X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fextract.c;h=9207ab76518cfe7c1f09d8cee96b5f249d3771c6;hb=83762ea76e5af65ccb4407c6b38053bc6491a875;hp=9b7f76bd42c8d1dd16edb56273082a6fd07ca097;hpb=ca19a15163ee74ee0c6765bfef4fd9794627e26a;p=idzebra-moved-to-github.git

diff --git a/index/extract.c b/index/extract.c
index 9b7f76b..9207ab7 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,55 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.10  1995-09-28 09:19:41  adam
+ * Revision 1.26  1995-11-20 11:56:24  adam
+ * Work on new traversal.
+ *
+ * Revision 1.25  1995/11/16  15:34:54  adam
+ * Uses new record management system in both indexer and server.
+ *
+ * Revision 1.24  1995/11/15  19:13:08  adam
+ * Work on record management.
+ *
+ * Revision 1.23  1995/10/27  14:00:10  adam
+ * Implemented detection of database availability.
+ *
+ * Revision 1.22  1995/10/17  18:02:07  adam
+ * New feature: databases. Implemented as prefix to words in dictionary.
+ *
+ * Revision 1.21  1995/10/10  12:24:38  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.20  1995/10/06  13:52:05  adam
+ * Bug fixes. Handler may abort further scanning.
+ *
+ * Revision 1.19  1995/10/04  12:55:16  adam
+ * Bug fix in ranked search. Use=Any keys inserted.
+ *
+ * Revision 1.18  1995/10/04  09:37:08  quinn
+ * Fixed bug.
+ *
+ * Revision 1.17  1995/10/03  14:28:57  adam
+ * Buffered read in extract works.
+ *
+ * Revision 1.16  1995/10/03  14:28:45  adam
+ * Work on more effecient read handler in extract.
+ *
+ * Revision 1.15  1995/10/02  15:42:53  adam
+ * Extract uses file descriptors instead of FILE pointers.
+ *
+ * Revision 1.14  1995/10/02  15:29:13  adam
+ * More logging in file_extract.
+ *
+ * Revision 1.13  1995/09/29  14:01:39  adam
+ * Bug fixes.
+ *
+ * Revision 1.12  1995/09/28  14:22:56  adam
+ * Sort uses smaller temporary files.
+ *
+ * Revision 1.11  1995/09/28  12:10:31  adam
+ * Bug fixes. Field prefix used in queries.
+ *
+ * Revision 1.10  1995/09/28  09:19:41  adam
  * xfree/xmalloc used everywhere.
  * Extract/retrieve method seems to work for text records.
  *
@@ -49,34 +97,51 @@
 #include <recctrl.h>
 #include "index.h"
 
+#if RECORD_BASE
+#include "recindex.h"
+#endif
+
 static Dict file_idx;
-static SYSNO sysno_next;
-static int key_fd = -1;
+
+
+#if RECORD_BASE
+static Records records = NULL;
+#else
 static int sys_idx_fd = -1;
-static char *key_buf;
-static int key_offset, key_buf_size;
+static SYSNO sysno_next;
+#endif
+
 static int key_cmd;
 static int key_sysno;
+static char *key_databaseName;
+static char **key_buf;
+static size_t ptr_top;
+static size_t ptr_i;
+static size_t kused;
+static int key_file_no;
 
-void key_open (const char *fname)
+void key_open (int mem)
 {
+#if !RECORD_BASE
     void *file_key;
-    if (key_fd != -1)
-        return;
-    if ((key_fd = open (fname, O_RDWR|O_CREAT, 0666)) == -1)
-    {
-        logf (LOG_FATAL|LOG_ERRNO, "open %s", fname);
-        exit (1);
-    }
-    logf (LOG_DEBUG, "key_open of %s", fname);
-    key_buf_size = 49100;
-    key_buf = xmalloc (key_buf_size);
-    key_offset = 0;
+#endif
+    if (mem < 50000)
+        mem = 50000;
+    key_buf = xmalloc (mem);
+    ptr_top = mem/sizeof(char*);
+    ptr_i = 0;
+    kused = 0;
+    key_file_no = 0;
+
     if (!(file_idx = dict_open (FNAME_FILE_DICT, 40, 1)))
     {
         logf (LOG_FATAL, "dict_open fail of %s", "fileidx");
         exit (1);
     }
+#if RECORD_BASE
+    assert (!records);
+    records = rec_open (1);
+#else
     file_key = dict_lookup (file_idx, ".");
     if (file_key)
         memcpy (&sysno_next, (char*)file_key+1, sizeof(sysno_next));
@@ -87,104 +152,262 @@ void key_open (const char *fname)
         logf (LOG_FATAL|LOG_ERRNO, "open %s", FNAME_SYS_IDX);
         exit (1);
     }
+#endif
 }
 
