write eval correction rate

1 files changed, 146 insertions, 1 deletions

diff --git a/utils/training/eval_correction_rate.cpp b/utils/training/eval_correction_rate.cpp
index 6009251..81aadaa 100644
--- a/utils/training/eval_correction_rate.cpp
+++ b/utils/training/eval_correction_rate.cpp
@@ -20,9 +20,154 @@
  */
 
 
-
 #include "pinyin.h"
 
+typedef MatchResults TokenVector;
+
+void print_help(){
+    printf("Usage: eval_correction_rate\n");
+}
+
+bool get_possible_pinyin(FacadePhraseIndex * phrase_index,
+                         TokenVector tokens, PinyinKeyVector pinyins){
+    PinyinKey buffer[MAX_PHRASE_LENGTH];
+    size_t pinyin_index; guint32 max_freq;
+    guint32 freq;
+    g_array_set_size(pinyins, 0);
+
+    for (size_t i = 0; i < tokens->len; ++i){
+        phrase_token_t * token = &g_array_index(tokens, phrase_token_t, i);
+        PhraseItem item;
+        phrase_index->get_phrase_item(*token, item);
+        pinyin_index = 0; max_freq = 0;
+        for ( size_t m = 0; m < item.get_n_pronunciation(); ++m ) {
+            freq = 0;
+            assert(item.get_nth_pronunciation(m, buffer, freq));
+            if ( freq > max_freq ) {
+                pinyin_index = m;
+                max_freq = freq;
+            }
+        }
+
+        assert(item.get_nth_pronunciation(pinyin_index, buffer, freq));
+        assert(max_freq == freq);
+        guint8 len = item.get_phrase_length();
+        g_array_append_vals(pinyins, buffer, len);
+    }
+    return true;
+}
+
+bool get_best_match(PinyinLookup * pinyin_lookup,
+                    PinyinKeyVector pinyins, TokenVector tokens){
+    /* initialize constraints. */
+    CandidateConstraints constraints = g_array_new
+        (FALSE, FALSE, sizeof(lookup_constraint_t));
+    g_array_set_size(constraints, pinyins->len);
+    for ( size_t i = 0; i < constraints->len; ++i ) {
+        lookup_constraint_t * constraint = &g_array_index
+            (constraints, lookup_constraint_t, i);
+        constraint->m_type = NO_CONSTRAINT;
+    }
+
+    return pinyin_lookup->get_best_match(pinyins, constraints, tokens);
+}
+
 int main(int argc, char * argv[]){
+    const char * evals_text = "../../data/evals.text";
+
+    PinyinCustomSettings custom;
+    PinyinLargeTable largetable(&custom);
+
+    MemoryChunk * new_chunk = new MemoryChunk;
+    new_chunk->load("../../data/pinyin_index.bin");
+    largetable.load(new_chunk);
+
+    FacadePhraseIndex phrase_index;
+    new_chunk = new MemoryChunk;
+    new_chunk->load("../../data/gb_char.bin");
+    phrase_index.load(1, new_chunk);
+    new_chunk = new MemoryChunk;
+    new_chunk->load("../../data/gbk_char.bin");
+    phrase_index.load(2, new_chunk);
+
+    PhraseLargeTable phrases;
+    new_chunk = new MemoryChunk;
+    new_chunk->load("../../data/phrase_index.bin");
+    phrases.load(new_chunk);
+
+    Bigram system_bigram;
+    system_bigram.attach("../../data/bigram.db", ATTACH_READONLY);
+    Bigram user_bigram;
+    user_bigram.attach(NULL, ATTACH_CREATE|ATTACH_READWRITE);
+
+    PinyinLookup pinyin_lookup(&custom, &largetable, &phrase_index,
+                               &system_bigram, &user_bigram);
+
+    /* open evals.text. */
+    FILE * evals_file = fopen(evals_text, "r");
+    if ( NULL == evals_file ) {
+        fprintf(stderr, "Can't open file:%s\n", evals_text);
+        exit(ENOENT);
+    }
+
+    /* Evaluates the correction rate of test text documents. */
+    size_t tested_count = 0; size_t passed_count = 0;
+    char* linebuf = NULL; size_t size = 0;
+    TokenVector tokens = g_array_new(FALSE, TRUE, sizeof(phrase_token_t));
+    TokenVector guessed_tokens = g_array_new
+        (FALSE, TRUE, sizeof(phrase_token_t));
+    PinyinKeyVector pinyins = g_array_new(FALSE, TRUE, sizeof(PinyinKey));
+
+    phrase_token_t token;
+    while( getline(&linebuf, &size, evals_file) ) {
+        if ( feof(evals_file) )
+            break;
+        if ( '\n' == linebuf[strlen(linebuf)-1] )
+            linebuf[strlen(linebuf)-1] = '\0';
+
+        glong phrase_len = 0;
+        utf16_t * phrase = g_utf8_to_utf16(linebuf, -1, NULL, &phrase_len, NULL);
+
+        token = 0;
+        if ( 0 != phrase_len ) {
+            int result = phrases.search( phrase_len, phrase, token);
+            if ( ! (result & SEARCH_OK) )
+                token = 0;
+            g_free(phrase);
+            phrase = NULL;
+        }
+
+        if ( 0 == token ) {
+            if ( tokens->len ) { /* one test. */
+                get_possible_pinyin(&phrase_index, tokens, pinyins);
+                get_best_match(&pinyin_lookup, pinyins, guessed_tokens);
+                /* compare the results */
+                char * sentence = NULL; char * guessed_sentence = NULL;
+                pinyin_lookup.convert_to_utf8(tokens, sentence);
+                pinyin_lookup.convert_to_utf8
+                    (guessed_tokens, guessed_sentence);
+
+                if (strcmp(sentence, guessed_sentence) != 0){
+                    fprintf(stderr, "test sentence:%s\n", sentence);
+                    fprintf(stderr, "guessed sentence:%s\n", guessed_sentence);
+                    fprintf(stderr, "the result mis-matches.\n");
+                    tested_count ++;
+                } else {
+                    tested_count ++; passed_count ++;
+                }
+
+                g_free(sentence); g_free(guessed_sentence);
+                g_array_set_size(tokens, 0);
+            }
+        } else {
+            g_array_append_val(tokens, token);
+        }
+    }
+
+    parameter_t rate = passed_count / (parameter_t) tested_count;
+    printf("correction rate:%f\n", rate);
+
+    fclose(evals_file);
+    free(linebuf);
     return 0;
 }