src/bin/mail_utils/mail2mbox.cpp

   1 /*
   2  * Copyright 2005-2009, Haiku Inc.
   3  * This file may be used under the terms of the MIT License.
   4  *
   5  * Originally public domain written by Alexander G. M. Smith.
   6  */
   7
   8
   9 /*!     BeMailToMBox is a utility program (requested by Frank Zschockelt) that
  10         converts BeOS e-mail files into Unix mailbox files (the kind that Pine
  11         uses).  All the files in the input directory are concatenated with the
  12         appropriate mbox header lines added between them, and trailing blank lines
  13         reduced.  The resulting text is written to standard output.  Command line
  14         driven.
  15 */
  16
  17 #include <ctype.h>
  18 #include <errno.h>
  19 #include <string.h>
  20 #include <stdio.h>
  21 #include <time.h>
  22
  23 #include <Application.h>
  24 #include <StorageKit.h>
  25 #include <SupportKit.h>
  26
  27
  28 extern const char* __progname;
  29 static const char* kProgramName = __progname;
  30
  31 time_t gDateStampTime;
  32         // Time value used for stamping each message header. Incremented by 1 second
  33         // for each message, starts out with the current local time.
  34
  35
  36 /*!     Global utility function to display an error message and return.  The message
  37         part describes the error, and if errorNumber is non-zero, gets the string
  38         ", error code $X (standard description)." appended to it.  If the message
  39         is NULL then it gets defaulted to "Something went wrong".
  40 */
  41 static void
  42 DisplayErrorMessage(const char* messageString = NULL, status_t errorNumber = 0,
  43         const char* titleString = NULL)
  44 {
  45         char errorBuffer[2048];
  46
  47         if (titleString == NULL)
  48                 titleString = "Error Message:";
  49
  50         if (messageString == NULL) {
  51                 if (errorNumber == B_OK)
  52                         messageString = "No error, no message, why bother?";
  53                 else
  54                         messageString = "Error";
  55         }
  56
  57         if (errorNumber != 0) {
  58                 snprintf(errorBuffer, sizeof(errorBuffer), "%s: %s (%lx)"
  59                         "has occured.", messageString, strerror(errorNumber), errorNumber);
  60                 messageString = errorBuffer;
  61         }
  62
  63         fputs(titleString, stderr);
  64         fputc('\n', stderr);
  65         fputs(messageString, stderr);
  66         fputc('\n', stderr);
  67 }
  68
  69
  70 /*!     Determine if a line of text is the start of another message.  Pine mailbox
  71         files have messages that start with a line that could say something like
  72         "From agmsmith@achilles.net Fri Oct 31 21:19:36 EST 1997" or maybe something
  73         like "From POPmail Mon Oct 20 21:12:36 1997" or in a more modern format,
  74         "From agmsmith@achilles.net Tue Sep 4 09:04:11 2001 -0400".  I generalise it
  75         to "From blah Day MMM NN XX:XX:XX TZONE1 YYYY TZONE2".  Blah is an e-mail
  76         address you can ignore (just treat it as a word separated by spaces).  Day
  77         is a 3 letter day of the week.  MMM is a 3 letter month name.  NN is the two
  78         digit day of the week, has a leading space if the day is less than 10.
  79         XX:XX:XX is the time, the X's are digits.  TZONE1 is the old style optional
  80         time zone of 3 capital letters.  YYYY is the four digit year.  TZONE2 is the
  81         optional modern time zone info, a plus or minus sign and 4 digits.  Returns
  82         true if the line of text (ended with a NUL byte, no line feed or carriage
  83         returns at the end) is the start of a message.
  84 */
  85 bool
  86 IsStartOfMailMessage(char* lineString)
  87 {
  88         // It starts with "From "
  89         if (memcmp("From ", lineString, 5) != 0)
  90                 return false;
  91
  92         char* string = lineString + 4;
  93         while (*string == ' ')
  94                 string++;
  95
  96         // Skip over the e-mail address (or stop at the end of string).
  97
  98         while (*string != ' ' && *string != 0)
  99                 string++;
 100         while (*string == ' ')
 101                 string++;
 102
 103         // TODO: improve this!!!
 104
 105         // Look for the 3 letter day of the week.
 106         if (memcmp(string, "Mon", 3) != 0 && memcmp(string, "Tue", 3) != 0
 107                 && memcmp(string, "Wed", 3) != 0 && memcmp(string, "Thu", 3) != 0
 108                 && memcmp(string, "Fri", 3) != 0 && memcmp(string, "Sat", 3) != 0
 109                 && memcmp(string, "Sun", 3) != 0) {
 110                 fprintf(stderr, "False alarm, not a valid day of the week in \"%s\""
 111                         ".\n", lineString);
 112                 return false;
 113         }
 114
 115         string += 3;
 116         while (*string == ' ')
 117                 string++;
 118
 119         // Look for the 3 letter month code.
