96abhishekarora
commited on
Commit
•
f35912b
1
Parent(s):
806a21e
testing_loader
Browse files- AmericanStories.py +146 -39
AmericanStories.py
CHANGED
@@ -1,60 +1,167 @@
|
|
1 |
import json
|
2 |
import tarfile
|
3 |
-
from datasets import DatasetInfo, DatasetBuilder, DownloadManager
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4 |
|
5 |
_CITATION = """\
|
6 |
Coming Soon
|
7 |
"""
|
8 |
|
9 |
_DESCRIPTION = """\
|
10 |
-
American Stories offers high-quality structured data from historical newspapers suitable for pre-training large language models to enhance the understanding of historical English and world knowledge. It can also be integrated into external databases of retrieval-augmented language models, enabling broader access to historical information, including interpretations of political events and intricate details about people's ancestors. Additionally, the structured article texts facilitate the application of transformer-based methods for popular tasks like detecting reproduced content, significantly improving accuracy compared to traditional OCR methods. American Stories serves as a substantial and valuable dataset for advancing multimodal layout analysis models and other multimodal applications.
|
11 |
-
|
|
|
12 |
|
13 |
-
class AmericanStories(DatasetBuilder):
|
14 |
-
VERSION = "0.0.1"
|
15 |
|
16 |
-
BUILDER_CONFIGS = [
|
17 |
-
DatasetBuilderConfig(
|
18 |
-
name="AmericanStories",
|
19 |
-
version=VERSION,
|
20 |
-
description=_DESCRIPTION,
|
21 |
-
),
|
22 |
-
]
|
23 |
|
24 |
-
|
25 |
-
|
26 |
-
|
27 |
-
|
28 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
29 |
|
30 |
return datasets.DatasetInfo(
|
|
|
31 |
description=_DESCRIPTION,
|
32 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
33 |
citation=_CITATION,
|
34 |
)
|
35 |
|
36 |
-
def _split_generators(self, dl_manager):
|
37 |
-
|
38 |
-
|
39 |
-
|
40 |
-
|
41 |
-
|
42 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
43 |
return [
|
44 |
-
|
45 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
46 |
]
|
|
|
|
|
|
|
47 |
|
48 |
-
|
49 |
-
|
50 |
-
|
51 |
-
|
52 |
-
|
53 |
-
|
54 |
-
|
55 |
-
|
56 |
-
|
57 |
-
|
58 |
-
|
59 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
60 |
}
|
|
|
1 |
import json
|
2 |
import tarfile
|
3 |
+
from datasets import DatasetInfo, DatasetBuilder, DownloadManager,BuilderConfig, SplitGenerator, Split, Version
|
4 |
+
import datasets
|
5 |
+
import os
|
6 |
+
import requests
|
7 |
+
import re
|
8 |
+
|
9 |
+
|
10 |
+
###GEt list of files
|
11 |
+
DATASET_URL="https://huggingface.co/datasets/dell-research-harvard/AmericanStories/blob/main/"
|
12 |
+
def get_list_of_files(url):
|
13 |
+
page = requests.get(url).text
|
14 |
+
links=re.findall(r'href=[\'"]?([^\'" >]+)', page)
|
15 |
+
###Get only links containing faro_
|
16 |
+
links=[link for link in links if link.startswith('faro_')]
|
17 |
+
return links
|
18 |
+
|
19 |
+
###Arrange into splits by year - files follow the format faro_YYYY.tar.gz
|
20 |
+
def get_splits(links):
|
21 |
+
splits={}
|
22 |
+
years=[]
|
23 |
+
for link in links:
|
24 |
+
year=link.split('_')[1].split('.')[0]
|
25 |
+
if year not in splits:
|
26 |
+
splits[year]=[]
|
27 |
+
splits[year].append(link)
|
28 |
+
years.append(year)
|
29 |
+
return splits,years
|
30 |
+
|
31 |
+
####data dir
|
32 |
+
DATA_DIR="."
