Mercurial > libpst

comparison pst2ldif.cpp @ 8:2b58cf15aaf7

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
pst is case sensitive, ldap is not
author carl
date Thu, 23 Dec 2004 15:49:19 -0800
parents
children 8c724896a28a
comparison
equal deleted inserted replaced
7:3e3d8b576630 8:2b58cf15aaf7
1 /*
2
3 Copyright (c) 2004 Carl Byington - 510 Software Group, released under
4 the GPL version 2 or any later version at your choice available at
5 http://www.fsf.org/licenses/gpl.txt
6
7 Based on readpst.c by David Smith
8
9 */
10
11 using namespace std;
12
13 #include <stdio.h>
14 #include <stdlib.h>
15 #include <time.h>
16 #include <string.h>
17 #include <ctype.h>
18 #include <limits.h>
19 #include <errno.h>
20
21 #ifndef _WIN32
22 # include <unistd.h>
23 # include <sys/stat.h> //mkdir
24
25 // for reading of directory and clearing in function mk_separate_dir
26 # include <sys/types.h>
27 # include <dirent.h>
28 #else
29 # include <direct.h>
30 # define chdir _chdir
31 # define int32_t __int32
32 #endif
33
34 #ifndef __GNUC__
35 # include "XGetopt.h"
36 #endif
37
38 // needed for std c++ collections
39 #include <set>
40
41 extern "C" {
42 #include "libstrfunc.h" // for base64_encoding
43 #include "define.h"
44 #include "libpst.h"
45 #include "common.h"
46 #include "timeconv.h"
47 #include "lzfu.h"
48 }
49
50 #define VERSION "0.2"
51
52 int32_t usage();
53 int32_t version();
54 char *my_stristr(char *haystack, char *needle);
55 char *check_filename(char *fname);
56 char *single(char *str);
57 char *folded(char *str);
58 void multi(char *fmt, char *str);
59 char *rfc2426_escape(char *str);
60 int32_t chr_count(char *str, char x);
61
62 char *prog_name;
63 pst_file pstfile;
64 char *ldap_base = NULL; // 'o=some.domain.tld, c=US'
65 char *ldap_class = NULL; // 'newPerson'
66 char *ldap_org = NULL; // 'o=some.domain.tld', computed from ldap_base
67
68
69 ////////////////////////////////////////////////
70 // define our ordering
71 struct ltstr {
72 bool operator()(char* s1, char* s2) const {
73 return strcasecmp(s1, s2) < 0;
74 }
75 };
76 // define our set
77 typedef set<char *, ltstr> string_set;
78 // make a static set to hold the cn values
79 static string_set all_strings;
80
81
82 ////////////////////////////////////////////////
83 // helper to register a string in a string set
84 //
85 static char* register_string(string_set &s, char *name);
86 static char* register_string(string_set &s, char *name) {
87 string_set::iterator i = s.find(name);
88 if (i != s.end()) return *i;
89 char *x = strdup(name);
90 s.insert(x);
91 return x;
92 }
93
94 ////////////////////////////////////////////////
95 // register a global string
96 //
97 static char* register_string(char *name);
98 static char* register_string(char *name) {
99 return register_string(all_strings, name);
100 }
101
102
103 ////////////////////////////////////////////////
104 // make a unique string
105 //
106 static char* unique_string(char *name);
107 static char* unique_string(char *name) {
108 int unique = 2;
109 string_set::iterator i = all_strings.find(name);
110 if (i == all_strings.end()) return register_string(name);
111 while (true) {
112 char n[strlen(name)+10];
113 snprintf(n, sizeof(n), "%s %d", name, unique++);
114 string_set::iterator i = all_strings.find(n);
115 if (i == all_strings.end()) return register_string(n);
116 }
117 }
118
119
120 ////////////////////////////////////////////////