 120         if (memcmp(string, "Jan", 3) != 0 && memcmp(string, "Feb", 3) != 0
 121                 && memcmp(string, "Mar", 3) != 0 && memcmp(string, "Apr", 3) != 0
 122                 && memcmp(string, "May", 3) != 0 && memcmp(string, "Jun", 3) != 0
 123                 && memcmp(string, "Jul", 3) != 0 && memcmp(string, "Aug", 3) != 0
 124                 && memcmp(string, "Sep", 3) != 0 && memcmp(string, "Oct", 3) != 0
 125                 && memcmp(string, "Nov", 3) != 0 && memcmp(string, "Dec", 3) != 0) {
 126                 fprintf(stderr, "False alarm, not a valid month name in \"%s\".\n",
 127                         lineString);
 128                 return false;
 129         }
 130
 131         string += 3;
 132         while (*string == ' ')
 133                 string++;
 134
 135         // Skip the day of the month.  Require at least one digit.
 136         if (*string < '0' || *string > '9') {
 137                 fprintf(stderr, "False alarm, not a valid day of the "
 138                         "month number in \"%s\".\n", lineString);
 139                 return false;
 140         }
 141
 142         while (*string >= '0' && *string <= '9')
 143                 string++;
 144         while (*string == ' ')
 145                 string++;
 146
 147         // Check the time.  Look for the sequence
 148         // digit-digit-colon-digit-digit-colon-digit-digit.
 149
 150         if (string[0] < '0' || string[0] > '9'
 151                 || string[1] < '0' || string[1] > '9'
 152                 || string[2] != ':'
 153                 || string[3] < '0' || string[3] > '9'
 154                 || string[4] < '0' || string[4] > '9'
 155                 || string[5] != ':'
 156                 || string[6] < '0' || string[6] > '9'
 157                 || string[7] < '0' || string[7] > '9') {
 158                 fprintf(stderr, "False alarm, not a valid time value in \"%s\".\n",
 159                         lineString);
 160                 return false;
 161         }
 162
 163         string += 8;
 164         while (*string == ' ')
 165                 string++;
 166
 167         // Look for the optional antique 3 capital letter time zone and skip it.
 168         if (string[0] >= 'A' && string[0] <= 'Z'
 169                 && string[1] >= 'A' && string[1] <= 'Z'
 170                 && string[2] >= 'A' && string[2] <= 'Z') {
 171                 string += 3;
 172                 while (*string == ' ')
 173                         string++;
 174         }
 175
 176         // Look for the 4 digit year.
 177         if (string[0] < '0' || string[0] > '9'
 178                 || string[1] < '0' || string[1] > '9'
 179                 || string[2] < '0' || string[2] > '9'
 180                 || string[3] < '0' || string[3] > '9') {
 181                 fprintf(stderr, "False alarm, not a valid 4 digit year in \"%s\".\n",
 182                         lineString);
 183                 return false;
 184         }
 185
 186         string += 4;
 187         while (*string == ' ')
 188                 string++;
 189
 190         // Look for the optional modern time zone and skip over it if present.
 191         if ((string[0] == '+' || string[0] == '-')
 192                 && string[1] >= '0' && string[1] <= '9'
 193                 && string[2] >= '0' && string[2] <= '9'
 194                 && string[3] >= '0' && string[3] <= '9'
 195                 && string[4] >= '0' && string[4] <= '9') {
 196                 string += 5;
 197                 while (*string == ' ')
 198                         string++;
 199         }
 200
 201         // Look for end of string.
 202         if (*string != 0) {
 203                 fprintf(stderr, "False alarm, extra stuff after the "
 204                         "year/time zone in \"%s\".\n", lineString);
 205                 return false;
 206         }
 207
 208         return true;
 209 }
 210
 211
 212 /*!     Read the input file, convert it to mbox format, and write it to standard
 213         output.  Returns zero if successful, a negative error code if an error
 214         occured.
 215 */
 216 status_t
 217 ProcessMessageFile(char* fileName)
 218 {
 219         fprintf(stdout, "Now processing: \"%s\"\n", fileName);
 220
 221         FILE* inputFile = fopen(fileName, "rb");
 222         if (inputFile == NULL) {
 223                 DisplayErrorMessage("Unable to open file", errno);
 224                 return errno;
 225         }
 226
 227         // Extract a text message from the Mail file.
 228
 229         BString messageText;
 230         int lineNumber = 0;
 231
 232         while (!feof(inputFile)) {
 233                 // First read in one line of text.