|
33 |
+
|
34 |
+
def make_year_file_splits(data_dir):
|
35 |
+
###Get list of files
|
36 |
+
data_files=os.listdir(data_dir)
|
37 |
+
###Get only files containing faro_
|
38 |
+
data_files=[file for file in data_files if file.startswith('faro_')]
|
39 |
+
|
40 |
+
###Arrange into splits by year - files follow the format faro_YYYY.tar.gz
|
41 |
+
splits={}
|
42 |
+
years=[]
|
43 |
+
for file in data_files:
|
44 |
+
year=file.split('_')[1].split('.')[0]
|
45 |
+
if year not in splits:
|
46 |
+
splits[year]=[]
|
47 |
+
splits[year].append(file)
|
48 |
+
years.append(year)
|
49 |
+
return splits, years
|
50 |
+
|
51 |
+
|
52 |
+
|
53 |
+
|
54 |
|
55 |
_CITATION = """\
|
56 |
Coming Soon
|
57 |
"""
|
58 |
|
59 |
_DESCRIPTION = """\
|
60 |
+
American Stories offers high-quality structured data from historical newspapers suitable for pre-training large language models to enhance the understanding of historical English and world knowledge. It can also be integrated into external databases of retrieval-augmented language models, enabling broader access to historical information, including interpretations of political events and intricate details about people's ancestors. Additionally, the structured article texts facilitate the application of transformer-based methods for popular tasks like detecting reproduced content, significantly improving accuracy compared to traditional OCR methods. American Stories serves as a substantial and valuable dataset for advancing multimodal layout analysis models and other multimodal applications. """
|
61 |
+
|
62 |
+
_FILE_DICT,_YEARS=make_year_file_splits(DATA_DIR)
|
63 |
|
|
|
|
|
64 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
65 |
|
66 |
+
|
67 |
+
class AmericanStories(datasets.GeneratorBasedBuilder):
|
68 |
+
"""TODO: Short description of my dataset."""
|
69 |
+
|
70 |
+
VERSION = datasets.Version("0.0.1")
|
71 |
+
|
72 |
+
# This is an example of a dataset with multiple configurations.
|
73 |
+
# If you don't want/need to define several sub-sets in your dataset,
|
74 |
+
# just remove the BUILDER_CONFIG_CLASS and the BUILDER_CONFIGS attributes.
|
75 |
+
|
76 |
+
# If you need to make complex sub-parts in the datasets with configurable options
|
77 |
+
# You can create your own builder configuration class to store attribute, inheriting from datasets.BuilderConfig
|
78 |
+
# BUILDER_CONFIG_CLASS = MyBuilderConfig
|
79 |
+
|
80 |
+
# You will be able to load one or the other configurations in the following list with
|
81 |
+
# data = datasets.load_dataset('my_dataset', 'first_domain')
|
82 |
+
# data = datasets.load_dataset('my_dataset', 'second_domain')
|
83 |
+
BUILDER_CONFIGS = [datasets.BuilderConfig(name="american_stories", version="0.0.1", description="This part of my dataset covers a first domain")]
|
84 |
+
|
85 |
+
|
86 |
+
def _info(self):
|
87 |
+
# TODO: This method specifies the datasets.DatasetInfo object which contains informations and typings for the dataset
|
88 |
+
features = datasets.Features(
|
89 |
+
{ "newspaper_name": datasets.Value("string"),
|
90 |
+
"edition": datasets.Value("string"),
|
91 |
+
"date": datasets.Value("string"),
|
92 |
+
"page": datasets.Value("string"),
|
93 |
+
"headline": datasets.Value("string"),
|
94 |
+
"byline": datasets.Value("string"),
|
95 |
+
"article": datasets.Value("string")
|
96 |
+
# These are the features of your dataset like images, labels ...
|
97 |
+
}
|
98 |
+
)
