X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;ds=sidebyside;f=index%2Fextract.c;h=b320cbc54812a4d14696025b679b9b94f1f6cd4b;hb=85df66537199c30a492ad54be4fbe25fa77e18c8;hp=285003d928f1862eef57e1cb129ced64c85df83f;hpb=162fa86044b5aa303c03fa26f1ecf140b34060d5;p=idzebra-moved-to-github.git diff --git a/index/extract.c b/index/extract.c index 285003d..b320cbc 100644 --- a/index/extract.c +++ b/index/extract.c @@ -4,7 +4,47 @@ * Sebastian Hammer, Adam Dickmeiss * * $Log: extract.c,v $ - * Revision 1.34 1995-11-28 09:09:38 adam + * Revision 1.44 1995-12-12 16:00:54 adam + * System call sync(2) used after update/commit. + * Locking (based on fcntl) uses F_EXLCK and F_SHLCK instead of F_WRLCK + * and F_RDLCK. + * + * Revision 1.43 1995/12/11 09:12:46 adam + * The rec_get function returns NULL if record doesn't exist - will + * happen in the server if the result set records have been deleted since + * the creation of the set (i.e. the search). + * The server saves a result temporarily if it is 'volatile', i.e. the + * set is register dependent. + * + * Revision 1.42 1995/12/07 17:38:46 adam + * Work locking mechanisms for concurrent updates/commit. + * + * Revision 1.41 1995/12/06 16:06:42 adam + * Better diagnostics. Work on 'real' dictionary deletion. + * + * Revision 1.40 1995/12/05 16:57:40 adam + * More work on regular patterns. + * + * Revision 1.39 1995/12/05 13:20:18 adam + * Bug fix: file_read sometimes returned early EOF. + * + * Revision 1.38 1995/12/04 17:59:21 adam + * More work on regular expression conversion. + * + * Revision 1.37 1995/12/04 14:22:27 adam + * Extra arg to recType_byName. + * Started work on new regular expression parsed input to + * structured records. + * + * Revision 1.36 1995/11/30 08:34:29 adam + * Started work on commit facility. + * Changed a few malloc/free to xmalloc/xfree. + * + * Revision 1.35 1995/11/28 14:26:21 adam + * Bug fix: recordId with constant wasn't right. + * Bug fix: recordId dictionary entry wasn't deleted when needed. + * + * Revision 1.34 1995/11/28 09:09:38 adam * Zebra config renamed. * Use setting 'recordId' to identify record now. * Bug fix in recindex.c: rec_release_blocks was invokeded even @@ -145,8 +185,6 @@ static int records_inserted = 0; static int records_updated = 0; static int records_deleted = 0; -#define MATCH_DICT "match" - void key_open (int mem) { if (mem < 50000) @@ -158,9 +196,9 @@ void key_open (int mem) key_buf_used = 0; key_file_no = 0; - if (!(matchDict = dict_open (MATCH_DICT, 20, 1))) + if (!(matchDict = dict_open (GMATCH_DICT, 50, 1))) { - logf (LOG_FATAL, "dict_open fail of %s", MATCH_DICT); + logf (LOG_FATAL, "dict_open fail of %s", GMATCH_DICT); exit (1); } assert (!records); @@ -311,10 +349,10 @@ static void addRecordKey (const RecWord *p) { char *b; - b = malloc (reckeys.buf_max += 65000); + b = xmalloc (reckeys.buf_max += 65000); if (reckeys.buf_used > 0) memcpy (b, reckeys.buf, reckeys.buf_used); - free (reckeys.buf); + xfree (reckeys.buf); reckeys.buf = b; } dst = reckeys.buf + reckeys.buf_used; @@ -450,29 +488,37 @@ static void addRecordKeyAny (const RecWord *p) addRecordKey (p); } -#define FILE_READ_BUFSIZE 4096 +#define FILE_READ_BUFSIZE 4096 + +static int file_noread; +#if FILE_READ_BUFSIZE static char *file_buf; static int file_offset; static int file_bufsize; -static int file_noread; +#endif static void file_read_start (int fd) { + file_noread = 0; +#if FILE_READ_BUFSIZE file_offset = 0; file_buf = xmalloc (FILE_READ_BUFSIZE); file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE); - file_noread = 0; +#endif } static void file_read_stop (int fd) { +#if FILE_READ_BUFSIZE xfree (file_buf); file_buf = NULL; +#endif } static int file_read (int fd, char *buf, size_t count) { +#if FILE_READ_BUFSIZE int l = file_bufsize - file_offset; if (count > l) @@ -481,7 +527,7 @@ static int file_read (int fd, char *buf, size_t count) if (l > 0) memcpy (buf, file_buf + file_offset, l); count = count-l; - if (count > file_bufsize) + if (count > FILE_READ_BUFSIZE) { if ((r = read (fd, buf + l, count)) == -1) { @@ -490,8 +536,8 @@ static int file_read (int fd, char *buf, size_t count) } file_bufsize = 0; file_offset = 0; - file_noread += r; - return r; + file_noread += l+r; + return l+r; } file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE); if (r == -1) @@ -503,8 +549,8 @@ static int file_read (int fd, char *buf, size_t count) { file_offset = r; memcpy (buf + l, file_buf, r); - file_noread += (l+r); - return l + r; + file_noread += l+r; + return l+r; } else { @@ -518,6 +564,13 @@ static int file_read (int fd, char *buf, size_t count) file_offset += count; file_noread += count; return count; +#else + int r; + r = read (fd, buf, count); + if (r > 0) + file_noread += r; + return r; +#endif } static int atois (const char **s) @@ -613,13 +666,13 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, special[spec_len] = '\0'; s = s1; - if (strcmp (special, "group")) + if (!strcmp (special, "group")) spec_src = rGroup->groupName; - else if (strcmp (special, "database")) + else if (!strcmp (special, "database")) spec_src = rGroup->databaseName; - else if (strcmp (special, "filename")) + else if (!strcmp (special, "filename")) spec_src = fname; - else if (strcmp (special, "type")) + else if (!strcmp (special, "type")) spec_src = rGroup->recordType; else spec_src = NULL; @@ -638,7 +691,7 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, while (*s && *s != stopMarker) { if (i < 63) - tmpString[i++] = *s; + tmpString[i++] = *s++; } if (*s) s++; @@ -665,8 +718,7 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup, static int recordExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup, int deleteFlag, - int fd, - RecType recType) + int fd, RecType recType, char *subType) { struct recExtractCtrl extractCtrl; int r; @@ -678,7 +730,7 @@ static int recordExtract (SYSNO *sysno, const char *fname, { extractCtrl.fd = fd; /* extract keys */ - extractCtrl.subType = ""; + extractCtrl.subType = subType; extractCtrl.init = wordInit; extractCtrl.add = addRecordKeyAny; @@ -725,15 +777,17 @@ static int recordExtract (SYSNO *sysno, const char *fname, { if (deleteFlag) { - logf (LOG_LOG, "? record %s", fname); + logf (LOG_LOG, "? %s", fname); return 1; } - logf (LOG_LOG, "add %s record %s", rGroup->recordType, fname); + logf (LOG_LOG, "add %s %s", rGroup->recordType, fname); rec = rec_new (records); *sysno = rec->sysno; if (matchStr) + { dict_insert (matchDict, matchStr, sizeof(*sysno), sysno); + } flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); records_inserted++; @@ -743,7 +797,7 @@ static int recordExtract (SYSNO *sysno, const char *fname, struct recKeys delkeys; rec = rec_get (records, *sysno); - + assert (rec); delkeys.buf_used = rec->size[recInfo_delKeys]; delkeys.buf = rec->info[recInfo_delKeys]; flushRecordKeys (*sysno, 0, &delkeys, rec->info[recInfo_databaseName]); @@ -751,44 +805,47 @@ static int recordExtract (SYSNO *sysno, const char *fname, { if (!delkeys.buf_used) { - logf (LOG_WARN, "cannot delete %s - no delete keys", + logf (LOG_WARN, "cannot delete %s: storeKeys false", fname); } else - logf (LOG_LOG, "delete %s record %s", rGroup->recordType, - fname); - records_deleted++; - rec_del (records, &rec); + { + logf (LOG_LOG, "delete %s %s", rGroup->recordType, fname); + records_deleted++; + if (matchStr) + dict_delete (matchDict, matchStr); + rec_del (records, &rec); + } return 1; } else { if (!delkeys.buf_used) { - logf (LOG_WARN, "cannot update %s - no delete keys", + logf (LOG_WARN, "cannot update %s: storeKeys false", fname); } else { - logf (LOG_LOG, "update %s record %s", rGroup->recordType, + logf (LOG_LOG, "update %s %s", rGroup->recordType, fname); flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); records_updated++; } } } - free (rec->info[recInfo_fileType]); + xfree (rec->info[recInfo_fileType]); rec->info[recInfo_fileType] = rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]); - free (rec->info[recInfo_filename]); + xfree (rec->info[recInfo_filename]); rec->info[recInfo_filename] = rec_strdup (fname, &rec->size[recInfo_filename]); - free (rec->info[recInfo_delKeys]); + xfree (rec->info[recInfo_delKeys]); if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1) { - rec->info[recInfo_delKeys] = malloc (reckeys.buf_used); + rec->info[recInfo_delKeys] = xmalloc (reckeys.buf_used); rec->size[recInfo_delKeys] = reckeys.buf_used; memcpy (rec->info[recInfo_delKeys], reckeys.buf, rec->size[recInfo_delKeys]); @@ -799,14 +856,16 @@ static int recordExtract (SYSNO *sysno, const char *fname, rec->size[recInfo_delKeys] = 0; } - free (rec->info[recInfo_storeData]); + xfree (rec->info[recInfo_storeData]); if (rGroup->flagStoreData == 1) { rec->size[recInfo_storeData] = file_noread; - rec->info[recInfo_storeData] = malloc (file_noread); + rec->info[recInfo_storeData] = xmalloc (file_noread); +#if FILE_READ_BUFSIZE if (file_noread < FILE_READ_BUFSIZE) memcpy (rec->info[recInfo_storeData], file_buf, file_noread); else +#endif { if (lseek (fd, 0L, SEEK_SET) < 0) { @@ -827,7 +886,7 @@ static int recordExtract (SYSNO *sysno, const char *fname, rec->info[recInfo_storeData] = NULL; rec->size[recInfo_storeData] = 0; } - free (rec->info[recInfo_databaseName]); + xfree (rec->info[recInfo_databaseName]); rec->info[recInfo_databaseName] = rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); @@ -842,6 +901,7 @@ int fileExtract (SYSNO *sysno, const char *fname, char gprefix[128]; char ext[128]; char ext_res[128]; + char subType[128]; RecType recType; struct recordGroup rGroupM; struct recordGroup *rGroup = &rGroupM; @@ -876,7 +936,7 @@ int fileExtract (SYSNO *sysno, const char *fname, sprintf (ext_res, "%srecordType", gprefix); if (!(rGroup->recordType = res_get (common_resource, ext_res))) { - logf (LOG_LOG, "? record %s", fname); + logf (LOG_LOG, "? %s", fname); return 0; } } @@ -886,7 +946,7 @@ int fileExtract (SYSNO *sysno, const char *fname, logf (LOG_LOG, "? record %s", fname); return 0; } - if (!(recType = recType_byName (rGroup->recordType))) + if (!(recType = recType_byName (rGroup->recordType, subType))) { logf (LOG_WARN, "No such record type: %s", rGroup->recordType); return 0; @@ -954,7 +1014,7 @@ int fileExtract (SYSNO *sysno, const char *fname, } } file_read_start (fd); - recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType); + recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType, subType); file_read_stop (fd); if (fd != -1) close (fd);