X-Git-Url: http://pileus.org/git/?a=blobdiff_plain;f=uid.c;h=8a775b9c63f48b60997f834f2d0f4f86f92bbe20;hb=e4dd196b137223195739b9e0f50ec2a8a02b3534;hp=2f388a23423ebcc659e821308b69431f52e99965;hpb=582012e400e28648a602804eeea6d701f2295306;p=~andy%2Ffetchmail diff --git a/uid.c b/uid.c index 2f388a23..8a775b9c 100644 --- a/uid.c +++ b/uid.c @@ -1,5 +1,5 @@ -/* - * uid.c -- UIDL handling for POP3 servers without LAST +/** + * \file uid.c -- UIDL handling for POP3 servers without LAST * * For license terms, see the file COPYING in this directory. */ @@ -19,11 +19,13 @@ #endif #include "fetchmail.h" +#include "i18n.h" +#include "sdump.h" /* * Machinery for handling UID lists live here. This is mainly to support - * RFC1725-conformant POP3 servers without a LAST command, but may also be - * useful for making the IMAP4 querying logic UID-oriented, if a future + * RFC1725/RFC1939-conformant POP3 servers without a LAST command, but may also + * be useful for making the IMAP4 querying logic UID-oriented, if a future * revision of IMAP forces me to. * * These functions are also used by the rest of the code to maintain @@ -38,7 +40,7 @@ * This list is initially set up by initialize_saved_list() from the * .fetchids file. * - * Early in the query, during the execution of the protocol-specific + * Early in the query, during the execution of the protocol-specific * getrange code, the driver expects that the host's `newsaved' member * will be filled with a list of UIDs and message numbers representing * the mailbox state. If this list is empty, the server did @@ -62,23 +64,63 @@ * be picked up by the next run. If there are no un-expunged * messages, the file is deleted. * - * Note: some comparisons (those used for DNS address lists) are caseblind! + * One disadvantage of UIDL is that all the UIDs have to be downloaded + * before a search for new messages can be done. Typically, new messages + * are appended to mailboxes. Hence, downloading all UIDs just to download + * a few new mails is a waste of bandwidth. If new messages are always at + * the end of the mailbox, fast UIDL will decrease the time required to + * download new mails. + * + * During fast UIDL, the UIDs of all messages are not downloaded! The first + * unseen message is searched for by using a binary search on UIDs. UIDs + * after the first unseen message are downloaded as and when needed. + * + * The advantages of fast UIDL are (this is noticeable only when the + * mailbox has too many mails): + * + * - There is no need to download the UIDs of all mails right at the start. + * - There is no need to save all the UIDs in memory separately in + * `newsaved' list. + * - There is no need to download the UIDs of seen mail (except for the + * first binary search). + * - The first new mail is downloaded considerably faster. + * + * The disadvantages are: + * + * - Since all UIDs are not downloaded, it is not possible to swap old and + * new list. The current state of the mailbox is essentially a merged state + * of old and new mails. + * - If an intermediate mail has been temporarily refused (say, due to 4xx + * code from the smtp server), this mail may not get downloaded. + * - If 'flush' is used, such intermediate mails will also get deleted. + * + * The first two disadvantages can be overcome by doing a linear search + * once in a while (say, every 10th poll). Also, with flush, fast UIDL + * should be disabled. + * + * Note: some comparisons (those used for DNS address lists) are