Re: Alan Hourihane: [bug #24687] implicit usage of mbsinit & mbrtowc

[Bruno Haible]

Hello Sergey,

> > This renders compilation of 4.13 inoperable on these systems, and you get 
> > the
> > build errors below....
> > 
> > display.o:display.o:(.text+0x786): undefined reference to `mbrtowc'
> > display.o:display.o:(.text+0x78c): undefined reference to `mbsinit'

The problem is that texinfo's info.h includes mbiter.h unconditionally,
whereas the module description in gnulib specifies this:

Include:
#if HAVE_MBRTOWC
#include "mbiter.h"
#endif

Now I see that you have 9 functions in 
texinfo/info/{display.c,session.c,window.c}
which use the mbiter facility unconditionally, and you probably have no
interest in adding fallback code for older systems for each of these functions.

I'll therefore extend the modules in a way that they work also on older
systems without HAVE_MBRTOWC.

Something like this. Probably also adding replacements for btowc and mbrtowc.

Bruno


--- modules/mbfile.orig 2008-10-30 12:42:19.000000000 +0100
+++ modules/mbfile      2008-10-30 11:22:00.000000000 +0100
@@ -18,9 +18,7 @@
 lib_SOURCES += mbfile.h
 
 Include:
-#if HAVE_MBRTOWC
 #include "mbfile.h"
-#endif
 
 License:
 LGPL
--- modules/mbiter.orig 2008-10-30 12:42:19.000000000 +0100
+++ modules/mbiter      2008-10-30 11:21:50.000000000 +0100
@@ -18,9 +18,7 @@
 lib_SOURCES += mbiter.h
 
 Include:
-#if HAVE_MBRTOWC
 #include "mbiter.h"
-#endif
 
 License:
 LGPL
--- modules/mbuiter.orig        2008-10-30 12:42:19.000000000 +0100
+++ modules/mbuiter     2008-10-30 11:21:40.000000000 +0100
@@ -19,9 +19,7 @@
 lib_SOURCES += mbuiter.h
 
 Include:
-#if HAVE_MBRTOWC
 #include "mbuiter.h"
-#endif
 
 License:
 LGPL
--- lib/mbfile.h.orig   2008-10-30 12:41:22.000000000 +0100
+++ lib/mbfile.h        2008-10-30 12:27:45.000000000 +0100
@@ -1,5 +1,5 @@
 /* Multibyte character I/O: macros for multi-byte encodings.
-   Copyright (C) 2001, 2005 Free Software Foundation, Inc.
+   Copyright (C) 2001, 2005, 2008 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -62,6 +62,42 @@
 
 #include "mbchar.h"
 
+/* ------------------ Implementation for unibyte locales ------------------ */
+
+typedef FILE * mbfile_unibyte;
+
+static inline void
+mbfile_unibyte_getc (struct mbchar *mbc, mbfile_unibyte mbf)
+{
+  int c = getc (mbf);
+  if (c != EOF)
+    {
+      mbc->buf[0] = (unsigned char) c;
+      mbc->ptr = &mbc->buf[0];
+      mbc->bytes = 1;
+      mbc->wc = btowc (c);
+      mbc->wc_valid = true;
+    }
+  else
+    {
+      /* An mbchar_t with bytes == 0 is used to indicate EOF.  */
+      mbc->ptr = NULL;
+      mbc->bytes = 0;
+      mbc->wc_valid = false;
+    }
+}
+
+static inline void
+mbfile_unibyte_ungetc (const struct mbchar *mbc, mbfile_unibyte mbf)
+{
+  if (mbc->bytes > 0)
+    ungetc ((unsigned char) mbc->ptr[0]);
+}
+
+/* ----------------- Implementation for multibyte locales ----------------- */
+
+#if HAVE_MBRTOWC
+
 struct mbfile_multi {
   FILE *fp;
   bool eof_seen;
@@ -222,6 +258,12 @@
   mbf->have_pushback = true;
 }
 
+#endif
+
+/* ------------------------------------------------------------------------ */
+
+#if HAVE_MBRTOWC /* systems with multibyte locales */
+
 typedef struct mbfile_multi mb_file_t;
 
 typedef mbchar_t mbf_char_t;
@@ -237,6 +279,20 @@
 
 #define mbf_ungetc(mbc, mbf) mbfile_multi_ungetc (&(mbc), &(mbf))
 
+#else /* systems without multibyte locales */
+
+typedef mbfile_unibyte mb_file_t;
+
+typedef mbchar_t mbf_char_t;
+
+#define mbf_init(mbf, stream) ((mbf) = (stream))
+
+#define mbf_getc(mbc, mbf) mbfile_unibyte_getc (&(mbc), (mbf))
+
+#define mbf_ungetc(mbc, mbf) mbfile_unibyte_ungetc (&(mbc), (mbf))
+
+#endif
+
 #define mb_iseof(mbc) ((mbc).bytes == 0)
 
 #endif /* _MBFILE_H */
--- lib/mbiter.h.orig   2008-10-30 12:41:22.000000000 +0100
+++ lib/mbiter.h        2008-10-30 12:27:51.000000000 +0100
@@ -1,5 +1,5 @@
 /* Iterating through multibyte strings: macros for multi-byte encodings.
-   Copyright (C) 2001, 2005, 2007 Free Software Foundation, Inc.
+   Copyright (C) 2001, 2005, 2007-2008 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -97,6 +97,31 @@
 
 #include "mbchar.h"
 
