ascii: Add support for multibyte characters.

[pspp-builds.git] / src / libpspp / str.c

diff --git a/src/libpspp/str.c b/src/libpspp/str.c
index e34c150df1c32506522155482c3285433c021c99..c630cf9ada761aa89b260197ca0718b1e9ff4344 100644 (file)
--- a/src/libpspp/str.c
+++ b/src/libpspp/str.c
@@ -347,15 +347,16 @@ ss_realloc (struct substring *ss, size_t size)
   ss->string = xrealloc (ss->string, size);
 }
 
-/* Makes a pool_alloc_unaligned()'d copy of the contents of OLD
-   in POOL, and stores it in NEW. */
+/* Makes a pool_alloc_unaligned()'d, null-terminated copy of the contents of
+   OLD in POOL, and stores it in NEW. */
 void
 ss_alloc_substring_pool (struct substring *new, struct substring old,
                          struct pool *pool)
 {
-  new->string = pool_alloc_unaligned (pool, old.length);
+  new->string = pool_alloc_unaligned (pool, old.length + 1);
   new->length = old.length;
   memcpy (new->string, old.string, old.length);
+  new->string[old.length] = '\0';
 }
 
 /* Allocates room for a CNT-byte string in NEW in POOL. */
@@ -427,6 +428,20 @@ ss_chomp_byte (struct substring *ss, char c)
     return false;
 }
 
+/* If SS ends with SUFFIX, removes it and returns true.
+   Otherwise, returns false without changing the string. */
+bool
+ss_chomp (struct substring *ss, struct substring suffix)
+{
+  if (ss_ends_with (*ss, suffix))
+    {
+      ss->length -= suffix.length;
+      return true;
+    }
+  else
+    return false;
+}
+
 /* Divides SS into tokens separated by any of the DELIMITERS.
    Each call replaces TOKEN by the next token in SS, or by an
    empty string if no tokens remain.  Returns true if a token was
@@ -655,6 +670,15 @@ ss_last (struct substring ss)
   return ss.length > 0 ? (unsigned char) ss.string[ss.length - 1] : EOF;
 }
 
+/* Returns true if SS ends with SUFFIX, false otherwise. */
+bool
+ss_ends_with (struct substring ss, struct substring suffix)
+{
+  return (ss.length >= suffix.length
+          && !memcmp (&ss.string[ss.length - suffix.length], suffix.string,
+                      suffix.length));
+}
+
 /* Returns the number of contiguous bytes at the beginning
    of SS that are in SKIP_SET. */
 size_t
@@ -1041,6 +1065,14 @@ ds_chomp_byte (struct string *st, char c)
   return ss_chomp_byte (&st->ss, c);
 }
 
+/* If ST ends with SUFFIX, removes it and returns true.
+   Otherwise, returns false without modifying ST. */
+bool
+ds_chomp (struct string *st, struct substring suffix)
+{
+  return ss_chomp (&st->ss, suffix);
+}
+
 /* Divides ST into tokens separated by any of the DELIMITERS.
    Each call replaces TOKEN by the next token in ST, or by an
    empty string if no tokens remain.  Returns true if a token was
@@ -1180,6 +1212,13 @@ ds_last (const struct string *st)
   return ss_last (ds_ss (st));
 }
 
+/* Returns true if ST ends with SUFFIX, false otherwise. */
+bool
+ds_ends_with (const struct string *st, struct substring suffix)
+{
+  return ss_ends_with (st->ss, suffix);
+}
+
 /* Returns the number of consecutive bytes at the beginning
    of ST that are in SKIP_SET. */
 size_t
@@ -1416,6 +1455,30 @@ ds_put_uninit (struct string *st, size_t incr)
   return end;
 }
 
+/* Moves the bytes in ST following offset OFS + OLD_LEN in ST to offset OFS +
+   NEW_LEN and returns the byte at offset OFS.  The first min(OLD_LEN, NEW_LEN)
+   bytes at the returned position are unchanged; if NEW_LEN > OLD_LEN then the
+   following NEW_LEN - OLD_LEN bytes are initially indeterminate.
+
+   The intention is that the caller should write NEW_LEN bytes at the returned
+   position, to effectively replace the OLD_LEN bytes previously at that
+   position. */
+char *
+ds_splice_uninit (struct string *st,
+                  size_t ofs, size_t old_len, size_t new_len)
+{
+  if (new_len != old_len)
+    {
+      if (new_len > old_len)
+        ds_extend (st, ds_length (st) + (new_len - old_len));
+      memmove (ds_data (st) + (ofs + new_len),
+               ds_data (st) + (ofs + old_len),
+               ds_length (st) - (ofs + old_len));
+      st->ss.length += new_len - old_len;
+    }
+  return ds_data (st) + ofs;
+}
+
 /* Formats FORMAT as a printf string and appends the result to ST. */
 void
 ds_put_format (struct string *st, const char *format, ...)