X-Git-Url: http://git.indexdata.com/?a=blobdiff_plain;f=index%2Fkinput.c;h=eeb610daf54ac138515c269183edee83d51c4c40;hb=0e56fa84bb4a5985c19a21926e86862c08d84689;hp=a5905873d594810f765568082a122efdbb51e27e;hpb=dbd6f3a01791401472b07250bf35dcbc3ead4f92;p=idzebra-moved-to-github.git

diff --git a/index/kinput.c b/index/kinput.c
index a590587..eeb610d 100644
--- a/index/kinput.c
+++ b/index/kinput.c
@@ -1,5 +1,5 @@
-/* $Id: kinput.c,v 1.56 2003-06-23 15:35:25 adam Exp $
-   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002
+/* $Id: kinput.c,v 1.63 2004-11-19 10:26:57 heikki Exp $
+   Copyright (C) 1995,1996,1997,1998,1999,2000,2001,2002,2003,2004
    Index Data Aps
 
 This file is part of the Zebra server.
@@ -19,8 +19,6 @@ along with Zebra; see the file LICENSE.zebra.  If not, write to the
 Free Software Foundation, 59 Temple Place - Suite 330, Boston, MA
 02111-1307, USA.
 */
-
-
  
 #include <fcntl.h>
 #ifdef WIN32
@@ -49,8 +47,7 @@ struct key_file {
     size_t chunk;        /* number of bytes allocated */
     size_t buf_ptr;      /* current position in buffer */
     char *prev_name;     /* last word read */
-    int   sysno;         /* last sysno */
-    int   seqno;         /* last seqno */
+    void *decode_handle;
     off_t length;        /* length of file */
                          /* handler invoked in each read */
     void (*readHandler)(struct key_file *keyp, void *rinfo);
@@ -85,21 +82,21 @@ void key_file_chunk_read (struct key_file *f)
     f->buf_size = 0;
     if (fd == -1)
     {
-        logf (LOG_WARN|LOG_ERRNO, "cannot open %s", fname);
+        yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot open %s", fname);
 	return ;
     }
     if (!f->length)
     {
         if ((f->length = lseek (fd, 0L, SEEK_END)) == (off_t) -1)
         {
-            logf (LOG_WARN|LOG_ERRNO, "cannot seek %s", fname);
+            yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot seek %s", fname);
 	    close (fd);
 	    return ;
         }
     }
     if (lseek (fd, f->offset, SEEK_SET) == -1)
     {
-        logf (LOG_WARN|LOG_ERRNO, "cannot seek %s", fname);
+        yaz_log (YLOG_WARN|YLOG_ERRNO, "cannot seek %s", fname);
 	close(fd);
 	return ;
     }
@@ -112,7 +109,7 @@ void key_file_chunk_read (struct key_file *f)
     }
     if (r == -1)
     {
-        logf (LOG_WARN|LOG_ERRNO, "read of %s", fname);
+        yaz_log (YLOG_WARN|YLOG_ERRNO, "read of %s", fname);
 	close (fd);
 	return;
     }
@@ -124,6 +121,7 @@ void key_file_chunk_read (struct key_file *f)
 
 void key_file_destroy (struct key_file *f)
 {
+    iscz1_stop(f->decode_handle);
     xfree (f->buf);
     xfree (f->prev_name);
     xfree (f);
@@ -135,8 +133,7 @@ struct key_file *key_file_init (int no, int chunk, Res res)
 
     f = (struct key_file *) xmalloc (sizeof(*f));
     f->res = res;
-    f->sysno = 0;
-    f->seqno = 0;
+    f->decode_handle = iscz1_start();
     f->no = no;
     f->chunk = chunk;
     f->offset = 0;
@@ -180,7 +177,7 @@ int key_file_decode (struct key_file *f)
         d = ((c&63) << 8) + (key_file_getc (f) & 0xff);
         d = (d << 8) + (key_file_getc (f) & 0xff);
         break;
-    case 192:
+    default: /* 192 */
         d = ((c&63) << 8) + (key_file_getc (f) & 0xff);
         d = (d << 8) + (key_file_getc (f) & 0xff);
         d = (d << 8) + (key_file_getc (f) & 0xff);
@@ -191,8 +188,11 @@ int key_file_decode (struct key_file *f)
 
