Rewrite system file reader code, to clean up and improve.

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

diff --git a/src/libpspp/str.c b/src/libpspp/str.c
index cf0069dd3d4e46e5ba292cdbb352fd4d4b89e48b..183a4022cc1280fd4f9664fb427c5da9a095f6a3 100644 (file)
--- a/src/libpspp/str.c
+++ b/src/libpspp/str.c
@@ -158,6 +158,22 @@ buf_copy_str_lpad (char *dst, size_t dst_size, const char *src)
     }
 }
 
+/* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
+   DST is truncated to DST_SIZE bytes or padded on the left with
+   spaces as needed. */
+void
+buf_copy_lpad (char *dst, size_t dst_size,
+               const char *src, size_t src_size)
+{
+  if (src_size >= dst_size)
+    memmove (dst, src, dst_size);
+  else
+    {
+      memset (dst, ' ', dst_size - src_size);
+      memmove (&dst[dst_size - src_size], src, src_size);
+    }
+}
+
 /* Copies buffer SRC, of SRC_SIZE bytes, to DST, of DST_SIZE bytes.
    DST is truncated to DST_SIZE bytes or padded on the right with
    spaces as needed. */
@@ -253,6 +269,15 @@ spprintf (char *dst, const char *format, ...)
 
   return dst + count;
 }
+
+/* Sets the SIZE bytes starting at BLOCK to C,
+   and returns the byte following BLOCK. */
+void *
+mempset (void *block, int c, size_t size) 
+{
+  memset (block, c, size);
+  return (char *) block + size;
+}
 \f
 /* Substrings. */
 
@@ -333,6 +358,25 @@ ss_alloc_uninit (struct substring *new, size_t cnt)
   new->length = cnt;
 }
 
+/* Makes a pool_alloc_unaligned()'d 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->length = old.length;
+  memcpy (new->string, old.string, old.length);
+}
+
+/* Allocates room for a CNT-character string in NEW in POOL. */
+void
+ss_alloc_uninit_pool (struct substring *new, size_t cnt, struct pool *pool) 
+{
+  new->string = pool_alloc_unaligned (pool, cnt);
+  new->length = cnt;
+}
+
 /* Frees the string that SS points to. */
 void
 ss_dealloc (struct substring *ss) 
@@ -428,7 +472,8 @@ ss_separate (struct substring ss, struct substring delimiters,
 /* Divides SS into tokens separated by any of the DELIMITERS,
    merging adjacent delimiters so that the empty string is never
    produced as a token.  Each call replaces TOKEN by the next
-   token in SS, or by an empty string if no tokens remain.
+   token in SS, or by an empty string if no tokens remain, and
+   then skips past the first delimiter following the token.
    Returns true if a token was obtained, false otherwise.
 
    Before the first call, initialize *SAVE_IDX to 0.  Do not
@@ -439,7 +484,8 @@ ss_tokenize (struct substring ss, struct substring delimiters,
 {
   ss_advance (&ss, *save_idx);
   *save_idx += ss_ltrim (&ss, delimiters);
-  *save_idx += ss_get_chars (&ss, ss_cspan (ss, delimiters), token);
+  ss_get_chars (&ss, ss_cspan (ss, delimiters), token);
+  *save_idx += ss_length (*token) + 1;
   return ss_length (*token) > 0;
 }
 
@@ -468,6 +514,21 @@ ss_match_char (struct substring *ss, char c)
     return false;
 }
 
+/* If SS begins with TARGET, removes it and returns true.
+   Otherwise, returns false without changing SS. */
+bool
+ss_match_string (struct substring *ss, const struct substring target)
+{
+  size_t length = ss_length (target);
+  if (ss_equals (ss_head (*ss, length), target))
+    {
+      ss_advance (ss, length);
+      return true;
+    }
+  else
+    return false;
+}
+
 /* Removes the first character from SS and returns it.
    If SS is empty, returns EOF without modifying SS. */
 int
@@ -504,6 +565,36 @@ ss_get_chars (struct substring *ss, size_t cnt, struct substring *out)
   return cnt;
 }
 
+/* Parses and removes an optionally signed decimal integer from
+   the beginning of SS.  Returns 0 if an error occurred,
+   otherwise the number of characters removed from SS.  Stores
+   the integer's value into *VALUE. */
+size_t
+ss_get_long (struct substring *ss, long *value) 
+{
+  char tmp[64];
+  size_t length;
+
+  length = ss_span (*ss, ss_cstr ("+-"));
+  length += ss_span (ss_substr (*ss, length, SIZE_MAX), ss_cstr (CC_DIGITS));
+  if (length > 0 && length < sizeof tmp) 
+    {
+      char *tail;
+
+      memcpy (tmp, ss_data (*ss), length);
+      tmp[length] = '\0';
+
+      *value = strtol (tmp, &tail, 10);
+      if (tail - tmp == length) 
+        {
+          ss_advance (ss, length);
+          return length;
+        }
+    }
+  *value = 0;
+  return 0;
+}
+
 /* Returns true if SS is empty (contains no characters),
    false otherwise. */
 bool
@@ -602,6 +693,34 @@ ss_compare (struct substring a, struct substring b)
   return retval;
 }
 
+/* Compares A and B case-insensitively and returns a
+   strcmp()-type comparison result. */
+int
+ss_compare_case (struct substring a, struct substring b)
+{
+  int retval = memcasecmp (a.string, b.string, MIN (a.length, b.length));
+  if (retval == 0)
+    retval = a.length < b.length ? -1 : a.length > b.length;
+  return retval;
+}
+
+/* Compares A and B and returns true if their contents are
+   identical, false otherwise. */
+int
+ss_equals (struct substring a, struct substring b)
+{
+  return a.length == b.length && !memcmp (a.string, b.string, a.length);
+}
+
+/* Compares A and B and returns true if their contents are
+   identical except possibly for case differences, false
+   otherwise. */
+int
+ss_equals_case (struct substring a, struct substring b)
+{
+  return a.length == b.length && !memcasecmp (a.string, b.string, a.length);
+}
+
 /* Returns the position in SS that the character at P occupies.
    P must point within SS or one past its end. */
 size_t