Implemented Sort.

[idzebra-moved-to-github.git] / recctrl / rectext.c

diff --git a/recctrl/rectext.c b/recctrl/rectext.c
index 321e414..9993d70 100644 (file)
--- a/recctrl/rectext.c
+++ b/recctrl/rectext.c
@@ -1,10 +1,28 @@
 /*
- * Copyright (C) 1994-1995, Index Data I/S 
+ * Copyright (C) 1994-1998, Index Data I/S 
  * All rights reserved.
  * Sebastian Hammer, Adam Dickmeiss
  *
  * $Log: rectext.c,v $
- * Revision 1.1  1996-10-11 10:57:28  adam
+ * Revision 1.6  1998-02-10 12:03:06  adam
+ * Implemented Sort.
+ *
+ * Revision 1.5  1997/10/27 14:33:06  adam
+ * Moved towards generic character mapping depending on "structure"
+ * field in abstract syntax file. Fixed a few memory leaks. Fixed
+ * bug with negative integers when doing searches with relational
+ * operators.
+ *
+ * Revision 1.4  1996/11/04 14:09:16  adam
+ * Minor changes.
+ *
+ * Revision 1.3  1996/11/01 09:00:33  adam
+ * This simple "text" format now supports element specs B and M.
+ *
+ * Revision 1.2  1996/10/29 14:02:45  adam
+ * Uses buffered read to speed up things.
+ *
+ * Revision 1.1  1996/10/11 10:57:28  adam
  * New module recctrl. Used to manage records (extract/retrieval).
  *
  * Revision 1.7  1996/01/17 14:57:55  adam
@@ -36,39 +54,80 @@
 #include <assert.h>
 #include <ctype.h>
 
-#include <alexutil.h>
+#include <zebrautl.h>
 #include "rectext.h"
 
 static void text_init (void)
 {
 }
 
+struct buf_info {
+    struct recExtractCtrl *p;
+    char *buf;
+    int offset;
+    int max;
+};
+
+struct buf_info *buf_open (struct recExtractCtrl *p)
+{
+    struct buf_info *fi = xmalloc (sizeof(*fi));
+
+    fi->p = p;
+    fi->buf = xmalloc (4096);
+    fi->offset = 1;
+    fi->max = 1;
+    return fi;
+}
+
+int buf_read (struct buf_info *fi, char *dst)
+{
+    if (fi->offset >= fi->max)
+    {
+        if (fi->max <= 0)
+            return 0;
+        fi->max = (*fi->p->readf)(fi->p->fh, fi->buf, 4096);
+        fi->offset = 0;
+        if (fi->max <= 0)
+            return 0;
+    }
+    *dst = fi->buf[(fi->offset)++];
+    return 1;
+}
+
+void buf_close (struct buf_info *fi)
+{
+    xfree (fi->buf);
+    xfree (fi);
+}
+
 static int text_extract (struct recExtractCtrl *p)
 {
-    char w[256];
+    char w[512];
     RecWord recWord;
     int r, seqno = 1;
+    struct buf_info *fi = buf_open (p);
 
-    (*p->init)(&recWord);
-    recWord.which = Word_String;
+    (*p->init)(p, &recWord);
+    recWord.reg_type = 'w';
     do
     {
         int i = 0;
-
-        r = (*p->readf)(p->fh, w, 1);
-        while (r > 0 && i < 255 && isalnum(w[i]))
+            
+        r = buf_read (fi, w);
+        while (r > 0 && i < 511 && w[i] != '\n' && w[i] != '\r')
         {
             i++;
-            r = (*p->readf) (p->fh, w + i, 1);
+            r = buf_read (fi, w + i);
         }
         if (i)
         {
-            w[i] = 0;
             recWord.seqno = seqno++;
-            recWord.u.string = w;
+            recWord.string = w;
+           recWord.length = i;
             (*p->add)(&recWord);
         }
     } while (r > 0);
+    buf_close (fi);
     return 0;
 }
 
@@ -78,6 +137,12 @@ static int text_retrieve (struct recRetrieveCtrl *p)
     static char *text_buf = NULL;
     static int text_size = 0;
     int start_flag = 1;
+    const char *elementSetName = NULL;
+    int no_lines = 0;
+
+    if (p->comp && p->comp->which == Z_RecordComp_simple &&
+        p->comp->u.simple->which == Z_ElementSetNames_generic)
+        elementSetName = p->comp->u.simple->u.generic;
 
     while (1)
     {
@@ -110,6 +175,27 @@ static int text_retrieve (struct recRetrieveCtrl *p)
             break;
         text_ptr += r;
     }
+    text_buf[text_ptr] = '\0';
+    if (elementSetName)
+    {
+        if (!strcmp (elementSetName, "B"))
+            no_lines = 4;
+        if (!strcmp (elementSetName, "M"))
+            no_lines = 20;
+    }
+    if (no_lines)
+    {
+        char *p = text_buf;
+        int i = 0;
+
+        while (++i <= no_lines && (p = strchr (p, '\n')))
+            p++;
+        if (p)
+        {
+            p[1] = '\0';
+            text_ptr = p-text_buf;
+        }
+    }
     p->output_format = VAL_SUTRS;
     p->rec_buf = text_buf;
     p->rec_len = text_ptr;