/* ** Subsystem: USENET Sources Archiver ** File Name: rkive.c ** ** usage: rkive [-?dgstuvV] [-f config_file] [-n newsgroup] ** [-B batchfile] [-S newsgroup] ** ** ** This software is Copyright (c) 1989, 1990, 1991 by Kent Landfield. ** ** Permission is hereby granted to copy, distribute or otherwise ** use any part of this package as long as you do not try to make ** money from it or pretend that you wrote it. This copyright ** notice must be maintained in any copy made. ** ** Use of this software constitutes acceptance for use in an AS IS ** condition. There are NO warranties with regard to this software. ** In no event shall the author be liable for any damages whatsoever ** arising out of or in connection with the use or performance of this ** software. Any use of this software is at the user's own risk. ** ** If you make modifications to this software that you feel ** increases it usefulness for the rest of the community, please ** email the changes, enhancements, bug fixes as well as any and ** all ideas to me. This software is going to be maintained and ** enhanced as deemed necessary by the community. ** ** Kent Landfield ** uunet!sparky!kent ** kent@sparky.imd.sterling.com */ #include #include #include #include "article.h" #include "cfg.h" #ifdef NNTP char sccsid[] = "@(#)rkive.c 2.3 5/9/91 - NNTP Version"; #else char sccsid[] = "@(#)rkive.c 2.3 5/9/91"; #endif /*!NNTP*/ /* ** This is necessary since the builtin makedir call uses ** mknod which is a superuser only call for directories. */ #ifndef MKDIR # ifndef USE_SYSMKDIR # define ROOT_ONLY # endif #endif char tmp_mailfile[] = "/usr/tmp/rkive.mail"; char global_mailfile[] = "/usr/tmp/gbl.mail"; char *batch_file = NULL; char article_name[MAXNAMLEN]; int retrieve = FROM_DISK; /* default archiving is done from local disk */ int overwrite; int status_only; struct stat sbuf; FILE *inputfp; int needs_to_be_archived(); char *save_article(); char *compress_file(); char *do_compress(); char *basename(); char *suffix(); char *format_output(); void archive(); void build_index(); void display_group_info(); void logit(); void log_activities(); void mail_file(); void notify_users(); void record_problem(); void set_ownership(); int fclose(); int stat(); int strcmp(); int strlen(); int system(); int unlink(); char *strcpy(); char *strcat(); char *strchr(); FILE *efopen(); void exit(); extern int yydebug; extern char inputstring[]; extern parser_return_value; extern int debug; extern int verbose; extern int test; extern int inum; extern int problem_article; extern char *nntp_server; extern int num_newsgroups; void usage() { (void)fprintf(stderr,"usage: %s [-?dgstuvV] [-f config_file]\n", progname); (void)fprintf(stderr," [-n newsgroup] [-B batchfile] [-S newsgroup]\n"); (void)fprintf(stderr,"options:\n"); (void)fprintf(stderr," -? Display this message.\n"); (void)fprintf(stderr," -d Debug, assumes verbose flag as well.\n"); (void)fprintf(stderr," -g Fill in default values with group display.\n"); (void)fprintf(stderr," Only effective with verbose mode flag.\n"); (void)fprintf(stderr," -s Display status of article archiving.\n"); (void)fprintf(stderr," -t Test mode, display what would be done but do no archiving.\n"); (void)fprintf(stderr," -u Unconditionally overwrite dupliclate articles.\n"); (void)fprintf(stderr," -v Verbose archiving information printed.\n"); (void)fprintf(stderr," -V Display %s software version information.\n", progname); (void)fprintf(stderr," -f config_file\n"); (void)fprintf(stderr," Specify alternate configuration file to be used.\n"); (void)fprintf(stderr," -n newsgroup\n"); (void)fprintf(stderr," Specify newsgroup to archive or display status for.\n"); (void)fprintf(stderr," -A newsgroup\n"); (void)fprintf(stderr," Take an article to archive from stdin.\n"); (void)fprintf(stderr," -B batchfile\n"); (void)fprintf(stderr," Read names of articles to archive from batchfile.