caseblind! */ -/* UIDs associated with un-queried hosts */ -static struct idlist *scratchlist; +int dofastuidl = 0; #ifdef POP3_ENABLE +/** UIDs associated with un-queried hosts */ +static struct idlist *scratchlist; + +/** Read saved IDs from \a idfile and attach to each host in \a hostlist. */ void initialize_saved_lists(struct query *hostlist, const char *idfile) -/* read file of saved IDs and attach to each host */ { struct stat statbuf; FILE *tmpfp; struct query *ctl; /* make sure lists are initially empty */ - for (ctl = hostlist; ctl; ctl = ctl->next) - ctl->skipped = ctl->oldsaved = ctl->newsaved = (struct idlist *)NULL; + for (ctl = hostlist; ctl; ctl = ctl->next) { + ctl->skipped = (struct idlist *)NULL; + ctl->oldsaved = (struct idlist *)NULL; + ctl->newsaved = (struct idlist *)NULL; + ctl->oldsavedend = &ctl->oldsaved; + } errno = 0; @@ -90,13 +132,13 @@ void initialize_saved_lists(struct query *hostlist, const char *idfile) * that all implementations of lstat() will return ENOTDIR * rather than plain ENOENT in this case... */ - if (lstat(idfile, &statbuf) < 0) { - if (errno == ENOTDIR) - { - report(stderr, "lstat: %s: %s\n", idfile, strerror(errno)); - exit(PS_IOERR); + if (lstat(idfile, &statbuf) < 0) { + if (errno == ENOTDIR) + { + report(stderr, "lstat: %s: %s\n", idfile, strerror(errno)); + exit(PS_IOERR); + } } - } /* let's get stored message UIDs from previous queries */ if ((tmpfp = fopen(idfile, "r")) != (FILE *)NULL) @@ -120,7 +162,7 @@ void initialize_saved_lists(struct query *hostlist, const char *idfile) * the rightmost '@'. This is not correct, as InterMail puts an * '@' in the UIDL. */ - + /* first, skip leading spaces */ user = buf + strspn(buf, " \t"); @@ -130,8 +172,18 @@ void initialize_saved_lists(struct query *hostlist, const char *idfile) * espescially if the POP server returns an X-UIDL header * instead of a Message-ID, as GMX's (www.gmx.net) POP3 * StreamProxy V1.0 does. + * + * this is one other trick. The userhost part + * may contain ' ' in the user part, at least in + * the lotus notes case. + * So we start looking for the '@' after which the + * host will follow with the ' ' separator with the id. + * + * XXX FIXME: There is a case this code cannot handle: + * the user name cannot have blanks after a '@'. */ - if ((id = strchr(user, ' ')) != NULL ) + if ((delimp1 = strchr(user, '@')) != NULL && + (id = strchr(delimp1,' ')) != NULL) { for (delimp1 = id; delimp1 >= user; delimp1--) if ((*delimp1 != ' ') && (*delimp1 != '\t')) @@ -145,41 +197,43 @@ void initialize_saved_lists(struct query *hostlist, const char *idfile) id = id + strspn(id, " "); delimp1++; /* but what if there is only white space ?!? */ - saveddelim1 = *delimp1; /* save char after token */ + /* we have at least one @, else we are not in this branch */ + saveddelim1 = *delimp1; /* save char after token */ *delimp1 = '\0'; /* delimit token with \0 */ - if (id != NULL) - { - /* now remove trailing white space chars from id */ - if ((delimp2 = strpbrk(id, " \t\n")) != NULL ) { - saveddelim2 = *delimp2; - *delimp2 = '\0'; - } - atsign = strrchr(user, '@'); - if (atsign) { - *atsign = '\0'; - host = atsign + 1; - } - for (ctl = hostlist; ctl; ctl = ctl->next) { - if (ctl->server.truename && - strcasecmp(host, ctl->server.truename) == 0 + /* now remove trailing white space chars from id */ + if ((delimp2 = strpbrk(id, " \t\n")) != NULL ) { + saveddelim2 = *delimp2; + *delimp2 = '\0'; + } + + atsign = strrchr(user, '@'); + /* we have