From 9fb32b9ec35a83d916e0227c751d4a7925301d34 Mon Sep 17 00:00:00 2001
From: Adam Dickmeiss <adam@indexdata.dk>
Date: Wed, 5 Mar 2008 21:21:22 +0000
Subject: [PATCH] For MARC-8 decoding avoid messing up G0 when G1 is selected.

Use two flags for G0/G1 selection so that selection of G1 does not
conflict with selection of G0. Bug #2115. Note that G1 is not
properly handled. We must add &127 for G1 decoding since those
characters have bit 7 set. A flag must be parsed to the conversion
routines probably.
---
 src/siconv.c    |   31 ++++++++++++++++++++++++-------
 test/tsticonv.c |   10 ++++++++--
 2 files changed, 32 insertions(+), 9 deletions(-)

diff --git a/src/siconv.c b/src/siconv.c
index 663d03a..e7a0de0 100644
--- a/src/siconv.c
+++ b/src/siconv.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: siconv.c,v 1.48 2007-10-15 20:45:05 adam Exp $
+ * $Id: siconv.c,v 1.49 2008-03-05 21:21:22 adam Exp $
  */
 /**
  * \file siconv.c
@@ -100,7 +100,8 @@ struct yaz_iconv_struct {
                            char **outbuf, size_t *outbytesleft);
     size_t (*flush_handle)(yaz_iconv_t cd,
                            char **outbuf, size_t *outbytesleft);
-    int marc8_esc_mode;
+    int g0_mode;
+    int g1_mode;
 
     int comb_offset;
     int comb_size;
@@ -1168,22 +1169,36 @@ static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp,
     *no_read = 0;
     while(inbytesleft >= 1 && inp[0] == 27)
     {
+        int ch;
         size_t inbytesleft0 = inbytesleft;
         inp++;
         inbytesleft--;
-        while(inbytesleft > 0 && strchr("(,$!)-", *inp))
+        if (inbytesleft > 0 && *inp == '$')
         {
             inbytesleft--;
             inp++;
         }
-        if (inbytesleft <= 0)
+        if (inbytesleft <= 1)
         {
             *no_read = 0;
             cd->my_errno = YAZ_ICONV_EINVAL;
             return 0;
         }
-        cd->marc8_esc_mode = *inp++;
         inbytesleft--;
+        ch = *inp++;
+        if (inbytesleft > 0 && (ch == '(' || ch == ','))
+        {
+            inbytesleft--;
+            cd->g0_mode = *inp++;
+        }
+        else if (inbytesleft > 0 && (ch == ')' || ch == '-'))
+        {
+            inbytesleft--;
+            cd->g1_mode = *inp++;
+        }
+        else
+            cd->g0_mode = ch;
+
         (*no_read) += inbytesleft0 - inbytesleft;
     }
     if (inbytesleft <= 0)
@@ -1197,9 +1212,10 @@ static unsigned long yaz_read_marc8_comb(yaz_iconv_t cd, unsigned char *inp,
     {
         unsigned long x;
         size_t no_read_sub = 0;
+        int mode = *inp < 128 ? cd->g0_mode : cd->g1_mode;
         *comb = 0;
 
-        switch(cd->marc8_esc_mode)
+        switch(mode)
         {
         case 'B':  /* Basic ASCII */
         case 's':  /* ASCII */
@@ -1896,7 +1912,8 @@ size_t yaz_iconv(yaz_iconv_t cd, char **inbuf, size_t *inbytesleft,
     if (cd->init_flag)
     {
         cd->my_errno = YAZ_ICONV_UNKNOWN;
-        cd->marc8_esc_mode = 'B';
+        cd->g0_mode = 'B';
+        cd->g1_mode = 'B';
         
         cd->comb_offset = cd->comb_size = 0;
         cd->compose_char = 0;
diff --git a/test/tsticonv.c b/test/tsticonv.c
index 843df37..b39923b 100644
--- a/test/tsticonv.c
+++ b/test/tsticonv.c
@@ -2,7 +2,7 @@
  * Copyright (C) 1995-2007, Index Data ApS
  * See the file LICENSE for details.
  *
- * $Id: tsticonv.c,v 1.33 2007-11-16 08:01:47 adam Exp $
+ * $Id: tsticonv.c,v 1.34 2008-03-05 21:21:22 adam Exp $
  */
 
 #if HAVE_CONFIG_H
@@ -17,6 +17,8 @@
 #include <yaz/yaz-util.h>
 #include <yaz/test.h>
 
+#define ESC "\x1b"
+
 static int compare_buffers(char *msg, int no,
                            int expect_len, const char *expect_buf,
                            int got_len, const char *got_buf)
@@ -386,16 +388,20 @@ static void tst_marc8_to_utf8(void)
     if (!cd)
         return;
 
+    /* bug #2115 */
+    YAZ_CHECK(tst_convert(cd, ESC "(N" ESC ")Qp" ESC "(B", "\xd0\x9f"));
+
+
     YAZ_CHECK(tst_convert(cd, "Cours de math", 
                           "Cours de math"));
     /* COMBINING ACUTE ACCENT */
     YAZ_CHECK(tst_convert(cd, "Cours de mathâe", 
                           "Cours de mathe\xcc\x81"));
 
-
     YAZ_CHECK(tst_convert(cd, "a\xea\x1e", "a\x1e\xcc\x8a"));
 
     YAZ_CHECK(tst_convert(cd, "a\xea", "a"));
+
     yaz_iconv_close(cd);
 }
 
-- 
1.7.10.4