+/* ------------------ Implementation for unibyte locales ------------------ */
+
+struct mbiter_unibyte
+{
+  const char *limit;   /* pointer to end of string */
+  struct mbchar cur;   /* the current character:
+       const char *cur.ptr             pointer to current character
+       The following are only valid after mbi_avail.
+       size_t cur.bytes                number of bytes of current character
+       bool cur.wc_valid               true if wc is a valid wide character
+       wchar_t cur.wc                  if wc_valid: the current character
+       */
+};
+
+static inline void
+mbiter_unibyte_copy (struct mbiter_unibyte *new_iter, const struct 
mbiter_unibyte *old_iter)
+{
+  new_iter->limit = old_iter->limit;
+  mb_copy (&new_iter->cur, &old_iter->cur);
+}
+
+/* ----------------- Implementation for multibyte locales ----------------- */
+
+#if HAVE_MBRTOWC
+
 struct mbiter_multi
 {
   const char *limit;   /* pointer to end of string */
@@ -191,6 +216,12 @@
   mb_copy (&new_iter->cur, &old_iter->cur);
 }
 
+#endif
+
+/* ------------------------------------------------------------------------ */
+
+#if HAVE_MBRTOWC /* systems with multibyte locales */
+
 /* Iteration macros.  */
 typedef struct mbiter_multi mbi_iterator_t;
 #define mbi_init(iter, startptr, length) \
@@ -212,4 +243,29 @@
 /* Copying an iterator.  */
 #define mbi_copy mbiter_multi_copy
 
+#else /* systems without multibyte locales */
+
+/* Iteration macros.  */
+typedef struct mbiter_unibyte mbi_iterator_t;
+#define mbi_init(iter, startptr, length) \
+  ((iter).cur.ptr = (startptr), (iter).limit = (iter).cur.ptr + (length), \
+   (iter).cur.bytes = 1, (iter).cur.wc_valid = true)
+#define mbi_avail(iter) \
+  ((iter).cur.ptr < (iter).limit \
+   && (((iter).cur.wc = btowc (*(iter).cur.ptr)), true))
+#define mbi_advance(iter) \
+  ((iter).cur.ptr++)
+
+/* Access to the current character.  */
+#define mbi_cur(iter) (iter).cur
+#define mbi_cur_ptr(iter) (iter).cur.ptr
+
+/* Relocation.  */
+#define mbi_reloc(iter, ptrdiff) (void) ((iter).cur.ptr += (ptrdiff))
+
+/* Copying an iterator.  */
+#define mbi_copy mbiter_unibyte_copy
+
+#endif
+
 #endif /* _MBITER_H */
--- lib/mbuiter.h.orig  2008-10-30 12:41:23.000000000 +0100
+++ lib/mbuiter.h       2008-10-30 12:27:57.000000000 +0100
@@ -1,5 +1,5 @@
 /* Iterating through multibyte strings: macros for multi-byte encodings.
-   Copyright (C) 2001, 2005, 2007 Free Software Foundation, Inc.
+   Copyright (C) 2001, 2005, 2007-2008 Free Software Foundation, Inc.
 
    This program is free software: you can redistribute it and/or modify
    it under the terms of the GNU General Public License as published by
@@ -106,6 +106,29 @@
 #include "mbchar.h"
 #include "strnlen1.h"
 
+/* ------------------ Implementation for unibyte locales ------------------ */
+
+struct mbuiter_unibyte
+{
+  struct mbchar cur;   /* the current character:
+       const char *cur.ptr             pointer to current character
+       The following are only valid after mbui_avail.
+       size_t cur.bytes                number of bytes of current character
+       bool cur.wc_valid               true if wc is a valid wide character
+       wchar_t cur.wc                  if wc_valid: the current character
+       */
+};
+
+static inline void
+mbuiter_unibyte_copy (struct mbuiter_unibyte *new_iter, const struct 
mbuiter_unibyte *old_iter)
+{
+  mb_copy (&new_iter->cur, &old_iter->cur);
+}
+
+/* ----------------- Implementation for multibyte locales ----------------- */
+
+#if HAVE_MBRTOWC
+
 struct mbuiter_multi
 {
   bool in_shift;       /* true if next byte may not be interpreted as ASCII */
@@ -198,6 +221,12 @@
   mb_copy (&new_iter->cur, &old_iter->cur);
 }
 
+#endif
+
+/* ------------------------------------------------------------------------ */
+
+#if HAVE_MBRTOWC /* systems with multibyte locales */
+
 /* Iteration macros.  */
 typedef struct mbuiter_multi mbui_iterator_t;
 #define mbui_init(iter, startptr) \
@@ -219,4 +248,29 @@
 /* Copying an iterator.  */
 #define mbui_copy mbuiter_multi_copy
 
+#else /* systems without multibyte locales */
+
+/* Iteration macros.  */
+typedef struct mbuiter_unibyte mbui_iterator_t;
+#define mbui_init(iter, startptr) \
+  ((iter).cur.ptr = (startptr), \
+   (iter).cur.bytes = 1, (iter).cur.wc_valid = true)
+#define mbui_avail(iter) \
+  (*(iter).cur.ptr != '\0' \
+   && (((iter).cur.wc = btowc (*(iter).cur.ptr)), true))
+#define mbui_advance(iter) \
+  ((iter).cur.ptr++)
+
+/* Access to the current character.  */
+#define mbui_cur(iter) (iter).cur
+#define mbui_cur_ptr(iter) (iter).cur.ptr
+
+/* Relocation.  */
+#define mbui_reloc(iter, ptrdiff) (void) ((iter).cur.ptr += (ptrdiff))
+
+/* Copying an iterator.  */
+#define mbui_copy mbuiter_unibyte_copy
+
+#endif
+
 #endif /* _MBUITER_H */