Usermode/Applications/wget_src/main.c

   1 /*
   2  * Acess2 Command-line HTTP Client (wget)
   3  * - By John Hodge
   4  *
   5  * main.c
   6  */
   7 #include <stdlib.h>
   8 #include <stdio.h>
   9 #include <net.h>
  10 #include <uri.h>
  11 #include <string.h>
  12 #include <unistd.h>
  13 #include <netdb.h>
  14
  15 enum eProcols
  16 {
  17         PROTO_NONE,
  18         PROTO_HTTP,
  19         PROTO_HTTPS
  20 };
  21
  22 const char      **gasURLs;
  23  int    giNumURLs;
  24
  25  int    main(int argc, char *argv[]);
  26  int    _ParseHeaderLine(char *Line, int State, size_t *Size);
  27 void    writef(int fd, const char *format, ...);
  28
  29 int main(int argc, char *argv[])
  30 {
  31          int    proto, rv;
  32         gasURLs = malloc( (argc - 1) * sizeof(*gasURLs) );
  33
  34         // Parse arguments
  35         for(int i = 1; i < argc; i ++ )
  36         {
  37                 char    *arg = argv[i];
  38                 if( arg[0] != '-' ) {
  39                         // URL
  40                         gasURLs[giNumURLs++] = arg;
  41                 }
  42                 else if( arg[1] != '-') {
  43                         // Short arg
  44                 }
  45                 else {
  46                         // Long arg
  47                 }
  48         }
  49
  50         // Do the download
  51         for( int i = 0; i < giNumURLs; i ++ )
  52         {
  53                 char    *outfile = NULL;
  54                 tURI    *uri = URI_Parse(gasURLs[i]);
  55                 struct addrinfo *addrinfo;
  56
  57                 if( !uri ) {
  58                         fprintf(stderr, "'%s' is not a valid URL", gasURLs[i]);
  59                         continue ;
  60                 }
  61
  62                 printf("Proto: %s, Host: %s, Path: %s\n", uri->Proto, uri->Host, uri->Path);
  63
  64                 if( uri->Path[0] == '\0' || uri->Path[strlen(uri->Path)-1] == '/' )
  65                         outfile = "index.html";
  66                 else {
  67                         outfile = strrchr(uri->Path, '/');
  68                         if( !outfile )
  69                                 outfile = uri->Path;
  70                         else
  71                                 outfile += 1;
  72                 }
  73
  74                 if( strcmp(uri->Proto, "http") == 0 ) {
  75                         proto = PROTO_HTTP;
  76                 }
  77                 else if( strcmp(uri->Proto, "https") == 0 ) {
  78                         proto = PROTO_HTTPS;
  79                 }
  80                 else {
  81                         // Unknown
  82                         fprintf(stderr, "Unknown protocol '%s'\n", uri->Proto);
  83                         free(uri);
  84                         continue ;
  85                 }
  86
  87                 if( proto != PROTO_HTTP ) {
  88                         fprintf(stderr, "TODO: Support protocols other than HTTP\n");
  89                         free(uri);
  90                         continue ;
  91                 }
  92
  93                 rv = getaddrinfo(uri->Host, "http", NULL, &addrinfo);
  94                 if( rv != 0 ) {
  95                         fprintf(stderr, "Unable to resolve %s: %s\n", uri->Host, gai_strerror(rv));
  96                         continue ;
  97                 }
  98
  99                 for( struct addrinfo *addr = addrinfo; addr != NULL; addr = addr->ai_next )
 100                 {
 101                          int    bSkipLine = 0;
 102                         // TODO: Convert to POSIX/BSD
 103                         // NOTE: using addr->ai_addr will break for IPv6, as there is more info before the address
 104                          int    sock;
 105
 106                         printf("Attempting [%s]:80\n", Net_PrintAddress(addr->ai_family, addr->ai_addr->sa_data));
 107
 108                         sock = Net_OpenSocket_TCPC(addr->ai_family, addr->ai_addr->sa_data, 80);
 109                         if( sock == -1 ) {
 110                                 continue ;
 111                         }
 112
 113                         _SysDebug("Connected as %i", sock);
 114
 115                         writef(sock, "GET /%s HTTP/1.1\r\n", uri->Path);
 116                         writef(sock, "Host: %s\r\n", uri->Host);
 117 //                      writef(sock, "Accept-Encodings: */*\r\n");
 118                         writef(sock, "User-Agent: awget/0.1 (Acess2)\r\n");
 119                         writef(sock, "\r\n");
 120
 121                         // Parse headers
 122                         char    inbuf[BUFSIZ+1];
 123                         size_t  offset = 0, len = 0;
 124                          int    state = 0;
 125                         size_t  bytes_seen = 0;
 126                         size_t  bytes_wanted = -1;      // invalid
 127
 128
 129                         inbuf[0] = '\0';
 130                         while( state == 0 || state == 1 )
