stringi-set: New functions for not necessarily null terminated strings.

[pspp] / src / libpspp / stringi-set.c

diff --git a/src/libpspp/stringi-set.c b/src/libpspp/stringi-set.c
index a7ae6994711606eb0c5a2e5acec67ecb124ea71c..a609150d9930b572bd367d311eaa4abb100a1338 100644 (file)
--- a/src/libpspp/stringi-set.c
+++ b/src/libpspp/stringi-set.c
@@ -1,5 +1,5 @@
 /* PSPP - a program for statistical analysis.
-   Copyright (C) 2009, 2010 Free Software Foundation, Inc.
+   Copyright (C) 2009, 2010, 2012 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
@@ -26,11 +26,14 @@
 
 #include "libpspp/cast.h"
 #include "libpspp/hash-functions.h"
+#include "libpspp/i18n.h"
 
 #include "gl/xalloc.h"
 
 static struct stringi_set_node *stringi_set_find_node__ (
   const struct stringi_set *, const char *, unsigned int hash);
+static struct stringi_set_node *stringi_set_find_node_len__ (
+  const struct stringi_set *, const char *, size_t length, unsigned int hash);
 static void stringi_set_insert__ (struct stringi_set *, char *,
                                  unsigned int hash);
 static bool stringi_set_delete__ (struct stringi_set *, const char *,
@@ -80,7 +83,16 @@ stringi_set_destroy (struct stringi_set *set)
 bool
 stringi_set_contains (const struct stringi_set *set, const char *s)
 {
-  return stringi_set_find_node (set, s) != NULL;
+  return stringi_set_contains_len (set, s, strlen (s));
+}
+
+/* Returns true if SET contains S with the given LENGTH (or a similar string
+   with different case), false otherwise. */
+bool
+stringi_set_contains_len (const struct stringi_set *set, const char *s,
+                          size_t length)
+{
+  return stringi_set_find_node_len (set, s, length) != NULL;
 }
 
 /* Returns the node in SET that contains S, or a null pointer if SET does not
@@ -88,7 +100,17 @@ stringi_set_contains (const struct stringi_set *set, const char *s)
 struct stringi_set_node *
 stringi_set_find_node (const struct stringi_set *set, const char *s)
 {
-  return stringi_set_find_node__ (set, s, hash_case_string (s, 0));
+  return stringi_set_find_node_len (set, s, strlen (s));
+}
+
+/* Returns the node in SET that contains S with the given LENGTH, or a null
+   pointer if SET does not contain S. */
+struct stringi_set_node *
+stringi_set_find_node_len (const struct stringi_set *set, const char *s,
+                           size_t length)
+{
+  return stringi_set_find_node_len__ (set, s, length,
+                                      utf8_hash_case_bytes (s, length, 0));
 }
 
 /* Inserts a copy of S into SET.  Returns true if successful, false if SET
@@ -96,7 +118,7 @@ stringi_set_find_node (const struct stringi_set *set, const char *s)
 bool
 stringi_set_insert (struct stringi_set *set, const char *s)
 {
-  unsigned int hash = hash_case_string (s, 0);
+  unsigned int hash = utf8_hash_case_string (s, 0);
   if (!stringi_set_find_node__ (set, s, hash))
     {
       stringi_set_insert__ (set, xstrdup (s), hash);
@@ -112,7 +134,7 @@ stringi_set_insert (struct stringi_set *set, const char *s)
 bool
 stringi_set_insert_nocopy (struct stringi_set *set, char *s)
 {
-  unsigned int hash = hash_case_string (s, 0);
+  unsigned int hash = utf8_hash_case_string (s, 0);
   if (!stringi_set_find_node__ (set, s, hash))
     {
       stringi_set_insert__ (set, s, hash);
@@ -130,7 +152,7 @@ stringi_set_insert_nocopy (struct stringi_set *set, char *s)
 bool
 stringi_set_delete (struct stringi_set *set, const char *s)
 {
-  return stringi_set_delete__ (set, s, hash_case_string (s, 0));
+  return stringi_set_delete__ (set, s, utf8_hash_case_string (s, 0));
 }
 
 /* Deletes NODE from SET, and frees NODE and its string. */
@@ -258,7 +280,7 @@ compare_strings (const void *a_, const void *b_)
 {
   const char *const *a = a_;
   const char *const *b = b_;
-  return strcasecmp (*a, *b);
+  return utf8_strcasecmp (*a, *b);
 }
 
 /* Allocates and returns an array that points to each of the strings in SET.
@@ -267,7 +289,7 @@ compare_strings (const void *a_, const void *b_)
    caller it is responsible for freeing the returned array itself (with
    free()).
 
-   The returned array is ordered according to strcasecmp(). */
+   The returned array is ordered according to utf8_strcasecmp(). */
 char **
 stringi_set_get_sorted_array (const struct stringi_set *set)
 {
@@ -280,13 +302,20 @@ stringi_set_get_sorted_array (const struct stringi_set *set)
 
 static struct stringi_set_node *
 stringi_set_find_node__ (const struct stringi_set *set, const char *s,
-                        unsigned int hash)
+                         unsigned int hash)
+{
+  return stringi_set_find_node_len__ (set, s, strlen (s), hash);
+}
+
+static struct stringi_set_node *
+stringi_set_find_node_len__ (const struct stringi_set *set, const char *s,
+                             size_t length, unsigned int hash)
 {
   struct stringi_set_node *node;
 
   HMAP_FOR_EACH_WITH_HASH (node, struct stringi_set_node, hmap_node,
                            hash, &set->hmap)
-    if (!strcasecmp (s, node->string))
+    if (!utf8_strncasecmp (s, length, node->string, strlen (node->string)))
       return node;
 
   return NULL;