| 1 | /* |
| 2 | ** Copyright 2000 Double Precision, Inc. |
| 3 | ** See COPYING for distribution information. |
| 4 | */ |
| 5 | |
| 6 | /* |
| 7 | ** $Id: imapsubj.c,v 1.10 2006/04/11 02:24:59 mrsam Exp $ |
| 8 | */ |
| 9 | #include <stdio.h> |
| 10 | #include <ctype.h> |
| 11 | #include <stdlib.h> |
| 12 | #include <string.h> |
| 13 | #include "rfc822.h" |
| 14 | #include "config.h" |
| 15 | |
| 16 | #if HAVE_STRCASECMP |
| 17 | |
| 18 | #else |
| 19 | #define strcasecmp stricmp |
| 20 | #endif |
| 21 | |
| 22 | #if HAVE_STRNCASECMP |
| 23 | |
| 24 | #else |
| 25 | #define strncasecmp strnicmp |
| 26 | #endif |
| 27 | |
| 28 | /* Skip over blobs */ |
| 29 | |
| 30 | static char *skipblob(char *p, char **save_blob_ptr) |
| 31 | { |
| 32 | char *q; |
| 33 | char *orig_p=p; |
| 34 | int isalldigits=1; |
| 35 | |
| 36 | if (*p == '[') |
| 37 | { |
| 38 | for (q= p+1; *q; q++) |
| 39 | if (*q == '[' || *q == ']') |
| 40 | break; |
| 41 | else if (strchr("0123456789", *q) == NULL) |
| 42 | isalldigits=0; |
| 43 | |
| 44 | if (*q == ']') |
| 45 | { |
| 46 | p=q+1; |
| 47 | |
| 48 | while (isspace((int)(unsigned char)*p)) |
| 49 | { |
| 50 | ++p; |
| 51 | } |
| 52 | |
| 53 | if (save_blob_ptr && *save_blob_ptr && !isalldigits) |
| 54 | { |
| 55 | while (orig_p != p) |
| 56 | *(*save_blob_ptr)++=*orig_p++; |
| 57 | } |
| 58 | |
| 59 | return (p); |
| 60 | } |
| 61 | } |
| 62 | return (p); |
| 63 | } |
| 64 | |
| 65 | static char *skipblobs(char *p, char **save_blob_ptr) |
| 66 | { |
| 67 | char *q=p; |
| 68 | |
| 69 | do |
| 70 | { |
| 71 | p=q; |
| 72 | q=skipblob(p, save_blob_ptr); |
| 73 | } while (q != p); |
| 74 | return (q); |
| 75 | } |
| 76 | |
| 77 | /* Remove artifacts from the subject header */ |
| 78 | |
| 79 | static void stripsubj(char *s, int *hasrefwd, char *save_blob_buf) |
| 80 | { |
| 81 | char *p; |
| 82 | char *q; |
| 83 | int doit; |
| 84 | |
| 85 | for (p=q=s; *p; p++) |
| 86 | { |
| 87 | if (!isspace((int)(unsigned char)*p)) |
| 88 | { |
| 89 | *q++=*p; |
| 90 | continue; |
| 91 | } |
| 92 | while (p[1] && isspace((int)(unsigned char)p[1])) |
| 93 | { |
| 94 | ++p; |
| 95 | } |
| 96 | *q++=' '; |
| 97 | } |
| 98 | *q=0; |
| 99 | |
| 100 | do |
| 101 | { |
| 102 | doit=0; |
| 103 | /* |
| 104 | ** |
| 105 | ** (2) Remove all trailing text of the subject that matches |
| 106 | ** the subj-trailer ABNF, repeat until no more matches are |
| 107 | ** possible. |
| 108 | ** |
| 109 | ** subj-trailer = "(fwd)" / WSP |
| 110 | */ |
| 111 | |
| 112 | for (p=s; *p; p++) |
| 113 | ; |
| 114 | while (p > s) |
| 115 | { |
| 116 | if ( isspace((int)(unsigned char)p[-1])) |
| 117 | { |
| 118 | --p; |
| 119 | continue; |
| 120 | } |
| 121 | if (p-s >= 5 && strncasecmp(p-5, "(FWD)", 5) == 0) |
| 122 | { |
| 123 | p -= 5; |
| 124 | *hasrefwd |= CORESUBJ_FWD; |
| 125 | continue; |
| 126 | } |
| 127 | break; |
| 128 | } |
| 129 | *p=0; |
| 130 | |
| 131 | for (p=s; *p; ) |
| 132 | { |
| 133 | for (;;) |
| 134 | { |
| 135 | char *orig_blob_ptr; |
| 136 | int flag=CORESUBJ_FWD; |
| 137 | |
| 138 | /* |
| 139 | ** |
| 140 | ** (3) Remove all prefix text of the subject |
| 141 | ** that matches the subj-leader ABNF. |
| 142 | ** |
| 143 | ** subj-leader = (*subj-blob subj-refwd) / WSP |
| 144 | ** |
| 145 | ** subj-blob = "[" *BLOBCHAR "]" *WSP |
| 146 | ** |
| 147 | ** subj-refwd = ("re" / ("fw" ["d"])) *WSP [subj-blob] ":" |
| 148 | ** |
| 149 | ** BLOBCHAR = %x01-5a / %x5c / %x5e-7f |
| 150 | ** ; any CHAR except '[' and ']' |
| 151 | */ |
| 152 | |
