File size: 3,192 Bytes
a6326c7 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 |
from utils import parse, write_jsonl_file, read_json_file
import os
def build_wiki_index(wiki_dir):
idx2path = dict()
for wiki_filename in os.listdir(wiki_dir):
wiki_path = os.path.join(wiki_dir, wiki_filename)
wiki_data = read_json_file(wiki_path)
wiki_idx = wiki_data["wikiDocumentIdx"]
idx2path[wiki_idx] = wiki_path
return idx2path
def merge_consecutive_turns(history):
new_history = []
cur_turn = None
for turn in history:
if cur_turn is None:
cur_turn = turn
cur_turn["docIdx"] = [cur_turn["docIdx"]]
else:
if cur_turn["uid"] == turn["uid"]:
cur_turn["text"] += " " + turn["text"].strip()
if turn["docIdx"] not in cur_turn["docIdx"]:
cur_turn["docIdx"].append(turn["docIdx"])
else:
new_history.append(cur_turn)
cur_turn = turn
cur_turn["docIdx"] = [cur_turn["docIdx"]]
if cur_turn is not None:
new_history.append(cur_turn)
return new_history
def load_train_files(args):
conv_dir = os.path.join(args.input_dir, "Conversations")
input_dir = os.path.join(conv_dir, "train")
filenames = [filename for filename in os.listdir(input_dir)]
return filenames
def preprocess(args, split, train_files=None):
wiki_dir = os.path.join(args.input_dir, "WikiData")
conv_dir = os.path.join(args.input_dir, "Conversations")
idx2path = build_wiki_index(wiki_dir)
input_dir = os.path.join(conv_dir, split)
if split == "valid":
split = "dev"
output_file = os.path.join(args.output_dir, f"{split}.jsonl")
processed_data = []
def flip_role(role):
if role == "user1":
return "user2"
elif role == "user2":
return "user1"
else:
raise ValueError(f"Unknown role: {role}")
for filename in os.listdir(input_dir):
if train_files is not None and filename in train_files:
continue
filepath = os.path.join(input_dir, filename)
data = read_json_file(filepath)
history = data["history"]
history = merge_consecutive_turns(history)
dialog = {"turn": "multi", "locale": "en", "dialog": []}
wikipath = idx2path[data["wikiDocumentIdx"]]
wikidata = read_json_file(wikipath)
flip = False
for idx, turn in enumerate(history):
if idx == 0 and turn["uid"] == "user2":
flip = True
dialog["dialog"].append(
{
"roles": [turn["uid"]] if not flip else [flip_role(turn["uid"])],
"utterance": turn["text"],
"knowledge_to_select": list(map(str, turn["docIdx"])),
}
)
dialog["knowledge"] = {"type": "dict", "value": wikidata}
processed_data.append(dialog)
write_jsonl_file(processed_data, output_file)
if __name__ == "__main__":
args = parse()
train_files = load_train_files(args)
preprocess(args, "train")
preprocess(args, "valid", train_files)
preprocess(args, "test", train_files)
|