X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=b320cbc54812a4d14696025b679b9b94f1f6cd4b;hb=85df66537199c30a492ad54be4fbe25fa77e18c8;hp=56bc94913d5a4414fb3e3e0eaaff158e93e029cc;hpb=dd5f8f31bc5018a7b4a00accdda256ce1eaef14b;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 56bc949..b320cbc 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,66 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.30  1995-11-22 17:19:16  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
+ * though the blocks were already released.
+ * File traversal properly deletes records when needed.
+ *
+ * Revision 1.33  1995/11/27  09:56:20  adam
+ * Record info elements better enumerated. Internal store of records.
+ *
+ * Revision 1.32  1995/11/25  10:24:05  adam
+ * More record fields - they are enumerated now.
+ * New options: flagStoreData flagStoreKey.
+ *
+ * Revision 1.31  1995/11/24  11:31:35  adam
+ * Commands add & del read filenames from stdin if source directory is
+ * empty.
+ * Match criteria supports 'constant' strings.
+ *
+ * Revision 1.30  1995/11/22  17:19:16  adam
  * Record management uses the bfile system.
  *
  * Revision 1.29  1995/11/21  15:01:14  adam
@@ -126,8 +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)
@@ -139,9 +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);
@@ -292,10 +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;
@@ -431,24 +488,37 @@ static void addRecordKeyAny (const RecWord *p)
     addRecordKey (p);
 }
 
+
+#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;
+#endif
 
 static void file_read_start (int fd)
 {
+    file_noread = 0;
+#if FILE_READ_BUFSIZE
     file_offset = 0;
-    file_buf = xmalloc (4096);
-    file_bufsize = read (fd, file_buf, 4096);
+    file_buf = xmalloc (FILE_READ_BUFSIZE);
+    file_bufsize = read (fd, file_buf, FILE_READ_BUFSIZE);
+#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)
@@ -457,7 +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)
             {
@@ -466,9 +536,10 @@ static int file_read (int fd, char *buf, size_t count)
             }
             file_bufsize = 0;
             file_offset = 0;
-            return r;
+            file_noread += l+r;
+            return l+r;
         }
-        file_bufsize = r = read (fd, file_buf, 4096);
+        file_bufsize = r = read (fd, file_buf, FILE_READ_BUFSIZE);
         if (r == -1)
         {
             logf (LOG_FATAL|LOG_ERRNO, "read");
@@ -478,18 +549,28 @@ static int file_read (int fd, char *buf, size_t count)
         {
             file_offset = r;
             memcpy (buf + l, file_buf, r);
-            return l + r;
+            file_noread += l+r;
+            return l+r;
         }
         else
         {
             file_offset = count;
             memcpy (buf + l, file_buf, count - l);
+            file_noread += count;
             return count;
         }
     }
     memcpy (buf, file_buf + file_offset, 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)
@@ -505,7 +586,6 @@ static int atois (const char **s)
 
 static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
                            const char *fname,
-                           const char *recordType,
                            const char *spec)
 {
     static char dstBuf[2048];
@@ -573,34 +653,51 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
         else if (*s == '$')
         {
             int spec_len;
-            char special[32];
+            char special[64];
             const char *spec_src = NULL;
             const char *s1 = ++s;
             while (*s1 && *s1 != ' ' && *s1 != '\t')
                 s1++;
 
             spec_len = s1 - s;
-            if (spec_len > 31)
-                spec_len = 31;
+            if (spec_len > 63)
+                spec_len = 63;
             memcpy (special, s, spec_len);
             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"))
-                spec_src = recordType;
+            else if (!strcmp (special, "type"))
+                spec_src = rGroup->recordType;
             else 
                 spec_src = NULL;
             if (spec_src)
             {
                 strcpy (dst, spec_src);
-                dst += strlen(spec_src);
+                dst += strlen (spec_src);
+            }
+        }
+        else if (*s == '\"' || *s == '\'')
+        {
+            int stopMarker = *s++;
+            char tmpString[64];
+            int i = 0;
+
+            while (*s && *s != stopMarker)
+            {
+                if (i < 63)
+                    tmpString[i++] = *s++;
             }
+            if (*s)
+                s++;
+            tmpString[i] = '\0';
+            strcpy (dst, tmpString);
+            dst += strlen (tmpString);
         }
         else
         {
@@ -619,27 +716,204 @@ static char *fileMatchStr (struct recKeys *reckeys, struct recordGroup *rGroup,
     return dstBuf;
 }
 