\n"); (void)fprintf(stderr," Note: Use of the -B option requires a newsgroup\n"); (void)fprintf(stderr," be specified using the -n option.\n"); (void)fprintf(stderr," -S newsgroup\n"); (void)fprintf(stderr," Take the name of articles to archive from stdin.\n"); } int main(argc, argv) int argc; char **argv; { int getopt(); void version(); void setup_defaults(); void init_article(); int c; extern char *optarg; extern int opterr; char *nwsg = NULL; yydebug = FALSE; opterr = 0; progname = argv[0]; inputfp = stdin; logfp = stdout; errfp = stderr; status_only = debug = verbose = 0; test = fill_in_defaults = 0; overwrite = 1; /* ** Setup the default config file to be used ** unless the user specifies otherwise. */ config_file = LOCATION; if (argc > 1) { while ((c = getopt(argc, argv, "?dgstuvVn:f:A:B:S:y")) != EOF) { switch (c) { case 'A': /* stdin article archiving */ retrieve = FROM_STDIN; nwsg = optarg; break; case 'B': /* take filenames from batch file */ retrieve = FROM_BATCHFILE; batch_file = optarg; break; case 'd': /* for extremely verbose output */ debug++; verbose++; break; case 'f': /* alternate configuration file */ config_file = optarg; break; case 'g': /* display defaults filled in */ fill_in_defaults++; break; case 'n': /* specify an individual newsgroup */ nwsg = optarg; break; case 's': /* display article archive status */ status_only++; break; case 'S': /* single article archiving */ retrieve = FROM_NAME; nwsg = optarg; break; case 't': /* test mode, show but do nothing */ test++; verbose++; break; case 'u': /* overwrite existing duplicates */ overwrite++; break; case 'v': /* display expanded output */ verbose++; break; case 'V': /* display software version only */ version(); break; case 'y': /* turn on yydebug facilities */ yydebug++; break; case '?': /* display usage information */ default: /* display usage, invalid option */ usage(); return(1); } } } /* ** If the user has specified that input is be filenames ** retrieved from a batch file, the user must specify a ** a newsgroup in addition to the batch filename. */ if (retrieve == FROM_BATCHFILE) { if (nwsg == NULL) { (void) fprintf(errfp, "%s: Must specify a newsgroup when using batchfile mode\n", progname); (void) fprintf(errfp,"Sample command line...\n"); (void) fprintf(errfp,"\t%s -B %s -n newsgroup-here\n", progname, batch_file); return(1); } } /* ** If the user has specified that the article is to be read from stdin, ** the user must specify a a newsgroup as well. A request for status ** on an article from stdin makes little sense to me... If I am wrong ** let me know... */ else if (retrieve == FROM_STDIN) { if (nwsg == NULL) { (void) fprintf(errfp, "%s: Must specify a newsgroup\n",progname); (void) fprintf(errfp,"Sample command line...\n"); (void) fprintf(errfp,"\t%s -A newsgroup-here\n",progname); return(1); } if (status_only != 0) { (void) fprintf(errfp,"%s: can't get status for an article on stdin\n", progname); (void) fprintf(errfp,"Sample command line...\n"); (void) fprintf(errfp,"\t%s -A newsgroup-here\n",progname); return(1); } } setup_defaults(); init_article(); for (c = 0; c <= num; c++) { newsgrp = &group[c]; /* ** Was a newsgroup specified on the command line ? */ if (nwsg != NULL) { if (strcmp(nwsg, newsgrp->ng_name) != 0) continue; } archive(); } if (!status_only) { /* ** Mail notification of the archived members to the ** list of users specified in the configuration file ** and remove the file containing the archived info. */ mail_file(mail, global_mailfile, "Rkive results"); (void) unlink(global_mailfile); } return(0); } void archive() { char *get_archived_rec(); int retrieve_article(); void get_header(); int cct; int first; int val; char *rp, *rec; char *new_member; char *archived_file; #ifdef NNTP extern int nntp_retrieve_article(); #endif #ifdef ROOT_ONLY /* ** check to assure that the user is root if ** actual archiving is to take place. This is necessary ** if there is no mkdir system call. */ if (!status_only && (getuid() != 0)) { (void) fprintf(errfp, "%s: Sorry, Must be root to rkive.