121 // remove leading and trailing blanks
122 //
123 static char *trim(char *name);
124 static char *trim(char *name) {
125 char *p;
126 while (*name == ' ') name++;
127 p = name + strlen(name) - 1;
128 while ((p >= name) && (*p == ' ')) *p-- = '\0';
129 return name;
130 }
131
132
133 static void process(pst_desc_ll *d_ptr);
134 static void process(pst_desc_ll *d_ptr) {
135 pst_item *item = NULL;
136 while (d_ptr) {
137 if (d_ptr->desc) {
138 item = (pst_item*)_pst_parse_item(&pstfile, d_ptr);
139 if (item) {
140 if (item->message_store) {
141 // there should only be one message_store, and we have already done it
142 DIE(("main: A second message_store has been found. Sorry, this must be an error.\n"));
143 }
144
145 if (item->folder && d_ptr->child && strcasecmp(item->file_as, "Deleted Items")) {
146 //if this is a non-empty folder other than deleted items, we want to recurse into it
147 fprintf(stderr, "entering folder %s\n", item->file_as);
148 process(d_ptr->child);
149 } else if (item->contact) {
150 // deal with a contact
151 if (item->type != PST_TYPE_CONTACT) {
152 DIE(("type should be contact\n"));
153 }
154 else if (item->contact == NULL) { // this is an incorrect situation. Inform user
155 DIE(("null item contact\n"));
156 } else {
157 char cn[1000];
158 snprintf(cn, sizeof(cn), "%s %s %s %s",
159 single(item->contact->display_name_prefix),
160 single(item->contact->first_name),
161 single(item->contact->surname),
162 single(item->contact->suffix));
163 if (strcmp(cn, " ")) {
164 // fprintf(stderr, "\n\n\n");
165 // fprintf(stderr, "access_method %s\n", item->contact->access_method);
166 // fprintf(stderr, "account_name %s\n", item->contact->account_name);
167 // fprintf(stderr, "address1 %s\n", item->contact->address1);
168 // fprintf(stderr, "address1_desc %s\n", item->contact->address1_desc);
169 // fprintf(stderr, "address1_transport %s\n", item->contact->address1_transport);
170 // fprintf(stderr, "address2 %s\n", item->contact->address2);
171 // fprintf(stderr, "address2_desc %s\n", item->contact->address2_desc);
172 // fprintf(stderr, "address2_transport %s\n", item->contact->address2_transport);
173 // fprintf(stderr, "address3 %s\n", item->contact->address3);
174 // fprintf(stderr, "address3_desc %s\n", item->contact->address3_desc);
175 // fprintf(stderr, "address3_transport %s\n", item->contact->address3_transport);
176 // fprintf(stderr, "assistant_name %s\n", item->contact->assistant_name);
177 // fprintf(stderr, "assistant_phone %s\n", item->contact->assistant_phone);
178 // fprintf(stderr, "billing_information %s\n", item->contact->billing_information);
179 // fprintf(stderr, "business_address %s\n", item->contact->business_address);
180 // fprintf(stderr, "business_city %s\n", item->contact->business_city);
181 // fprintf(stderr, "business_country %s\n", item->contact->business_country);
182 // fprintf(stderr, "business_fax %s\n", item->contact->business_fax);
183 // fprintf(stderr, "business_homepage %s\n", item->contact->business_homepage);
184 // fprintf(stderr, "business_phone %s\n", item->contact->business_phone);
185 // fprintf(stderr, "business_phone2 %s\n", item->contact->business_phone2);
186 // fprintf(stderr, "business_po_box %s\n", item->contact->business_po_box);
187 // fprintf(stderr, "business_postal_code %s\n", item->contact->business_postal_code);
188 // fprintf(stderr, "business_state %s\n", item->contact->business_state);
