maint: update the mbsalign module

* gl/lib/mbsalign.c (mbsalign):  Support the MBA_UNIBYTE_FALLBACK
flag which reverts to unibyte mode if one can't allocate memory
or if there are invalid multibyte characters present.
Note memory is no longer dynamically allocated in unibyte mode so
one can assume that mbsalign() will not return an error if this
flag is present.  Don't calculate twice, the number of spaces,
when centering.  Suppress a signed/unsigned comparison warning.
(ambsalign): A new wrapper function to dynamically allocate
the minimum memory required to hold the aligned string.
* gl/lib/mbsalign.h: Add the MBA_UNIBYTE_FALLBACK flag and
also document others that may be implemented in future.
(ambsalign): A prototype for the new wrapper.
* gl/tests/test-mbsalign.c (main): New test program.
* gl/modules/mbsalign-tests: A new index to reference the tests.
* .x-sc_program_name: Exclude test-mbsalign.c from this check.

2 files changed, 103 insertions, 29 deletions

diff --git a/gl/lib/mbsalign.c b/gl/lib/mbsalign.c
index be259568d..5b55ec2d0 100644
--- a/gl/lib/mbsalign.c
+++ b/gl/lib/mbsalign.c
@@ -32,6 +32,7 @@
 #endif
 
 /* Replace non printable chars.
+   Note \t and \n etc. are non printable.
    Return 1 if replacement made, 0 otherwise.  */
 
 static bool
@@ -119,17 +120,17 @@ mbs_align_pad (char *dest, const char* dest_end, size_t n_spaces)
    ALIGNMENT specifies whether to left- or right-justify or to center.
    If SRC requires more than *WIDTH columns, truncate it to fit.
    When centering, the number of trailing spaces may be one less than the
-   number of leading spaces. The FLAGS parameter is unused at present.
+   number of leading spaces.
    Return the length in bytes required for the final result, not counting
    the trailing NUL.  A return value of DEST_SIZE or larger means there
    wasn't enough space.  DEST will be NUL terminated in any case.
    Return (size_t) -1 upon error (invalid multi-byte sequence in SRC,
-   or malloc failure).
+   or malloc failure), unless MBA_UNIBYTE_FALLBACK is specified.
    Update *WIDTH to indicate how many columns were used before padding.  */
 
 size_t
 mbsalign (const char *src, char *dest, size_t dest_size,
-          size_t *width, mbs_align_t align, int flags _UNUSED_PARAMETER_)
+          size_t *width, mbs_align_t align, int flags)
 {
   size_t ret = -1;
   size_t src_size = strlen (src) + 1;
@@ -149,12 +150,22 @@ mbsalign (const char *src, char *dest, size_t dest_size,
     {
       size_t src_chars = mbstowcs (NULL, src, 0);
       if (src_chars == (size_t) -1)
-        goto mbsalign_cleanup;
+        {
+          if (flags & MBA_UNIBYTE_FALLBACK)
+            goto mbsalign_unibyte;
+          else
+            goto mbsalign_cleanup;
+        }
       src_chars += 1; /* make space for NUL */
       str_wc = malloc (src_chars * sizeof (wchar_t));
       if (str_wc == NULL)
-        goto mbsalign_cleanup;
-      if (mbstowcs (str_wc, src, src_chars) > 0)
+        {
+          if (flags & MBA_UNIBYTE_FALLBACK)
+            goto mbsalign_unibyte;
+          else
+            goto mbsalign_cleanup;
+        }
+      if (mbstowcs (str_wc, src, src_chars) != 0)
         {
           str_wc[src_chars - 1] = L'\0';
           wc_enabled = true;
@@ -165,27 +176,30 @@ mbsalign (const char *src, char *dest, size_t dest_size,
 
   /* If we transformed or need to truncate the source string
      then create a modified copy of it.  */
-  if (conversion || (n_cols > *width))
+  if (wc_enabled && (conversion || (n_cols > *width)))
     {
-      newstr = malloc (src_size);
-      if (newstr == NULL)
-        goto mbsalign_cleanup;
-      str_to_print = newstr;
-      if (wc_enabled)
+        newstr = malloc (src_size);
+        if (newstr == NULL)
         {
-          n_cols = wc_truncate (str_wc, *width);
-          n_used_bytes = wcstombs (newstr, str_wc, src_size);
-        }
-      else
-        {
-          n_cols = *width;
-          n_used_bytes = n_cols;
-          memcpy (newstr, src, n_cols);
-          newstr[n_cols] = '\0';
+          if (flags & MBA_UNIBYTE_FALLBACK)
+            goto mbsalign_unibyte;
+          else
+            goto mbsalign_cleanup;
         }
+        str_to_print = newstr;
+        n_cols = wc_truncate (str_wc, *width);
+        n_used_bytes = wcstombs (newstr, str_wc, src_size);
+    }
+
+mbsalign_unibyte:
+
+  if (n_cols > *width) /* Unibyte truncation required.  */
+    {
+      n_cols = *width;
+      n_used_bytes = n_cols;
     }
 
