Varun Hemachandran
commited on
Commit
·
a16b08b
1
Parent(s):
0cd38af
Update dataset files with actual PDF metadata
Browse files- dataset_dict.json +6 -6
- default/train.json +0 -0
- generate_dataset_files.py +63 -0
dataset_dict.json
CHANGED
@@ -3,15 +3,15 @@
|
|
3 |
"splits": {
|
4 |
"train": {
|
5 |
"name": "train",
|
6 |
-
"num_bytes":
|
7 |
-
"num_examples":
|
8 |
"dataset_name": "jfk-files-2025"
|
9 |
}
|
10 |
},
|
11 |
"download_checksums": {},
|
12 |
-
"download_size":
|
13 |
"post_processing_size": null,
|
14 |
-
"dataset_size":
|
15 |
-
"size_in_bytes":
|
16 |
}
|
17 |
-
}
|
|
|
3 |
"splits": {
|
4 |
"train": {
|
5 |
"name": "train",
|
6 |
+
"num_bytes": 6468037461,
|
7 |
+
"num_examples": 2181,
|
8 |
"dataset_name": "jfk-files-2025"
|
9 |
}
|
10 |
},
|
11 |
"download_checksums": {},
|
12 |
+
"download_size": 6468037461,
|
13 |
"post_processing_size": null,
|
14 |
+
"dataset_size": 6468037461,
|
15 |
+
"size_in_bytes": 6468037461
|
16 |
}
|
17 |
+
}
|
default/train.json
CHANGED
The diff for this file is too large to render.
See raw diff
|
|
generate_dataset_files.py
ADDED
@@ -0,0 +1,63 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
#!/usr/bin/env python3
|
2 |
+
import os
|
3 |
+
import json
|
4 |
+
import glob
|
5 |
+
from tqdm import tqdm
|
6 |
+
|
7 |
+
# Directory containing the PDF files
|
8 |
+
data_dir = "data"
|
9 |
+
|
10 |
+
# Get all PDF files
|
11 |
+
pdf_files = glob.glob(os.path.join(data_dir, "*.pdf"))
|
12 |
+
|
13 |
+
# Create dataset structure
|
14 |
+
dataset = {
|
15 |
+
"file_name": [],
|
16 |
+
"file_path": [],
|
17 |
+
"file_size": [],
|
18 |
+
"content_type": []
|
19 |
+
}
|
20 |
+
|
21 |
+
# Add information for each PDF file
|
22 |
+
for pdf_file in tqdm(pdf_files, desc="Processing PDF files"):
|
23 |
+
file_name = os.path.basename(pdf_file)
|
24 |
+
file_path = pdf_file
|
25 |
+
file_size = os.path.getsize(pdf_file)
|
26 |
+
|
27 |
+
dataset["file_name"].append(file_name)
|
28 |
+
dataset["file_path"].append(file_path)
|
29 |
+
dataset["file_size"].append(file_size)
|
30 |
+
dataset["content_type"].append("application/pdf")
|
31 |
+
|
32 |
+
# Create the default directory if it doesn't exist
|
33 |
+
os.makedirs("default", exist_ok=True)
|
34 |
+
|
35 |
+
# Write the dataset to a JSON file
|
36 |
+
with open("default/train.json", "w") as f:
|
37 |
+
json.dump(dataset, f, indent=2)
|
38 |
+
|
39 |
+
# Update dataset_dict.json with correct information
|
40 |
+
dataset_dict = {
|
41 |
+
"default": {
|
42 |
+
"splits": {
|
43 |
+
"train": {
|
44 |
+
"name": "train",
|
45 |
+
"num_bytes": sum(dataset["file_size"]),
|
46 |
+
"num_examples": len(dataset["file_name"]),
|
47 |
+
"dataset_name": "jfk-files-2025"
|
48 |
+
}
|
49 |
+
},
|
50 |
+
"download_checksums": {},
|
51 |
+
"download_size": sum(dataset["file_size"]),
|
52 |
+
"post_processing_size": None,
|
53 |
+
"dataset_size": sum(dataset["file_size"]),
|
54 |
+
"size_in_bytes": sum(dataset["file_size"])
|
55 |
+
}
|
56 |
+
}
|
57 |
+
|
58 |
+
# Write the dataset_dict.json file
|
59 |
+
with open("dataset_dict.json", "w") as f:
|
60 |
+
json.dump(dataset_dict, f, indent=2)
|
61 |
+
|
62 |
+
print(f"Successfully processed {len(dataset['file_name'])} PDF files")
|
63 |
+
print(f"Total dataset size: {sum(dataset['file_size']) / (1024*1024):.2f} MB")
|