 int key_file_read (struct key_file *f, char *key)
 {
-    int i, d, c;
-    struct it_key itkey;
+    int i, c;
+    char srcbuf[128];
+    const char *src = srcbuf;
+    char *dst;
+    int j;
 
     c = key_file_getc (f);
     if (c == 0)
@@ -209,22 +209,16 @@ int key_file_read (struct key_file *f, char *key)
         while ((key[i++] = key_file_getc (f)))
             ;
         strcpy (f->prev_name, key);
-        f->sysno = 0;
-    }
-    d = key_file_decode (f);
-    key[i++] = d & 1;
-    d = d >> 1;
-    itkey.sysno = d + f->sysno;
-    if (d) 
-    {
-        f->sysno = itkey.sysno;
-        f->seqno = 0;
+	iscz1_reset(f->decode_handle);
     }
-    d = key_file_decode (f);
-    itkey.seqno = d + f->seqno;
-    f->seqno = itkey.seqno;
-    memcpy (key + i, &itkey, sizeof(struct it_key));
-    return i + sizeof (struct it_key);
+    c = key_file_getc(f); /* length +  insert/delete combined */
+    key[i++] = c & 128;
+    c = c & 127;
+    for (j = 0; j < c; j++)
+	srcbuf[j] = key_file_getc(f);
+    dst = key + i;
+    iscz1_decode(f->decode_handle, &dst, &src);
+    return i + sizeof(struct it_key);
 }
 
 struct heap_info {
@@ -236,7 +230,7 @@ struct heap_info {
     int    *ptr;
     int    (*cmp)(const void *p1, const void *p2);
     struct zebra_register *reg;
-
+    ZebraHandle zh; /* only used for raw reading that bypasses the heaps */
     int no_diffs;
     int no_updates;
     int no_deletions;
@@ -244,17 +238,34 @@ struct heap_info {
     int no_iterations;
 };
 
+static struct heap_info *key_heap_malloc()
+{  /* malloc and clear it */
+    struct heap_info *hi;
+    hi = (struct heap_info *) xmalloc (sizeof(*hi));
+    hi->info.file = 0;
+    hi->info.buf = 0;
+    hi->heapnum = 0;
+    hi->ptr = 0;
+    hi->zh=0;
+    hi->no_diffs = 0;
+    hi->no_diffs = 0;
+    hi->no_updates = 0;
+    hi->no_deletions = 0;
+    hi->no_insertions = 0;
+    hi->no_iterations = 0;
+    return hi;
+}
+
 struct heap_info *key_heap_init (int nkeys,
                                  int (*cmp)(const void *p1, const void *p2))
 {
     struct heap_info *hi;
     int i;
 
-    hi = (struct heap_info *) xmalloc (sizeof(*hi));
+    hi = key_heap_malloc();
     hi->info.file = (struct key_file **)
-	xmalloc (sizeof(*hi->info.file) * (1+nkeys));
+        xmalloc (sizeof(*hi->info.file) * (1+nkeys));
     hi->info.buf = (char **) xmalloc (sizeof(*hi->info.buf) * (1+nkeys));
-    hi->heapnum = 0;
     hi->ptr = (int *) xmalloc (sizeof(*hi->ptr) * (1+nkeys));
     hi->cmp = cmp;
     for (i = 0; i<= nkeys; i++)
@@ -262,21 +273,26 @@ struct heap_info *key_heap_init (int nkeys,
         hi->ptr[i] = i;
         hi->info.buf[i] = (char *) xmalloc (INP_NAME_MAX);
     }
-    hi->no_diffs = 0;
-    hi->no_diffs = 0;
-    hi->no_updates = 0;
-    hi->no_deletions = 0;
-    hi->no_insertions = 0;
-    hi->no_iterations = 0;
+    return hi;
+}
+
+struct heap_info *key_heap_init_buff ( ZebraHandle zh,
+                                 int (*cmp)(const void *p1, const void *p2))
+{
+    struct heap_info *hi=key_heap_malloc();
+    hi->cmp=cmp;
+    hi->zh=zh;
     return hi;
 }
 
