wrote export k mixture model

3 files changed, 111 insertions, 12 deletions

diff --git a/utils/storage/export_interpolation.cpp b/utils/storage/export_interpolation.cpp
index bcc90b8..23e4d4a 100644
--- a/utils/storage/export_interpolation.cpp
+++ b/utils/storage/export_interpolation.cpp
@@ -30,12 +30,12 @@
 void gen_unigram(FILE * output, FacadePhraseIndex * phrase_index);
 void gen_bigram(FILE * output, FacadePhraseIndex * phrase_index, Bigram * bigram);
 
-void begin_data(FILE * file){
-    fprintf(file, "\\data model interpolation\n");
+void begin_data(FILE * output){
+    fprintf(output, "\\data model interpolation\n");
 }
 
-void end_data(FILE * file){
-    fprintf(file, "\\end\n");
+void end_data(FILE * output){
+    fprintf(output, "\\end\n");
 }
 
 int main(int argc, char * argv[]){
diff --git a/utils/training/Makefile.am b/utils/training/Makefile.am
index 44e4dad..adc0c30 100644
--- a/utils/training/Makefile.am
+++ b/utils/training/Makefile.am
@@ -20,7 +20,8 @@ MAINTAINERCLEANFILES    = Makefile.in
 INCLUDES		= -I$(top_srcdir)/src \
 			  -I$(top_srcdir)/src/include \
 			  -I$(top_srcdir)/src/storage \
-                          -I$(top_srcdir)/src/lookup \
+              -I$(top_srcdir)/src/lookup \
+			  -I$(top_srcdir)/utils/storage \
 			  @GLIB2_CPPFLAGS@
 
 noinst_HEADERS		= k_mixture_model.h
@@ -32,7 +33,8 @@ noinst_PROGRAMS		= gen_ngram \
 			  estimate_interpolation \
 			  estimate_k_mixture_model \
 			  merge_k_mixture_model \
-			  prune_k_mixture_model
+			  prune_k_mixture_model \
+			  export_k_mixture_model
 
 gen_ngram_SOURCES	= gen_ngram.cpp
 
@@ -64,4 +66,8 @@ merge_k_mixture_model_LDADD = ../../src/libpinyin.la @GLIB2_LDFLAGS@
 
 prune_k_mixture_model_SOURCES = prune_k_mixture_model.cpp
 
-prune_k_mixture_model_LDADD = ../../src/libpinyin.la @GLIB2_LDFLAGS@
\ No newline at end of file
+prune_k_mixture_model_LDADD = ../../src/libpinyin.la @GLIB2_LDFLAGS@
+
+export_k_mixture_model_SOURCES = export_k_mixture_model.cpp
+
+export_k_mixture_model_LDADD = ../storage/libtagutils.la ../../src/libpinyin.la @GLIB2_LDFLAGS@
\ No newline at end of file
diff --git a/utils/training/export_k_mixture_model.cpp b/utils/training/export_k_mixture_model.cpp
index b167394..7802e4d 100644
--- a/utils/training/export_k_mixture_model.cpp
+++ b/utils/training/export_k_mixture_model.cpp
@@ -20,20 +20,29 @@
  */
 
 #include "pinyin.h"
+#include "tag_utility.h"
+#include "k_mixture_model.h"
 
-bool print_k_mixture_model_magic_header(KMixtureModelBigram * bigram){
+void print_help(){
+    printf("export_k_mixture_model [--k-mixture-model-file <FILENAME>]\n");
+}
+
+bool print_k_mixture_model_magic_header(FILE * output,
+                                        KMixtureModelBigram * bigram){
     KMixtureModelMagicHeader magic_header;
     if ( !bigram->get_magic_header(magic_header) ){
         fprintf(stderr, "no magic header in k mixture model.\n");
         exit(ENODATA);
     }
-    printf("\\data model \"k mixture model\" count %d N %d\n",
+    fprintf(output, "\\data model \"k mixture model\" count %d N %d\n",
            magic_header.m_WC, magic_header.m_N);
     return true;
 }
 
