From ca746304e3e14795401733d8358f75e586b1a737 Mon Sep 17 00:00:00 2001
From: Al <albarrentine@gmail.com>
Date: Thu, 4 Jun 2015 13:20:14 -0400
Subject: [PATCH] [utils] Adding a few methods to string_utils for finding
 utf8proc category groups

---
 src/string_utils.c | 39 +++++++++++++++++++++++++++++++++------
 src/string_utils.h |  5 +++++
 2 files changed, 38 insertions(+), 6 deletions(-)

diff --git a/src/string_utils.c b/src/string_utils.c
index 836c9f63..bd63b056 100644
--- a/src/string_utils.c
+++ b/src/string_utils.c
@@ -142,15 +142,42 @@ error_free_output:
     return NULL;
 }
 
-bool utf8_is_letter(int32_t ch) {
-    const utf8proc_property_t *props = utf8proc_get_property(ch);
-    utf8proc_propval_t cat = props->category;
+inline bool utf8_is_letter(int32_t ch) {
+    int cat = utf8proc_category(ch);
     return cat == UTF8PROC_CATEGORY_LL || cat == UTF8PROC_CATEGORY_LU        \
-            || cat == UTF8PROC_CATEGORY_LT || cat == UTF8PROC_CATEGORY_LO || \
-            cat == UTF8PROC_CATEGORY_LM;
-
+            || cat == UTF8PROC_CATEGORY_LT || cat == UTF8PROC_CATEGORY_LO    \
+            || cat == UTF8PROC_CATEGORY_LM;
 }
 
+inline bool utf8_is_number(int32_t ch) {
+    int cat = utf8proc_category(ch);
+    return cat == UTF8PROC_CATEGORY_ND || cat == UTF8PROC_CATEGORY_NL || cat == UTF8PROC_CATEGORY_NO;
+}
+
+inline bool utf8_is_hyphen(int32_t ch) {
+    int cat = utf8proc_category(ch);
+    return cat == UTF8PROC_CATEGORY_PD || c == 0x2212;
+}
+
+inline bool utf8_is_punctuation(int32_t ch) {
+    int cat = utf8proc_category(ch);
+    return cat == UTF8PROC_CATEGORY_PD || cat == UTF8PROC_CATEGORY_PE        \
+           || cat == UTF8PROC_CATEGORY_PF || cat == UTF8PROC_CATEGORY_PI    \
+           || cat == UTF8PROC_CATEGORY_PO || cat == UTF8PROC_CATEGORY_PS;
+}
+
+inline bool utf8_is_symbol(int32_t ch) {
+    int cat = utf8proc_category(ch);
+    return cat == UTF8PROC_CATEGORY_SK || cat == UTF8PROC_CATEGORY_SC       \
+           || cat == UTF8PROC_CATEGORY_SM || cat == UTF8PROC_CATEGORY_SO;
+}
+
+inline bool utf8_is_separator(int32_t ch) {
+    int cat = utf8proc_category(ch);
+    return cat == UTF8PROC_CATEGORY_ZS || cat == UTF8PROC_CATEGORY_ZL || cat == UTF8PROC_CATEGORY_ZP;
+}
+
+
 size_t string_rtrim(char *str) {
     size_t spaces = 0;
 
diff --git a/src/string_utils.h b/src/string_utils.h
index f244811b..d1bc634c 100644
--- a/src/string_utils.h
+++ b/src/string_utils.h
@@ -31,6 +31,11 @@ uint string_translate(char *str, size_t len, char *word_chars, char *word_repls,
 char *utf8_reversed_string(const char *s); // returns a copy, caller frees
 ssize_t utf8proc_iterate_reversed(const uint8_t *str, ssize_t start, int32_t *dst);
 bool utf8_is_letter(int32_t ch);
+bool utf8_is_number(int32_t ch);
+bool utf8_is_hyphen(int32_t ch);
+bool utf8_is_punctuation(int32_t ch);
+bool utf8_is_symbol(int32_t ch);
+bool utf8_is_separator(int32_t ch);
 
 size_t string_ltrim(char *str);
 size_t string_rtrim(char *str);