 void key_heap_destroy (struct heap_info *hi, int nkeys)
 {
     int i;
-    yaz_log (LOG_DEBUG, "key_heap_destroy");
-    for (i = 0; i<=nkeys; i++)
-        xfree (hi->info.buf[i]);
+    yaz_log (YLOG_DEBUG, "key_heap_destroy");
+    yaz_log (YLOG_DEBUG, "key_heap_destroy nk=%d",nkeys);
+    if (!hi->zh)
+        for (i = 0; i<=nkeys; i++)
+            xfree (hi->info.buf[i]);
     
     xfree (hi->info.buf);
     xfree (hi->ptr);
@@ -338,12 +354,32 @@ static void key_heap_insert (struct heap_info *hi, const char *buf, int nbytes,
     }
 }
 
+static int heap_read_one_raw (struct heap_info *hi, char *name, char *key)
+{
+    ZebraHandle zh=hi->zh;
+    size_t ptr_i = zh->reg->ptr_i;
+    char *cp;
+    if (!ptr_i)
+        return 0;
+    --(zh->reg->ptr_i);
+    cp=(zh->reg->key_buf)[zh->reg->ptr_top - ptr_i];
+    yaz_log (YLOG_DEBUG, " raw: i=%ld top=%ld cp=%p", (long) ptr_i,
+	  (long) zh->reg->ptr_top, cp);
+    strcpy(name, cp);
+    memcpy(key, cp+strlen(name)+1, KEY_SIZE);
+    hi->no_iterations++;
+    return 1;
+}
+
 static int heap_read_one (struct heap_info *hi, char *name, char *key)
 {
     int n, r;
     char rbuf[INP_NAME_MAX];
     struct key_file *kf;
 
+    if (hi->zh) /* bypass the heap stuff, we have a readymade buffer */
+        return heap_read_one_raw(hi, name, key);
+
     if (!hi->heapnum)
         return 0;
     n = hi->ptr[1];
@@ -567,7 +603,7 @@ static void print_dict_item (ZebraMaps zm, const char *s)
                 *to++ = *res++;
     }
     *to = '\0';
-    yaz_log (LOG_LOG, "%s", keybuf);
+    yaz_log (YLOG_LOG, "%s", keybuf);
 }
 #endif
 
