From: Adam Dickmeiss <adam@indexdata.dk>
Date: Tue, 10 Oct 1995 12:24:38 +0000 (+0000)
Subject: Temporary sort files are compressed.
X-Git-Tag: ZEBRA.1.0~710
X-Git-Url: http://git.indexdata.com/?p=idzebra-moved-to-github.git;a=commitdiff_plain;h=fc7107844c9ec8ea23e680ca4f3231923db4e9c5

Temporary sort files are compressed.
---

diff --git a/index/extract.c b/index/extract.c
index ea31d07..1830850 100644
--- a/index/extract.c
+++ b/index/extract.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: extract.c,v $
- * Revision 1.20  1995-10-06 13:52:05  adam
+ * 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
@@ -119,12 +122,73 @@ void key_open (int mem)
         exit (1);
     }
 }
+
+struct encode_info {
+    int  sysno;
+    int  seqno;
+    char buf[512];
+};
+
+void encode_key_init (struct encode_info *i)
+{
+    i->sysno = 0;
+    i->seqno = 0;
+}
+
+char *encode_key_int (int d, char *bp)
+{
+    if (d <= 63)
+        *bp++ = d;
+    else if (d <= 16383)
+    {
+        *bp++ = 64 + (d>>8);
+        *bp++ = d  & 255;
+    }
+    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 encode_key_write (char *k, struct encode_info *i, FILE *outf)
+{
+    struct it_key key;
+    char *bp = i->buf;
+
+    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)
+    {
+        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;
@@ -142,33 +206,19 @@ void key_flush (void)
     logf (LOG_LOG, "writing section %d", key_file_no);
     prevcp = cp = key_buf[ptr_top-ptr_i];
     
-    if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1, outf) != 1)
-    {
-        logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-        exit (1);
-    }
+    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))
         {
-            if (fwrite (cp, strlen (cp)+2+sizeof(struct it_key), 1,
-                        outf) != 1)
-            {
-                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-                exit (1);
-            }
+            encode_key_init (&encode_info);
+            encode_key_write (cp, &encode_info, outf);
             prevcp = cp;
         }
         else
-        {
-            cp = strlen (cp) + cp;
-            if (fwrite (cp, 2+sizeof(struct it_key), 1, outf) != 1)
-            {
-                logf (LOG_FATAL|LOG_ERRNO, "fwrite %s", out_fname);
-                exit (1);
-            }
-        }
+            encode_key_write (cp + strlen(cp), &encode_info, outf);
     }
     if (fclose (outf))
     {
diff --git a/index/index.h b/index/index.h
index 08ced19..ece96af 100644
--- a/index/index.h
+++ b/index/index.h
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: index.h,v $
- * Revision 1.15  1995-10-04 16:57:19  adam
+ * Revision 1.16  1995-10-10 12:24:38  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.15  1995/10/04  16:57:19  adam
  * Key input and merge sort in one pass.
  *
  * Revision 1.14  1995/09/29  14:01:40  adam
@@ -84,9 +87,7 @@ void key_write (int cmd, struct it_key *k, const char *str);
 int key_compare (const void *p1, const void *p2);
 int key_qsort_compare (const void *p1, const void *p2);
 void key_logdump (int mask, const void *p);
-void key_input (const char *dict_fname, const char *isam_fname, 
-                const char *key_fname, int cache);
-void key_input2 (const char *dict_fname, const char *isam_fname,
+void key_input (const char *dict_fname, const char *isam_fname,
                  int nkeys, int cache);
 int merge_sort (char **buf, int from, int to);
 
diff --git a/index/kdump.c b/index/kdump.c
index 899f508..93344e6 100644
--- a/index/kdump.c
+++ b/index/kdump.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: kdump.c,v $
- * Revision 1.6  1995-09-29 14:01:42  adam
+ * Revision 1.7  1995-10-10 12:24:38  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.6  1995/09/29  14:01:42  adam
  * Bug fixes.
  *
  * Revision 1.5  1995/09/11  13:09:35  adam
@@ -28,24 +31,68 @@
 #include <stdio.h>
 #include <assert.h>
 #include <unistd.h>
+#include <assert.h>
 
 #include <alexutil.h>
 #include "index.h"
 
 char *prog;
 
-static int read_one (FILE *inf, char *name, char *key)
+
+int key_file_decode (FILE *f)
+{
+    int c, d;
+
+    c = getc (f);
+    switch (c & 192) 
+    {
+    case 0:
+        d = c;
+        break;
+    case 64:
+        d = ((c&63) << 8) + (getc (f) & 0xff);
+        break;
+    case 128:
+        d = ((c&63) << 8) + (getc (f) & 0xff);
+        d = (d << 8) + (getc (f) & 0xff);
+        break;
+    case 192:
+        d = ((c&63) << 8) + (getc (f) & 0xff);
+        d = (d << 8) + (getc (f) & 0xff);
+        d = (d << 8) + (getc (f) & 0xff);
+        break;
+    }
+    return d;
+}
+
+
+static int read_one (FILE *inf, char *name, char *key, struct it_key *prevk)
 {
     int c;
     int i = 0;
+    struct it_key itkey;
     do
     {
         if ((c=getc(inf)) == EOF)
             return 0;
         name[i++] = c;
     } while (c);
-    for (i = 0; i<sizeof(struct it_key)+1; i++)
-        ((char *)key)[i] = getc (inf);
+    if (i > 1)
+        prevk->sysno = 0;
+    c = key_file_decode (inf);
+    key[0] = c & 1;
+    c = c >> 1;
+    itkey.sysno = c + prevk->sysno;
+    if (c)
+    {
+        prevk->sysno = itkey.sysno;
+        prevk->seqno = 0;
+    }
+    c = key_file_decode (inf);
+    itkey.seqno = c + prevk->seqno;
+    prevk->seqno = itkey.seqno;
+
+    memcpy (key+1, &itkey, sizeof(itkey));
     return 1;
 }
 