at least one @, else we are not in this branch */ + *atsign = '\0'; + host = atsign + 1; + + /* find proper list and save it */ + for (ctl = hostlist; ctl; ctl = ctl->next) { + if (strcasecmp(host, ctl->server.queryname) == 0 && strcasecmp(user, ctl->remotename) == 0) { - - save_str(&ctl->oldsaved, id, UID_SEEN); - break; - } + save_str(&ctl->oldsaved, id, UID_SEEN); + break; } - /* if it's not in a host we're querying, - ** save it anyway */ - if (ctl == (struct query *)NULL) { - /* restore string */ - *delimp1 = saveddelim1; - *atsign = '@'; - if (delimp2 != NULL) { - *delimp2 = saveddelim2; - } - save_str(&scratchlist, buf, UID_SEEN); + } + /* + * If it's not in a host we're querying, + * save it anyway. Otherwise we'd lose UIDL + * information any time we queried an explicit + * subset of hosts. + */ + if (ctl == (struct query *)NULL) { + /* restore string */ + *delimp1 = saveddelim1; + *atsign = '@'; + if (delimp2 != NULL) { + *delimp2 = saveddelim2; } + save_str(&scratchlist, buf, UID_SEEN); } } } @@ -189,238 +243,91 @@ void initialize_saved_lists(struct query *hostlist, const char *idfile) if (outlevel >= O_DEBUG) { struct idlist *idp; - int uidlcount = 0; for (ctl = hostlist; ctl; ctl = ctl->next) - if (ctl->server.uidl) { - report_build(stdout, "Old UID list from %s:",ctl->server.pollname); - for (idp = ctl->oldsaved; idp; idp = idp->next) - report_build(stdout, " %s", idp->id); + report_build(stdout, GT_("Old UID list from %s:"), + ctl->server.pollname); + idp = ctl->oldsaved; if (!idp) - report_build(stdout, " "); + report_build(stdout, GT_(" ")); + else for (idp = ctl->oldsaved; idp; idp = idp->next) { + char *t = sdump(idp->id, strlen(idp->id)-1); + report_build(stdout, " %s\n", t); + free(t); + } report_complete(stdout, "\n"); - uidlcount++; } - if (uidlcount) - { - report_build(stdout, "Scratch list of UIDs:"); - for (idp = scratchlist; idp; idp = idp->next) - report_build(stdout, " %s", idp->id); - if (!idp) - report_build(stdout, " "); - report_complete(stdout, "\n"); + report_build(stdout, GT_("Scratch list of UIDs:")); + if (!scratchlist) + report_build(stdout, GT_(" ")); + else for (idp = scratchlist; idp; idp = idp->next) { + char *t = sdump(idp->id, strlen(idp->id)-1); + report_build(stdout, " %s\n", t); + free(t); } + report_complete(stdout, "\n"); } } -#endif /* POP3_ENABLE */ - -struct idlist *save_str(struct idlist **idl, const char *str, flag status) -/* save a number/UID pair on the given UID list */ -{ - struct idlist **end; - - /* do it nonrecursively so the list is in the right order */ - for (end = idl; *end; end = &(*end)->next) - continue; - - *end = (struct idlist *)xmalloc(sizeof(struct idlist)); - (*end)->val.status.mark = status; - (*end)->id = str ? xstrdup(str) : (char *)NULL; - (*end)->next = NULL; - - return(*end); -} - -void free_str_list(struct idlist **idl) -/* free the given UID list */ -{ - if (*idl == (struct idlist *)NULL) - return; - - free_str_list(&(*idl)->next); - free ((*idl)->id); - free(*idl); - *idl = (struct idlist *)NULL; -} - -void save_str_pair(struct idlist **idl, const char *str1, const char *str2) -/* save an ID pair on the given list */ -{ - struct idlist **end; - - /* do it nonrecursively so the list is in the right order */ - for (end = idl; *end; end = &(*end)->next) - continue; - - *end = (struct idlist *)xmalloc(sizeof(struct idlist)); - (*end)->id = str1 ? xstrdup(str1) : (char *)NULL; - if (str2) - (*end)->val.id2 = xstrdup(str2); - else - (*end)->val.id2 = (char *)NULL; - (*end)->next = (struct idlist *)NULL; -} - -#ifdef __UNUSED__ -void free_str_pair_list(struct idlist **idl) -/* free the given ID pair list */ -{ - if (*idl == (struct idlist *)NULL) - return; - - free_idpair_list(&(*idl)->next); - free ((*idl)->id); - free ((*idl)->val.id2); - free(*idl); - *idl = (struct idlist *)NULL; -} -#endif - -int str_in_list(struct idlist **idl, const char *str, const flag caseblind) -/* is a given ID in the given list? (comparison may be caseblind) */ -{ - if (*idl == (struct idlist *)NULL || str == (char *) NULL) - return(0); - else if (!caseblind && strcmp(str, (*idl)->id) == 0) - return(1); - else if (caseblind && strcasecmp(str, (*idl)->id) == 0) - return(1); - else - return(str_in_list(&(*idl)->next, str, caseblind)); -} - -int str_nr_in_list( struct idlist **idl, const char *str ) - /* return the position of str in idl */ -{ - int nr; - struct idlist *walk; - if ( !str ) - return -1; - for( walk = *idl, nr = 0; walk; nr ++, walk = walk->next ) - if( strcmp( str, walk->id) == 0 ) - return nr; - return -1; -} - -int str_nr_last_in_list( struct idlist **idl, const char *str) -/* return the last position of str in idl */ -{ - int nr, ret = -1; - struct idlist *walk; - if ( !str ) - return -1; - for( walk = *idl, nr = 0; walk; nr ++, walk = walk->next ) - if( strcmp( str, walk->id) == 0 ) - ret = nr; - return ret; -} - -void str_set_mark( struct idlist **idl, const char *str, const flag val) -/* update the mark on an of an id to given value */ -{ - int nr; - struct idlist *walk; - if (!str) - return; - for(walk = *idl, nr = 0; walk; nr ++, walk = walk->next) - if (strcmp(str, walk->id) == 0) - walk->val.status.mark = val; -} - -int count_list( struct idlist **idl) -/* count the number of elements in the list */ -{ - if( !*idl ) - return 0; - return 1 + count_list( &(*idl)->next ); -} -char *str_from_nr_list(struct idlist **idl, int number) -/* return the number'th string in idl */ -{ - if( !*idl || number < 0) - return 0; - if( number == 0 ) - return (*idl)->id; - return str_from_nr_list(&(*idl)->next, number-1); -} - - -char *str_find(struct idlist **idl, int number) -/* return the id of the given number in the given list. */ +/** Assert that all UIDs marked deleted in query \a ctl have actually been +expunged. */ +void expunge_uids(struct query *ctl) { - if (*idl == (struct idlist *) 0) - return((char *) 0); - else if (number == (*idl)->val.status.num) - return((*idl)->id); - else - return(str_find(&(*idl)->next, number)); -} + struct idlist *idl; -char *idpair_find(struct idlist **idl, const char *id) -/* return the id of the given id in the given list (caseblind comparison) */ -{ - if (*idl == (struct idlist *) 0) - return((char *) 0); - else if (strcasecmp(id, (*idl)->id) == 0) - return((*idl)->val.id2 ? (*idl)->val.id2 : (*idl)->id); - else - return(idpair_find(&(*idl)->next, id)); + for (idl = dofastuidl ? ctl->oldsaved : ctl->newsaved; idl; idl = idl->next) + if (idl->val.status.mark == UID_DELETED) + idl->val.status.mark = UID_EXPUNGED; } -int delete_str(struct idlist **idl, int num) -/* delete given message from given list */ +static const char *str_uidmark(int mark) { - struct idlist *idp; - - for (idp = *idl; idp; idp = idp->next) - if (idp->val.status.num == num) - { - idp->val.status.mark = UID_DELETED; - return(1); + static char buf[20]; + + switch(mark) { + case UID_UNSEEN: + return "UNSEEN"; + case UID_SEEN: + return "SEEN"; + case UID_EXPUNGED: + return "EXPUNGED"; + case UID_DELETED: + return "DELETED"; + default: + if (snprintf(buf, sizeof(buf), "MARK=%d", mark) < 0) + return "ERROR"; + else + return buf; } - return(0); -} - -void append_str_list(struct idlist **idl, struct idlist **nidl) -/* append nidl to idl (does not copy *) */ -{ - if ((*nidl) == (struct idlist *)NULL || *nidl == *idl) - return; - else if ((*idl) == (struct idlist *)NULL) - *idl = *nidl; - else if ((*idl)->next == (struct idlist *)NULL) - (*idl)->next = *nidl; - else if ((*idl)->next != *nidl) - append_str_list(&(*idl)->next, nidl); } -#ifdef POP3_ENABLE -void expunge_uids(struct query *ctl) -/* assert that all UIDs marked deleted have actually been expunged */ +static void dump_list(const struct idlist *idp) { - struct idlist *idl; - - for (idl = ctl->newsaved; idl; idl = idl->next) - if (idl->val.status.mark == UID_DELETED) - idl->val.status.mark = UID_EXPUNGED; + if (!idp) { + report_build(stdout, GT_(" ")); + } else while (idp) { + char *t = sdump(idp->id, strlen(idp->id)); + report_build(stdout, " %s = %s%s", t, str_uidmark(idp->val.status.mark), idp->next ? "," : ""); + free(t); + idp = idp->next; + } } -void uid_end_query(struct query *ctl) /* finish a query */ +void uid_swap_lists(struct query *ctl) { /* debugging code */ - if (ctl->server.uidl && outlevel >= O_DEBUG) + if (outlevel >= O_DEBUG) { - struct idlist *idp; - - report_build(stdout, "New UID list from %s:", ctl->server.pollname); - for (idp = ctl->newsaved; idp; idp = idp->next) - report_build(stdout, " %s = %d", idp->id, idp->val.status.mark); - if (!idp) - report_build(stdout, " "); + if (dofastuidl) { + report_build(stdout, GT_("Merged UID list from %s:"), ctl->server.pollname); + dump_list(ctl->oldsaved); + } else { + report_build(stdout, GT_("New UID list from %s:"), ctl->server.pollname); + dump_list(ctl->newsaved); + } report_complete(stdout, "\n"); } @@ -428,26 +335,69 @@ void uid_end_query(struct query *ctl) * Don't swap UID lists unless we've actually seen UIDLs. * This is necessary in order to keep UIDL information * from being heedlessly deleted later on. + * + * Older versions of fetchmail did + * + * free_str_list(&scratchlist); + * + * after swap. This was wrong; we need to preserve the UIDL information + * from unqueried hosts. Unfortunately, not doing this means that + * under some circumstances UIDLs can end up being stored forever -- + * specifically, if a user description is removed from .fetchmailrc + * with UIDLs from that account in .fetchids, there is no way for + * them to ever get garbage-collected. */ - if (ctl->have_uids) + if (ctl->newsaved) { /* old state of mailbox may now be irrelevant */ + struct idlist *temp = ctl->oldsaved; if (outlevel >= O_DEBUG) - report(stdout, "swapping UID lists\n"); - free_str_list(&ctl->oldsaved); - free_str_list(&scratchlist); + report(stdout, GT_("swapping UID lists\n")); ctl->oldsaved = ctl->newsaved; ctl->newsaved = (struct idlist *) NULL; - ctl->have_uids = FALSE; + free_str_list(&temp); } - else if (outlevel >= O_DEBUG) - report(stdout, "not swapping UID lists, no UIDs seen this query\n"); + /* in fast uidl, there is no need to swap lists: the old state of + * mailbox cannot be discarded! */ + else if (outlevel >= O_DEBUG && !dofastuidl) + report(stdout, GT_("not swapping UID lists, no UIDs seen this query\n")); } +/* finish a query which had errors */ +void uid_discard_new_list(struct query *ctl) +{ + /* debugging code */ + if (outlevel >= O_DEBUG) + { + /* this is now a merged