Reformat some source for ICU

diff --git a/src/icu_casemap.c b/src/icu_casemap.c
index 4204f8a..f06f64e 100644 (file)
--- a/src/icu_casemap.c
+++ b/src/icu_casemap.c
@@ -30,9 +30,9 @@ struct icu_casemap
     char action;
 };
 
-struct icu_casemap * icu_casemap_create(char action, UErrorCode *status)
+struct icu_casemap *icu_casemap_create(char action, UErrorCode *status)
 {
-    struct icu_casemap * casemap
+    struct icu_casemap *casemap
         = (struct icu_casemap *) xmalloc(sizeof(struct icu_casemap));
     casemap->action = action;
 
@@ -67,9 +67,9 @@ void icu_casemap_destroy(struct icu_casemap * casemap)
     xfree(casemap);
 }
 
-int icu_casemap_casemap(struct icu_casemap * casemap,
-                        struct icu_buf_utf16 * dest16,
-                        struct icu_buf_utf16 * src16,
+int icu_casemap_casemap(struct icu_casemap *casemap,
+                        struct icu_buf_utf16 *dest16,
+                        struct icu_buf_utf16 *src16,
                         UErrorCode *status,
                         const char *locale)
 {
@@ -81,8 +81,8 @@ int icu_casemap_casemap(struct icu_casemap * casemap,
 }
 
 
-static uint32_t icu_utf16_sub(struct icu_buf_utf16 * dest16,
-                          struct icu_buf_utf16 * src16,
+static uint32_t icu_utf16_sub(struct icu_buf_utf16 *dest16,
+                          struct icu_buf_utf16 *src16,
                           const char *locale, char action,
                           UErrorCode *status)
 {
@@ -119,8 +119,8 @@ static uint32_t icu_utf16_sub(struct icu_buf_utf16 * dest16,
 }
 
 
-int icu_utf16_casemap(struct icu_buf_utf16 * dest16,
-                      struct icu_buf_utf16 * src16,
+int icu_utf16_casemap(struct icu_buf_utf16 *dest16,
+                      struct icu_buf_utf16 *src16,
                       const char *locale, char action,
                       UErrorCode *status)
 {

diff --git a/src/icu_sortkey.c b/src/icu_sortkey.c
index 69c9f57..e4decac 100644 (file)
--- a/src/icu_sortkey.c
+++ b/src/icu_sortkey.c
@@ -27,8 +27,8 @@
 #include <unicode/uchar.h>    /* char names           */
 
 void icu_sortkey8_from_utf16(UCollator *coll,
-                             struct icu_buf_utf8 * dest8,
-                             struct icu_buf_utf16 * src16,
+                             struct icu_buf_utf8 *dest8,
+                             struct icu_buf_utf16 *src16,
                              UErrorCode * status)
 {
     int32_t sortkey_len = 0;

diff --git a/src/icu_tokenizer.c b/src/icu_tokenizer.c
index e09960f..67246ea 100644 (file)
--- a/src/icu_tokenizer.c
+++ b/src/icu_tokenizer.c
@@ -78,7 +78,7 @@ struct icu_tokenizer *icu_tokenizer_clone(struct icu_tokenizer *old)
 struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action,
                                            UErrorCode *status)
 {
-    struct icu_tokenizer * tokenizer
+    struct icu_tokenizer *tokenizer
         = (struct icu_tokenizer *) xmalloc(sizeof(struct icu_tokenizer));
 
     icu_tokenizer_reset(tokenizer, action);
@@ -119,7 +119,7 @@ struct icu_tokenizer *icu_tokenizer_create(const char *locale, char action,
     return 0;
 }
 
-void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
+void icu_tokenizer_destroy(struct icu_tokenizer *tokenizer)
 {
     if (tokenizer)
     {
@@ -130,8 +130,8 @@ void icu_tokenizer_destroy(struct icu_tokenizer * tokenizer)
     }
 }
 
-int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
-                         struct icu_buf_utf16 * src16,
+int icu_tokenizer_attach(struct icu_tokenizer *tokenizer,
+                         struct icu_buf_utf16 *src16,
                          UErrorCode *status)
 {
     if (!tokenizer || !tokenizer->bi || !src16)
@@ -153,8 +153,8 @@ int icu_tokenizer_attach(struct icu_tokenizer * tokenizer,
     return 1;
 }
 