189 // fprintf(stderr, "business_street %s\n", item->contact->business_street);
190 // fprintf(stderr, "callback_phone %s\n", item->contact->callback_phone);
191 // fprintf(stderr, "car_phone %s\n", item->contact->car_phone);
192 // fprintf(stderr, "company_main_phone %s\n", item->contact->company_main_phone);
193 // fprintf(stderr, "company_name %s\n", item->contact->company_name);
194 // fprintf(stderr, "computer_name %s\n", item->contact->computer_name);
195 // fprintf(stderr, "customer_id %s\n", item->contact->customer_id);
196 // fprintf(stderr, "def_postal_address %s\n", item->contact->def_postal_address);
197 // fprintf(stderr, "department %s\n", item->contact->department);
198 // fprintf(stderr, "display_name_prefix %s\n", item->contact->display_name_prefix);
199 // fprintf(stderr, "first_name %s\n", item->contact->first_name);
200 // fprintf(stderr, "followup %s\n", item->contact->followup);
201 // fprintf(stderr, "free_busy_address %s\n", item->contact->free_busy_address);
202 // fprintf(stderr, "ftp_site %s\n", item->contact->ftp_site);
203 // fprintf(stderr, "fullname %s\n", item->contact->fullname);
204 // fprintf(stderr, "gov_id %s\n", item->contact->gov_id);
205 // fprintf(stderr, "hobbies %s\n", item->contact->hobbies);
206 // fprintf(stderr, "home_address %s\n", item->contact->home_address);
207 // fprintf(stderr, "home_city %s\n", item->contact->home_city);
208 // fprintf(stderr, "home_country %s\n", item->contact->home_country);
209 // fprintf(stderr, "home_fax %s\n", item->contact->home_fax);
210 // fprintf(stderr, "home_phone %s\n", item->contact->home_phone);
211 // fprintf(stderr, "home_phone2 %s\n", item->contact->home_phone2);
212 // fprintf(stderr, "home_po_box %s\n", item->contact->home_po_box);
213 // fprintf(stderr, "home_postal_code %s\n", item->contact->home_postal_code);
214 // fprintf(stderr, "home_state %s\n", item->contact->home_state);
215 // fprintf(stderr, "home_street %s\n", item->contact->home_street);
216 // fprintf(stderr, "initials %s\n", item->contact->initials);
217 // fprintf(stderr, "isdn_phone %s\n", item->contact->isdn_phone);
218 // fprintf(stderr, "job_title %s\n", item->contact->job_title);
219 // fprintf(stderr, "keyword %s\n", item->contact->keyword);
220 // fprintf(stderr, "language %s\n", item->contact->language);
221 // fprintf(stderr, "location %s\n", item->contact->location);
222 // fprintf(stderr, "manager_name %s\n", item->contact->manager_name);
223 // fprintf(stderr, "middle_name %s\n", item->contact->middle_name);
224 // fprintf(stderr, "mileage %s\n", item->contact->mileage);
225 // fprintf(stderr, "mobile_phone %s\n", item->contact->mobile_phone);
226 // fprintf(stderr, "nickname %s\n", item->contact->nickname);
227 // fprintf(stderr, "office_loc %s\n", item->contact->office_loc);
228 // fprintf(stderr, "org_id %s\n", item->contact->org_id);
229 // fprintf(stderr, "other_address %s\n", item->contact->other_address);
230 // fprintf(stderr, "other_city %s\n", item->contact->other_city);
231 // fprintf(stderr, "other_country %s\n", item->contact->other_country);
232 // fprintf(stderr, "other_phone %s\n", item->contact->other_phone);
233 // fprintf(stderr, "other_po_box %s\n", item->contact->other_po_box);
234 // fprintf(stderr, "other_postal_code %s\n", item->contact->other_postal_code);
235 // fprintf(stderr, "other_state %s\n", item->contact->other_state);
236 // fprintf(stderr, "other_street %s\n", item->contact->other_street);
