contrib/pullart.c

   1 /*
   2 June 14, 1999
   3
   4 Recover text articles from cyclic buffers
   5 Articles start with  "\0Path:"
   6 and end with "\r\n.\r\n"
   7
   8 Tested with INND 2.2 under AIX 4.2
   9
  10 rifkin@uconn.edu
  11 */
  12 /*
  13 (1) Pull 16 bytes at a time
  14 (2) Last 7 bytes must be \000\000\000Path
  15 (3) When found, print "\nPath";
  16 (4) print subsequent bytes until \r\n.\r\n found
  17 */
  18
  19 #include "config.h"
  20 #include <stdio.h>
  21 #include <stdlib.h>
  22 #include <string.h>
  23
  24 #define INFILE     1
  25 #define FILEPREFIX 2
  26 #define HEADER     3
  27 #define STRING     4
  28
  29 /*  String buffer size  */
  30 #define NBUFF 512
  31
  32 #define MAX_ART_SIZE 2200000
  33
  34
  35 #define WRITEMSG printf ("File %s line %i\n", __FILE__, __LINE__); \
  36         fflush(stdout);
  37
  38 #define WRITEVAR(VAR_NAME,VAR_TYPE) \
  39       { \
  40       printf ("FILE %s LINE %i :", __FILE__, __LINE__); \
  41       printf ("%s = ", #VAR_NAME); \
  42       printf (#VAR_TYPE, (VAR_NAME) ); \
  43       printf ("\n"); \
  44       }
  45
  46 #define WRITETXT(TEXT)  \
  47         printf ("FILE %s LINE %i \"%s\"\n", __FILE__, __LINE__, TEXT); \
  48         fflush(stdout);
  49
  50 #if 0
  51 #define WRITEMSG
  52 #define WRITEVAR(X,Y)
  53 #endif
  54
  55
  56 int WriteArticle (char *, int, char *, char *, char *, int);
  57
  58
  59 char ArtHead[7] = {0, 0, 0, 'P', 'a', 't', 'h'};
  60 char ArtTail[5] = {'\r', '\n', '.', '\r', '\n'};
  61 int  LenTail    = 5;
  62
  63 int main (int argc, char *argv[])
  64         {
  65         FILE *Infile;
  66         int   NumTailCharFound;
  67         bool  ReadingArticle = false;
  68         char  buffer[32];
  69         char *obuffer = NULL;
  70         char *header  = NULL;
  71         char *string  = NULL;
  72         int   osize = MAX_ART_SIZE;
  73         int   opos  = 0;
  74         int   i;
  75         int   nchar;
  76         int   fileno = 0;
  77         int   artno  = 0;
  78
  79         /*  Check number of args  */
  80         if (argc<3)
  81                 {
  82                 printf ("Usage: pullart <cycbuff> <fileprefix> [<header> <string>]\n");
  83                 printf ("  Read cycbuffer <cycbuff> and print all articles whose\n");
  84                 printf ("   article header <header> contains <string>.\n");
  85                 printf ("  Articles are written to files name <fileprefix>.nnnnnn\n");
  86                 printf ("   where nnnnnn is numbered sequentially from 0.\n");
  87                 printf ("  If <header> and <string> not specified, all articles\n");
  88                 printf ("   are written.\n");
  89                 printf (" Examples:\n");
  90                 printf ("  pullart /news3/cycbuff.3 alt.rec  Newsgroup: alt.rec\n");
  91                 printf ("  pullart /news3/cycbuff.3  all\n");
  92                 printf ("  pullart firstbuff  article Subject bluejay\n");
  93                 return 0;
  94                 }
  95
  96         /*  Allocate output buffer  */
  97         obuffer = (char *) calloc (osize+1, sizeof(char));
  98         if (obuffer==NULL)
  99                 {
 100                 printf ("Cannot allocate obuffer[]\n");
 101                 return 1;
 102                 }
 103
 104
 105         /*  Open input file  */
 106         Infile = fopen (argv[INFILE], "rb");
 107         if (Infile==NULL)
 108                 {
 109                 printf ("Cannot open input file.\n");
 110                 return 1;
 111                 }
 112
 113
 114 if (argc>=4) header = argv[HEADER];
 115 if (argc>=5) string = argv[STRING];
 116 if (*header=='\0') header=NULL;
 117 if (*string=='\0') string=NULL;
 118
 119 /*test*/
 120 printf ("filename <%s>\n", argv[INFILE]);
 121 printf ("fileprefix <%s>\n", argv[FILEPREFIX]);
 122 printf ("header <%s>\n", header);
 123 printf ("string <%s>\n", string);
 124
 125
 126         /*  Skip first 0x38000 16byte buffers  */
 127         i = fseek (Infile, 0x38000L, SEEK_SET);
 128
 129         /*  Read following 16 byte buffers  */
 130         ReadingArticle = false;
 131         NumTailCharFound = 0;
 132         nchar=0;
 133         artno=0;
 134         while ( 0!=fread(buffer, 16, 1, Infile) )
 135                 {
 136
 137                 nchar+=16;
 138
 139                 /*  Found start of article, start writing to obuffer  */
 140                 if (0==memcmp(buffer+9, ArtHead, 7))
 141                         {
 142                         ReadingArticle = true;
 143                         memcpy (obuffer, "Path", 4);
 144                         opos = 4;
 145                         continue;
 146                         }
 147
 148                 /*  Currnetly reading article  */
 149                 if (ReadingArticle)
 150                         {
 151                         for (i=0; i<16; i++)
 152                                 {
 153
 154                                 /*  Article too big, drop it and move on  */
 155                                 if (opos>=osize)
 156                                         {
 157                                         printf
 158                                                 ("article number %i bigger than buffer size %i.\n",
 159                                                 artno+1, osize);