-int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
-                                 struct icu_buf_utf16 * tkn16,
+int32_t icu_tokenizer_next_token(struct icu_tokenizer *tokenizer,
+                                 struct icu_buf_utf16 *tkn16,
                                  UErrorCode *status)
 {
     int32_t tkn_start = 0;
@@ -217,7 +217,7 @@ int32_t icu_tokenizer_next_token(struct icu_tokenizer * tokenizer,
     return tkn_len;
 }
 
-int32_t icu_tokenizer_token_count(struct icu_tokenizer * tokenizer)
+int32_t icu_tokenizer_token_count(struct icu_tokenizer *tokenizer)
 {
     return tokenizer->token_count;
 }

diff --git a/src/icu_transform.c b/src/icu_transform.c
index 6879bc4..05214b7 100644 (file)
--- a/src/icu_transform.c
+++ b/src/icu_transform.c
@@ -30,7 +30,7 @@ struct icu_transform
 {
     char action;
     UParseError parse_error;
-    UTransliterator * trans;
+    UTransliterator *trans;
 };
 
 struct icu_transform *icu_transform_clone(struct icu_transform *old)
@@ -46,9 +46,9 @@ struct icu_transform *icu_transform_clone(struct icu_transform *old)
     return transform;
 }
 
-struct icu_transform * icu_transform_create(const char *id, char action,
-                                            const char *rules,
-                                            UErrorCode *status)
+struct icu_transform *icu_transform_create(const char *id, char action,
+                                           const char *rules,
+                                           UErrorCode *status)
 {
     struct icu_buf_utf16 *id16 = icu_buf_utf16_create(0);
     struct icu_buf_utf16 *rules16 = icu_buf_utf16_create(0);
@@ -102,7 +102,7 @@ struct icu_transform * icu_transform_create(const char *id, char action,
     return 0;
 }
 
-void icu_transform_destroy(struct icu_transform * transform)
+void icu_transform_destroy(struct icu_transform *transform)
 {
     if (transform)
     {
@@ -112,9 +112,9 @@ void icu_transform_destroy(struct icu_transform * transform)
     }
 }
 
-int icu_transform_trans(struct icu_transform * transform,
-                        struct icu_buf_utf16 * dest16,
-                        const struct icu_buf_utf16 * src16,
+int icu_transform_trans(struct icu_transform *transform,
+                        struct icu_buf_utf16 *dest16,
+                        const struct icu_buf_utf16 *src16,
                         UErrorCode *status)
 {
     if (!transform || !transform->trans

diff --git a/src/icu_utf16.c b/src/icu_utf16.c
index 466f7af..5bb74ec 100644 (file)
--- a/src/icu_utf16.c
+++ b/src/icu_utf16.c
@@ -27,9 +27,9 @@
 #include <unicode/ustring.h>  /* some more string fcns*/
 #include <unicode/uchar.h>    /* char names           */
 
-struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity)
+struct icu_buf_utf16 *icu_buf_utf16_create(size_t capacity)
 {
-    struct icu_buf_utf16 * buf16
+    struct icu_buf_utf16 *buf16
         = (struct icu_buf_utf16 *) xmalloc(sizeof(struct icu_buf_utf16));
 
     buf16->utf16 = 0;
@@ -45,7 +45,7 @@ struct icu_buf_utf16 * icu_buf_utf16_create(size_t capacity)
     return buf16;
 }
 
-struct icu_buf_utf16 * icu_buf_utf16_clear(struct icu_buf_utf16 * buf16)
+struct icu_buf_utf16 *icu_buf_utf16_clear(struct icu_buf_utf16 *buf16)
 {
     if (buf16)
     {
@@ -56,8 +56,8 @@ struct icu_buf_utf16 * icu_buf_utf16_clear(struct icu_buf_utf16 * buf16)
     return buf16;
 }
 
-struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
-                                            size_t capacity)
+struct icu_buf_utf16 *icu_buf_utf16_resize(struct icu_buf_utf16 *buf16,
+                                           size_t capacity)
 {
     if (!buf16)
         return 0;
@@ -81,8 +81,8 @@ struct icu_buf_utf16 * icu_buf_utf16_resize(struct icu_buf_utf16 * buf16,
 }
 
 
-struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
-                                          const struct icu_buf_utf16 * src16)
+struct icu_buf_utf16 *icu_buf_utf16_copy(struct icu_buf_utf16 *dest16,
+                                         const struct icu_buf_utf16 *src16)
 {
     if (!dest16 || !src16 || dest16 == src16)
         return 0;
@@ -98,7 +98,7 @@ struct icu_buf_utf16 * icu_buf_utf16_copy(struct icu_buf_utf16 * dest16,
 
 
 struct icu_buf_utf16 *icu_buf_utf16_append(struct icu_buf_utf16 *dest16,
-                                           const struct icu_buf_utf16 * src16)
+                                           const struct icu_buf_utf16 *src16)
 {
     assert(dest16);
     if (!src16)
@@ -117,7 +117,7 @@ struct icu_buf_utf16 *icu_buf_utf16_append(struct icu_buf_utf16 *dest16,
 }
 