|
99 |
|
100 |
return datasets.DatasetInfo(
|
101 |
+
# This is the description that will appear on the datasets page.
|
102 |
description=_DESCRIPTION,
|
103 |
+
# This defines the different columns of the dataset and their types
|
104 |
+
features=features, # Here we define them above because they are different between the two configurations
|
105 |
+
# If there's a common (input, target) tuple from the features, uncomment supervised_keys line below and
|
106 |
+
# specify them. They'll be used if as_supervised=True in builder.as_dataset.
|
107 |
+
# supervised_keys=("sentence", "label"),
|
108 |
+
# Homepage of the dataset for documentation
|
109 |
+
# License for the dataset if available
|
110 |
+
# Citation for the dataset
|
111 |
citation=_CITATION,
|
112 |
)
|
113 |
|
114 |
+
def _split_generators(self, dl_manager,online=False):
|
115 |
+
# TODO: This method is tasked with downloading/extracting the data and defining the splits depending on the configuration
|
116 |
+
# If several configurations are possible (listed in BUILDER_CONFIGS), the configuration selected by the user is in self.config.name
|
117 |
+
|
118 |
+
# dl_manager is a datasets.download.DownloadManager that can be used to download and extract URLS
|
119 |
+
# It can accept any type or nested list/dict and will give back the same structure with the url replaced with path to local files.
|
120 |
+
# By default the archives will be extracted and a path to a cached folder where they are extracted is returned instead of the archive
|
121 |
+
if not online:
|
122 |
+
urls = _FILE_DICT
|
123 |
+
else:
|
124 |
+
_URL_DICT,year_list=get_splits(get_list_of_files(DATASET_URL))
|
125 |
+
urls = _URL_DICT
|
126 |
+
year_list=_YEARS
|
127 |
+
data_dir = dl_manager.download_and_extract(urls)
|
128 |
+
|
129 |
+
###REturn a list of splits - but each split is for a year!
|
130 |
return [
|
131 |
+
datasets.SplitGenerator(
|
132 |
+
name=year,
|
133 |
+
# These kwargs will be passed to _generate_examples
|
134 |
+
gen_kwargs={
|
135 |
+
"year_dir": os.path.join(data_dir[year][0], "mnt/122a7683-fa4b-45dd-9f13-b18cc4f4a187/ca_rule_based_fa_clean/faro_"+year),
|
136 |
+
"split": year,
|
137 |
+
},
|
138 |
+
) for year in year_list
|
139 |
]
|
140 |
+
|
141 |
+
|
142 |
+
|
143 |
|
144 |
+
# method parameters are unpacked from `gen_kwargs` as given in `_split_generators`
|
145 |
+
def _generate_examples(self, year_dir, split):
|
146 |
+
# TODO: This method handles input defined in _split_generators to yield (key, example) tuples from the dataset.
|
147 |
+
# The `key` is for legacy reasons (tfds) and is not important in itself, but must be unique for each example.
|
148 |
+
for filepath in os.listdir(year_dir):
|
149 |
+
with open(os.path.join(year_dir,filepath), encoding="utf-8") as f:
|
150 |
+
data = json.load(f)
|
151 |
+
scan_id=filepath.split('.')[0]
|
152 |
+
scan_date=filepath.split("_")[0]
|
153 |
+
scan_page=filepath.split("_")[1]
|
154 |
+
scan_edition=filepath.split("_")[-2][8:]
|
155 |
+
newspaper_name=data["lccn"]["title"]
|
156 |
+
full_articles_in_data=data["full articles"]
|
157 |
+
for article in full_articles_in_data:
|
158 |
+
article_id=str(article["full_article_id"]) +"_" +scan_id
|
159 |
+
yield article_id, {
|
160 |
+
"newspaper_name": newspaper_name,
|
161 |
+
"edition": scan_edition,
|
162 |
+
"date": scan_date,
|
163 |
+
"page": scan_page,
|
164 |
+
"headline": article["headline"],
|
165 |
+
"byline": article["byline"],
|
166 |
+
"article": article["article"]
|
167 |
}
|