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 date[1000];
  14 static char name[1000];
  15 static char email[1000];
  16 static char subject[1000];
  17
  18 static char *sanity_check(char *name, char *email)
  19 {
  20         int len = strlen(name);
  21         if (len < 3 || len > 60)
  22                 return email;
  23         if (strchr(name, '@') || strchr(name, '<') || strchr(name, '>'))
  24                 return email;
  25         return name;
  26 }
  27
  28 static int handle_from(char *line)
  29 {
  30         char *at = strchr(line, '@');
  31         char *dst;
  32
  33         if (!at)
  34                 return 0;
  35
  36         /*
  37          * If we already have one email, don't take any confusing lines
  38          */
  39         if (*email && strchr(at+1, '@'))
  40                 return 0;
  41
  42         while (at > line) {
  43                 char c = at[-1];
  44                 if (isspace(c) || c == '<')
  45                         break;
  46                 at--;
  47         }
  48         dst = email;
  49         for (;;) {
  50                 unsigned char c = *at;
  51                 if (!c || c == '>' || isspace(c))
  52                         break;
  53                 *at++ = ' ';
  54                 *dst++ = c;
  55         }
  56         *dst++ = 0;
  57
  58         at = line + strlen(line);
  59         while (at > line) {
  60                 unsigned char c = *--at;
  61                 if (isalnum(c))
  62                         break;
  63                 *at = 0;
  64         }
  65
  66         at = line;
  67         for (;;) {
  68                 unsigned char c = *at;
  69                 if (!c)
  70                         break;
  71                 if (isalnum(c))
  72                         break;
  73                 at++;
  74         }
  75
  76         at = sanity_check(at, email);
  77
  78         strcpy(name, at);
  79         return 1;
  80 }
  81
  82 static void handle_date(char *line)
  83 {
  84         strcpy(date, line);
  85 }
  86
  87 static void handle_subject(char *line)
  88 {
  89         strcpy(subject, line);
  90 }
  91
  92 static void add_subject_line(char *line)
  93 {
  94         while (isspace(*line))
  95                 line++;
  96         *--line = ' ';
  97         strcat(subject, line);
  98 }
  99
 100 static void check_line(char *line, int len)
 101 {
 102         static int cont = -1;
 103         if (!memcmp(line, "From:", 5) && isspace(line[5])) {
 104                 handle_from(line+6);
 105                 cont = 0;
 106                 return;
 107         }
 108         if (!memcmp(line, "Date:", 5) && isspace(line[5])) {
 109                 handle_date(line+6);
 110                 cont = 0;
 111                 return;
 112         }
 113         if (!memcmp(line, "Subject:", 8) && isspace(line[8])) {
 114                 handle_subject(line+9);
 115                 cont = 1;
 116                 return;
 117         }
 118         if (isspace(*line)) {
 119                 switch (cont) {
 120                 case 0:
 121                         fprintf(stderr, "I don't do 'Date:' or 'From:' line continuations\n");
 122                         break;
 123                 case 1:
 124                         add_subject_line(line);
 125                         return;
 126                 default:
 127                         break;
 128                 }
 129         }
 130         cont = -1;
 131 }
 132
 133 static char * cleanup_subject(char *subject)
 134 {
 135         for (;;) {
 136                 char *p;
 137                 int len, remove;
 138                 switch (*subject) {
 139                 case 'r': case 'R':
 140                         if (!memcmp("e:", subject+1, 2)) {
 141                                 subject +=3;
 142                                 continue;
 143                         }
 144                         break;
 145                 case ' ': case '\t': case ':':
 146                         subject++;
 147                         continue;
 148
 149                 case '[':
 150                         p = strchr(subject, ']');
 151                         if (!p) {
 152                                 subject++;
 153                                 continue;
 154                         }
 155                         len = strlen(p);
 156                         remove = p - subject;
 157                         if (remove <= len *2) {
 158                                 subject = p+1;
 159                                 continue;
 160                         }
 161                         break;
 162                 }
 163                 return subject;
 164         }
 165 }
 166
 167 static void cleanup_space(char *buf)
 168 {
 169         unsigned char c;
 170         while ((c = *buf) != 0) {
 171                 buf++;
 172                 if (isspace(c)) {
 173                         buf[-1] = ' ';
 174                         c = *buf;
 175                         while (isspace(c)) {
 176                                 int len = strlen(buf);
 177                                 memmove(buf, buf+1, len);
 178                                 c = *buf;
 179                         }
 180                 }
 181         }
 182 }
 183
 184 /*
 185  * Hacky hacky. This depends not only on -p1, but on
 186  * filenames not having some special characters in them,
 187  * like tilde.