 
-void icu_buf_utf16_destroy(struct icu_buf_utf16 * buf16)
+void icu_buf_utf16_destroy(struct icu_buf_utf16 *buf16)
 {
     if (buf16)
         xfree(buf16->utf16);

diff --git a/src/icu_utf8.c b/src/icu_utf8.c
index ff9e9d6..dd392d1 100644 (file)
--- a/src/icu_utf8.c
+++ b/src/icu_utf8.c
@@ -28,7 +28,7 @@
 
 struct icu_buf_utf8 *icu_buf_utf8_create(size_t capacity)
 {
-    struct icu_buf_utf8 * buf8
+    struct icu_buf_utf8 *buf8
         = (struct icu_buf_utf8 *) xmalloc(sizeof(struct icu_buf_utf8));
 
     buf8->utf8 = 0;
@@ -44,7 +44,7 @@ struct icu_buf_utf8 *icu_buf_utf8_create(size_t capacity)
     return buf8;
 }
 
-struct icu_buf_utf8 * icu_buf_utf8_clear(struct icu_buf_utf8 * buf8)
+struct icu_buf_utf8 *icu_buf_utf8_clear(struct icu_buf_utf8 *buf8)
 {
     if (buf8)
     {
@@ -55,8 +55,8 @@ struct icu_buf_utf8 * icu_buf_utf8_clear(struct icu_buf_utf8 * buf8)
     return buf8;
 }
 
-struct icu_buf_utf8 * icu_buf_utf8_resize(struct icu_buf_utf8 * buf8,
-                                          size_t capacity)
+struct icu_buf_utf8 *icu_buf_utf8_resize(struct icu_buf_utf8 *buf8,
+                                         size_t capacity)
 {
     if (!buf8)
         return 0;
@@ -95,16 +95,16 @@ const char *icu_buf_utf8_to_cstr(struct icu_buf_utf8 *src8)
     return (const char *) src8->utf8;
 }
 
-void icu_buf_utf8_destroy(struct icu_buf_utf8 * buf8)
+void icu_buf_utf8_destroy(struct icu_buf_utf8 *buf8)
 {
     if (buf8)
         xfree(buf8->utf8);
     xfree(buf8);
 }
 
-UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
-                                    const char * src8cstr,
-                                    UErrorCode * status)
+UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 *dest16,
+                                    const char *src8cstr,
+                                    UErrorCode *status)
 {
     size_t src8cstr_len = 0;
     int32_t utf16_len = 0;
@@ -136,7 +136,7 @@ UErrorCode icu_utf16_from_utf8_cstr(struct icu_buf_utf16 * dest16,
 
 UErrorCode icu_utf16_to_utf8(struct icu_buf_utf8 *dest8,
                              const struct icu_buf_utf16 *src16,
-                             UErrorCode * status)
+                             UErrorCode *status)
 {
     int32_t utf8_len = 0;
 

diff --git a/test/test_icu.c b/test/test_icu.c
index 391e66e..b1de907 100644 (file)
--- a/test/test_icu.c
+++ b/test/test_icu.c
@@ -51,16 +51,16 @@ static int icu_termmap_cmp(const void *vp1, const void *vp2)
 }
 
 
-static int test_icu_casemap(const char * locale, char action,
-                            const char * src8cstr, const char * chk8cstr)
+static int test_icu_casemap(const char *locale, char action,
+                            const char *src8cstr, const char *chk8cstr)
 {
     int success = 0;
     UErrorCode status = U_ZERO_ERROR;
 
-    struct icu_buf_utf8 * src8 = icu_buf_utf8_create(0);
-    struct icu_buf_utf8 * dest8 = icu_buf_utf8_create(0);
-    struct icu_buf_utf16 * src16 = icu_buf_utf16_create(0);
-    struct icu_buf_utf16 * dest16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf8 *src8 = icu_buf_utf8_create(0);
+    struct icu_buf_utf8 *dest8 = icu_buf_utf8_create(0);
+    struct icu_buf_utf16 *src16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf16 *dest16 = icu_buf_utf16_create(0);
 
 
     int src8cstr_len = strlen(src8cstr);
@@ -165,19 +165,19 @@ static void check_icu_casemap(void)
 
