                         }
 298                 }
 299                 *outbuf++ = c;
 300         }
 301         *outbuf++='\0'; //add nullterm to string
 302 }
 303
 304 static void
 305 url_escape_string_part(char *outbuf, const char *inbuf) {
 306         unsigned char c,c1,c2;
 307         int i,len=strlen(inbuf);
 308
 309         for  (i=0;i<len;i++) {
 310                 c = inbuf[i];
 311                 if ((c=='%') && i<len-2 ) { //need 2 more characters
 312                     c1=toupper(inbuf[i+1]); c2=toupper(inbuf[i+2]); // need uppercase chars
 313                    } else {
 314                     c1=129; c2=129; //not escape chars
 315                    }
 316
 317                 if(     (c >= 'A' && c <= 'Z') ||
 318                         (c >= 'a' && c <= 'z') ||
 319                         (c >= '0' && c <= '9') ||
 320                         (c >= 0x7f)) {
 321                         *outbuf++ = c;
 322                 } else if ( c=='%' && ((c1 >= '0' && c1 <= '9') || (c1 >= 'A' && c1 <= 'F')) &&
 323                            ((c2 >= '0' && c2 <= '9') || (c2 >= 'A' && c2 <= 'F'))) {
 324                                                               // check if part of an escape sequence
 325                             *outbuf++=c;                      // already
 326
 327                                                               // dont escape again
 328                             mp_tmsg(MSGT_NETWORK,MSGL_ERR,"String appears to be already escaped in url_escape %c%c1%c2\n",c,c1,c2);
 329                                                               // error as this should not happen against RFC 2396
 330                                                               // to escape a string twice
 331                 } else {
 332                         /* all others will be escaped */
 333                         c1 = ((c & 0xf0) >> 4);
 334                         c2 = (c & 0x0f);
 335                         if (c1 < 10) c1+='0';
 336                         else c1+='A'-10;
 337                         if (c2 < 10) c2+='0';
 338                         else c2+='A'-10;
 339                         *outbuf++ = '%';
 340                         *outbuf++ = c1;
 341                         *outbuf++ = c2;
 342                 }
 343         }
 344         *outbuf++='\0';
 345 }
 346
 347 /* Replace specific characters in the URL string by an escape sequence */
 348 /* works like strcpy(), but without return argument */
 349 void
 350 url_escape_string(char *outbuf, const char *inbuf) {
 351         unsigned char c;
 352         int i = 0,j,len = strlen(inbuf);
 353         char* tmp,*unesc = NULL, *in;
 354
 355         // Look if we have an ip6 address, if so skip it there is
 356         // no need to escape anything in there.
 357         tmp = strstr(inbuf,"://[");
 358         if(tmp) {
 359                 tmp = strchr(tmp+4,']');
 360                 if(tmp && (tmp[1] == '/' || tmp[1] == ':' ||
 361                            tmp[1] == '\0')) {
 362                         i = tmp+1-inbuf;
 363                         strncpy(outbuf,inbuf,i);
 364                         outbuf += i;
 365                         tmp = NULL;
 366                 }
 367         }
 368
 369         tmp = NULL;
 370         while(i < len) {
 371                 // look for the next char that must be kept
 372                 for  (j=i;j<len;j++) {
 373                         c = inbuf[j];
 374                         if(c=='-' || c=='_' || c=='.' || c=='!' || c=='~' ||    /* mark characters */
 375                            c=='*' || c=='\'' || c=='(' || c==')' ||             /* do not touch escape character */
 376                            c==';' || c=='/' || c=='?' || c==':' || c=='@' ||    /* reserved characters */
 377                            c=='&' || c=='=' || c=='+' || c=='$' || c==',')      /* see RFC 2396 */
 378                                 break;
 379                 }
 380                 // we are on a reserved char, write it out
 381                 if(j == i) {
 382                         *outbuf++ = c;
 383                         i++;
 384                         continue;
 385                 }
 386                 // we found one, take that part of the string
 387                 if(j < len) {
 388                         if(!tmp) tmp = malloc(len+1);
 389                         strncpy(tmp,inbuf+i,j-i);
 390                         tmp[j-i] = '\0';
 391                         in = tmp;
 392                 } else // take the rest of the string
 393                         in = (char*)inbuf+i;
 394
 395                 if(!unesc) unesc = malloc(len+1);
 396                 // unescape first to avoid escaping escape
 397                 url_unescape_string(unesc,in);
 398                 // then escape, including mark and other reserved chars
 399                 // that can come from escape sequences
 400                 url_escape_string_part(outbuf,unesc);
 401                 outbuf += strlen(outbuf);
 402                 i += strlen(in);
 403         }
 404         *outbuf = '\0';
 405         free(tmp);
 406         free(unesc);
 407 }
 408
 409 #ifdef URL_DEBUG
 410 void
 411 url_debug(const URL_t *url) {
 412         if( url==NULL ) {
 413                 printf("URL pointer NULL\n");
 414                 return;
 415         }
 416         if( url->url!=NULL ) {
 417                 printf("url=%s\n", url->url );
 418         }
 419         if( url->protocol!=NULL ) {
 420                 printf("protocol=%s\n", url->protocol );
 421         }
 422         if( url->hostname!=NULL ) {
 423                 printf("hostname=%s\n", url->hostname );
 424         }
 425         printf("port=%d\n", url->port );
 426         if( url->file!=NULL ) {
 427                 printf("file=%s\n", url->file );
 428         }
 429         if( url->username!=NULL ) {
 430                 printf("username=%s\n", url->username );
 431         }
 432         if( url->password!=NULL ) {
 433                 printf("password=%s\n", url->password );
 434         }
 435 }
 436 #endif /* URL_DEBUG */