4 files changed, 294 insertions, 0 deletions
diff --git a/kaldi_io/src/cwrapper_kaldi.cpp b/kaldi_io/src/cwrapper_kaldi.cpp
new file mode 100644
index 0000000..f48d343
--- /dev/null
+++ b/kaldi_io/src/cwrapper_kaldi.cpp
@@ -0,0 +1,111 @@
+#include <string>
+#include "base/kaldi-common.h"
+#include "hmm/posterior.h"
+#include "util/table-types.h"
+typedef kaldi::BaseFloat BaseFloat;
+
+extern "C" {
+#include "cwrapper_kaldi.h"
+#include "string.h"
+#include "assert.h"
+#include "nerv/common.h"
+
+    extern Matrix *nerv_matrix_host_float_create(long nrow, long ncol, Status *status);
+    extern Matrix *nerv_matrix_host_double_create(long nrow, long ncol, Status *status);
+
+    struct KaldiFeatureRepo {
+        kaldi::SequentialBaseFloatMatrixReader* feature_reader;
+        string utt;
+    };
+
+    KaldiFeatureRepo *kaldi_feature_repo_new(const char *feature_rspecifier) {
+        KaldiFeatureRepo *repo = new KaldiFeatureRepo();
+        repo->feature_reader = new kaldi::SequentialBaseFloatMatrixReader(string(feature_rspecifier));
+        return repo;
+    }
+
+    Matrix *kaldi_feature_repo_read_utterance(KaldiFeatureRepo *repo, lua_State *L, int debug) {
+        Matrix *mat;                    /* nerv implementation */
+
+        repo->utt = repo->feature_reader->Key();
+        kaldi::Matrix<BaseFloat> kmat = repo->feature_reader->Value();
+
+        int n = kmat.NumRows();
+        int m = kmat.NumCols();
+        Status status;
+        assert(sizeof(BaseFloat) == sizeof(float));
+        if(sizeof(BaseFloat) == sizeof(float))
+            mat = nerv_matrix_host_float_create(n, m, &status);
+        else if(sizeof(BaseFloat) == sizeof(double))
+            mat = nerv_matrix_host_double_create(n, m, &status);
+        NERV_LUA_CHECK_STATUS(L, status);
+        size_t stride = mat->stride;
+        if (debug)
+            fprintf(stderr, "[kaldi] feature: %s %d %d\n", repo->utt.c_str(), n, m);
+
+        for (int i = 0; i < n; i++)
+        {
+            const BaseFloat *row = kmat.RowData(i);
+            BaseFloat *nerv_row = (BaseFloat *)((char *)mat->data.f + i * stride);
+            /* use memmove to copy the row, since KaldiLib uses compact storage */
+            memmove(nerv_row, row, sizeof(BaseFloat) * m);
+        }
+        return mat;
+    }
+
+    void kaldi_feature_repo_next(KaldiFeatureRepo *repo) {
+        repo->feature_reader->Next();
+    }
+
+    int kaldi_feature_repo_is_end(KaldiFeatureRepo *repo) {
+        return repo->feature_reader->Done();
+    }
+
+    void kaldi_feature_repo_destroy(KaldiFeatureRepo *repo) {
+        if (repo->feature_reader)
+            delete repo->feature_reader;
+        delete repo;
+    }
+
+    struct KaldiLabelRepo {
+        kaldi::RandomAccessPosteriorReader *targets_reader;
+    };
+
+    KaldiLabelRepo *kaldi_label_repo_new(const char *targets_rspecifier, const char *fmt) {
+        KaldiLabelRepo *repo = new KaldiLabelRepo();
+        repo->targets_reader = new kaldi::RandomAccessPosteriorReader(string(targets_rspecifier));
+        return repo;
+    }
+
+    Matrix *kaldi_label_repo_read_utterance(KaldiLabelRepo *repo, KaldiFeatureRepo *frepo, int frm_ext, int nframes,
+                                            lua_State *L,
+                                            int debug) {
+        Matrix *mat;
+        kaldi::Posterior targets = repo->targets_reader->Value(frepo->utt);
+
+        int n = targets.size() < nframes ? targets.size() : nframes;
+        int m = (int)targets[0].size();
+
+        Status status;
+        assert(sizeof(BaseFloat) == sizeof(float));
+        if(sizeof(BaseFloat) == sizeof(float))
+            mat = nerv_matrix_host_float_create(n, m, &status);