@@ -632,158 +668,6 @@ int heap_inpb (struct heap_info *hi)
     return 0;
 } 
 
-int heap_inpd (struct heap_info *hi)
-{
-    struct heap_cread_info hci;
-    ISAMD_I isamd_i = (ISAMD_I) xmalloc (sizeof(*isamd_i));
-
-    hci.key = (char *) xmalloc (KEY_SIZE);
-    hci.key_1 = (char *) xmalloc (KEY_SIZE);
-    hci.key_2 = (char *) xmalloc (KEY_SIZE);
-    hci.ret = -1;
-    hci.first_in_list = 1;
-    hci.hi = hi;
-    hci.more = heap_read_one (hi, hci.cur_name, hci.key);
-
-    isamd_i->clientData = &hci;
-    isamd_i->read_item = heap_cread_item;
-
-    while (hci.more)
-    {
-        char this_name[INP_NAME_MAX];
-        char *dict_info;
-        char dictentry[ISAMD_MAX_DICT_LEN+1];
-        char dictlen;
-
-        strcpy (this_name, hci.cur_name);
-        
-        /* print_dict_item (hi->reg->zebra_maps, hci.cur_name); */
-        /*!*/ /* FIXME: depend on isamd-debug */
-
-	assert (hci.cur_name[1]);
-        hi->no_diffs++;
-        if ((dict_info = dict_lookup (hi->reg->dict, hci.cur_name)))
-        {
-            dictlen=dict_info[0];
-            memcpy (dictentry, dict_info+1, dictlen );
-#ifdef SKIPTHIS
-            logf(LOG_LOG,"dictentry before. len=%d: %d %d %d %d %d %d %d %d %d",
-               dictlen,dictentry[0], dictentry[1], dictentry[2],
-                       dictentry[3], dictentry[4], dictentry[5],
-                       dictentry[6], dictentry[7], dictentry[8]); /*!*/
-#endif
-            dictlen= isamd_append(hi->reg->isamd, dictentry, dictlen, isamd_i);
-             /* logf dictentry after */
-            if (dictlen)
-            {
-                hi->no_updates++;
-                if ( (dictlen!=dict_info[0]) ||
-                     (0!=memcmp(dictentry, dict_info+1, dictlen)) )
-                {
-                    dict_insert(hi->reg->dict, this_name,
-                                dictlen,dictentry);
-                }
-            }
-            else
-            {
-                hi->no_deletions++;
-                if (!dict_delete (hi->reg->dict, this_name)) 
-                {
-	            logf (LOG_FATAL, "dict_delete failed");
-                    abort();
-                }
-            }
-        } 
-        else
-        {
-            dictlen=0;
-            memset (dictentry, '\0', ISAMD_MAX_DICT_LEN);
-            dictlen= isamd_append(hi->reg->isamd, dictentry, dictlen, isamd_i);
-             /* logf dictentry first */
-            hi->no_insertions++;
-            if (dictlen)
-                dict_insert(hi->reg->dict, this_name,
-                                dictlen,dictentry);
-        }
-    }
-    xfree (isamd_i);
-    xfree (hci.key);
-    xfree (hci.key_1);
-    xfree (hci.key_2);
-    return 0;
-} 
-
-int heap_inp (struct heap_info *hi)
-{
-    char *info;
-    char next_name[INP_NAME_MAX];
-    char cur_name[INP_NAME_MAX];
-    int key_buf_size = INP_BUF_START;
-    int key_buf_ptr;
-    char *next_key;
-    char *key_buf;
-    int more;
-    
-    next_key = (char *) xmalloc (KEY_SIZE);
-    key_buf = (char *) xmalloc (key_buf_size);
-    more = heap_read_one (hi, cur_name, key_buf);