 234                 char line[102400];
 235                 if (fgets(line, sizeof(line), inputFile) == NULL) {
 236                         if (ferror(inputFile)) {
 237                                 char errorString[2048];
 238                                 snprintf(errorString, sizeof(errorString),
 239                                         "Error while reading from \"%s\"", fileName);
 240                                 DisplayErrorMessage(errorString, errno);
 241                                 fclose(inputFile);
 242                                 return errno;
 243                         }
 244                         break;
 245                                 // No error, just end of file.
 246                 }
 247
 248                 // Remove any trailing control characters (line feed usually, or CRLF).
 249                 // Might also nuke trailing tabs too. Doesn't usually matter. The main
 250                 // thing is to allow input files with both LF and CRLF endings (and
 251                 // even CR endings if you come from the Macintosh world).
 252
 253                 char* string = line + strlen(line) - 1;
 254                 while (string >= line && *string < 32)
 255                         string--;
 256                 *(++string) = 0;
 257
 258                 if (lineNumber == 0 && line[0] == 0) {
 259                         // Skip leading blank lines.
 260                         continue;
 261                 }
 262                 lineNumber++;
 263
 264                 // Prepend the new mbox message header, if the first line of the message
 265                 // doesn't already have one.
 266                 if (lineNumber == 1 && !IsStartOfMailMessage(line)) {
 267                         time_t timestamp = gDateStampTime++;
 268                         messageText.Append("From baron@be.com ");
 269                         messageText.Append(ctime(&timestamp));
 270                 }
 271
 272                 // Append the line to the current message text.
 273                 messageText.Append(line);
 274                 messageText.Append("\n");
 275         }
 276
 277         // Remove blank lines from the end of the message (a pet peeve of mine), but
 278         // end the message with two new lines to separate it from the next message.
 279         int i = messageText.Length();
 280         while (i > 0 && (messageText[i - 1] == '\n' || messageText[i - 1] == '\r'))
 281                 i--;
 282         messageText.Truncate(i);
 283         messageText.Append("\n\n");
 284
 285         // Write the message out.
 286
 287         status_t status = B_OK;
 288
 289         if (puts(messageText.String()) < 0) {
 290                 DisplayErrorMessage ("Error while writing the message", errno);
 291                 status = errno;
 292         }
 293
 294         fclose(inputFile);
 295         return status;
 296 }
 297
 298
 299 int
 300 main(int argc, char** argv)
 301 {
 302         BApplication app("application/x-vnd.Haiku-mail2mbox");
 303
 304         if (argc <= 1 || argc >= 3) {
 305                 printf("%s is a utility for converting Mail e-mail\n", argv[0]);
 306                 printf("files to Unix Pine style e-mail files.  It could well\n");
 307                 printf("work with other Unix style mailbox files.  Each message in\n");
 308                 printf("the input directory is converted and sent to the standard\n");
 309                 printf("output.  Usage:\n\n");
 310                 printf("%s InputDirectory >OutputFile\n\n", kProgramName);
 311                 printf("Public domain, by Alexander G. M. Smith.\n");
 312                 return -10;
 313         }
 314
 315         // Set the date stamp to the current time.
 316         gDateStampTime = time (NULL);
 317
 318         // Try to open the input directory.
 319         char inputPathName[B_PATH_NAME_LENGTH];
 320         strlcpy(inputPathName, argv[1], sizeof(inputPathName) - 2);
 321
 322         char tempString[2048];
 323
 324         DIR* dir = opendir(inputPathName);
 325         if (dir == NULL) {
 326                 sprintf(tempString, "Problems opening directory named \"%s\".",
 327                         inputPathName);
 328                 DisplayErrorMessage(tempString, errno);
 329                 return 1;
 330         }
 331
 332         // Append a trailing slash to the directory name, if it needs one.
 333         if (inputPathName[strlen(inputPathName) - 1] != '/')
 334                 strcat(inputPathName, "/");
 335
 336         int messagesDoneCount = 0;
 337         status_t status = B_OK;
 338
 339         while (dirent_t* entry = readdir(dir)) {
 340                 // skip '.' and '..'
 341                 if (!strcmp(entry->d_name, ".") || !strcmp(entry->d_name, ".."))
 342                         break;
 343
 344                 strlcpy(tempString, inputPathName, sizeof(tempString));
 345                 strlcat(tempString, entry->d_name, sizeof(tempString));
 346
 347                 status = ProcessMessageFile(tempString);
 348                 if (status != B_OK)
 349                         break;
 350
 351                 messagesDoneCount++;
 352         }
 353
 354         closedir(dir);
 355
 356         if (status != B_OK) {
 357                 DisplayErrorMessage("Stopping early because an error occured", status);
 358                 return status;
 359         }
 360
 361         fprintf(stderr, "Did %d messages successfully.\n", messagesDoneCount);
 362         return 0;
 363 }