-int key_close (void)
+struct encode_info {
+    int  sysno;
+    int  seqno;
+    char buf[512];
+};
+
+void encode_key_init (struct encode_info *i)
 {
-    if (key_fd == -1)
+    i->sysno = 0;
+    i->seqno = 0;
+}
+
+char *encode_key_int (int d, char *bp)
+{
+    if (d <= 63)
+        *bp++ = d;
+    else if (d <= 16383)
     {
-        logf (LOG_DEBUG, "key_close - but no file");
-        return 0;
+        *bp++ = 64 + (d>>8);
+        *bp++ = d  & 255;
     }
-    close (key_fd);
-    close (sys_idx_fd);
-    dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
-    dict_close (file_idx);
-    key_fd = -1;
-    xfree (key_buf);
-    return 1;
+    else if (d <= 4194303)
+    {
+        *bp++ = 128 + (d>>16);
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
+    }
+    else
+    {
+        *bp++ = 192 + (d>>24);
+        *bp++ = (d>>16) & 255;
+        *bp++ = (d>>8) & 255;
+        *bp++ = d & 255;
+    }
+    return bp;
 }
 
-void wordFlush (int sysno)
+void encode_key_write (char *k, struct encode_info *i, FILE *outf)
 {
-    size_t i = 0;
-    int w;
+    struct it_key key;
+    char *bp = i->buf;
 
-    if (key_fd == -1)
-	return; 
-    while (i < key_offset)
+    while ((*bp++ = *k++))
+        ;
+    memcpy (&key, k+1, sizeof(struct it_key));
+    bp = encode_key_int ( (key.sysno - i->sysno) * 2 + *k, bp);
+    if (i->sysno != key.sysno)
     {
-        w = write (key_fd, key_buf + i, key_offset - i);
-        if (w == -1)
+        i->sysno = key.sysno;
+        i->seqno = 0;
+    }
+    bp = encode_key_int (key.seqno - i->seqno, bp);
+    i->seqno = key.seqno;
+    if (fwrite (i->buf, bp - i->buf, 1, outf) != 1)
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fwrite");
+        exit (1);
+    }
+}
+
+void key_flush (void)
+{
+    FILE *outf;
+    char out_fname[200];
+    char *prevcp, *cp;
+    struct encode_info encode_info;
+    
+    if (ptr_i <= 0)
+        return;
+
+    key_file_no++;
+    logf (LOG_LOG, "sorting section %d", key_file_no);
+    qsort (key_buf + ptr_top-ptr_i, ptr_i, sizeof(char*), key_qsort_compare);
+    sprintf (out_fname, TEMP_FNAME, key_file_no);
+
+    if (!(outf = fopen (out_fname, "w")))
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fopen (4) %s", out_fname);
+        exit (1);
+    }
+    logf (LOG_LOG, "writing section %d", key_file_no);
+    prevcp = cp = key_buf[ptr_top-ptr_i];
+    
+    encode_key_init (&encode_info);
+    encode_key_write (cp, &encode_info, outf);
+    while (--ptr_i > 0)
+    {
+        cp = key_buf[ptr_top-ptr_i];
+        if (strcmp (cp, prevcp))
         {
-            logf (LOG_FATAL|LOG_ERRNO, "Write key fail");
-            exit (1);
+            encode_key_init (&encode_info);
+            encode_key_write (cp, &encode_info, outf);
+            prevcp = cp;
         }
-        i += w;
+        else
+            encode_key_write (cp + strlen(cp), &encode_info, outf);
     }
-    key_offset = 0;
+    if (fclose (outf))
+    {
+        logf (LOG_FATAL|LOG_ERRNO, "fclose %s", out_fname);
+        exit (1);
+    }
+    logf (LOG_LOG, "finished section %d", key_file_no);
+    ptr_i = 0;
+    kused = 0;
+}
+
+int key_close (void)
+{
+    key_flush ();
+    xfree (key_buf);
+#if RECORD_BASE
+    rec_close (&records);
+#else
+    close (sys_idx_fd);
+    dict_insert (file_idx, ".", sizeof(sysno_next), &sysno_next);
+#endif
+    dict_close (file_idx);
+    return key_file_no;
 }
 
