Fix MARC8 conversion

diff --git a/CHANGELOG b/CHANGELOG
index d7331c2..dcf97c0 100644 (file)
--- a/CHANGELOG
+++ b/CHANGELOG
@@ -1,5 +1,7 @@
 Possible compatibility problems with earlier versions marked with '*'.
 
 Possible compatibility problems with earlier versions marked with '*'.
 

+Bug fix: some MARC8 sequences were not converted.
+

 New ZOOM option "step" which specifies number of records to be
 retrieved in one chunk. Used in conjunction with "start" and "count".
 
 New ZOOM option "step" which specifies number of records to be
 retrieved in one chunk. Used in conjunction with "start" and "count".
 

diff --git a/util/Makefile.am b/util/Makefile.am
index 8fb85d8..3d72fc8 100644 (file)
--- a/util/Makefile.am
+++ b/util/Makefile.am
@@ -1,10 +1,12 @@
 ## Copyright (C) 1994-2003, Index Data
 ## All rights reserved.
 ## Copyright (C) 1994-2003, Index Data
 ## All rights reserved.

-## $Id: Makefile.am,v 1.18 2003-05-06 10:07:33 adam Exp $
+## $Id: Makefile.am,v 1.19 2003-05-22 22:44:50 adam Exp $

 
 noinst_LTLIBRARIES = libutil.la
 
 
 noinst_LTLIBRARIES = libutil.la
 

-TESTS = tsticonv tstnmem tstmatchstr tstwrbuf
+check_PROGRAMS = tsticonv tstnmem tstmatchstr tstwrbuf
+
+TESTS = $(check_PROGRAMS)

 
 bin_SCRIPTS = yaz-comp
 
 
 bin_SCRIPTS = yaz-comp
 

@@ -14,8 +16,6 @@ AM_CPPFLAGS=-I$(top_srcdir)/include
 
 noinst_PROGRAMS = marcdump yaziconv
 
 
 noinst_PROGRAMS = marcdump yaziconv
 

-EXTRA_PROGRAMS = tsticonv tstnmem tstmatchstr tstwrbuf
-

 # MARC dumper utility
 marcdump_LDADD = libutil.la 
 marcdump_SOURCES = marcdump.c
 # MARC dumper utility
 marcdump_LDADD = libutil.la 
 marcdump_SOURCES = marcdump.c

diff --git a/util/charconv.tcl b/util/charconv.tcl
index befad0f..ae4e15e 100755 (executable)
--- a/util/charconv.tcl
+++ b/util/charconv.tcl
@@ -2,7 +2,7 @@
 # the next line restats using tclsh \
 exec tclsh "$0" "$@"
 #
 # the next line restats using tclsh \
 exec tclsh "$0" "$@"
 #

-# $Id: charconv.tcl,v 1.1 2002-12-16 13:13:53 adam Exp $
+# $Id: charconv.tcl,v 1.2 2003-05-22 22:44:50 adam Exp $

 
 proc usage {} {
     puts {charconv.tcl: [-p prefix] [-s split] input output}
 
 proc usage {} {
     puts {charconv.tcl: [-p prefix] [-s split] input output}

@@ -103,7 +103,7 @@ proc dump_trie {ofile} {
             foreach m $trie($this,content) {
                 puts -nonewline $f "  \{\""
                 foreach d [lindex $m 0] {
             foreach m $trie($this,content) {
                 puts -nonewline $f "  \{\""
                 foreach d [lindex $m 0] {

-                    puts -nonewline $f "\\0x$d"
+                    puts -nonewline $f "\\x$d"

                 }
                 puts -nonewline $f "\", 0x[lindex $m 1]"
                 puts $f "\},"
                 }
                 puts -nonewline $f "\", 0x[lindex $m 1]"
                 puts $f "\},"

@@ -175,7 +175,7 @@ proc dump_trie {ofile} {
                 while (flat->from)
                 {
                     size_t len = strlen(flat->from);
                 while (flat->from)
                 {
                     size_t len = strlen(flat->from);

-                    if (len >= inbytesleft)
+                    if (len <= inbytesleft)

                     {
                         if (memcmp(flat->from, inp, len) == 0)
                         {
                     {
                         if (memcmp(flat->from, inp, len) == 0)
                         {

diff --git a/util/tsticonv.c b/util/tsticonv.c
index 8f4653f..f5b71dc 100644 (file)
--- a/util/tsticonv.c
+++ b/util/tsticonv.c
@@ -2,7 +2,7 @@
  * Copyright (c) 2002-2003, Index Data
  * See the file LICENSE for details.
  *
  * Copyright (c) 2002-2003, Index Data
  * See the file LICENSE for details.
  *

- * $Id: tsticonv.c,v 1.2 2003-05-06 10:07:33 adam Exp $
+ * $Id: tsticonv.c,v 1.3 2003-05-22 22:44:50 adam Exp $

  */
 
 #if HAVE_CONFIG_H
  */
 
 #if HAVE_CONFIG_H

@@ -17,10 +17,58 @@
 
 /* some test strings in ISO-8859-1 format */
 const char *buf[] = {
 
 /* some test strings in ISO-8859-1 format */
 const char *buf[] = {

-       "ax" ,
-       "\330",
-       "eneb\346r",
-               0 };
+    "ax" ,
+    "\330",
+    "eneb\346r",
+    "\xfc",
+    "\xfb",
+    "\xfbr",
+    0 };
+
+/* some test strings in MARC-8 format */
+const char *marc8_strings[] = {
+    "ax",   
+    "\xa2",          /* latin capital letter o with stroke */
+    "eneb\xb5r",     /* latin small letter ae */
+    "\xe8\x75",      /* latin small letter u with umlaut */
+    "\xe3\x75",      /* latin small letter u with circumflex */
+    "\xe3\x75r",     /* latin small letter u with circumflex */
+    0
+};
+
+static marc8_tst()
+{
+    int i;
+    yaz_iconv_t cd;
+
+    cd = yaz_iconv_open("ISO-8859-1", "MARC8");
+    for (i = 0; buf[i]; i++)
+    {
+        size_t r;
+        char *inbuf= (char*) marc8_strings[i];
+        size_t inbytesleft = strlen(inbuf);
+        char outbuf0[24];
+        char *outbuf = outbuf0;
+        size_t outbytesleft = sizeof(outbuf0);
+
+        r = yaz_iconv(cd, &inbuf, &inbytesleft, &outbuf, &outbytesleft);
+        if (r == (size_t) (-1))
+        {
+            int e = yaz_iconv_error(cd);
+
+            printf ("tsticonv 6 i=%d e=%d\n", i, e);
+           exit(6);
+        }
+        if ((outbuf - outbuf0) != strlen(buf[i]) 
+            || memcmp(outbuf0, buf[i], strlen(buf[i])))
+        {
+            printf ("tsticonv 7 i=%d\n", i);
+            printf ("buf=%s   out=%s\n", buf[i], outbuf0);
+           exit(7);
+        }
+    }
+    yaz_iconv_close(cd);
+}

 
 static dconvert(int mandatory, const char *tmpcode)
 {
 
 static dconvert(int mandatory, const char *tmpcode)
 {

@@ -91,5 +139,6 @@ int main (int argc, char **argv)
     dconvert(1, "ISO-8859-1");
     dconvert(1, "UCS4");
     dconvert(0, "CP865");
     dconvert(1, "ISO-8859-1");
     dconvert(1, "UCS4");
     dconvert(0, "CP865");

+    marc8_tst();

     exit (0);
 }
     exit (0);
 }