-    while (more)                   /* EOF ? */
-    {
-        int nmemb;
-        key_buf_ptr = KEY_SIZE;
-        while (1)
-        {
-            if (!(more = heap_read_one (hi, next_name, next_key)))
-                break;
-            if (*next_name && strcmp (next_name, cur_name))
-                break;
-            memcpy (key_buf + key_buf_ptr, next_key, KEY_SIZE);
-            key_buf_ptr += KEY_SIZE;
-            if (key_buf_ptr+(int) KEY_SIZE >= key_buf_size)
-            {
-                char *new_key_buf;
-                new_key_buf = (char *) xmalloc (key_buf_size + INP_BUF_ADD);
-                memcpy (new_key_buf, key_buf, key_buf_size);
-                key_buf_size += INP_BUF_ADD;
-                xfree (key_buf);
-                key_buf = new_key_buf;
-            }
-        }
-        hi->no_diffs++;
-        nmemb = key_buf_ptr / KEY_SIZE;
-        assert (nmemb * (int) KEY_SIZE == key_buf_ptr);
-        if ((info = dict_lookup (hi->reg->dict, cur_name)))
-        {
-            ISAM_P isam_p, isam_p2;
-            memcpy (&isam_p, info+1, sizeof(ISAM_P));
-            isam_p2 = is_merge (hi->reg->isam, isam_p, nmemb, key_buf);
-            if (!isam_p2)
-            {
-                hi->no_deletions++;
-                if (!dict_delete (hi->reg->dict, cur_name))
-                    abort ();
-            }
-            else 
-            {
-                hi->no_updates++;
-                if (isam_p2 != isam_p)
-                    dict_insert (hi->reg->dict, cur_name,
-                                 sizeof(ISAM_P), &isam_p2);
-            }
-        }
-        else
-        {
-            ISAM_P isam_p;
-            hi->no_insertions++;
-            isam_p = is_merge (hi->reg->isam, 0, nmemb, key_buf);
-            dict_insert (hi->reg->dict, cur_name, sizeof(ISAM_P), &isam_p);
-        }
-        memcpy (key_buf, next_key, KEY_SIZE);
-        strcpy (cur_name, next_name);
-    }
-    return 0;
-}
-
 int heap_inps (struct heap_info *hi)
 {
     struct heap_cread_info hci;
@@ -817,7 +701,7 @@ int heap_inps (struct heap_info *hi)
         }
 	else
 	{
-	    logf (LOG_FATAL, "isams doesn't support this kind of update");
+	    yaz_log (YLOG_FATAL, "isams doesn't support this kind of update");
 	    break;
 	}
     }
@@ -847,10 +731,10 @@ void progressFunc (struct key_file *keyp, void *info)
         remaining = (time_t) ((now - p->startTime)*
             ((double) p->totalBytes/p->totalOffset - 1.0));
         if (remaining <= 130)
-            logf (LOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining",
+            yaz_log (YLOG_LOG, "Merge %2.1f%% completed; %ld seconds remaining",
                  (100.0*p->totalOffset) / p->totalBytes, (long) remaining);
         else
-            logf (LOG_LOG, "Merge %2.1f%% completed; %ld minutes remaining",
+            yaz_log (YLOG_LOG, "Merge %2.1f%% completed; %ld minutes remaining",
 	         (100.0*p->totalOffset) / p->totalBytes, (long) remaining/60);
     }
     p->totalOffset += keyp->buf_size;
@@ -862,73 +746,89 @@ void progressFunc (struct key_file *keyp, void *info)
 