-  if (*width > n_cols)
+  if (*width > n_cols) /* Padding required.  */
     n_spaces = *width - n_cols;
 
   /* indicate to caller how many cells needed (not including padding).  */
@@ -197,16 +211,11 @@ mbsalign (const char *src, char *dest, size_t dest_size,
   /* Write as much NUL terminated output to DEST as possible.  */
   if (dest_size != 0)
     {
+      size_t start_spaces, end_spaces;
       char *dest_end = dest + dest_size - 1;
-      size_t start_spaces = n_spaces / 2 + n_spaces % 2;
-      size_t end_spaces = n_spaces / 2;
 
       switch (align)
         {
-        case MBS_ALIGN_CENTER:
-          start_spaces = n_spaces / 2 + n_spaces % 2;
-          end_spaces = n_spaces / 2;
-          break;
         case MBS_ALIGN_LEFT:
           start_spaces = 0;
           end_spaces = n_spaces;
@@ -215,10 +224,16 @@ mbsalign (const char *src, char *dest, size_t dest_size,
           start_spaces = n_spaces;
           end_spaces = 0;
           break;
+        case MBS_ALIGN_CENTER:
+        default:
+          start_spaces = n_spaces / 2 + n_spaces % 2;
+          end_spaces = n_spaces / 2;
+          break;
         }
 
       dest = mbs_align_pad (dest, dest_end, start_spaces);
-      dest = mempcpy(dest, str_to_print, MIN (n_used_bytes, dest_end - dest));
+      size_t space_left = dest_end - dest;
+      dest = mempcpy (dest, str_to_print, MIN (n_used_bytes, space_left));
       mbs_align_pad (dest, dest_end, end_spaces);
     }
 
@@ -229,3 +244,39 @@ mbsalign_cleanup:
 
   return ret;
 }
+
+/* A wrapper around mbsalign() to dynamically allocate the
+   minimum amount of memory to store the result.
+   Return NULL on failure.  */
+
+char *
+ambsalign (const char *src, size_t *width, mbs_align_t align, int flags)
+{
+  size_t orig_width = *width;
+  size_t size = *width;         /* Start with enough for unibyte mode.  */
+  size_t req = size;
+  char *buf = NULL;
+
+  while (req >= size)
+    {
+      size = req + 1;           /* Space for NUL.  */
+      char *nbuf = realloc (buf, size);
+      if (nbuf == NULL)
+        {
+          free (buf);
+          buf = NULL;
+          break;
+        }
+      buf = nbuf;
+      *width = orig_width;
+      req = mbsalign (src, buf, size, width, align, flags);
+      if (req == (size_t) -1)
+        {
+          free (buf);
+          buf = NULL;
+          break;
+        }
+    }
+
+  return buf;
+}
diff --git a/gl/lib/mbsalign.h b/gl/lib/mbsalign.h
index a4ec69395..41bd49092 100644
--- a/gl/lib/mbsalign.h
+++ b/gl/lib/mbsalign.h
@@ -18,6 +18,29 @@
 
 typedef enum { MBS_ALIGN_LEFT, MBS_ALIGN_RIGHT, MBS_ALIGN_CENTER } mbs_align_t;
 
+enum {
+  /* Use unibyte mode for invalid multibyte strings or
+     or when heap memory is exhausted.  */
+  MBA_UNIBYTE_FALLBACK = 0x0001,
+
+#if 0 /* Other possible options.  */
+  /* Skip invalid multibyte chars rather than failing  */
+  MBA_IGNORE_INVALID   = 0x0002,
+
+  /* Align multibyte strings using "figure space" (\u2007)  */
+  MBA_USE_FIGURE_SPACE = 0x0004,
+
+  /* Don't add any padding  */
+  MBA_TRUNCATE_ONLY    = 0x0008,
+
+  /* Don't truncate  */
+  MBA_PAD_ONLY         = 0x0010,
+#endif
+};
+
 size_t
 mbsalign (const char *src, char *dest, size_t dest_size,
           size_t *width, mbs_align_t align, int flags);
+
+char *
+ambsalign (const char *src, size_t *width, mbs_align_t align, int flags);