\n", progname); exit(1); } #endif /* ROOT_ONLY */ inum = 0; /* initialize chronological issue counter */ /* Remove any existing temporary mail file */ (void) unlink(tmp_mailfile); cct = 0; /* counter for newsgroup message in global mail */ /* ** Assure that there something specified in the ** archive location variable... */ if (!*newsgrp->location) { (void) fprintf(errfp, "SKIPPING %s: No archive location specified..\n", newsgrp->ng_name); return; } /* ** print out the appropriate ** header for the newsgroup. ** If the user wishes to see the newsgroup info as well as all the ** information about a file archived, the user must specify ** rkive -svv to see it all... */ if (debug || (verbose == 2 && status_only)) { (void) fprintf(logfp,"\n\n"); display_group_info(newsgrp); (void) fprintf(logfp,"\n"); } else if (status_only || verbose == 2) (void) fprintf(logfp, "%s\n",newsgrp->ng_name); /* ** Create a path to the .archived file for the newsgroup's archive. ** This file is used to determine if an article has already been ** archived. If the user did not specify one ... */ if (!*newsgrp->arc_done) { free(newsgrp->arc_done); newsgrp->arc_done = (char *) malloc(strlen(newsgrp->location) + 11); (void) sprintf(newsgrp->arc_done,"%s/.archived",newsgrp->location); } /* ** Create a path to the .patchlog file for the newsgroup's archive. ** This file is used to record patches to posted software so that ** it can easily be determined what the full set of software is. */ if (!*newsgrp->patchlog) { free(newsgrp->patchlog); newsgrp->patchlog = (char *) malloc(strlen(newsgrp->location) + 11); (void) sprintf(newsgrp->patchlog,"%s/.patchlog",newsgrp->location); } /* ** first indicates that archiving has just begun for this group ** so that any necessary setup can be done in retrieve_article(). */ first = 1; #ifdef NNTP /* ** Determine if the newsgroup being archived is to be accessed via ** nntp or via local disk access. */ if (*newsgrp->nntp) { nntp_server = newsgrp->nntp; retrieve = FROM_NNTP; } else if (*nntp) { nntp_server = nntp; retrieve = FROM_NNTP; } #endif /*NNTP*/ while ((val = retrieve_article(article_name,first)) != DONE) { first = 0; if (val == ERROR_ENCOUNTERED) return; else if (val != RETRIEVED) { /* ** Invalid return value encountered, we're in truble now.. */ (void) fprintf(errfp,"Invalid return from retrieve_article..?? Skipping %s\n", newsgrp->ng_path); return; } /* ** Read the news article file to extract the ** header information and fill appropriate ** data structures. */ get_header(article_name); /* ** If the user has specified that a quick status ** listing should be produced then hop to it.... ** FROM_STDIN is not allowed to have status requests. */ if (status_only) { if ((rec = get_archived_rec(header.ident)) == NULL) { /* ** Assure that the article that we are going ** to give status on is one that would be archived ** if archiving was occuring. (e.g. does it match ** the user supplied criteria for archiving...) */ if (*newsgrp->match) { parser_return_value = 0; (void) strcpy(inputstring, newsgrp->match); if (yyparse()) (void) fprintf(logfp, "Parse failed and returned %d\n", parser_return_value); else { /* Did match string produce true as result? */ if (debug) (void) fprintf(logfp, "Parser returned %d\n", parser_return_value); } if (!parser_return_value) { #ifdef NNTP /* ** Since they only want status, remove the ** nntp transfer tmpfile if newsgroup is ** being archived via nntp. */ if (retrieve == FROM_NNTP) (void) unlink(article_name); #endif /*NNTP*/ continue; } } #ifdef NNTP if (retrieve == FROM_NNTP) (void) fprintf(logfp, "\t[%s] Awaiting Archiving\n",header.ident); else (void) fprintf(logfp, "\t[%s] Awaiting Archiving\n",article_name); #else /*!NNTP*/ (void) fprintf(logfp,"\t[%s] Awaiting Archiving\n",article_name); #endif /*NNTP*/ } else if (verbose) { if ((rp = strchr(rec,' ')) == NULL) { #ifdef NNTP if (retrieve == FROM_NNTP) (void) fprintf(logfp,"\t[%s] Archived\n",header.ident); else (void) fprintf(logfp,"\t[%s] Archived\n",article_name); #else /*!NNTP*/ (void) fprintf(logfp,"\t[%s] Archived\n",article_name); #endif /*NNTP*/ } else { rp++; *(rp-1) = '\0'; (void) fprintf(logfp,"\t%s Archived as %s\n",rec,rp); } } #ifdef NNTP /* ** Since they only want status, remove the ** nntp transfer tmpfile if newsgroup is ** being archived via nntp. */ if (retrieve == FROM_NNTP) (void) unlink(article_name); #endif /*NNTP*/ continue; } /* End of status only... */ /* ** We have located a file that may need to be archived. First ** read the header information from the found file. Next, ** check if the article needs to be archived according to the ** configuration file. Finally, determine if the file needs to ** be archived by doing a linear search of of the contents of ** the .archived file comparing the message-id of the new ** article with the message-ids recorded in the .archived ** file. If the new message-id is not specified in the ** .archived file, it has not been archived before and we can ** proceed with the archiving. Can we say slow... maybe dbm in ** the future ??? */ if (*newsgrp->match) { (void) strcpy(inputstring, newsgrp->match); if (yyparse()) { (void) fprintf(logfp, "Parse failed and returned %d\n", parser_return_value); #ifdef NNTP /* ** Remove the nntp transfer tmpfile if the ** newsgroup is being archived via nntp. */ if (retrieve == FROM_NNTP) (void) unlink(article_name); #endif /*NNTP*/ /* ** Remove the tmpfile if the article came from stdin. */ if (retrieve == FROM_STDIN) (void) unlink(article_name); continue; } /* Did match string produce true as result? */ if (debug) (void) fprintf(logfp, "Parser returned %d\n", parser_return_value); if (!parser_return_value) { #ifdef NNTP /* ** Remove the nntp transfer tmpfile if the ** newsgroup is being archived via nntp. */ if (retrieve == FROM_NNTP) (void) unlink(article_name); #endif /*NNTP*/ /* ** Remove the tmpfile if the article came from stdin. */ if (retrieve == FROM_STDIN) (void) unlink(article_name); continue; } } if (!needs_to_be_archived(header.ident)) { #ifdef NNTP /* ** Remove the nntp transfer tmpfile if the ** newsgroup is being archived via nntp. */ if (retrieve == FROM_NNTP) (void) unlink(article_name); #endif /*NNTP*/ /* ** Remove the tmpfile if the article came from stdin. */ if (retrieve == FROM_STDIN) (void) unlink(article_name); continue; } #ifdef NNTP if (nntp_retrieve_article() == ERROR_ENCOUNTERED) { fprintf(logfp, "Error retrieving NNTP article.\n"); return; } #endif /* ** Archiving from here on out. */ if ((new_member = save_article(article_name,newsgrp)) != NULL) { /* ** To not do any compression of setting of ownership ** for newsgroup articles that are archived by an ** external application. */ if (newsgrp->type != EXTERNAL_COMMAND) { archived_file = compress_file(new_member,newsgrp); set_ownership(archived_file,new_member,newsgrp); } else archived_file = new_member; /* ** If a problem has been encountered, ** the function do_problem handles ** the logging, and notifying. */ if (!problem_article) { log_activities(archived_file,newsgrp); build_index(archived_file,newsgrp); notify_users(archived_file,newsgrp,cct++); } } else { if (newsgrp->type != ONLY_ARCHIVE_NAME) { (void) fprintf(logfp,"Unable to archive %s/%s!!!