+        else if(sizeof(BaseFloat) == sizeof(double))
+            mat = nerv_matrix_host_double_create(n, m, &status);
+        NERV_LUA_CHECK_STATUS(L, status);
+        size_t stride = mat->stride;
+
+        if (debug)
+            fprintf(stderr, "[kaldi] label: %s %d %d\n", frepo->utt.c_str(), n, m);
+        for (int i = 0; i < n; i++)
+            for(int j = 0; j < m; j++)
+                *((BaseFloat *)((char *)mat->data.f + (i * stride + j))) = (BaseFloat)targets[i][j].first;
+        return mat;
+    }
+
+    void kaldi_label_repo_destroy(KaldiLabelRepo *repo) {
+        if(repo->targets_reader)
+            delete repo->targets_reader;
+        delete repo;
+    }
+}
diff --git a/kaldi_io/src/cwrapper_kaldi.h b/kaldi_io/src/cwrapper_kaldi.h
new file mode 100644
index 0000000..e34cb5a
--- /dev/null
+++ b/kaldi_io/src/cwrapper_kaldi.h
@@ -0,0 +1,29 @@
+#ifndef NERV_kaldi_KALDI_IO_CWRAPPER
+#define NERV_kaldi_KALDI_IO_CWRAPPER
+#include "nerv/matrix/matrix.h"
+#include "nerv/common.h"
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+    typedef struct KaldiFeatureRepo KaldiFeatureRepo;
+
+    KaldiFeatureRepo *kaldi_feature_repo_new(const char *);
+    Matrix *kaldi_feature_repo_read_utterance(KaldiFeatureRepo *repo, lua_State *L, int debug);
+    void kaldi_feature_repo_next(KaldiFeatureRepo *repo);
+    int kaldi_feature_repo_is_end(KaldiFeatureRepo *repo);
+    void kaldi_feature_repo_destroy(KaldiFeatureRepo *repo);
+
+    typedef struct KaldiLabelRepo KaldiLabelRepo;
+
+    KaldiLabelRepo *kaldi_label_repo_new(const char *, const char *fmt);
+
+    Matrix *kaldi_label_repo_read_utterance(KaldiLabelRepo *repo, KaldiFeatureRepo *, int, int,
+                                            lua_State *L,
+                                            int debug);
+
+    void kaldi_label_repo_destroy(KaldiLabelRepo *repo);
+#ifdef __cplusplus
+}
+#endif
+#endif
diff --git a/kaldi_io/src/init.c b/kaldi_io/src/init.c
new file mode 100644
index 0000000..413452c
--- /dev/null
+++ b/kaldi_io/src/init.c
@@ -0,0 +1,106 @@
+#include "nerv/common.h"
+#include "cwrapper_kaldi.h"
+#include <stdio.h>
+
+const char *nerv_kaldi_feat_repo_tname = "nerv.KaldiFeatureRepo";
+const char *nerv_kaldi_label_repo_tname = "nerv.KaldiLabelRepo";
+const char *nerv_matrix_host_float_tname = "nerv.MMatrixFloat";
+
+static int feat_repo_new(lua_State *L) {
+    const char *feature_rsepcifier = luaL_checkstring(L, 1);
+    KaldiFeatureRepo *repo = kaldi_feature_repo_new(feature_rsepcifier);
+    luaT_pushudata(L, repo, nerv_kaldi_feat_repo_tname);
+    return 1;
+}
+
+static int feat_repo_destroy(lua_State *L) {
+    KaldiFeatureRepo *repo = luaT_checkudata(L, 1, nerv_kaldi_feat_repo_tname);
+    kaldi_feature_repo_destroy(repo);
+    return 0;
+}
+
+static int feat_repo_current_utterance(lua_State *L) {
+    KaldiFeatureRepo *repo = luaT_checkudata(L, 1, nerv_kaldi_feat_repo_tname);
+    int debug;
+    if (!lua_isboolean(L, 2))
+        nerv_error(L, "debug flag should be a boolean");
+    debug = lua_toboolean(L, 2);
+    Matrix *utter = kaldi_feature_repo_read_utterance(repo, L, debug);
+    luaT_pushudata(L, utter, nerv_matrix_host_float_tname);
+    return 1;
+}
+
+static int feat_repo_next(lua_State *L) {
+    KaldiFeatureRepo *repo = luaT_checkudata(L, 1, nerv_kaldi_feat_repo_tname);
+    kaldi_feature_repo_next(repo);
+    return 0;
+}
+
+static int feat_repo_is_end(lua_State *L) {
+    KaldiFeatureRepo *repo = luaT_checkudata(L, 1, nerv_kaldi_feat_repo_tname);
+    lua_pushboolean(L, kaldi_feature_repo_is_end(repo));
+    return 1;
+}
+
+static const luaL_Reg feat_repo_methods[] = {
+    {"cur_utter", feat_repo_current_utterance},