237 // fprintf(stderr, "pager_phone %s\n", item->contact->pager_phone);
238 // fprintf(stderr, "personal_homepage %s\n", item->contact->personal_homepage);
239 // fprintf(stderr, "pref_name %s\n", item->contact->pref_name);
240 // fprintf(stderr, "primary_fax %s\n", item->contact->primary_fax);
241 // fprintf(stderr, "primary_phone %s\n", item->contact->primary_phone);
242 // fprintf(stderr, "profession %s\n", item->contact->profession);
243 // fprintf(stderr, "radio_phone %s\n", item->contact->radio_phone);
244 // fprintf(stderr, "spouse_name %s\n", item->contact->spouse_name);
245 // fprintf(stderr, "suffix %s\n", item->contact->suffix);
246 // fprintf(stderr, "surname %s\n", item->contact->surname);
247 // fprintf(stderr, "telex %s\n", item->contact->telex);
248 // fprintf(stderr, "transmittable_display_name %s\n", item->contact->transmittable_display_name);
249 // fprintf(stderr, "ttytdd_phone %s\n", item->contact->ttytdd_phone);
250 // have a valid cn
251 char *ucn = unique_string(folded(trim(cn)));
252 printf("dn: cn=%s, %s\n", ucn, ldap_base);
253 printf("cn: %s\n", ucn);
254 if (item->contact->first_name) {
255 snprintf(cn, sizeof(cn), "%s %s",
256 single(item->contact->display_name_prefix),
257 single(item->contact->first_name));
258 printf("givenName: %s\n", trim(cn));
259 }
260 if (item->contact->surname) {
261 snprintf(cn, sizeof(cn), "%s %s",
262 single(item->contact->surname),
263 single(item->contact->suffix));
264 printf("sn: %s\n", trim(cn));
265 }
266 else if (item->contact->company_name) {
267 printf("sn: %s\n", single(item->contact->company_name));
268 }
269 else
270 printf("sn: %s\n", ucn); // use cn as sn if we cannot find something better
271
272 if (item->contact->job_title)
273 printf("personalTitle: %s\n", single(item->contact->job_title));
274 if (item->contact->company_name)
275 printf("company: %s\n", single(item->contact->company_name));
276 if (item->contact->address1)
277 printf("mail: %s\n", single(item->contact->address1));
278 if (item->contact->address2)
279 printf("mail: %s\n", single(item->contact->address2));
280 if (item->contact->address3)
281 printf("mail: %s\n", single(item->contact->address3));
282 if (item->contact->business_address) {
283 if (item->contact->business_po_box)
284 printf("postalAddress: %s\n", single(item->contact->business_po_box));
285 if (item->contact->business_street)
286 multi("postalAddress: %s\n", item->contact->business_street);
287 if (item->contact->business_city)
288 printf("l: %s\n", single(item->contact->business_city));
289 if (item->contact->business_state)
290 printf("st: %s\n", single(item->contact->business_state));
291 if (item->contact->business_postal_code)
292 printf("postalCode: %s\n", single(item->contact->business_postal_code));
293 }
294 else if (item->contact->home_address) {
295 if (item->contact->home_po_box)
296 printf("postalAddress: %s\n", single(item->contact->home_po_box));
297 if (item->contact->home_street)
298 multi("postalAddress: %s\n", item->contact->home_street);
299 if (item->contact->home_city)
300 printf("l: %s\n", single(item->contact->home_city));
301 if (item->contact->home_state)
302 printf("st: %s\n", single(item->contact->home_state));
303 if (item->contact->home_postal_code)
304 printf("postalCode: %s\n", single(item->contact->home_postal_code));
305 }
306 else if (item->contact->other_address) {
307 if (item->contact->other_po_box)