 160                                         artno++;
 161                                         ReadingArticle=false;
 162                                         break;
 163                                         }
 164
 165                                 /*  Add current character to output buffer, but remove \r  */
 166                                 if ('\r' != buffer[i])
 167                                         obuffer[opos++] = buffer[i];
 168
 169                                 /*  Check for article ending sequence  */
 170                                 if (buffer[i]==ArtTail[NumTailCharFound])
 171                                         {
 172                                         NumTailCharFound++;
 173                                         }
 174                                 else
 175                                         NumTailCharFound=0;
 176
 177                                 /*  End found, write article, reset for next  */
 178                                 if (NumTailCharFound==LenTail)
 179                                         {
 180                                         ReadingArticle = false;
 181                                         NumTailCharFound = 0;
 182
 183                                         /*  Add trailing \0 to buffer  */
 184                                         obuffer[opos+1] = '\0';
 185
 186                                         fileno += WriteArticle
 187                                                 (obuffer, opos, argv[FILEPREFIX],
 188                                                 header, string, fileno);
 189                                         artno++;
 190                                         break;
 191                                         }
 192                                 }
 193
 194                         }
 195
 196                 }
 197
 198         close (Infile);
 199
 200         return 0;
 201         }
 202
 203
 204
 205 /*
 206 Writes article stored in buff[] if it has a
 207 "Newsgroups:" header line which contains *newsgroup
 208 Write to a file named  fileprefix.fileno
 209 */
 210 int
 211 WriteArticle
 212 (char *buff, int n, char *fileprefix, char *headerin, char *string, int fileno)
 213         {
 214         char *begptr;
 215         char *endptr;
 216         char *newsptr;
 217         char savechar;
 218         char header[NBUFF];
 219         char filename[NBUFF];
 220         FILE *outfile;
 221
 222
 223         /*  Prevent buffer overflow due to fileprefix too long  */
 224         if (strlen(fileprefix)>384)
 225                 {
 226                 printf
 227                 ("program error: cannot have file prefix greater then 384 characters\n");
 228                 exit(1);
 229                 }
 230
 231         /*
 232         Is header here?  Search if header string requested, leave if not found
 233         */
 234         if (headerin!=NULL)
 235                 {
 236                 /*  Find \nHEADER  */
 237                 strlcpy(header, "\n", sizeof(header));
 238                 strlcat(header, headerin, sizeof(header));
 239
 240                 begptr = strstr (buff, header);
 241
 242                 /*  return if Header name not found  */
 243                 if (begptr==NULL)
 244                         {
 245                         return 0;
 246                         }
 247
 248                 /*
 249                 Header found. What about string?
 250                 Search if string requested, leave if not found
 251                 */
 252                 if (string!=NULL)
 253                         {
 254                         /*  Find end of header line  */
 255                         begptr++;
 256                         endptr = strchr (begptr, '\n');
 257
 258                         /*  Something is wrong, end of header not found, do not write
 259                          *  article
 260                         */
 261                         if (endptr==NULL)
 262                                 return 0;
 263
 264                         /*  Temporarily make string end a null char  */
 265                         savechar = *endptr;
 266                         *endptr = '\0';
 267                         newsptr = strstr (begptr, string);
 268
 269                         /*  Requested newsgroup not found  */
 270                         if (newsptr==NULL)
 271                                 return 0;
 272
 273                         /*  Restore character at end of header string  */
 274                         *endptr = savechar;
 275                         }
 276                         /*  No string specified  */
 277
 278                 }
 279         /*  No header specified  */
 280
 281         /*  Open file, write buffer, close file  */
 282         snprintf (filename, sizeof(filename), "%s.%06i", fileprefix, fileno);
 283
 284         outfile = fopen (filename, "wt");
 285         if (outfile==NULL) {
 286                 printf ("Cannot open file name %s\n", filename);
 287                 exit(1);
 288                 }
 289
 290         while (n--)
 291                 fprintf (outfile, "%c", *buff++);
 292
 293         close (outfile);
 294
 295         /*  Return number of files written  */
 296         return 1;
 297         }