 void zebra_index_merge (ZebraHandle zh)
 {
-    struct key_file **kf;
+    struct key_file **kf = 0;
     char rbuf[1024];
     int i, r;
     struct heap_info *hi;
     struct progressInfo progressInfo;
     int nkeys = zh->reg->key_file_no;
+    int usefile; 
+    
+    yaz_log (YLOG_DEBUG, " index_merge called with nk=%d b=%p", 
+                    nkeys, zh->reg->key_buf);
+    if ( (nkeys==0) && (zh->reg->key_buf==0) )
+        return; /* nothing to merge - probably flush after end-trans */
     
-    if (nkeys < 0)
+    usefile = (nkeys!=0); 
+
+    if (usefile)
     {
-        char fname[1024];
-        nkeys = 0;
-        while (1)
+        if (nkeys < 0)
         {
-            extract_get_fname_tmp  (zh, fname, nkeys+1);
-            if (access (fname, R_OK) == -1)
-                break;
-            nkeys++;
+            char fname[1024];
+            nkeys = 0;
+            while (1)
+            {
+                extract_get_fname_tmp  (zh, fname, nkeys+1);
+                if (access (fname, R_OK) == -1)
+                        break;
+                nkeys++;
+            }
+            if (!nkeys)
+                return ;
         }
-        if (!nkeys)
-            return ;
-    }
-    kf = (struct key_file **) xmalloc ((1+nkeys) * sizeof(*kf));
-    progressInfo.totalBytes = 0;
-    progressInfo.totalOffset = 0;
-    time (&progressInfo.startTime);
-    time (&progressInfo.lastTime);
-    for (i = 1; i<=nkeys; i++)
-    {
-        kf[i] = key_file_init (i, 8192, zh->res);
-        kf[i]->readHandler = progressFunc;
-        kf[i]->readInfo = &progressInfo;
-        progressInfo.totalBytes += kf[i]->length;
-        progressInfo.totalOffset += kf[i]->buf_size;
+        kf = (struct key_file **) xmalloc ((1+nkeys) * sizeof(*kf));
+        progressInfo.totalBytes = 0;
+        progressInfo.totalOffset = 0;
+        time (&progressInfo.startTime);
+        time (&progressInfo.lastTime);
+        for (i = 1; i<=nkeys; i++)
+        {
+            kf[i] = key_file_init (i, 8192, zh->res);
+            kf[i]->readHandler = progressFunc;
+            kf[i]->readInfo = &progressInfo;
+            progressInfo.totalBytes += kf[i]->length;
+            progressInfo.totalOffset += kf[i]->buf_size;
+        }
+        hi = key_heap_init (nkeys, key_qsort_compare);
+        hi->reg = zh->reg;
+        
+        for (i = 1; i<=nkeys; i++)
+            if ((r = key_file_read (kf[i], rbuf)))
+                key_heap_insert (hi, rbuf, r, kf[i]);
+    }  /* use file */
+    else 
+    { /* do not use file, read straight from buffer */
+        hi = key_heap_init_buff (zh,key_qsort_compare);
+        hi->reg = zh->reg;
     }
-    hi = key_heap_init (nkeys, key_qsort_compare);
-    hi->reg = zh->reg;
-    
-    for (i = 1; i<=nkeys; i++)
-        if ((r = key_file_read (kf[i], rbuf)))
-            key_heap_insert (hi, rbuf, r, kf[i]);
     if (zh->reg->isams)
 	heap_inps (hi);
     if (zh->reg->isamc)
         heap_inpc (hi);
-    if (zh->reg->isam)
-	heap_inp (hi);
-    if (zh->reg->isamd)
-	heap_inpd (hi);
     if (zh->reg->isamb)
 	heap_inpb (hi);
 	
-    for (i = 1; i<=nkeys; i++)
+    if (usefile)
     {
-        extract_get_fname_tmp  (zh, rbuf, i);
-        unlink (rbuf);
+        for (i = 1; i<=nkeys; i++)
+        {
+            extract_get_fname_tmp  (zh, rbuf, i);
+            unlink (rbuf);
+        }
+        for (i = 1; i<=nkeys; i++)
+            key_file_destroy (kf[i]);
+        xfree (kf);
+    }
+    if (hi->no_iterations)
+    { /* do not log if nothing happened */
+        yaz_log (YLOG_LOG, "Iterations . . .%7d", hi->no_iterations);
+        yaz_log (YLOG_LOG, "Distinct words .%7d", hi->no_diffs);
+        yaz_log (YLOG_LOG, "Updates. . . . .%7d", hi->no_updates);
+        yaz_log (YLOG_LOG, "Deletions. . . .%7d", hi->no_deletions);
+        yaz_log (YLOG_LOG, "Insertions . . .%7d", hi->no_insertions);
     }
-    logf (LOG_LOG, "Iterations . . .%7d", hi->no_iterations);
-    logf (LOG_LOG, "Distinct words .%7d", hi->no_diffs);
-    logf (LOG_LOG, "Updates. . . . .%7d", hi->no_updates);
-    logf (LOG_LOG, "Deletions. . . .%7d", hi->no_deletions);
-    logf (LOG_LOG, "Insertions . . .%7d", hi->no_insertions);
     zh->reg->key_file_no = 0;
 
     key_heap_destroy (hi, nkeys);
-    for (i = 1; i<=nkeys; i++)
-        key_file_destroy (kf[i]);
-    xfree (kf);
 }
-
-