@@ -57,6 +104,10 @@ int main (int argc, char **argv)
     char key_string[IT_MAX_WORD];
     char key_info[256];
     FILE *inf;
+    struct it_key prevk;
+
+    prevk.sysno = 0;
+    prevk.seqno = 0;
 
     prog = *argv;
     while ((ret = options ("v:", argv, argc, &arg)) != -2)
@@ -85,21 +136,15 @@ int main (int argc, char **argv)
         logf (LOG_FATAL|LOG_ERRNO, "fopen %s", key_fname);
         exit (1);
     }
-    while (read_one (inf, key_string, key_info))
+    while (read_one (inf, key_string, key_info, &prevk))
     {
         struct it_key k;
         int op;
 
         op = key_info[0];
         memcpy (&k, 1+key_info, sizeof(k));
-#if IT_KEY_HAVE_SEQNO
         printf ("%7d op=%d s=%-5d %s\n", k.sysno, op, k.seqno,
                 key_string);
-#else
-        printf ("%7d op=%d f=%-3d %s\n", k.sysno, op, k.freq,
-                key_string);
-
-#endif
     }
     if (fclose (inf))
     {
diff --git a/index/kinput.c b/index/kinput.c
index 71f5372..e9d467a 100644
--- a/index/kinput.c
+++ b/index/kinput.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: kinput.c,v $
- * Revision 1.8  1995-10-04 16:57:19  adam
+ * Revision 1.9  1995-10-10 12:24:39  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.8  1995/10/04  16:57:19  adam
  * Key input and merge sort in one pass.
  *
  * Revision 1.7  1995/10/02  15:18:52  adam
@@ -52,120 +55,6 @@ static int no_updates = 0;
 static int no_insertions = 0;
 static int no_iterations = 0;
 
-static int read_one (FILE *inf, char *name, char *key)
-{
-    int c;
-    int i = 0;
-    do
-    {
-        if ((c=getc(inf)) == EOF)
-            return 0;
-        name[i++] = c;
-    } while (c);
-    for (i = 0; i<KEY_SIZE; i++)
-        ((char *)key)[i] = getc (inf);
-    ++no_iterations;
-    return 1;
-}
-
-static int inp (Dict dict, ISAM isam, const char *name)
-{
-    FILE *inf;
-    char *info;
-    char next_name[INP_NAME_MAX+1];
-    char cur_name[INP_NAME_MAX+1];
-    int key_buf_size = INP_BUF_START;
-    int key_buf_ptr;
-    char *next_key;
-    char *key_buf;
-    int more;
-    
-    next_key = xmalloc (KEY_SIZE);
-    key_buf = xmalloc (key_buf_size * (KEY_SIZE));
-    if (!(inf = fopen (name, "r")))
-    {
-        logf (LOG_FATAL|LOG_ERRNO, "cannot open `%s'", name);
-        exit (1);
-    }
-    more = read_one (inf, cur_name, key_buf);
-    while (more)                   /* EOF ? */
-    {
-        int nmemb;
-        key_buf_ptr = KEY_SIZE;
-        while (1)
-        {
-            if (!(more = read_one (inf, 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+KEY_SIZE >= key_buf_size)
-            {
-                char *new_key_buf;
-                new_key_buf = 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;
-            }
-        }
-        no_diffs++;
-        nmemb = key_buf_ptr / KEY_SIZE;
-        assert (nmemb*KEY_SIZE == key_buf_ptr);
-        if ((info = dict_lookup (dict, cur_name)))
-        {
-            ISAM_P isam_p, isam_p2;
-            logf (LOG_DEBUG, "updating %s", cur_name);
-            no_updates++;
-            memcpy (&isam_p, info+1, sizeof(ISAM_P));
-            isam_p2 = is_merge (isam, isam_p, nmemb, key_buf);
-            if (isam_p2 != isam_p)
-                dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p2);
-        }
-        else
-        {
-            ISAM_P isam_p;
-            logf (LOG_DEBUG, "inserting %s", cur_name);
-            no_insertions++;
-            isam_p = is_merge (isam, 0, nmemb, key_buf);
-            dict_insert (dict, cur_name, sizeof(ISAM_P), &isam_p);
-        }
-        memcpy (key_buf, next_key, KEY_SIZE);
-        strcpy (cur_name, next_name);
-    }
-    fclose (inf);
-    return 0;
-}
-
-void key_input (const char *dict_fname, const char *isam_fname,
-                const char *key_fname, int cache)
-{
-    Dict dict;
-    ISAM isam;
-
-    dict = dict_open (dict_fname, cache, 1);
-    if (!dict)
-    {
-        logf (LOG_FATAL, "dict_open fail of `%s'", dict_fname);
-        exit (1);
-    }
-    isam = is_open (isam_fname, key_compare, 1, sizeof(struct it_key));
-    if (!isam)
-    {
-        logf (LOG_FATAL, "is_open fail of `%s'", isam_fname);
-        exit (1);
-    }
-    inp (dict, isam, key_fname);
-    dict_close (dict);
-    is_close (isam);
-    logf (LOG_LOG, "Iterations . . .%7d", no_iterations);
-    logf (LOG_LOG, "Distinct words .%7d", no_diffs);
-    logf (LOG_LOG, "Updates. . . . .%7d", no_updates);
-    logf (LOG_LOG, "Insertions . . .%7d", no_insertions);
-}
-
-
 struct key_file {
     int   no;            /* file no */
     off_t offset;        /* file offset */
@@ -174,6 +63,8 @@ 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 key_file_chunk_read (struct key_file *f)
@@ -215,6 +106,8 @@ struct key_file *key_file_init (int no, int chunk)
     struct key_file *f;
 
