mailinfo.c

   1 /*
   2  * Another stupid program, this one parsing the headers of an
   3  * email to figure out authorship and subject
   4  */
   5 #include <stdio.h>
   6 #include <stdlib.h>
   7 #include <string.h>
   8 #include <ctype.h>
   9
  10 static FILE *cmitmsg, *patchfile, *filelist;
  11
  12 static char line[1000];
  13 static char name[1000];
  14 static char email[1000];
  15 static char subject[1000];
  16
  17 static char *sanity_check(char *name, char *email)
  18 {
  19         int len = strlen(name);
  20         if (len < 3 || len > 60)
  21                 return email;
  22         if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
  23                 return email;
  24         return name;
  25 }
  26
  27 static int handle_from(char *line)
  28 {
  29         char *at = strchr(line, '@');
  30         char *dst;
  31
  32         if (!at)
  33                 return 0;
  34
  35         /*
  36          * If we already have one email, don't take any confusing lines
  37          */
  38         if (*email && strchr(at+1, '@'))
  39                 return 0;
  40
  41         while (at > line) {
  42                 char c = at[-1];
  43                 if (isspace(c) || c == '<')
  44                         break;
  45                 at--;
  46         }
  47         dst = email;
  48         for (;;) {
  49                 unsigned char c = *at;
  50                 if (!c || c == '>' || isspace(c))
  51                         break;
  52                 *at++ = ' ';
  53                 *dst++ = c;
  54         }
  55         *dst++ = 0;
  56
  57         at = line + strlen(line);
  58         while (at > line) {
  59                 unsigned char c = *--at;
  60                 if (isalnum(c))
  61                         break;
  62                 *at = 0;
  63         }
  64
  65         at = line;
  66         for (;;) {
  67                 unsigned char c = *at;
  68                 if (!c)
  69                         break;
  70                 if (isalnum(c))
  71                         break;
  72                 at++;
  73         }
  74
  75         at = sanity_check(at, email);
  76
  77         strcpy(name, at);
  78         return 1;
  79 }
  80
  81 static void handle_subject(char *line)
  82 {
  83         strcpy(subject, line);
  84 }
  85
  86 static void add_subject_line(char *line)
  87 {
  88         while (isspace(*line))
  89                 line++;
  90         *--line = ' ';
  91         strcat(subject, line);
  92 }
  93
  94 static void check_line(char *line, int len)
  95 {
  96         static int cont = -1;
  97         if (!memcmp(line, "From:", 5) && isspace(line[5])) {
  98                 handle_from(line+6);
  99                 cont = 0;
 100                 return;
 101         }
 102         if (!memcmp(line, "Subject:", 8) && isspace(line[8])) {
 103                 handle_subject(line+9);
 104                 cont = 1;
 105                 return;
 106         }
 107         if (isspace(*line)) {
 108                 switch (cont) {
 109                 case 0:
 110                         fprintf(stderr, "I don't do 'From:' line continuations\n");
 111                         break;
 112                 case 1:
 113                         add_subject_line(line);
 114                         return;
 115                 default:
 116                         break;
 117                 }
 118         }
 119         cont = -1;
 120 }
 121
 122 static char * cleanup_subject(char *subject)
 123 {
 124         for (;;) {
 125                 char *p;
 126                 int len, remove;
 127                 switch (*subject) {
 128                 case 'r': case 'R':
 129                         if (!memcmp("e:", subject+1, 2)) {
 130                                 subject +=3;
 131                                 continue;
 132                         }
 133                         break;
 134                 case ' ': case '\t': case ':':
 135                         subject++;
 136                         continue;
 137
 138                 case '[':
 139                         p = strchr(subject, ']');
 140                         if (!p) {
 141                                 subject++;
 142                                 continue;
 143                         }
 144                         len = strlen(p);
 145                         remove = p - subject;
 146                         if (remove <= len *2) {
 147                                 subject = p+1;
 148                                 continue;
 149                         }
 150                         break;
 151                 }
 152                 return subject;
 153         }
 154 }
 155
 156 static void cleanup_space(char *buf)
 157 {
 158         unsigned char c;
 159         while ((c = *buf) != 0) {
 160                 buf++;
 161                 if (isspace(c)) {
 162                         buf[-1] = ' ';
 163                         c = *buf;
 164                         while (isspace(c)) {
 165                                 int len = strlen(buf);
 166                                 memmove(buf, buf+1, len);
 167                                 c = *buf;
 168                         }
 169                 }
 170         }
 171 }
 172
 173 /*
 174  * Hacky hacky. This depends not only on -p1, but on
 175  * filenames not having some special characters in them,
 176  * like tilde.