 131                         {
 132                                 if( offset == BUFSIZ ) {
 133                                         bSkipLine = 1;
 134                                         offset = 0;
 135                                 }
 136                                 inbuf[len] = '\0';
 137
 138                                 char    *eol = strchr(inbuf, '\n');
 139                                 // No end of line char? read some more
 140                                 if( eol == NULL ) {
 141                                         // TODO: Handle -1 return
 142                                         len += read(sock, inbuf + offset, BUFSIZ - 1 - offset);
 143                                         continue ;
 144                                 }
 145
 146                                 // abuse offset as the end of the string
 147                                 offset = (eol - inbuf) + 1;
 148
 149                                 // Clear EOL bytes
 150                                 *eol = '\0';
 151                                 // Nuke the \r
 152                                 if( eol - 1 >= inbuf )
 153                                         eol[-1] = '\0';
 154
 155                                 if( !bSkipLine )
 156                                         state = _ParseHeaderLine(inbuf, state, &bytes_wanted);
 157
 158                                 // Move unused data down in memory
 159                                 len -= offset;
 160                                 memmove( inbuf, inbuf + offset, BUFSIZ - offset );
 161                                 offset = len;
 162                         }
 163
 164                         if( state == 2 )
 165                         {
 166                                 _SysDebug("RXing %i bytes to '%s'", bytes_wanted, outfile);
 167                                  int    outfd = open(outfile, O_WR|O_CREAT, 0666);
 168                                 if( outfd == -1 ) {
 169                                         fprintf(stderr, "Unable to open '%s' for writing\n", outfile);
 170                                 }
 171                                 else
 172                                 {
 173                                         // Write the remainder of the buffer
 174                                         do
 175                                         {
 176                                                 write(outfd, inbuf, len);
 177                                                 bytes_seen += len;
 178                                                 _SysDebug("%i/%i bytes done", bytes_seen, bytes_wanted);
 179                                         } while( bytes_seen < bytes_wanted && (len = read(sock, inbuf, sizeof(inbuf))) > 0 );
 180                                         close(outfd);
 181                                 }
 182                         }
 183
 184                         _SysDebug("Closing socket");
 185                         close(sock);
 186                         break ;
 187                 }
 188
 189                 free(uri);
 190         }
 191
 192         return 0;
 193 }
 194
 195 int _ParseHeaderLine(char *Line, int State, size_t *Size)
 196 {
 197         _SysDebug("Header - %s", Line);
 198         // First line (Status and version)
 199         if( State == 0 )
 200         {
 201                 // HACK - assumes HTTP/1.1 (well, 9 chars before status)
 202                 switch( atoi(Line + 9) )
 203                 {
 204                 case 200:
 205                         // All good!
 206                         return 1;
 207                 default:
 208                         fprintf(stderr, "Unknown HTTP status - %s\n", Line + 9);
 209                         return -1;
 210                 }
 211         }
 212         // Last line?
 213         else if( Line[0] == '\0' )
 214         {
 215                 return 2;
 216         }
 217         // Body lines
 218         else
 219         {
 220                 char    *value;
 221                 char    *colon = strchr(Line, ':');
 222                 if(colon == NULL)       return 1;
 223
 224                 *colon = '\0';
 225                 value = colon + 2;
 226                 if( strcmp(Line, "Content-Length") == 0 ) {
 227                         *Size = atoi(value);
 228                 }
 229                 else {
 230                         printf("Ignorning header '%s' = '%s'\n", Line, value);
 231                 }
 232
 233                 return 1;
 234         }
 235 }
 236
 237 void writef(int fd, const char *format, ...)
 238 {
 239         va_list args;
 240         size_t  len;
 241
 242         va_start(args, format);
 243         len = vsnprintf(NULL, 0, format, args);
 244         va_end(args);
 245
 246         char data[len + 1];
 247         va_start(args, format);
 248         vsnprintf(data, len+1, format, args);
 249         va_end(args);
 250
 251         write(fd, data, len);
 252 }
 253