begin to write full pinyin parser2 parse_one_key

3 files changed, 80 insertions, 6 deletions

diff --git a/src/storage/chewing_key.h b/src/storage/chewing_key.h
index ea086b7..5d708e0 100644
--- a/src/storage/chewing_key.h
+++ b/src/storage/chewing_key.h
@@ -181,14 +181,14 @@ struct ChewingKey
 struct ChewingKeyRest
 {
     guint16 m_index;           /* the index in pinyin parser table. */
-    guint16 m_pinyin_begin;    /* the begin of pinyin in raw input. */
-    guint16 m_pinyin_end;      /* the end of pinyin in raw input. */
+    guint16 m_raw_begin;           /* the begin of the raw input. */
+    guint16 m_raw_end;             /* the end of the raw input. */
 
     ChewingKeyRest() {
         /* the 0th item in pinyin parser table is reserved for invalid. */
         m_index = 0;
-        m_pinyin_begin = 0;
-        m_pinyin_end = 0;
+        m_raw_begin = 0;
+        m_raw_end = 0;
     }
 
     const char * get_pinyin_string();
diff --git a/src/storage/pinyin_parser2.cpp b/src/storage/pinyin_parser2.cpp
index ea180fe..c7bf71b 100644
--- a/src/storage/pinyin_parser2.cpp
+++ b/src/storage/pinyin_parser2.cpp
@@ -19,7 +19,11 @@
  *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA  02110-1301, USA.
  */
 
+
+#include <ctype.h>
 #include <assert.h>
+#include <string.h>
+#include "stl_lite.h"
 #include "pinyin_custom2.h"
 #include "chewing_key.h"
 #include "pinyin_parser2.h"
@@ -28,7 +32,7 @@
 
 using namespace pinyin;
 
-static bool check_pinyin_options(guint32 options, pinyin_index_item_t * item) {
+static bool check_pinyin_options(guint32 options, const pinyin_index_item_t * item) {
     guint32 flags = item->m_flags;
     assert (flags & IS_PINYIN);
 
@@ -50,7 +54,7 @@ static bool check_pinyin_options(guint32 options, pinyin_index_item_t * item) {
     return true;
 }
 
-static bool check_chewing_options(guint32 options, chewing_index_item_t * item) {
+static bool check_chewing_options(guint32 options, const chewing_index_item_t * item) {
     guint32 flags = item->m_flags;
     assert (flags & IS_CHEWING);
 
@@ -63,6 +67,8 @@ static bool check_chewing_options(guint32 options, chewing_index_item_t * item)
     return true;
 }
 
+
+/* methods for Chewing Keys to access pinyin parser table. */
 const char * ChewingKeyRest::get_pinyin_string(){
     if (m_index == 0)
         return NULL;
@@ -80,3 +86,68 @@ const char * ChewingKeyRest::get_chewing_string(){
     assert(m_index < G_N_ELEMENTS(content_table));
     return content_table[m_index].m_chewing_str;
 }
+
+
+static bool compare_less_than(const pinyin_index_item_t & lhs,
+                              const pinyin_index_item_t & rhs){
+    return 0 > strcmp(lhs.m_pinyin_input, rhs.m_pinyin_input);
+}
+
+int FullPinyinParser2::parse_one_key (guint32 options, ChewingKey & key,
+                                      ChewingKeyRest & key_rest,
+                                      const char * pinyin, int len) const {
+    /* "'" are not accepted in parse_one_key. */
+    assert(NULL == strchr(pinyin, '\''));
+    gchar * input = g_strndup(pinyin, len);
+
+    guint16 tone = CHEWING_ZERO_TONE; guint16 tone_pos = 0;
+    guint16 parsed_len = len;
+    key = ChewingKey(); key_rest = ChewingKeyRest();
+
+    /* find the tone in the last character. */
+    char chr = input[parsed_len - 1];
+    if ( '0' < chr && chr <= '5' ) {
+        tone = chr - '0';
+        parsed_len --;
+        tone_pos = parsed_len;
+    }
+
+    /* parse pinyin core staff here. */
+    pinyin_index_item_t item;
+    memset(&item, 0, sizeof(item));
+
+    for (; parsed_len > 0; --parsed_len) {
+        input[parsed_len] = '\0';
+        item.m_pinyin_input = input;
+        std_lite::pair<const pinyin_index_item_t *,
+                       const pinyin_index_item_t *> range;
+        range = std_lite::equal_range
+            (pinyin_index, pinyin_index + G_N_ELEMENTS(pinyin_index),
+             item, compare_less_than);
+
+        guint16 len = range.second - range.first;
+        assert (len <= 1);
+        if ( len == 1 ) {
+            const pinyin_index_item_t * index = range.first;
+
+            if (!check_pinyin_options(options, index))
+                continue;
+
+            key_rest.m_index = index->m_table_index;
+            key = content_table[key_rest.m_index].m_chewing_key;
+            break;
+        }
+    }
+
+    /* post processing tone. */
+    if ( parsed_len == tone_pos ) {
+        if (tone != CHEWING_ZERO_TONE) {
+            key.m_tone = tone;
+            parsed_len ++;
+        }
+    }
+
+    key_rest.m_raw_begin = 0; key_rest.m_raw_end = parsed_len;
+    g_free(input);
+    return parsed_len;
+}
diff --git a/src/storage/pinyin_parser2.h b/src/storage/pinyin_parser2.h
index 588a4f1..67bf0b0 100644
--- a/src/storage/pinyin_parser2.h
+++ b/src/storage/pinyin_parser2.h
@@ -146,12 +146,15 @@ public:
  * @brief Class to parse Chewing input string
  *
  * Several keyboard scheme are supported:
+ * * ZHUYIN_ZHUYIN    Parse original ZhuYin string, such as ㄅㄧㄢ
  * * Chewing_STANDARD  Standard ZhuYin keyboard, which maps 1 to Bo(ㄅ), q to Po(ㄆ) etc.
  * * Chewing_HSU       Hsu ZhuYin keyboard, which uses a-z (except q) chars.
  * * Chewing_IBM       IBM ZhuYin keyboard, which maps 1 to Bo(ㄅ), 2 to Po(ㄆ) etc.
  * * Chewing_GIN_YIEH  Gin-Yieh ZhuYin keyboard.
  * * Chewing_ET        Eten (倚天) ZhuYin keyboard.
  * * Chewing_ET26      Eten (倚天) ZhuYin keyboard, which only uses a-z chars.
+ * UTF-8 string is used in ZhuYin Parser, because the requirement of supporting original ZhuYin strings.
+ * So that the length of inputted string is calculated in number of utf8 chars instead of bytes.
  */
 class ChewingParser2 : public PinyinParser2
 {