#!/usr/bin/env python
"""Fetch the two EmoTweetID files this dataset is derived from (CC BY 4.0, Mendeley Data jzgnjsff9f).

  data/file1.csv : tweet, label          - the 2,243 manually annotated tweets (6 Ekman emotions)
  data/file2.csv : tweet_en              - the same rows, English translation, aligned on `index`

Both files are the `file_downloaded` endpoints of the dataset's public file ids; the row order and
index are identical, which `label_anger.py` asserts before it joins them.
"""
import os
import sys
import urllib.request

FILES = {
    "data/file1.csv": "https://data.mendeley.com/public-files/datasets/jzgnjsff9f/files/"
                      "56002c37-6fad-4949-b821-c803e9d5b31d/file_downloaded",
    "data/file2.csv": "https://data.mendeley.com/public-files/datasets/jzgnjsff9f/files/"
                      "c6503598-7b82-4d5e-906a-93386d73239a/file_downloaded",
}

def main():
    for path, url in FILES.items():
        if os.path.exists(path) and os.path.getsize(path) > 10_000:
            print("[fetch] %s already present (%d bytes)" % (path, os.path.getsize(path)))
            continue
        os.makedirs(os.path.dirname(path), exist_ok=True)
        print("[fetch] %s <- %s" % (path, url))
        with urllib.request.urlopen(url, timeout=120) as r, open(path + ".part", "wb") as fh:
            fh.write(r.read())
        os.replace(path + ".part", path)
    import pandas as pd
    f1, f2 = pd.read_csv("data/file1.csv", index_col=0), pd.read_csv("data/file2.csv", index_col=0)
    print("[fetch] file1 %s %s | file2 %s | aligned: %s | anger rows: %d" % (
        f1.shape, f1.columns.tolist(), f2.shape, (f1.index == f2.index).all(),
        int((f1["label"] == "anger").sum())))

if __name__ == "__main__":
    sys.exit(main())