308 printf("postalAddress: %s\n", single(item->contact->other_po_box));
309 if (item->contact->other_street)
310 multi("postalAddress: %s\n", item->contact->other_street);
311 if (item->contact->other_city)
312 printf("l: %s\n", single(item->contact->other_city));
313 if (item->contact->other_state)
314 printf("st: %s\n", single(item->contact->other_state));
315 if (item->contact->other_postal_code)
316 printf("postalCode: %s\n", single(item->contact->other_postal_code));
317 }
318 if (item->contact->business_fax)
319 printf("facsimileTelephoneNumber: %s\n", single(item->contact->business_fax));
320 else if (item->contact->home_fax)
321 printf("facsimileTelephoneNumber: %s\n", single(item->contact->home_fax));
322
323 if (item->contact->business_phone)
324 printf("telephoneNumber: %s\n", single(item->contact->business_phone));
325 if (item->contact->home_phone)
326 printf("homePhone: %s\n", single(item->contact->home_phone));
327
328 if (item->contact->car_phone)
329 printf("mobile: %s\n", single(item->contact->car_phone));
330 else if (item->contact->mobile_phone)
331 printf("mobile: %s\n", single(item->contact->mobile_phone));
332 else if (item->contact->other_phone)
333 printf("mobile: %s\n", single(item->contact->other_phone));
334
335
336 if (item->comment)
337 printf("description: %s\n", single(item->comment));
338
339 printf("objectClass: %s\n\n", ldap_class);
340 }
341 }
342 }
343 }
344 _pst_freeItem(item);
345 }
346 d_ptr = d_ptr->next;
347 }
348 }
349
350
351 int main(int argc, char** argv) {
352 pst_desc_ll *d_ptr;
353 char *fname = NULL;
354 char *temp = NULL; //temporary char pointer
355 char c;
356 prog_name = argv[0];
357 pst_item *item = NULL;
358
359 while ((c = getopt(argc, argv, "b:c:Vh"))!= -1) {
360 switch (c) {
361 case 'b':
362 ldap_base = optarg;
363 temp = strchr(ldap_base, ',');
364 if (temp) {
365 *temp = '\0';
366 ldap_org = strdup(ldap_base);
367 *temp = ',';
368 }
369 break;
370 case 'c':
371 ldap_class = optarg;
372 break;
373 case 'h':
374 usage();
375 exit(0);
376 break;
377 case 'V':
378 version();
379 exit(0);
380 break;
381 default:
382 usage();
383 exit(1);
384 break;
385 }
386 }
387
388 if ((argc > optind) && (ldap_base) && (ldap_class) && (ldap_org)) {
389 fname = argv[optind];
390 } else {
391 usage();
392 exit(2);
393 }
394
395 DEBUG_INIT("pst2ldif.log");
396 DEBUG_REGISTER_CLOSE();
397 DEBUG_ENT("main");
398 RET_DERROR(pst_open(&pstfile, fname, "r"), 1, ("Error opening File\n"));
399 RET_DERROR(pst_load_index(&pstfile), 2, ("Index Error\n"));
400
401 pst_load_extended_attributes(&pstfile);
402
403 d_ptr = pstfile.d_head; // first record is main record
404 item = (pst_item*)_pst_parse_item(&pstfile, d_ptr);
405 if (!item || !item->message_store) {
406 DIE(("main: Could not get root record\n"));
407 }
408
409 d_ptr = pst_getTopOfFolders(&pstfile, item);
410 if (!d_ptr) {
411 DIE(("Top of folders record not found. Cannot continue\n"));
412 }
413
414 _pst_freeItem(item);
415
416 // write the ldap header
417 printf("dn: %s\n", ldap_base);
418 printf("o: %s\n", ldap_org);
419 printf("objectClass: organization\n\n");
420 printf("dn: cn=root, %s\n", ldap_base);
421 printf("cn: root\n");
422 printf("objectClass: %s\n\n", ldap_class);
423
424 process(d_ptr->child); // do the children of TOPF
425 pst_close(&pstfile);
426 return 0;
427 }
428
429
430 int usage() {
431 version();
432 printf("Usage: %s [OPTIONS] {PST FILENAME}\n", prog_name);