 static void wordInit (RecWord *p)
 {
     p->attrSet = 1;
-    p->attrUse = 1;
+    p->attrUse = 1016;
     p->which = Word_String;
 }
 
 static void wordAdd (const RecWord *p)
 {
     struct it_key key;
-    char x;
     size_t i;
-    char wordPrefix[8];
 
-    if (key_offset + 1000 > key_buf_size)
-    {
-        char *new_key_buf;
-
-        key_buf_size *= 2;
-        new_key_buf = xmalloc (2*key_buf_size);
-        memcpy (new_key_buf, key_buf, key_offset);
-        xfree (key_buf);
-        key_buf = new_key_buf;
-    }
-    sprintf (wordPrefix, "%c%04d", p->attrSet + '0', p->attrUse);
-    strcpy (key_buf + key_offset, wordPrefix);
-    key_offset += strlen (wordPrefix);
+    if (kused + 1024 > (ptr_top-ptr_i)*sizeof(char*))
+        key_flush ();
+    ++ptr_i;
+    key_buf[ptr_top-ptr_i] = (char*)key_buf + kused;
+    kused += index_word_prefix ((char*)key_buf + kused,
+                                p->attrSet, p->attrUse,
+                                key_databaseName);
     switch (p->which)
     {
     case Word_String:
         for (i = 0; p->u.string[i]; i++)
-            key_buf[key_offset++] = index_char_cvt (p->u.string[i]);
-        key_buf[key_offset++] = '\0';
+            ((char*)key_buf) [kused++] = index_char_cvt (p->u.string[i]);
+        ((char*)key_buf) [kused++] = '\0';
         break;
     default:
         return ;
     }
-    x = (key_cmd == 'a') ? 1 : 0;
-    memcpy (key_buf + key_offset, &x, 1);
-    key_offset++;
-
+    ((char*) key_buf)[kused++] = ((key_cmd == 'a') ? 1 : 0);
     key.sysno = key_sysno;
-    key.seqno   = p->seqno;
-    memcpy (key_buf + key_offset, &key, sizeof(key));
-    key_offset += sizeof(key);
+    key.seqno = p->seqno;
+    memcpy ((char*)key_buf + kused, &key, sizeof(key));
+    kused += sizeof(key);
+}
+
+static void wordAddAny (const RecWord *p)
+{
+    if (p->attrSet != 1 || p->attrUse != 1016)
+    {
+        RecWord w;
+
+        memcpy (&w, p, sizeof(w));
+        w.attrSet = 1;
+        w.attrUse = 1016;
+        wordAdd (&w);
+    }
+    wordAdd (p);
+}
+
+
+#define FILE_READ_BUF 1
+#if FILE_READ_BUF
+static char *file_buf;
+static int file_offset;
+static int file_bufsize;
+
+static void file_read_start (int fd)
+{
+    file_offset = 0;
+    file_buf = xmalloc (4096);
+    file_bufsize = read (fd, file_buf, 4096);
+}
+
+static void file_read_stop (int fd)
+{
+    xfree (file_buf);
 }
 