 188  */
 189 static void show_filename(char *line)
 190 {
 191         int len;
 192         char *name = strchr(line, '/');
 193
 194         if (!name || !isspace(*line))
 195                 return;
 196         name++;
 197         len = 0;
 198         for (;;) {
 199                 unsigned char c = name[len];
 200                 switch (c) {
 201                 default:
 202                         len++;
 203                         continue;
 204
 205                 case 0: case ' ':
 206                 case '\t': case '\n':
 207                         break;
 208
 209                 /* patch tends to special-case these things.. */
 210                 case '~':
 211                         break;
 212                 }
 213                 break;
 214         }
 215         /* remove ".orig" from the end - common patch behaviour */
 216         if (len > 5 && !memcmp(name+len-5, ".orig", 5))
 217                 len -=5;
 218         if (!len)
 219                 return;
 220         fprintf(filelist, "%.*s\n", len, name);
 221 }
 222
 223 static void handle_rest(void)
 224 {
 225         char *sub = cleanup_subject(subject);
 226         cleanup_space(name);
 227         cleanup_space(date);
 228         cleanup_space(email);
 229         cleanup_space(sub);
 230         printf("Author: %s\nEmail: %s\nSubject: %s\nDate: %s\n\n", name, email, sub, date);
 231         FILE *out = cmitmsg;
 232
 233         do {
 234                 /* Track filename information from the patch.. */
 235                 if (!memcmp("---", line, 3)) {
 236                         out = patchfile;
 237                         show_filename(line+3);
 238                 }
 239
 240                 if (!memcmp("+++", line, 3))
 241                         show_filename(line+3);
 242
 243                 fputs(line, out);
 244         } while (fgets(line, sizeof(line), stdin) != NULL);
 245
 246         if (out == cmitmsg) {
 247                 fprintf(stderr, "No patch found\n");
 248                 exit(1);
 249         }
 250
 251         fclose(cmitmsg);
 252         fclose(patchfile);
 253 }
 254
 255 static int eatspace(char *line)
 256 {
 257         int len = strlen(line);
 258         while (len > 0 && isspace(line[len-1]))
 259                 line[--len] = 0;
 260         return len;
 261 }
 262
 263 static void handle_body(void)
 264 {
 265         int has_from = 0;
 266
 267         /* First line of body can be a From: */
 268         while (fgets(line, sizeof(line), stdin) != NULL) {
 269                 int len = eatspace(line);
 270                 if (!len)
 271                         continue;
 272                 if (!memcmp("From:", line, 5) && isspace(line[5])) {
 273                         if (!has_from && handle_from(line+6)) {
 274                                 has_from = 1;
 275                                 continue;
 276                         }
 277                 }
 278                 line[len] = '\n';
 279                 handle_rest();
 280                 break;
 281         }
 282 }
 283
 284 static void usage(void)
 285 {
 286         fprintf(stderr, "mailinfo msg-file path-file filelist-file < email\n");
 287         exit(1);
 288 }
 289
 290 int main(int argc, char ** argv)
 291 {
 292         if (argc != 4)
 293                 usage();
 294         cmitmsg = fopen(argv[1], "w");
 295         if (!cmitmsg) {
 296                 perror(argv[1]);
 297                 exit(1);
 298         }
 299         patchfile = fopen(argv[2], "w");
 300         if (!patchfile) {
 301                 perror(argv[2]);
 302                 exit(1);
 303         }
 304         filelist = fopen(argv[3], "w");
 305         if (!filelist) {
 306                 perror(argv[3]);
 307                 exit(1);
 308         }
 309         while (fgets(line, sizeof(line), stdin) != NULL) {
 310                 int len = eatspace(line);
 311                 if (!len) {
 312                         handle_body();
 313                         break;
 314                 }
 315                 check_line(line, len);
 316         }
 317         return 0;
 318 }