Update m4

[idzebra-moved-to-github.git] / util / charmap.c

diff --git a/util/charmap.c b/util/charmap.c
index 85c9239..c67cae9 100644 (file)
--- a/util/charmap.c
+++ b/util/charmap.c
@@ -1,8 +1,5 @@
-/* $Id: charmap.c,v 1.45 2007-05-25 13:46:01 adam Exp $
-   Copyright (C) 1995-2007
-   Index Data ApS
-
-This file is part of the Zebra server.
+/* This file is part of the Zebra server.
+   Copyright (C) 1994-2011 Index Data
 
 Zebra is free software; you can redistribute it and/or modify it under
 the terms of the GNU General Public License as published by the Free
@@ -21,12 +18,17 @@ Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */
 
 
-
-/*
+/**
+ * \file charmap.c
+ * \brief character conversions (.chr)
+ *  
  * Support module to handle character-conversions into and out of the
  * Zebra dictionary.
  */
 
+#if HAVE_CONFIG_H
+#include <config.h>
+#endif
 #include <ctype.h>
 #include <string.h>
 #include <stdlib.h>
@@ -136,25 +138,6 @@ static chr_t_entry *set_map_string(chr_t_entry *root, NMEM nmem,
     return root;
 }
 
-static chr_t_entry *find_entry(chr_t_entry *t, const char **from, int len)
-{
-    chr_t_entry *res;
-
-    if (len && t->children && t->children[(unsigned char) **from])
-    {
-       const char *pos = *from;
-
-       (*from)++;
-       if ((res = find_entry(t->children[(unsigned char) *pos],
-                              from, len - 1)))
-           return res;
-       /* no match */
-       *from = pos;
-    }
-    /* no children match. use ourselves, if we have a target */
-    return t->target ? t : 0;
-}
-
 static chr_t_entry *find_entry_x(chr_t_entry *t, const char **from, int *len, int first)
 {
     chr_t_entry *res;
@@ -259,7 +242,6 @@ ucs4_t zebra_prim_w(ucs4_t **s)
     ucs4_t i = 0;
     char fmtstr[8];
 
-    yaz_log(YLOG_DEBUG, "prim_w %.3s", (char *) *s);
     if (**s == '\\' && 1[*s])
     {
        (*s)++;
@@ -420,6 +402,10 @@ static int scan_to_utf8(yaz_iconv_t t, ucs4_t *from, size_t inlen,
     else
     {
         ret = yaz_iconv(t, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+        if (ret != (size_t) (-1))
+            ret = yaz_iconv(t, 0, 0, &outbuf, &outbytesleft);
+        
+            
         if (ret == (size_t) (-1))
         {
            yaz_log(YLOG_LOG, "from: %2X %2X %2X %2X",
@@ -441,7 +427,7 @@ static int scan_string(char *s_native,
 
     ucs4_t arg[512];
     ucs4_t arg_prim[512];
-    ucs4_t *s0, *s = arg;
+    ucs4_t *s = arg;
     ucs4_t c, begin, end;
     size_t i;
 
@@ -454,6 +440,9 @@ static int scan_string(char *s_native,
         size_t ret;            
        ret = yaz_iconv(t_unicode, &inbuf, &inbytesleft,
                         &outbuf, &outbytesleft);
+        if (ret != (size_t)(-1))
+            ret = yaz_iconv(t_unicode, 0, 0, &outbuf, &outbytesleft);
+            
         if (ret == (size_t)(-1))
             return -1;
         i = (outbuf - (char*) arg)/sizeof(ucs4_t);
@@ -495,9 +484,16 @@ static int scan_string(char *s_native,
            break;
        case '(':
             ++s;
-           s0 = s; i = 0;
+           i = 0;
            while (*s != ')' || s[-1] == '\\')
+            {
+                if (*s == '\0')
+                {
+                    yaz_log(YLOG_FATAL, "Missing ) in charmap");
+                    return -1;
+                }
                arg_prim[i++] = zebra_prim_w(&s);
+            }
            arg_prim[i] = 0;
             if (scan_to_utf8(t_utf8, arg_prim, zebra_ucs4_strlen(arg_prim), str, sizeof(str)-1))
                 return -1;
@@ -521,6 +517,7 @@ chrmaptab chrmaptab_create(const char *tabpath, const char *name,
     char line[512], *argv[50];
     chrmaptab res;
     int lineno = 0;
+    int no_directives = 0;
     int errors = 0;
     int argc, num = (int) *CHR_BASE, i;
     NMEM nmem;
@@ -573,6 +570,8 @@ chrmaptab chrmaptab_create(const char *tabpath, const char *name,
     res->base_uppercase = 0;
 
     while (!errors && (argc = readconf_line(f, &lineno, line, 512, argv, 50)))
+    {
+        no_directives++;
        if (!yaz_matchstr(argv[0], "lowercase"))
        {
            if (argc != 2)
@@ -717,41 +716,28 @@ chrmaptab chrmaptab_create(const char *tabpath, const char *name,
        }
         else if (!yaz_matchstr(argv[0], "encoding"))
         {
-           /*
-            * Fix me. When t_unicode==0 and use encoding directive in *.chr file the beheviour of the
-            * zebra need to comment next part of code.
-            */
-
-           /* Original code */
-#if 1
             if (t_unicode != 0)
                 yaz_iconv_close(t_unicode);
             t_unicode = yaz_iconv_open(ucs4_native, argv[1]);
-#endif
-           /*
-            * Fix me. It is additional staff for conversion of characters from local encoding
-            * of *.chr file to UTF-8 (internal encoding).
-            * NOTE: The derective encoding must be first directive in *.chr file.
-            */
-           /* For whatever reason Oleg enabled this.. */
-#if 0
-           if (t_utf8 != 0)
-               yaz_iconv_close(t_utf8);
-           t_utf8 = yaz_iconv_open("UTF-8", argv[1]);
-#endif
         }
        else
        {
            yaz_log(YLOG_WARN, "Syntax error at '%s' in %s", line, name);
+            errors++;
        }
-    
+    }   
     yaz_fclose(f);
+    if (no_directives == 0)
+    {
+        yaz_log(YLOG_WARN, "No directives in '%s'", name);
+        errors++;
+    }
     if (errors)
     {
        chrmaptab_destroy(res);
        res = 0;
     }
-    yaz_log(YLOG_DEBUG, "maptab %s close %d errors", name, errors);
+    yaz_log(YLOG_DEBUG, "maptab %s num=%d close %d errors", name, num, errors);
     if (t_utf8 != 0)
         yaz_iconv_close(t_utf8);
     if (t_unicode != 0)
@@ -769,6 +755,7 @@ void chrmaptab_destroy(chrmaptab tab)
 /*
  * Local variables:
  * c-basic-offset: 4
+ * c-file-style: "Stroustrup"
  * indent-tabs-mode: nil
  * End:
  * vim: shiftwidth=4 tabstop=8 expandtab