-int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup,
-                 int deleteFlag)
+static int recordExtract (SYSNO *sysno, const char *fname,
+                          struct recordGroup *rGroup, int deleteFlag,
+                          int fd, RecType recType, char *subType)
 {
+    struct recExtractCtrl extractCtrl;
+    int r;
+    char *matchStr;
     SYSNO sysnotmp;
-    int i, r;
+    Record rec;
+
+    if (fd != -1)
+    {
+        extractCtrl.fd = fd;
+        /* extract keys */
+        extractCtrl.subType = subType;
+        extractCtrl.init = wordInit;
+        extractCtrl.add = addRecordKeyAny;
+
+        reckeys.buf_used = 0;
+        extractCtrl.readf = file_read;
+        r = (*recType->extract)(&extractCtrl);
+  
+        if (r)      
+        {
+            logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
+            return 0;
+        }
+    }
+
+    /* perform match if sysno not known and if match criteria is specified */
+       
+    matchStr = NULL;
+    if (!sysno) 
+    {
+        sysnotmp = 0;
+        sysno = &sysnotmp;
+        if (rGroup->recordId && *rGroup->recordId)
+        {
+            char *rinfo;
+        
+            matchStr = fileMatchStr (&reckeys, rGroup, fname, 
+                                     rGroup->recordId);
+            if (matchStr)
+            {
+                rinfo = dict_lookup (matchDict, matchStr);
+                if (rinfo)
+                    memcpy (sysno, rinfo+1, sizeof(*sysno));
+            }
+            else
+            {
+                logf (LOG_WARN, "Record not inserted");
+                return 0;
+            }
+        }
+    }
+
+    /* new record ? */
+    if (! *sysno)
+    {
+        if (deleteFlag)
+        {
+            logf (LOG_LOG, "? %s", fname);
+            return 1;
+        }
+        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++;
+    }
+    else
+    {
+        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]);
+        if (deleteFlag)
+        {
+            if (!delkeys.buf_used)
+            {
+                logf (LOG_WARN, "cannot delete %s: storeKeys false",
+                      fname);
+            }
+            else
+            {
+                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: storeKeys false",
+                      fname);
+            }
+            else
+            {
+                logf (LOG_LOG, "update %s %s", rGroup->recordType,
+                      fname);
+                flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); 
+                records_updated++;
+            }
+        }
+    }
+    xfree (rec->info[recInfo_fileType]);
+    rec->info[recInfo_fileType] =
+        rec_strdup (rGroup->recordType, &rec->size[recInfo_fileType]);
+
+    xfree (rec->info[recInfo_filename]);
+    rec->info[recInfo_filename] =
+        rec_strdup (fname, &rec->size[recInfo_filename]);
+
+    xfree (rec->info[recInfo_delKeys]);
+    if (reckeys.buf_used > 0 && rGroup->flagStoreKeys == 1)
+    {
+        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]);
+    }
+    else
+    {
+        rec->info[recInfo_delKeys] = NULL;
+        rec->size[recInfo_delKeys] = 0;
+    }
+
+    xfree (rec->info[recInfo_storeData]);
+    if (rGroup->flagStoreData == 1)
+    {
+        rec->size[recInfo_storeData] = 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)
+            {
+                logf (LOG_ERRNO|LOG_FATAL, "seek to 0 in %s", fname);
+                exit (1);
+            }
+            if (read (fd, rec->info[recInfo_storeData], file_noread) 
+                < file_noread)
+            {
+                logf (LOG_ERRNO|LOG_FATAL, "read %d bytes of %s",
+                      file_noread, fname);
+                exit (1);
+            }
+        }
+    }
+    else
+    {
+        rec->info[recInfo_storeData] = NULL;
+        rec->size[recInfo_storeData] = 0;
+    }
+    xfree (rec->info[recInfo_databaseName]);
+    rec->info[recInfo_databaseName] =
+        rec_strdup (rGroup->databaseName, &rec->size[recInfo_databaseName]); 
+
+    rec_put (records, &rec);
+    return 1;
+}
+
+int fileExtract (SYSNO *sysno, const char *fname, 
+                 const struct recordGroup *rGroupP, int deleteFlag)
+{
+    int i, fd;
     char gprefix[128];
     char ext[128];
     char ext_res[128];
-    const char *file_type;
-    const char *file_match;
-    struct recExtractCtrl extractCtrl;
+    char subType[128];
     RecType recType;
-    Record rec;
-    char *matchStr;
+    struct recordGroup rGroupM;
+    struct recordGroup *rGroup = &rGroupM;
 