433 printf("OPTIONS:\n");
434 printf("\t-h\t- Help. This screen\n");
435 printf("\t-V\t- Version. Display program version\n");
436 printf("\t-b ldapbase\t- set the ldap base value\n");
437 printf("\t-c class \t- set the class of the ldap objects\n");
438 return 0;
439 }
440
441
442 int version() {
443 printf("pst2ldif v%s using LibPST v%s\n", VERSION, PST_VERSION);
444 #if BYTE_ORDER == BIG_ENDIAN
445 printf("Big Endian implementation being used.\n");
446 #elif BYTE_ORDER == LITTLE_ENDIAN
447 printf("Little Endian implementation being used.\n");
448 #else
449 # error "Byte order not supported by this library"
450 #endif
451 #ifdef __GNUC__
452 printf("GCC %d.%d : %s %s\n", __GNUC__, __GNUC_MINOR__, __DATE__, __TIME__);
453 #endif
454 return 0;
455 }
456
457
458 // my_stristr varies from strstr in that its searches are case-insensitive
459 char * my_stristr(char *haystack, char *needle) {
460 char *x=haystack, *y=needle, *z = NULL;
461 if (haystack == NULL || needle == NULL)
462 return NULL;
463 while (*y != '\0' && *x != '\0') {
464 if (tolower(*y) == tolower(*x)) {
465 // move y on one
466 y++;
467 if (z == NULL) {
468 z = x; // store first position in haystack where a match is made
469 }
470 } else {
471 y = needle; // reset y to the beginning of the needle
472 z = NULL; // reset the haystack storage point
473 }
474 x++; // advance the search in the haystack
475 }
476 return z;
477 }
478
479
480 char *check_filename(char *fname) {
481 char *t = fname;
482 if (t == NULL) {
483 return fname;
484 }
485 while ((t = strpbrk(t, "/\\:"))) {
486 // while there are characters in the second string that we don't want
487 *t = '_'; //replace them with an underscore
488 }
489 return fname;
490 }
491
492
493 char *single(char *str) {
494 if (!str) return "";
495 char *ret = rfc2426_escape(str);
496 char *n = strchr(ret, '\n');
497 if (n) *n = '\0';
498 return ret;
499 }
500
501
502 char *folded(char *str) {
503 if (!str) return "";
504 char *ret = rfc2426_escape(str);
505 char *n = ret;
506 while (n = strchr(n, '\n')) {
507 *n = ' ';
508 }
509 n = ret;
510 while (n = strchr(n, ',')) {
511 *n = ' ';
512 }
513 return ret;
514 }
515
516
517 void multi(char *fmt, char *str) {
518 if (!str) return;
519 char *ret = rfc2426_escape(str);
520 char *n = ret;
521 while (n = strchr(ret, '\n')) {
522 *n = '\0';
523 printf(fmt, ret);
524 ret = n+1;
525 }
526 if (*ret) printf(fmt, ret);
527 }
528
529
530 char *rfc2426_escape(char *str) {
531 static char* buf = NULL;
532 char *ret, *a, *b;
533 int x = 0, y, z;
534 if (str == NULL)
535 ret = str;
536 else {
537
538 // calculate space required to escape all the following characters
539 x = strlen(str) +(y=(chr_count(str, ',')*2) + (chr_count(str, '\\')*2) + (chr_count(str, ';')*2) + (chr_count(str, '\n')*2));
540 z = chr_count(str, '\r');
541 if (y == 0 && z == 0)
542 // there isn't any extra space required
543 ret = str;
544 else {
545 buf = (char*) realloc(buf, x+1);
546 a = str;
547 b = buf;
548 while (*a != '\0') {
549 switch(*a) {
550 // case ',' :
551 case '\\':
552 case ';' :
553 // case '\n':
554 *(b++)='\\';
555 *b=*a;
556 break;
557 case '\r':
558 break;
559 default:
560 *b=*a;
561 }
562 b++;
563 a++;
564 }
565 *b = '\0';
566 ret = buf;
567 }
568 }
569 return ret;
570 }
571
572
573 int chr_count(char *str, char x) {
574 int r = 0;
575 while (*str != '\0') {
576 if (*str == x)
577 r++;
578 str++;
579 }
580 return r;
581 }
582