-void file_extract (int cmd, const char *fname, const char *kname)
+static int file_read (int fd, char *buf, size_t count)
 {
-    int i;
+    int l = file_bufsize - file_offset;
+
+    if (count > l)
+    {
+        int r;
+        if (l > 0)
+            memcpy (buf, file_buf + file_offset, l);
+        count = count-l;
+        if (count > file_bufsize)
+        {
+            if ((r = read (fd, buf + l, count)) == -1)
+            {
+                logf (LOG_FATAL|LOG_ERRNO, "read");
+                exit (1);
+            }
+            file_bufsize = 0;
+            file_offset = 0;
+            return r;
+        }
+        file_bufsize = r = read (fd, file_buf, 4096);
+        if (r == -1)
+        {
+            logf (LOG_FATAL|LOG_ERRNO, "read");
+            exit (1);
+        }
+        else if (r <= count)
+        {
+            file_offset = r;
+            memcpy (buf + l, file_buf, r);
+            return l + r;
+        }
+        else
+        {
+            file_offset = count;
+            memcpy (buf + l, file_buf, count - l);
+            return count;
+        }
+    }
+    memcpy (buf, file_buf + file_offset, count);
+    file_offset += count;
+    return count;
+}
+#else
+static int file_read (int fd, char *buf, size_t count)
+{
+    return read (fd, buf, count);
+}
+#endif
+SYSNO file_extract (int cmd, const char *fname, const char *kname,
+                    char *databaseName)
+{
+    int i, r;
     char ext[128];
     SYSNO sysno;
     char ext_res[128];
     const char *file_type;
     void *file_info;
-    FILE *inf;
     struct recExtractCtrl extractCtrl;
     RecType rt;
 
-    logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
+    key_databaseName = databaseName;
     for (i = strlen(fname); --i >= 0; )
         if (fname[i] == '/')
         {
@@ -198,35 +421,53 @@ void file_extract (int cmd, const char *fname, const char *kname)
         }
     sprintf (ext_res, "fileExtension.%s", ext);
     if (!(file_type = res_get (common_resource, ext_res)))
-        return;
+        return 0;
     if (!(rt = recType_byName (file_type)))
-        return;
+        return 0;
+    logf (LOG_DEBUG, "%c %s k=%s", cmd, fname, kname);
     file_info = dict_lookup (file_idx, kname);
     if (!file_info)
     {
+#if RECORD_BASE
+        Record rec = rec_new (records);
+
+        sysno = rec->sysno;
+        dict_insert (file_idx, kname, sizeof(sysno), &sysno);
+        rec->info[0] = rec_strdup (file_type);
+        rec->info[1] = rec_strdup (kname);
+        rec_put (records, rec);
+        rec_rm (rec);
+#else
         sysno = sysno_next++;
         dict_insert (file_idx, kname, sizeof(sysno), &sysno);
         lseek (sys_idx_fd, sysno * SYS_IDX_ENTRY_LEN, SEEK_SET);
         write (sys_idx_fd, file_type, strlen (file_type)+1);
         write (sys_idx_fd, kname, strlen(kname)+1);
+#endif
     }
     else
         memcpy (&sysno, (char*) file_info+1, sizeof(sysno));
 
-    if (!(inf = fopen (fname, "r")))
+    if ((extractCtrl.fd = open (fname, O_RDONLY)) == -1)
     {
         logf (LOG_WARN|LOG_ERRNO, "open %s", fname);
-        return;
+        return 0;
     }
-    extractCtrl.inf = inf;
     extractCtrl.subType = "";
     extractCtrl.init = wordInit;
-    extractCtrl.add = wordAdd;
+    extractCtrl.add = wordAddAny;
+#if FILE_READ_BUF
+    file_read_start (extractCtrl.fd);
+#endif
+    extractCtrl.readf = file_read;
     key_sysno = sysno;
     key_cmd = cmd;
-    (*rt->extract)(&extractCtrl);
-    fclose (inf);
-    wordFlush (sysno);
+    r = (*rt->extract)(&extractCtrl);
+#if FILE_READ_BUF
+    file_read_stop (extractCtrl.fd);
+#endif
+    close (extractCtrl.fd);
+    if (r)
+        logf (LOG_WARN, "Couldn't extract file %s, code %d", fname, r);
+    return sysno;
 }
-
-