+    memcpy (rGroup, rGroupP, sizeof(*rGroupP));
+   
     if (!rGroup->groupName || !*rGroup->groupName)
         *gprefix = '\0';
     else
         sprintf (gprefix, "%s.", rGroup->groupName);
 
-    logf (LOG_DEBUG, "fileExtractAdd %s", fname);
+    logf (LOG_DEBUG, "fileExtract %s", fname);
 
     /* determine file extension */
     for (i = strlen(fname); --i >= 0; )
@@ -654,19 +928,35 @@ int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup,
             break;
         }
     /* determine file type - depending on extension */
-    sprintf (ext_res, "%sfileExtension.%s", gprefix, ext);
-    if (!(file_type = res_get (common_resource, ext_res)))
+    if (!rGroup->recordType)
+    {
+        sprintf (ext_res, "%srecordType.%s", gprefix, ext);
+        if (!(rGroup->recordType = res_get (common_resource, ext_res)))
+        {
+            sprintf (ext_res, "%srecordType", gprefix);
+            if (!(rGroup->recordType = res_get (common_resource, ext_res)))
+            {
+                logf (LOG_LOG, "? %s", fname);
+                return 0;
+            }
+        }
+    }
+    if (!rGroup->recordType)
+    {
+        logf (LOG_LOG, "? record %s", fname);
         return 0;
-    if (!(recType = recType_byName (file_type)))
+    }
+    if (!(recType = recType_byName (rGroup->recordType, subType)))
+    {
+        logf (LOG_WARN, "No such record type: %s", rGroup->recordType);
         return 0;
+    }
 
     /* determine match criteria */
-    sprintf (ext_res, "%sfileMatch.%s", gprefix, ext);
-    file_match = res_get (common_resource, ext_res);
-    if (!file_match)
+    if (!rGroup->recordId)
     {
-        sprintf (ext_res, "%sfileMatch", gprefix);
-        file_match = res_get (common_resource, ext_res);
+        sprintf (ext_res, "%srecordId.%s", gprefix, ext);
+        rGroup->recordId = res_get (common_resource, ext_res);
     }
 
     /* determine database name */
@@ -682,118 +972,52 @@ int fileExtract (SYSNO *sysno, const char *fname, struct recordGroup *rGroup,
     if (!rGroup->databaseName)
         rGroup->databaseName = "Default";
 