-bool print_k_mixture_model_array_header(KMixtureModelBigram * bigram){
-    printf("\1-gram\n");
+bool print_k_mixture_model_array_headers(FILE * output,
+                                         KMixtureModelBigram * bigram,
+                                         FacadePhraseIndex * phrase_index){
+    fprintf(output, "\\1-gram\n");
     GArray * items = g_array_new(FALSE, FALSE, sizeof(phrase_token_t));
     bigram->get_all_items(items);
 
@@ -41,11 +50,95 @@ bool print_k_mixture_model_array_header(KMixtureModelBigram * bigram){
         phrase_token_t * token = &g_array_index(items, phrase_token_t, i);
         KMixtureModelArrayHeader array_header;
         bigram->get_array_header(*token, array_header);
-        
+        char * phrase = taglib_token_to_string(phrase_index, *token);
+        if ( phrase )
+            fprintf(output, "\\item %s count %d\n", phrase, array_header.m_WC);
+
+        g_free(phrase);
     }
+    return true;
+}
+
+bool print_k_mixture_model_array_items(FILE * output,
+                                       KMixtureModelBigram * bigram,
+                                       FacadePhraseIndex * phrase_index){
+    fprintf(output, "\\2-gram\n");
+    GArray * items = g_array_new(FALSE, FALSE, sizeof(phrase_token_t));
+    bigram->get_all_items(items);
+
+    for (size_t i = 0; i < items->len; ++i) {
+        phrase_token_t * token = &g_array_index(items, phrase_token_t, i);
+        KMixtureModelSingleGram * single_gram = NULL;
+        assert(bigram->load(*token, single_gram));
+        FlexibleBigramPhraseArray array = g_array_new
+            (FALSE, FALSE, sizeof(KMixtureModelArrayItemWithToken));
+        single_gram->retrieve_all(array);
+
+        for (size_t m = 0; m < array->len; ++m){
+            KMixtureModelArrayItemWithToken * item = &g_array_index(array, KMixtureModelArrayItemWithToken, m);
+            char * word1 = taglib_token_to_string(phrase_index, *token);
+            char * word2 = taglib_token_to_string(phrase_index, item->m_token);
+
+            if (word1 && word2)
+                fprintf(output, "\\item %s %s count %d T %d N_n_0 %d Mr %d\n",
+                        word1, word2, item->m_item.m_WC, item->m_item.m_WC,
+                        item->m_item.m_N_n_0, item->m_item.m_Mr);
+
+            g_free(word1); g_free(word2);
+        }
 
+        g_array_free(array, TRUE);
+        delete single_gram;
+    }
+
+    g_array_free(items, TRUE);
+    return true;
+}
+
+bool end_data(FILE * output){
+    fprintf(output, "\\end\n");
     return true;
 }
 
 int main(int argc, char * argv[]){
+    int i = 1;
+    const char * k_mixture_model_filename = NULL;
+    FILE * output = stdout;
+
+    while ( i < argc ){
+        if ( strcmp ("--help", argv[i]) == 0 ){
+            print_help();
+            exit(0);
+        } else if ( strcmp ("--k-mixture-model-file", argv[i]) == 0 ){
+            if ( ++i > argc ){
+                print_help();
+                exit(EINVAL);
+            }
+            k_mixture_model_filename = argv[i];
+        } else {
+            print_help();
+            exit(EINVAL);
+        }
+    }
+
+    FacadePhraseIndex phrase_index;
+
+    //gb_char binary file
+    MemoryChunk * chunk = new MemoryChunk;
+    chunk->load("../../data/gb_char.bin");
+    phrase_index.load(1, chunk);
+
+    //gbk_char binary file
+    chunk = new MemoryChunk;
+    chunk->load("../../data/gbk_char.bin");
+    phrase_index.load(2, chunk);
+
+    KMixtureModelBigram bigram(K_MIXTURE_MODEL_MAGIC_NUMBER);
+    bigram.attach(k_mixture_model_filename, ATTACH_READONLY);
+
+    print_k_mixture_model_magic_header(output, &bigram);
+    print_k_mixture_model_array_headers(output, &bigram, &phrase_index);
+    print_k_mixture_model_array_items(output, &bigram, &phrase_index);
+
+    return 0;
 }