\n", newsgrp->ng_path, article_name); } } #ifdef NNTP /* ** Remove the nntp transfer tmpfile if the ** newsgroup is being archived via nntp. */ if (retrieve == FROM_NNTP) (void) unlink(article_name); #endif /*NNTP*/ /* ** Remove the tmpfile if the article came from stdin. */ if (retrieve == FROM_STDIN) (void) unlink(article_name); } if (!status_only) { /* Mail notification of the archived members to the */ /* list of users specified in the configuration file */ /* and remove the file containing the archived info. */ mail_file(newsgrp->mail_list, tmp_mailfile, newsgrp->ng_name); (void) unlink(tmp_mailfile); } return; } /* ** Notify Users of Archiving. ** If users have been specified to be informed, check to see ** if they have requested a specific logging format. If so ** use the specified format to notify the user. If not, use ** "file archived at path" message. */ void notify_users(filename,ng,num_msgs) char *filename; struct group_archive *ng; int num_msgs; { /* ** Are there users specified in the ** newsgroup section ? */ if ( *(ng->mail_list) ) { if ( *(ng->logformat) ) logit(tmp_mailfile, ng->logformat, filename); else logit(tmp_mailfile, DEFAULT_LOG_FORMAT, filename); } /* ** Are there users specified in the ** global section ? */ if ( *mail ) { if (num_msgs == 0) /* print the newsgroup name out */ logit(global_mailfile, "\n\t\t:%G:\n",filename); if (*log_format) logit(global_mailfile, log_format,filename); else logit(global_mailfile, DEFAULT_LOG_FORMAT, filename); } } /* ** Log_activities ** ** There are two possible logfiles that need to be written. ** The group specific logfile (ng->logfile) and the global ** log. If it has been configured to use a specific format ** for the logging, do so. Else, just record the fact the ** file was sucessfully archived and the date. */ void log_activities(filename,ng) char *filename; struct group_archive *ng; { long clock; long time(); char *ctime(); char logbuf[BUFSIZ]; char dms_date[30]; if ( !*(ng->logformat) || !*log_format) { clock = time((long *)0); (void) strcpy(dms_date, ctime(&clock)); *(dms_date+(strlen(dms_date)-1)) = '\0'; (void) sprintf(logbuf,"%s archived %s",filename, dms_date); } if ( *(ng->logformat) ) logit(ng->logfile, ng->logformat, filename); else logit(ng->logfile, logbuf, filename); if ( *log_format ) logit(log, log_format, filename); else logit(log, logbuf, filename); } /* ** logit ** ** This function is used to append a logfile record ** if there is a logfile name specified. ** */ void logit(filename, format_of_log, arch_file) char *filename; char *format_of_log; char *arch_file; { FILE *fp, *fopen(); char *qp; if (!test) { if ( *(filename) ) { /* Is a logfile specified ? */ if ((fp = fopen(filename,"a")) != NULL) { qp = format_output(format_of_log, arch_file, ARCHIVE); (void) fprintf(fp,"%s\n",qp); (void) fclose(fp); } } } } /* ** Set_ownership ** ** This functions is responsible for setting the owner, group ** and modes of a file just put into the archive. Two file names ** are passed to this function. "filename" contains the compression ** suffix if the file is to be compressed. "flname" is the name of ** the file without the compression suffix. If the ownership or ** modes can not be set on the target "filename", this function ** then tries to set the original file, "flname". In this manner, ** the file permissions get set even if the compression routine fails. */ void set_ownership(filename,flname,ng) char *filename; /* filename with compression suffix */ char *flname; /* filename without compression suffix */ struct group_archive *ng; { int chmod(); int chown(); if (verbose) /* Print out the actions about to be preformed */ (void) fprintf(logfp,"chmod\t<%o> <%s>\n", ng->modes, filename); if (!test) { /* change the file modes to the specified modes */ if (chmod(filename,ng->modes) != 0) { /* Assume the compress failed and try the original... */ if (chmod(flname,ng->modes) != 0) record_problem("Can't change modes of %O", filename, ng); } } if (verbose) { /* Print out the actions about to be preformed */ (void) fprintf(logfp,"chown\t<%d> <%s>\n", ng->owner, filename); (void) fprintf(logfp,"chgrp\t<%d> <%s>\n", ng->group, filename); } if (!test) { /* chown the owner/group to the desired values */ if (chown(filename, ng->owner, ng->group) != 0) { /* Assume the compress failed and try the original... */ if (chown(flname, ng->owner, ng->group) != 0) { /* ** Are we on a system that has a braindamaged chown ** and does not let a general user give files away ? ** Assume so. Quota be gone!! */ if (getuid() == 0) record_problem("Can't change ownership of %O",filename,ng); } } } } void mail_file(user_list, file_to_mail, nwsgrp) char *user_list; char *file_to_mail; char *nwsgrp; { char *list, *name; char cmdstr[80]; /* Is there a list of users to mail to ? */ if ( !