 }
 
-static int test_icu_sortmap(const char * locale, int src_list_len,
-                            const char ** src_list, const char ** chk_list)
+static int test_icu_sortmap(const char *locale, int src_list_len,
+                            const char **src_list, const char **chk_list)
 {
     int success = 1;
 
     UErrorCode status = U_ZERO_ERROR;
 
-    struct icu_buf_utf8 * buf8 = icu_buf_utf8_create(0);
-    struct icu_buf_utf16 * buf16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf8 *buf8 = icu_buf_utf8_create(0);
+    struct icu_buf_utf16 *buf16 = icu_buf_utf16_create(0);
 
     int i;
 
-    struct icu_termmap * list[src_list_len];
+    struct icu_termmap *list[src_list_len];
 
     UCollator *coll = ucol_open(locale, &status);
     icu_check_status(status);
@@ -249,8 +249,8 @@ static void check_icu_sortmap(void)
 {
     /* successful tests */
     size_t en_1_len = 6;
-    const char * en_1_src[6] = {"z", "K", "a", "A", "Z", "k"};
-    const char * en_1_cck[6] = {"a", "A", "k", "K", "z", "Z"};
+    const char *en_1_src[6] = {"z", "K", "a", "A", "Z", "k"};
+    const char *en_1_cck[6] = {"a", "A", "k", "K", "z", "Z"};
     YAZ_CHECK(test_icu_sortmap("en", en_1_len, en_1_src, en_1_cck));
     YAZ_CHECK(test_icu_sortmap("en_AU", en_1_len, en_1_src, en_1_cck));
     YAZ_CHECK(test_icu_sortmap("en_CA", en_1_len, en_1_src, en_1_cck));
@@ -260,34 +260,34 @@ static void check_icu_sortmap(void)
     /* successful tests */
     {
         size_t da_1_len = 6;
-        const char * da_1_src[6] = {"z", "å", "o", "æ", "a", "ø"};
-        const char * da_1_cck[6] = {"a", "o", "z", "æ", "ø", "å"};
+        const char *da_1_src[6] = {"z", "å", "o", "æ", "a", "ø"};
+        const char *da_1_cck[6] = {"a", "o", "z", "æ", "ø", "å"};
         YAZ_CHECK(test_icu_sortmap("da", da_1_len, da_1_src, da_1_cck));
         YAZ_CHECK(test_icu_sortmap("da_DK", da_1_len, da_1_src, da_1_cck));
     }
     /* successful tests */
     {
         size_t de_1_len = 9;
-        const char * de_1_src[9] = {"u", "ä", "o", "t", "s", "ß", "ü", "ö", "a"};
-        const char * de_1_cck[9] = {"a","ä", "o", "ö", "s", "ß", "t", "u", "ü"};
+        const char *de_1_src[9] = {"u", "ä", "o", "t", "s", "ß", "ü", "ö", "a"};
+        const char *de_1_cck[9] = {"a","ä", "o", "ö", "s", "ß", "t", "u", "ü"};
         YAZ_CHECK(test_icu_sortmap("de", de_1_len, de_1_src, de_1_cck));
         YAZ_CHECK(test_icu_sortmap("de_AT", de_1_len, de_1_src, de_1_cck));
         YAZ_CHECK(test_icu_sortmap("de_DE", de_1_len, de_1_src, de_1_cck));
     }
 }
 
-static int test_icu_normalizer(const char * rules8cstr,
-                               const char * src8cstr,
-                               const char * chk8cstr)
+static int test_icu_normalizer(const char *rules8cstr,
+                               const char *src8cstr,
+                               const char *chk8cstr)
 {
     int success = 0;
 
     UErrorCode status = U_ZERO_ERROR;
 
-    struct icu_buf_utf16 * src16 = icu_buf_utf16_create(0);
-    struct icu_buf_utf16 * dest16 = icu_buf_utf16_create(0);
-    struct icu_buf_utf8 * dest8 = icu_buf_utf8_create(0);
-    struct icu_transform * transform
+    struct icu_buf_utf16 *src16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf16 *dest16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf8 *dest8 = icu_buf_utf8_create(0);
+    struct icu_transform *transform
         = icu_transform_create(rules8cstr, 'f', 0, &status);
     icu_check_status(status);
 
@@ -354,16 +354,16 @@ static void check_icu_normalizer(void)
                                   "a cote de l'alcove ovoide"));
 }
 
-static int test_icu_tokenizer(const char * locale, char action,
-                              const char * src8cstr, int count)
+static int test_icu_tokenizer(const char *locale, char action,
+                              const char *src8cstr, int count)
 {
     int success = 1;
 
