test/unit/test_ngram/test_lm_write.c

   1 #include <ngram_model.h>
   2 #include <logmath.h>
   3 #include <strfuncs.h>
   4 #include <err.h>
   5
   6 #include "test_macros.h"
   7
   8 #include <stdio.h>
   9 #include <stdlib.h>
  10 #include <string.h>
  11
  12 static int
  13 test_lm_vals(ngram_model_t *model)
  14 {
  15         int32 n_used;
  16
  17         TEST_ASSERT(model);
  18         TEST_EQUAL(ngram_wid(model, "<UNK>"), 0);
  19         TEST_EQUAL(strcmp(ngram_word(model, 0), "<UNK>"), 0);
  20         TEST_EQUAL(ngram_wid(model, "absolute"), 13);
  21         TEST_EQUAL(strcmp(ngram_word(model, 13), "absolute"), 0);
  22         /* Test unigrams. */
  23         TEST_EQUAL_LOG(ngram_score(model, "<UNK>", NULL), -75346);
  24         TEST_EQUAL_LOG(ngram_bg_score(model, ngram_wid(model, "<UNK>"),
  25                                   NGRAM_INVALID_WID, &n_used), -75346);
  26         TEST_EQUAL(n_used, 1);
  27         TEST_EQUAL_LOG(ngram_score(model, "sphinxtrain", NULL), -64208);
  28         TEST_EQUAL_LOG(ngram_bg_score(model, ngram_wid(model, "sphinxtrain"),
  29                                   NGRAM_INVALID_WID, &n_used), -64208);
  30         TEST_EQUAL(n_used, 1);
  31         printf("FOO %d\n", ngram_score(model, "huggins", "david", NULL));
  32         printf("FOO %d\n", ngram_score(model, "daines", "huggins", "david", NULL));
  33         /* Test bigrams. */
  34         TEST_EQUAL_LOG(ngram_score(model, "huggins", "david", NULL), -831);
  35         /* Test trigrams. */
  36         TEST_EQUAL_LOG(ngram_score(model, "daines", "huggins", "david", NULL), -9450);
  37         return 0;
  38 }
  39
  40 int
  41 main(int argc, char *argv[])
  42 {
  43         logmath_t *lmath;
  44         ngram_model_t *model;
  45
  46         /* Initialize a logmath object to pass to ngram_read */
  47         lmath = logmath_init(1.0001, 0, 0);
  48
  49         /* Convert ARPA to DMP */
  50         E_INFO("Converting ARPA to DMP\n");
  51         model = ngram_model_read(NULL, LMDIR "/100.arpa.bz2", NGRAM_ARPA, lmath);
  52         test_lm_vals(model);
  53         TEST_EQUAL(0, ngram_model_write(model, "100.tmp.DMP", NGRAM_DMP));
  54         ngram_model_free(model);
  55
  56         /* Convert DMP to ARPA */
  57         E_INFO("Converting DMP to ARPA\n");
  58         model = ngram_model_read(NULL, LMDIR "/100.arpa.DMP", NGRAM_DMP, lmath);
  59         test_lm_vals(model);
  60         TEST_EQUAL(0, ngram_model_write(model, "100.tmp.arpa", NGRAM_ARPA));
  61         ngram_model_free(model);
  62
  63         /* Test converted DMP */
  64         E_INFO("Testing converted DMP\n");
  65         model = ngram_model_read(NULL, "100.tmp.DMP", NGRAM_DMP, lmath);
  66         test_lm_vals(model);
  67         ngram_model_free(model);
  68
  69         /* Test converted ARPA */
  70         E_INFO("Testing converted ARPA\n");
  71         model = ngram_model_read(NULL, "100.tmp.arpa", NGRAM_ARPA, lmath);
  72         test_lm_vals(model);
  73         ngram_model_free(model);
  74
  75         /* Convert DMP back to ARPA*/
  76         E_INFO("Converting ARPA back to DMP\n");
  77         model = ngram_model_read(NULL, "100.tmp.arpa", NGRAM_ARPA, lmath);
  78         test_lm_vals(model);
  79         TEST_EQUAL(0, ngram_model_write(model, "100.tmp.DMP", NGRAM_DMP));
  80         ngram_model_free(model);
  81
  82         /* Convert ARPA back to DMP */
  83         E_INFO("Converting DMP back to ARPA\n");
  84         model = ngram_model_read(NULL, "100.tmp.DMP", NGRAM_DMP, lmath);
  85         test_lm_vals(model);
  86         TEST_EQUAL(0, ngram_model_write(model, "100.tmp.arpa", NGRAM_ARPA));
  87         ngram_model_free(model);
  88
  89         logmath_free(lmath);
  90         return 0;
  91 }