| 153 | if (isspace((int)(unsigned char)*p)) |
| 154 | { |
| 155 | ++p; |
| 156 | continue; |
| 157 | } |
| 158 | |
| 159 | q=skipblobs(p, NULL); |
| 160 | |
| 161 | if (strncasecmp(q, "RE", 2) == 0) |
| 162 | { |
| 163 | flag=CORESUBJ_RE; |
| 164 | q += 2; |
| 165 | } |
| 166 | else if (strncasecmp(q, "FWD", 3) == 0) |
| 167 | { |
| 168 | q += 3; |
| 169 | } |
| 170 | else if (strncasecmp(q, "FW", 2) == 0) |
| 171 | { |
| 172 | q += 2; |
| 173 | } |
| 174 | else q=0; |
| 175 | |
| 176 | if (q) |
| 177 | { |
| 178 | orig_blob_ptr=save_blob_buf; |
| 179 | |
| 180 | q=skipblob(q, &save_blob_buf); |
| 181 | if (*q == ':') |
| 182 | { |
| 183 | p=q+1; |
| 184 | *hasrefwd |= flag; |
| 185 | continue; |
| 186 | } |
| 187 | |
| 188 | save_blob_buf=orig_blob_ptr; |
| 189 | } |
| 190 | |
| 191 | |
| 192 | /* |
| 193 | ** (4) If there is prefix text of the subject |
| 194 | ** that matches the subj-blob ABNF, and |
| 195 | ** removing that prefix leaves a non-empty |
| 196 | ** subj-base, then remove the prefix text. |
| 197 | ** |
| 198 | ** subj-base = NONWSP *([*WSP] NONWSP) |
| 199 | ** ; can be a subj-blob |
| 200 | */ |
| 201 | |
| 202 | orig_blob_ptr=save_blob_buf; |
| 203 | |
| 204 | q=skipblob(p, &save_blob_buf); |
| 205 | |
| 206 | if (q != p && *q) |
| 207 | { |
| 208 | p=q; |
| 209 | continue; |
| 210 | } |
| 211 | save_blob_buf=orig_blob_ptr; |
| 212 | break; |
| 213 | } |
| 214 | |
| 215 | /* |
| 216 | ** |
| 217 | ** (6) If the resulting text begins with the |
| 218 | ** subj-fwd-hdr ABNF and ends with the subj-fwd-trl |
| 219 | ** ABNF, remove the subj-fwd-hdr and subj-fwd-trl and |
| 220 | ** repeat from step (2). |
| 221 | ** |
| 222 | ** subj-fwd-hdr = "[fwd:" |
| 223 | ** |
| 224 | ** subj-fwd-trl = "]" |
| 225 | */ |
| 226 | |
| 227 | if (strncasecmp(p, "[FWD:", 5) == 0) |
| 228 | { |
| 229 | q=strrchr(p, ']'); |
| 230 | if (q && q[1] == 0) |
| 231 | { |
| 232 | *q=0; |
| 233 | p += 5; |
| 234 | *hasrefwd |= CORESUBJ_FWD; |
| 235 | |
| 236 | for (q=s; (*q++=*p++) != 0; ) |
| 237 | ; |
| 238 | doit=1; |
| 239 | } |
| 240 | } |
| 241 | break; |
| 242 | } |
| 243 | } while (doit); |
| 244 | |
| 245 | q=s; |
| 246 | while ( (*q++ = *p++) != 0) |
| 247 | ; |
| 248 | if (save_blob_buf) |
| 249 | *save_blob_buf=0; |
| 250 | } |
| 251 | |
| 252 | char *rfc822_coresubj(const char *s, int *hasrefwd) |
| 253 | { |
| 254 | char *q=strdup(s), *r; |
| 255 | int dummy; |
| 256 | |
| 257 | if (!hasrefwd) |
| 258 | hasrefwd= &dummy; |
| 259 | |
| 260 | *hasrefwd=0; |
| 261 | if (!q) return (0); |
| 262 | |
| 263 | for (r=q; *r; r++) |
| 264 | if ((*r & 0x80) == 0) /* Just US-ASCII casing, thanks */ |
| 265 | *r=toupper((int)(unsigned char)*r); |
| 266 | stripsubj(q, hasrefwd, 0); |
| 267 | return (q); |
| 268 | } |
| 269 | |
| 270 | char *rfc822_coresubj_nouc(const char *s, int *hasrefwd) |
| 271 | { |
| 272 | char *q=strdup(s); |
| 273 | int dummy; |
| 274 | |
| 275 | if (!hasrefwd) |
| 276 | hasrefwd= &dummy; |
| 277 | |
| 278 | *hasrefwd=0; |
| 279 | if (!q) return (0); |
| 280 | |
| 281 | stripsubj(q, hasrefwd, 0); |
| 282 | return (q); |
| 283 | } |
| 284 | |
| 285 | char *rfc822_coresubj_keepblobs(const char *s) |
| 286 | { |
| 287 | char *q=strdup(s), *r; |
| 288 | int dummy; |
| 289 | |
| 290 | if (!q) return (0); |
| 291 | |
| 292 | r=strdup(s); |
| 293 | if (!r) |
| 294 | { |
| 295 | free(q); |
| 296 | return (0); |
| 297 | } |
| 298 | |
| 299 | stripsubj(q, &dummy, r); |
| 300 | strcat(r, q); |
| 301 | free(q); |
| 302 | return (r); |
| 303 | } |