zh_step.py

from datasets import load_from_disk, load_dataset, Dataset
import pandas as pd
import numpy as np
from functools import reduce
import pickle as pkl

file_path = "../graph.jsonl"
ds = load_dataset('json', data_files=file_path, )

df = pd.DataFrame(list(ds["train"]))
inference_l = df["inference"].map(lambda x: list(x.values())).map(
    lambda x: reduce(lambda a, b: a + b , map(lambda y: list(y.items()) ,x))
).map(
    lambda x: reduce(lambda a, b: a + b ,map(lambda y: y[1], x))
).explode().dropna().drop_duplicates().values.tolist()

event_l = df["event"].dropna().drop_duplicates().values.tolist()
len(inference_l), len(event_l)

with open("event_inference_ja.pkl", "wb") as f:
    pkl.dump(
        {
            "event_l": event_l,
            "inference_l": inference_l
        }, f
    )

eil = list(set(event_l + inference_l))
peil = [NEED_PREFIX, EFFECT_PREFIX, INTENT_PREFIX, REACT_PREFIX] + eil
with open("peil_ja.pkl", "wb") as f:
    pkl.dump(peil, f)

pd.Series(peil).to_csv("peil_ja.txt", header = None, index = False)

def read_file_to_lines(path):
    with open(path, "r") as f:
        return pd.Series(f.readlines()).map(lambda x: x.replace("\n", ""))

peil_zh = read_file_to_lines("../peil_ja zh.txt")

assert len(peil) == len(peil_zh)
d = dict(zip(peil, peil_zh))

with open("peil_ja_zh_d.pkl", "wb") as f:
    pkl.dump(d, f)

df["zh_event"] = df["event"].map(lambda x: d[x])

def one_infe_ele_map(inf_d):
    req = {}
    for k, v in inf_d.items():
        req[k] = {}
        for kk, vv in v.items():
            assert type(vv) == type([])
            req[k][kk] = list(map(lambda x: d[x], vv))
    return req

df["zh_inference"] = df["inference"].map(one_infe_ele_map)

with open("graph_ja_zh_df.pkl", "wb") as f:
    pkl.dump(df, f)

from huggingface_hub import HfApi
api = HfApi()

api.upload_file(
    path_or_fileobj="graph_ja_zh_df.pkl",
    path_in_repo="graph_ja_zh_df.pkl",
    repo_id="svjack/comet-atomic-ja-zh",
    repo_type="dataset",
)

zh_ds = Dataset.from_pandas(
    df[["zh_event", "zh_inference"]].rename(
    columns = {
        "zh_event": "event",
        "zh_inference": "inference"
    }
)
)

zh_ds.save_to_disk("graph_zh_ds_dir")