list! the mails which were seen in this + * poll are marked here. */ + report_build(stdout, GT_("Merged UID list from %s:"), ctl->server.pollname); + dump_list(ctl->oldsaved); + report_complete(stdout, "\n"); + } + + if (ctl->newsaved) + { + /* new state of mailbox is not reliable */ + if (outlevel >= O_DEBUG) + report(stdout, GT_("discarding new UID list\n")); + free_str_list(&ctl->newsaved); + ctl->newsaved = (struct idlist *) NULL; + } +} + +/** Reset the number associated with each id */ +void uid_reset_num(struct query *ctl) +{ + struct idlist *idp; + for (idp = ctl->oldsaved; idp; idp = idp->next) + idp->val.status.num = 0; +} + +/** Write list of seen messages, at end of run. */ void write_saved_lists(struct query *hostlist, const char *idfile) -/* perform end-of-run write of seen-messages list */ { - int idcount; + long idcount; FILE *tmpfp; struct query *ctl; struct idlist *idp; @@ -455,33 +405,68 @@ void write_saved_lists(struct query *hostlist, const char *idfile) /* if all lists are empty, nuke the file */ idcount = 0; for (ctl = hostlist; ctl; ctl = ctl->next) { - if (ctl->oldsaved) - idcount++; + for (idp = ctl->oldsaved; idp; idp = idp->next) + if (idp->val.status.mark == UID_SEEN + || idp->val.status.mark == UID_DELETED) + idcount++; } /* either nuke the file or write updated last-seen IDs */ if (!idcount && !scratchlist) { + if (outlevel >= O_DEBUG) { + if (access(idfile, F_OK) == 0) + report(stdout, GT_("Deleting fetchids file.\n")); + } + if (unlink(idfile) && errno != ENOENT) + report(stderr, GT_("Error deleting %s: %s\n"), idfile, strerror(errno)); + } else { + char *newnam = (char *)xmalloc(strlen(idfile) + 2); + strcpy(newnam, idfile); + strcat(newnam, "_"); if (outlevel >= O_DEBUG) - report(stdout, "Deleting fetchids file.\n"); - unlink(idfile); - } - else - { - if (outlevel >= O_DEBUG) - report(stdout, "Writing fetchids file.\n"); - if ((tmpfp = fopen(idfile, "w")) != (FILE *)NULL) { + report(stdout, GT_("Writing fetchids file.\n")); + (void)unlink(newnam); /* remove file/link first */ + if ((tmpfp = fopen(newnam, "w")) != (FILE *)NULL) { + int errflg = 0; for (ctl = hostlist; ctl; ctl = ctl->next) { for (idp = ctl->oldsaved; idp; idp = idp->next) if (idp->val.status.mark == UID_SEEN || idp->val.status.mark == UID_DELETED) - fprintf(tmpfp, "%s@%s %s\n", - ctl->remotename, ctl->server.truename, idp->id); + if (fprintf(tmpfp, "%s@%s %s\n", + ctl->remotename, ctl->server.queryname, idp->id) < 0) { + int e = errno; + report(stderr, GT_("Write error on fetchids file %s: %s\n"), newnam, strerror(e)); + errflg = 1; + goto bailout; + } } for (idp = scratchlist; idp; idp = idp->next) - fputs(idp->id, tmpfp); + if (EOF == fputs(idp->id, tmpfp)) { + int e = errno; + report(stderr, GT_("Write error on fetchids file %s: %s\n"), newnam, strerror(e)); + errflg = 1; + goto bailout; + } + +bailout: + (void)fflush(tmpfp); /* return code ignored, we check ferror instead */ + errflg |= ferror(tmpfp); fclose(tmpfp); + /* if we could write successfully, move into place; + * otherwise, drop */ + if (errflg) { + report(stderr, GT_("Error writing to fetchids file %s, old file left in place.\n"), newnam); + unlink(newnam); + } else { + if (rename(newnam, idfile)) { + report(stderr, GT_("Cannot rename fetchids file %s to %s: %s\n"), newnam, idfile, strerror(errno)); + } + } + } else { + report(stderr, GT_("Cannot open fetchids file %s for writing: %s\n"), newnam, strerror(errno)); } + free(newnam); } } #endif /* POP3_ENABLE */