 177  */
 178 static void show_filename(char *line)
 179 {
 180         int len;
 181         char *name = strchr(line, '/');
 182
 183         if (!name || !isspace(*line))
 184                 return;
 185         name++;
 186         len = 0;
 187         for (;;) {
 188                 unsigned char c = name[len];
 189                 switch (c) {
 190                 default:
 191                         len++;
 192                         continue;
 193
 194                 case 0: case ' ':
 195                 case '\t': case '\n':
 196                         break;
 197
 198                 /* patch tends to special-case these things.. */
 199                 case '~':
 200                         break;
 201                 }
 202                 break;
 203         }
 204         /* remove ".orig" from the end - common patch behaviour */
 205         if (len > 5 && !memcmp(name+len-5, ".orig", 5))
 206                 len -=5;
 207         if (!len)
 208                 return;
 209         fprintf(filelist, "%.*s\n", len, name);
 210 }
 211
 212 static void handle_rest(void)
 213 {
 214         char *sub = cleanup_subject(subject);
 215         cleanup_space(name);
 216         cleanup_space(email);
 217         cleanup_space(sub);
 218         printf("Author: %s\nEmail: %s\nSubject: %s\n\n", name, email, sub);
 219         FILE *out = cmitmsg;
 220
 221         do {
 222                 /* Track filename information from the patch.. */
 223                 if (!memcmp("---", line, 3)) {
 224                         out = patchfile;
 225                         show_filename(line+3);
 226                 }
 227
 228                 if (!memcmp("+++", line, 3))
 229                         show_filename(line+3);
 230
 231                 fputs(line, out);
 232         } while (fgets(line, sizeof(line), stdin) != NULL);
 233
 234         if (out == cmitmsg) {
 235                 fprintf(stderr, "No patch found\n");
 236                 exit(1);
 237         }
 238
 239         fclose(cmitmsg);
 240         fclose(patchfile);
 241 }
 242
 243 static int eatspace(char *line)
 244 {
 245         int len = strlen(line);
 246         while (len > 0 && isspace(line[len-1]))
 247                 line[--len] = 0;
 248         return len;
 249 }
 250
 251 static void handle_body(void)
 252 {
 253         int has_from = 0;
 254
 255         /* First line of body can be a From: */
 256         while (fgets(line, sizeof(line), stdin) != NULL) {
 257                 int len = eatspace(line);
 258                 if (!len)
 259                         continue;
 260                 if (!memcmp("From:", line, 5) && isspace(line[5])) {
 261                         if (!has_from && handle_from(line+6)) {
 262                                 has_from = 1;
 263                                 continue;
 264                         }
 265                 }
 266                 line[len] = '\n';
 267                 handle_rest();
 268                 break;
 269         }
 270 }
 271
 272 static void usage(void)
 273 {
 274         fprintf(stderr, "mailinfo msg-file path-file filelist-file < email\n");
 275         exit(1);
 276 }
 277
 278 int main(int argc, char ** argv)
 279 {
 280         if (argc != 4)
 281                 usage();
 282         cmitmsg = fopen(argv[1], "w");
 283         if (!cmitmsg) {
 284                 perror(argv[1]);
 285                 exit(1);
 286         }
 287         patchfile = fopen(argv[2], "w");
 288         if (!patchfile) {
 289                 perror(argv[2]);
 290                 exit(1);
 291         }
 292         filelist = fopen(argv[3], "w");
 293         if (!filelist) {
 294                 perror(argv[3]);
 295                 exit(1);
 296         }
 297         while (fgets(line, sizeof(line), stdin) != NULL) {
 298                 int len = eatspace(line);
 299                 if (!len) {
 300                         handle_body();
 301                         break;
 302                 }
 303                 check_line(line, len);
 304         }
 305         return 0;
 306 }