Datasets:

AstroCLIMB / code /wasp26_lib /build_base_df.py
fgrezes's picture
notebooks and code for wasp 2026
4528bac verified
Raw History Blame Contribute Delete
2.26 kB
import pandas as pd
from tqdm.auto import tqdm
from .encode_search_string import encode_search_string
from .get_all_image_links import get_all_image_links
from .extract_image_info import extract_image_info
def build_base_df(save_df_path=None):
"""
Builds the WASP 2026 - Shared task 3 base dataframe, images will be downloaded later.
If save_df is set to a path, saves to disk every loop (no checking, could be overwriting important files)
"""
# build dataframe
cols =["Image ID","Paper DOI", "Paper Title", "Image Caption", "Caption Title", "Image Bibliographic", "Image Copyright", "Image Authors", "Image Link",]
image_df = pd.DataFrame(columns=cols)
# total of 1092 pages for 2020 per page
# (bugs out after 50? 10K images)
# since searches are limited to 10K images returned (for now)
# go through by publication and year
for year in [2022, 2023, 2024, 2025]:
for pub in ["The Astrophysical Journal", "The Astronomical Journal", "The Astrophysical Journal Supplement Series", "The Astrophysical Journal Letters", "The Planetary Science Journal","Research Notes of the AAS"]:
print(f'year:{year} pub:{pub}')
for page in tqdm(range(1, 51)):
# Encode the proper 64 bit search url for a given page
url = encode_search_string(publication=pub,
fromYear=year,
toYear=year,
page=page)
# extract all image links
image_ids = get_all_image_links(url)
if image_ids:
# Loop through all the hyperlink tags, access the image details page and
# extract the relevant info and place in a dataframe
for image_id in image_ids:
out_dict = extract_image_info(image_id)
image_df = pd.concat([image_df, pd.DataFrame([out_dict])], ignore_index=True)
print('Dataframe size:', image_df.shape)
if save_df_path is not None:
image_df.to_csv(save_df_path, index=False)
return(image_df)