     f = xmalloc (sizeof(*f));
+    f->sysno = 0;
+    f->seqno = 0;
     f->no = no;
     f->chunk = chunk;
     f->offset = 0;
@@ -239,9 +132,36 @@ int key_file_getc (struct key_file *f)
         return EOF;
 }
 
+int key_file_decode (struct key_file *f)
+{
+    int c, d;
+
+    c = key_file_getc (f);
+    switch (c & 192) 
+    {
+    case 0:
+        d = c;
+        break;
+    case 64:
+        d = ((c&63) << 8) + (key_file_getc (f) & 0xff);
+        break;
+    case 128:
+        d = ((c&63) << 8) + (key_file_getc (f) & 0xff);
+        d = (d << 8) + (key_file_getc (f) & 0xff);
+        break;
+    case 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);
+        break;
+    }
+    return d;
+}
+
 int key_file_read (struct key_file *f, char *key)
 {
-    int i, j, c;
+    int i, d, c;
+    struct it_key itkey;
 
     c = key_file_getc (f);
     if (c == 0)
@@ -258,10 +178,22 @@ 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;
     }
-    for (j = KEY_SIZE; --j >= 0; )
-        key[i++] = key_file_getc (f);
-    return i;
+    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;
+    }
+    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);
 }
 
 struct heap_info {
@@ -310,12 +242,8 @@ static void key_heap_delete (struct heap_info *hi)
 
     assert (hi->heapnum > 0);
 
-#if 1
     key_heap_swap (hi, 1, hi->heapnum);
     hi->heapnum--;
-#else
-    hi->ptr[1] = hi->ptr[hi->heapnum--];
-#endif
     while (child <= hi->heapnum) {
         if (child < hi->heapnum &&
             (*hi->cmp)(&hi->info.buf[hi->ptr[child]],
@@ -435,7 +363,7 @@ int heap_inp (Dict dict, ISAM isam, struct heap_info *hi)
     return 0;
 }
 
-void key_input2 (const char *dict_fname, const char *isam_fname,
+void key_input (const char *dict_fname, const char *isam_fname,
                  int nkeys, int cache)
 {
     Dict dict;
diff --git a/index/main.c b/index/main.c
index 7c56296..e0fa497 100644
--- a/index/main.c
+++ b/index/main.c
@@ -4,7 +4,10 @@
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: main.c,v $
- * Revision 1.12  1995-10-04 16:57:20  adam
+ * Revision 1.13  1995-10-10 12:24:39  adam
+ * Temporary sort files are compressed.
+ *
+ * Revision 1.12  1995/10/04  16:57:20  adam
  * Key input and merge sort in one pass.
  *
  * Revision 1.11  1995/09/29  14:01:45  adam
@@ -61,7 +64,6 @@ int main (int argc, char **argv)
     char *base_name = NULL;
     char *base_path = NULL;
     int nsections;
-    char **mbuf;
 
     prog = *argv;
     while ((ret = options ("r:v:m:", argv, argc, &arg)) != -2)
@@ -130,17 +132,8 @@ int main (int argc, char **argv)
     nsections = key_close ();
     if (!nsections)
         exit (0);
-#if 0
-    logf (LOG_LOG, "Merge sorting");
-    mbuf = xmalloc (100000);
-    merge_sort (mbuf, 1, nsections+1);
-    xfree (mbuf);
-    logf (LOG_LOG, "Input");
-    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, "keys1.tmp", 60);
-#else
     logf (LOG_LOG, "Input");
-    key_input2 (FNAME_WORD_DICT, FNAME_WORD_ISAM, nsections, 60);
-#endif
+    key_input (FNAME_WORD_DICT, FNAME_WORD_ISAM, nsections, 60);
     exit (0);
 }