From 2defab0636aff5f37a63bcf1a9697ef70bcd53bb Mon Sep 17 00:00:00 2001
From: Navaneeth K N <navaneethkn@gmail.com>
Date: Mon, 9 Jun 2014 01:26:14 +0530
Subject: [PATCH] Added strbuf_chars() to strbuf which returns each character
 in string

This function is unicode aware and works well for UTF-8 encoded texts.
---
 CMakeLists.txt     |  1 +
 strbuf.c           | 24 ++++++++++++++++++++++++
 tests/strbuftest.c | 42 ++++++++++++++++++++++++++++++++++++++++++
 util.h             |  1 +
 4 files changed, 68 insertions(+)

diff --git a/CMakeLists.txt b/CMakeLists.txt
index 6c07d2d..4a4c921 100644
--- a/CMakeLists.txt
+++ b/CMakeLists.txt
@@ -110,6 +110,7 @@ SET_TARGET_PROPERTIES(${VARNAM_LIBRARY_NAME} PROPERTIES
 
 # a custom target to compile scheme files
 add_custom_target (vst command ./varnamc --compile schemes/ml && ./varnamc --compile schemes/hi)
+add_custom_target (test command cd tests && ./runtests)
 
 # Makes the distro
 add_custom_target (distro COMMAND ./makedistro.sh "${VARNAM_VERSION_MAJOR}.${VARNAM_VERSION_MINOR}.${VARNAM_VERSION_PATCH}")
diff --git a/strbuf.c b/strbuf.c
index fd5c396..15c43be 100644
--- a/strbuf.c
+++ b/strbuf.c
@@ -177,6 +177,30 @@ int strbuf_addvf(struct strbuf *string, const char *format, va_list args)
 
 }
 
+/*
+ * Gets each unicode character in this string
+ * returned result should be destroyed
+ * */
+varray*
+strbuf_chars(strbuf *b)
+{
+  const unsigned char *ustring; const char *inputcopy;
+  int bytes_read;
+  varray *chars;
+  strbuf *tmp;
+
+  inputcopy = b->buffer;
+  chars = varray_init();
+  while (*inputcopy != '\0') {
+    READ_A_UTF8_CHAR (ustring, inputcopy, bytes_read);
+    tmp = strbuf_init(8);
+    strbuf_add_bytes (tmp, inputcopy - bytes_read, bytes_read);
+    varray_push (chars, strbuf_detach(tmp));
+    bytes_read = 0;
+  }
+  return chars;
+}
+
 void strbuf_destroy(void *s)
 {
     strbuf *string;
diff --git a/tests/strbuftest.c b/tests/strbuftest.c
index 643ec98..933943f 100644
--- a/tests/strbuftest.c
+++ b/tests/strbuftest.c
@@ -112,6 +112,46 @@ START_TEST (addfln_should_add_newline)
 }
 END_TEST
 
+START_TEST (get_each_character)
+{
+  varray *chars;
+  strbuf *str = strbuf_init (20);
+  strbuf_add(str, "test");
+  chars = strbuf_chars(str);
+  ck_assert_int_eq (4, varray_length(chars));
+  ck_assert_str_eq ("t", (const char*)  varray_get(chars, 0));
+  ck_assert_str_eq ("e", (const char*)  varray_get(chars, 1));
+  ck_assert_str_eq ("s", (const char*)  varray_get(chars, 2));
+  ck_assert_str_eq ("t", (const char*)  varray_get(chars, 3));
+
+  strbuf_clear (str);
+  strbuf_add(str, "t");
+  chars = strbuf_chars(str);
+  ck_assert_int_eq (1, varray_length(chars));
+  ck_assert_str_eq ("t", (const char*)  varray_get(chars, 0));
+
+  strbuf_clear (str);
+  chars = strbuf_chars(str);
+  ck_assert_int_eq (0, varray_length(chars));
+}
+END_TEST
+
+START_TEST (get_each_character_should_be_unicode_aware)
+{
+  varray *chars;
+  strbuf *str = strbuf_init (20);
+  strbuf_add(str, "മലയാളം");
+  chars = strbuf_chars(str);
+  ck_assert_int_eq (6, varray_length(chars));
+  ck_assert_str_eq ("മ",  (const char*)  varray_get(chars, 0));
+  ck_assert_str_eq ("ല",  (const char*)  varray_get(chars, 1));
+  ck_assert_str_eq ("യ",  (const char*)  varray_get(chars, 2));
+  ck_assert_str_eq ("ാ", (const char*)  varray_get(chars, 3));
+  ck_assert_str_eq ("ള", (const char*)  varray_get(chars, 4));
+  ck_assert_str_eq ("ം", (const char*)  varray_get(chars, 5));
+}
+END_TEST
+
 TCase* get_strbuf_tests()
 {
     TCase* tcase = tcase_create("strbuf");
@@ -123,5 +163,7 @@ TCase* get_strbuf_tests()
     tcase_add_test (tcase, split_string);
     tcase_add_test (tcase, addf_should_not_add_newline);
     tcase_add_test (tcase, addfln_should_add_newline);
+    tcase_add_test (tcase, get_each_character);
+    tcase_add_test (tcase, get_each_character_should_be_unicode_aware);
     return tcase;
 }
diff --git a/util.h b/util.h
index dc76aab..23beafa 100644
--- a/util.h
+++ b/util.h
@@ -161,6 +161,7 @@ VARNAM_EXPORT int strbuf_addvf(struct strbuf *string, const char *format, va_lis
 VARNAM_EXPORT void strbuf_destroy(void *s);
 VARNAM_EXPORT char* strbuf_detach(struct strbuf *string);
 VARNAM_EXPORT const char* strbuf_to_s(struct strbuf *string);
+VARNAM_EXPORT struct varray_t* strbuf_chars(strbuf *b);
 VARNAM_EXPORT void strbuf_clear(struct strbuf *string);
 VARNAM_EXPORT int strbuf_is_blank(struct strbuf *string);
 VARNAM_EXPORT int strbuf_endswith(struct strbuf *string, const char *str);