| 42 | |
| 43 | |
| 44 | def convert(jsonf, dic, refs, hyps, srcs, dic_src): |
| 45 | # logging info |
| 46 | logfmt = "%(asctime)s (%(module)s:%(lineno)d) %(levelname)s: %(message)s" |
| 47 | logging.basicConfig(level=logging.INFO, format=logfmt) |
| 48 | logging.info(get_commandline_args()) |
| 49 | |
| 50 | logging.info("reading %s", jsonf) |
| 51 | with codecs.open(jsonf, "r", encoding="utf-8") as f: |
| 52 | j = json.load(f) |
| 53 | |
| 54 | # target dictionary |
| 55 | logging.info("reading %s", dic) |
| 56 | with codecs.open(dic, "r", encoding="utf-8") as f: |
| 57 | dictionary = f.readlines() |
| 58 | char_list_tgt = [entry.split(" ")[0] for entry in dictionary] |
| 59 | char_list_tgt.insert(0, "<blank>") |
| 60 | char_list_tgt.append("<eos>") |
| 61 | |
| 62 | # source dictionary |
| 63 | logging.info("reading %s", dic_src) |
| 64 | if dic_src: |
| 65 | with codecs.open(dic_src, "r", encoding="utf-8") as f: |
| 66 | dictionary = f.readlines() |
| 67 | char_list_src = [entry.split(" ")[0] for entry in dictionary] |
| 68 | char_list_src.insert(0, "<blank>") |
| 69 | char_list_src.append("<eos>") |
| 70 | |
| 71 | if hyps: |
| 72 | hyp_file = codecs.open(hyps[0], "w", encoding="utf-8") |
| 73 | ref_file = codecs.open(refs[0], "w", encoding="utf-8") |
| 74 | if srcs: |
| 75 | src_file = codecs.open(srcs[0], "w", encoding="utf-8") |
| 76 | |
| 77 | for x in j["utts"]: |
| 78 | # hyps |
| 79 | if hyps: |
| 80 | hyp_file.write(j["utts"][x]["output"][0]["rec_text"].replace("<eos>", "")), |
| 81 | |
| 82 | hyp_file.write( |
| 83 | " (" + j["utts"][x]["utt2spk"].replace("-", "_") + "-" + x + ")\n" |
| 84 | ) |
| 85 | |
| 86 | # ref |
| 87 | ref_file.write(j["utts"][x]["output"][0]["text"]), |
| 88 | |
| 89 | ref_file.write( |
| 90 | " (" + j["utts"][x]["utt2spk"].replace("-", "_") + "-" + x + ")\n" |
| 91 | ) |
| 92 | |
| 93 | # src |
| 94 | if "tokenid_src" in j["utts"][x]["output"][0].keys(): |
| 95 | if dic_src: |
| 96 | seq = [ |
| 97 | char_list_src[int(i)] |
| 98 | for i in j["utts"][x]["output"][0]["tokenid_src"].split() |
| 99 | ] |
| 100 | else: |
| 101 | seq = [ |