From cc0401a8d1a1ede1f15e609eb81a043eb6e8fa39 Mon Sep 17 00:00:00 2001
From: Al <albarrentine@gmail.com>
Date: Sun, 28 Jun 2015 19:37:53 -0400
Subject: [PATCH] [utf8] Adding a boolean struct member for string_script_t
 return values, set to true if the string is ASCII (no transliteration needed,
 should be frequent for English addresses)

---
 src/unicode_scripts.c | 8 +++++++-
 src/unicode_scripts.h | 1 +
 2 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/src/unicode_scripts.c b/src/unicode_scripts.c
index d9e2306c..69c1e719 100644
--- a/src/unicode_scripts.c
+++ b/src/unicode_scripts.c
@@ -2,6 +2,8 @@
 
 #include "unicode_scripts_data.c"
 
+#define MAX_ASCII 128
+
 inline script_t get_char_script(uint32_t ch) {
     if (ch > NUM_CODEPOINTS - 1) return SCRIPT_UNKNOWN;
     return char_scripts[ch];
@@ -21,6 +23,8 @@ string_script_t get_string_script(char *str, size_t len) {
     size_t script_len = 0;
     size_t idx = 0;
 
+    bool is_ascii = true;
+
     while (idx < len) {
         ssize_t char_len = utf8proc_iterate(ptr, -1, &ch);
 
@@ -36,6 +40,8 @@ string_script_t get_string_script(char *str, size_t len) {
             break;
         }
 
+        is_ascii = is_ascii && ch < MAX_ASCII;
+
         ptr += char_len;
         idx += char_len;
         script_len += char_len;
@@ -46,5 +52,5 @@ string_script_t get_string_script(char *str, size_t len) {
     
     }
 
-    return (string_script_t) {last_script, script_len};
+    return (string_script_t) {last_script, script_len, is_ascii};
 }
\ No newline at end of file
diff --git a/src/unicode_scripts.h b/src/unicode_scripts.h
index e290f012..ba797eea 100644
--- a/src/unicode_scripts.h
+++ b/src/unicode_scripts.h
@@ -18,6 +18,7 @@ typedef struct script_languages {
 typedef struct string_script {
     script_t script;
     size_t len;
+    bool ascii;
 } string_script_t;
 
 script_t get_char_script(uint32_t ch);