-    /* open input file */
-    if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
-    {
-        logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
-        return 0;
-    }
-
-    /* extract keys */
-    extractCtrl.subType = "";
-    extractCtrl.init = wordInit;
-    extractCtrl.add = addRecordKeyAny;
-
-    reckeys.buf_used = 0;
-    file_read_start (extractCtrl.fd);
-    extractCtrl.readf = file_read;
-    r = (*recType->extract)(&extractCtrl);
-    file_read_stop (extractCtrl.fd);
-    close (extractCtrl.fd);
-  
-    if (r)      
+    if (rGroup->flagStoreData == -1)
     {
-        logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
-        return 0;
-    }
-
-    /* perform match if sysno not known and if match criteria is specified */
-       
-    matchStr = NULL;
-    if (!sysno && file_match)
-    {
-        char *rinfo;
-        
-        sysno = &sysnotmp;
-        matchStr = fileMatchStr(&reckeys, rGroup, fname, file_type,
-                                file_match);
-        if (matchStr)
-        {
-            rinfo = dict_lookup (matchDict, matchStr);
-            if (rinfo)
-                memcpy (sysno, rinfo+1, sizeof(*sysno));
-            else
-                *sysno = 0;
-        }
-        else
+        const char *sval;
+        sprintf (ext_res, "%sstoreData.%s", gprefix, ext);
+        if (!(sval = res_get (common_resource, ext_res)))
         {
-            logf (LOG_WARN, "Record not inserted");
-            return 0;
+            sprintf (ext_res, "%sstoreData", gprefix);
+            sval = res_get (common_resource, ext_res);
         }
+        if (sval)
+            rGroup->flagStoreData = atoi (sval);
     }
+    if (rGroup->flagStoreData == -1)
+        rGroup->flagStoreData = 0;
 
-    /* new record ? */
-    if (! *sysno)
+    if (rGroup->flagStoreKeys == -1)
     {
-        if (deleteFlag)
-        {
-            logf (LOG_LOG, "? record %s", fname);
-            return 1;
-        }
-        logf (LOG_LOG, "add record %s", fname);
-        rec = rec_new (records);
-        *sysno = rec->sysno;
+        const char *sval;
 
-        if (matchStr)
-            dict_insert (matchDict, matchStr, sizeof(*sysno), sysno);
-        flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName);
-
-        records_inserted++;
-    }
-    else
-    {
-        struct recKeys delkeys;
-
-        rec = rec_get (records, *sysno);
-
-        delkeys.buf_used = rec->size[2];
-	delkeys.buf = rec->info[2];
-        flushRecordKeys (*sysno, 0, &delkeys, rec->info[3]);
-        if (deleteFlag)
+        sprintf (ext_res, "%sstoreKeys.%s", gprefix, ext);
+        if (!(sval = res_get (common_resource, ext_res)))
         {
-            logf (LOG_LOG, "delete record %s", fname);
-            records_deleted++;
-            rec_del (records, &rec);
-            return 1;
-        }
-        else
-        {
-            logf (LOG_LOG, "update record %s", fname);
-            flushRecordKeys (*sysno, 1, &reckeys, rGroup->databaseName); 
-            records_updated++;
+            sprintf (ext_res, "%sstoreKeys", gprefix);
+            sval = res_get (common_resource, ext_res);
         }
+        if (sval)
+            rGroup->flagStoreKeys = atoi (sval);
     }
-    free (rec->info[0]);
-    rec->info[0] = rec_strdup (file_type, &rec->size[0]);
-
-    free (rec->info[1]);
-    rec->info[1] = rec_strdup (fname, &rec->size[1]);
+    if (rGroup->flagStoreKeys == -1)
+        rGroup->flagStoreKeys = 0;
 
-    free (rec->info[2]);
-    if (reckeys.buf_used > 0)
-    {
-        rec->info[2] = malloc (reckeys.buf_used);
-        rec->size[2] = reckeys.buf_used;
-        memcpy (rec->info[2], reckeys.buf, rec->size[2]);
-    }
+    if (sysno && deleteFlag)
+        fd = -1;
     else
     {
-        rec->info[2] = NULL;
-        rec->size[2] = 0;
+        if ((fd = open (fname, O_RDONLY)) == -1)
+        {
+            logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
+            return 0;
+        }
     }
-    free (rec->info[3]);
-    rec->info[3] = rec_strdup (rGroup->databaseName, &rec->size[3]); 
-
-    rec_put (records, &rec);
+    file_read_start (fd);
+    recordExtract (sysno, fname, rGroup, deleteFlag, fd, recType, subType);
+    file_read_stop (fd);
+    if (fd != -1)
+        close (fd);
     return 1;
 }
+