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_buf_utf16 * src16 = icu_buf_utf16_create(0);
-    struct icu_buf_utf16 * tkn16 = icu_buf_utf16_create(0);
-    struct icu_buf_utf8 * tkn8 = icu_buf_utf8_create(0);
-    struct icu_tokenizer * tokenizer = 0;
+    struct icu_buf_utf16 *src16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf16 *tkn16 = icu_buf_utf16_create(0);
+    struct icu_buf_utf8 *tkn8 = icu_buf_utf8_create(0);
+    struct icu_tokenizer *tokenizer = 0;
 
     /* transforming to UTF16 */
     icu_utf16_from_utf8_cstr(src16, src8cstr, &status);
@@ -406,7 +406,7 @@ static int test_icu_tokenizer(const char * locale, char action,
 
 static void check_icu_tokenizer(void)
 {
-    const char * en_str
+    const char *en_str
         = "O Romeo, Romeo! wherefore art thou Romeo?";
 
     YAZ_CHECK(test_icu_tokenizer("en", 's', en_str, 2));
@@ -415,7 +415,7 @@ static void check_icu_tokenizer(void)
     YAZ_CHECK(test_icu_tokenizer("en", 'c', en_str, 41));
 
     {
-        const char * da_str
+        const char *da_str
             = "Blåbærtærte. Denne kage stammer fra Finland. "
             "Den er med blåbær, men alle sommerens forskellige bær kan bruges.";
 
@@ -428,13 +428,13 @@ static void check_icu_tokenizer(void)
 
 static void check_icu_chain(void)
 {
-    const char * en_str
+    const char *en_str
         = "O Romeo, Romeo! wherefore art thou\t Romeo?";
 
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
         "<transform rule=\"[:Control:] Any-Remove\"/>"
         "<tokenize rule=\"l\"/>"
         "<transform rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
@@ -487,9 +487,9 @@ static void check_icu_chain(void)
 static void check_bug_1140(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
 
         /* if the first rule is normalize instead. Then it works */
 #if 0
@@ -550,9 +550,9 @@ static void check_bug_1140(void)
 static void check_chain_empty_token(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
         "<tokenize rule=\"w\"/>"
         "<transform rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
         "</icu>";
@@ -587,13 +587,13 @@ static void check_chain_empty_token(void)
 static void check_chain_empty_chain(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
         "</icu>";
 
-    const char * src8 = "some 5487 weired !¤%&(/& sTuFf";
-    char * dest8 = 0;
+    const char *src8 = "some 5487 weired !¤%&(/& sTuFf";
+    char *dest8 = 0;
 
     xmlDoc *doc = xmlParseMemory(xml_str, strlen(xml_str));
     xmlNode *xml_node = xmlDocGetRootElement(doc);
@@ -628,11 +628,11 @@ static void check_chain_empty_chain(void)
 static void check_icu_iter1(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
     xmlNode *xml_node;
     yaz_icu_iter_t iter;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
         "<tokenize rule=\"w\"/>"
         "<transform rule=\"[[:WhiteSpace:][:Punctuation:]] Remove\"/>"
         "</icu>";
@@ -767,10 +767,10 @@ static void check_iter_threads(struct icu_chain *chain)
 static void check_icu_iter2(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
     xmlNode *xml_node;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
         "<transform rule=\"[:Control:] Any-Remove\"/>"
         "<tokenize rule=\"l\"/>"
         "<tokenize rule=\"w\"/>"
@@ -809,10 +809,10 @@ static void check_icu_iter2(void)
 static void check_icu_iter3(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
     xmlNode *xml_node;
 
-    const char * xml_str =
+    const char *xml_str =
         "<icu_chain id=\"sort\" locale=\"el\">\n"
         "<transform rule=\"[:Control:] Any-Remove\"/>\n"
         "<transform rule=\"[[:Control:][:WhiteSpace:][:Punctuation:]] Remove\"/>\n"
@@ -850,10 +850,10 @@ static void check_icu_iter3(void)
 static void check_icu_iter4(void)
 {
     UErrorCode status = U_ZERO_ERROR;
-    struct icu_chain * chain = 0;
+    struct icu_chain *chain = 0;
     xmlNode *xml_node;
 
-    const char * xml_str = "<icu locale=\"en\">"
+    const char *xml_str = "<icu locale=\"en\">"
         "<transform rule=\"[:Control:] Any-Remove\"/>"
         "<tokenize rule=\"l\"/>"
         "<tokenize rule=\"w\"/>"