+    {"next", feat_repo_next},
+    {"is_end", feat_repo_is_end},
+    {NULL, NULL}
+};
+
+static int label_repo_new(lua_State *L) {
+    const char *targets_rspecifier = luaL_checkstring(L, 1);
+    const char *fmt = luaL_checkstring(L, 2);
+    KaldiLabelRepo *repo = kaldi_label_repo_new(targets_rspecifier, fmt);
+    luaT_pushudata(L, repo, nerv_kaldi_label_repo_tname);
+    return 1;
+}
+
+static int label_repo_read_utterance(lua_State *L) {
+    KaldiLabelRepo *repo = luaT_checkudata(L, 1, nerv_kaldi_label_repo_tname);
+    KaldiFeatureRepo *feat_repo = luaT_checkudata(L, 2, nerv_kaldi_feat_repo_tname);
+    int frm_ext, nframes, debug;
+    if (!lua_isnumber(L, 3))
+        nerv_error(L, "frm_ext should be a number");
+    frm_ext = lua_tonumber(L, 3);
+    if (!lua_isnumber(L, 4))
+        nerv_error(L, "nframes should be a number");
+    nframes = lua_tonumber(L, 4);
+    if (!lua_isboolean(L, 5))
+        nerv_error(L, "debug flag should be a boolean");
+    debug = lua_toboolean(L, 5);
+    Matrix *utter = kaldi_label_repo_read_utterance(repo, feat_repo, frm_ext, nframes, L, debug);
+    luaT_pushudata(L, utter, nerv_matrix_host_float_tname);
+    return 1;
+}
+
+static int label_repo_destroy(lua_State *L) {
+    KaldiLabelRepo *repo = luaT_checkudata(L, 1, nerv_kaldi_label_repo_tname);
+    kaldi_label_repo_destroy(repo);
+    return 0;
+}
+
+static const luaL_Reg label_repo_methods[] = {
+    {"get_utter", label_repo_read_utterance},
+    {NULL, NULL}
+};
+
+static void feat_repo_init(lua_State *L) {
+    luaT_newmetatable(L, nerv_kaldi_feat_repo_tname, NULL,
+                        feat_repo_new, feat_repo_destroy, NULL);
+    luaL_register(L, NULL, feat_repo_methods);
+    lua_pop(L, 1);
+}
+
+static void label_repo_init(lua_State *L) {
+    luaT_newmetatable(L, nerv_kaldi_label_repo_tname, NULL,
+                        label_repo_new, label_repo_destroy, NULL);
+    luaL_register(L, NULL, label_repo_methods);
+    lua_pop(L, 1);
+}
+
+void kaldi_io_init(lua_State *L) {
+    feat_repo_init(L);
+    label_repo_init(L);
+}
diff --git a/kaldi_io/src/test.c b/kaldi_io/src/test.c
new file mode 100644
index 0000000..e92b4c9
--- /dev/null
+++ b/kaldi_io/src/test.c
@@ -0,0 +1,48 @@
+/*********************************************************************************
+*     File Name           :     test.c
+*     Created By          :     YIMMON, [email protected]
+*     Creation Date       :     [2015-08-05 17:39]
+*     Last Modified       :     [2015-08-06 14:28]
+*     Description         :
+**********************************************************************************/
+
+#include "cwrapper_kaldi.h"
+#include <stdio.h>
+
+char feature_rspecifier[] = {"ark:/slfs6/users/ymz09/kaldi/src/featbin/copy-feats scp:/slfs6/users/ymz09/swb_ivec/train_bp.scp ark:- |"};
+
+void print_nerv_matrix(Matrix *mat) {
+    int n = mat->nrow;
+    int m = mat->ncol;
+    int i, j;
+    size_t stride = mat->stride;
+    for (i = 0; i < n; i++)
+    {
+        float *nerv_row = (float *)((char *)mat->data.f + i * stride);
+        for (j = 0; j < m; j++)
+            printf("%.8f ", nerv_row[j]);
+        puts("");
+    }
+}
+
+int main(int argc, char *argv[])
+{
+    Matrix *mat;
+    KaldiFeatureRepo *repo = kaldi_feature_repo_new(feature_rspecifier);
+
+    mat = kaldi_feature_repo_read_utterance(repo, NULL, 1);
+    printf("1st uttrance\n");
+    print_nerv_matrix(mat);
+
+    kaldi_feature_repo_next(repo);
+
+    mat = kaldi_feature_repo_read_utterance(repo, NULL, 1);
+    printf("2nd uttrance\n");
+    print_nerv_matrix(mat);
+
+    printf("is end: %d\n", kaldi_feature_repo_is_end(repo));
+
+    kaldi_feature_repo_destroy(repo);
+
+    return 0;
+}