*user_list || (strlen(user_list) == 0)) return; /* Was there a notification file created ? */ if (stat(file_to_mail, &sbuf) != 0) return; name = user_list; do { if ((list = strchr(name,',')) != NULL) { list++; *(list-1) = '\0'; } #ifdef SUBJECT_LINE (void) sprintf(cmdstr, "%s -s '%s' %s < %s", MAIL, nwsgrp, name, file_to_mail); #else (void) sprintf(cmdstr, "%s %s < %s", MAIL, name, file_to_mail); #endif if (verbose) (void) fprintf(logfp,"Mailing %s to %s\n", nwsgrp, name); if (!test) (void) system(cmdstr); name = list; } while (name != NULL); return; } void build_index(filename,ng) char *filename; struct group_archive *ng; { if (*(ng->index)) { /* Is there a newsgroup index file ? */ if (*(ng->indformat)) /* Yes, Is there a index file format? */ logit(ng->index, ng->indformat, filename); else if (*index_format) /* No, is there a global format ? */ logit(ng->index, index_format, filename); else /* No, use the default index format */ logit(ng->index, DEFAULT_INDEX_FORMAT, filename); } if (*mindex) { /* Is there a global index file ? */ if (*index_format) /* Yes, Is there a global file format ? */ logit(mindex, index_format, filename); else /* No, so use the default index format */ logit(ng->index, DEFAULT_INDEX_FORMAT , filename); } } char *compress_file(filename,ng) char *filename; struct group_archive *ng; { static char compressed[MAXNAMLEN]; (void) strcpy(compressed, filename); /* store the filename */ /* Check to see if a group specific compress was specified. */ /* If so, then execute the command with the filename passed in. */ /* Else check to see if a global compress was specified. If so, */ /* then execute the command with the filename passed in. */ /* If both are NULL, no compression is done. */ if (*(ng->compress) || (*compress)) { if (*(ng->compress)) (void) strcat(compressed, do_compress(ng->compress, filename)); else if (*compress) (void) strcat(compressed, do_compress(compress, filename)); /* Check to see if the compression worked. If not, return the */ /* original file name passed to this function. The check is */ /* done by assuring the compressed file exists. If not then */ /* it is assumed that it failed. */ if (stat(compressed, &sbuf) == -1) (void) strcpy(compressed, filename); /* restore filename */ } return(compressed); } char *do_compress(packit,filename) char *packit; char *filename; { char *comp_cmd; char cmd[BUFSIZ]; (void) sprintf(cmd,"%s %s", packit, filename); /* ** get the basename of the command to use. */ comp_cmd = basename(packit); if (verbose) (void) fprintf(logfp,"%s %s\n", comp_cmd, filename); if (!test) (void) system(cmd); return(suffix(comp_cmd)); } /* ** Record_problem() ** This function is used to log problems encountered ** to the designated parties. */ void record_problem(msg_fmt,filename,ng) char *msg_fmt; char *filename; struct group_archive *ng; { /* ** This function is used in the event that a problem ** has occurred during archiving. It mails a message ** to the newsgroup specified list and it mails a ** message to the globally specified users. ** ** It then logs the fact into both the newsgroup ** and the global logfiles if they have been specified. */ if ( *(ng->mail_list) ) logit(tmp_mailfile, msg_fmt, filename); if ( *mail ) logit(global_mailfile, msg_fmt,filename); /* ** Assure that the file you are logging information to ** has a directory to exist in.. If it fails, oh well.. */ (void)mkparents(ng->logfile); logit(ng->logfile, msg_fmt, filename); (void)mkparents(log); logit(log, msg_fmt, filename); }