Spaces:
Running
Running
π¨ format
Browse filesSigned-off-by: peter szemraj <[email protected]>
- app.py +3 -1
- pdf2text.py +7 -2
app.py
CHANGED
|
@@ -146,7 +146,9 @@ if __name__ == "__main__":
|
|
| 146 |
|
| 147 |
gr.Markdown("## Load Inputs")
|
| 148 |
gr.Markdown("Upload your own file & replace the default")
|
| 149 |
-
gr.Markdown(
|
|
|
|
|
|
|
| 150 |
|
| 151 |
uploaded_file = gr.File(
|
| 152 |
label="Upload a PDF file",
|
|
|
|
| 146 |
|
| 147 |
gr.Markdown("## Load Inputs")
|
| 148 |
gr.Markdown("Upload your own file & replace the default")
|
| 149 |
+
gr.Markdown(
|
| 150 |
+
"_If no file is uploaded, a sample PDF will be used. PDFs are truncated to 20 pages._"
|
| 151 |
+
)
|
| 152 |
|
| 153 |
uploaded_file = gr.File(
|
| 154 |
label="Upload a PDF file",
|
pdf2text.py
CHANGED
|
@@ -32,12 +32,12 @@ from spellchecker import SpellChecker
|
|
| 32 |
from tqdm.auto import tqdm
|
| 33 |
|
| 34 |
|
| 35 |
-
|
| 36 |
def simple_rename(filepath, target_ext=".txt"):
|
| 37 |
_fp = Path(filepath)
|
| 38 |
basename = _fp.stem
|
| 39 |
return f"OCR_{basename}_{target_ext}"
|
| 40 |
|
|
|
|
| 41 |
def rm_local_text_files(name_contains="RESULT_"):
|
| 42 |
"""
|
| 43 |
rm_local_text_files - remove local text files
|
|
@@ -45,12 +45,17 @@ def rm_local_text_files(name_contains="RESULT_"):
|
|
| 45 |
Args:
|
| 46 |
name_contains (str, optional): [description]. Defaults to "OCR_".
|
| 47 |
"""
|
| 48 |
-
files = [
|
|
|
|
|
|
|
|
|
|
|
|
|
| 49 |
logging.info(f"removing {len(files)} text files")
|
| 50 |
for f in files:
|
| 51 |
os.remove(f)
|
| 52 |
logging.info("done")
|
| 53 |
|
|
|
|
| 54 |
def corr(
|
| 55 |
s: str,
|
| 56 |
add_space_when_numerics=False,
|
|
|
|
| 32 |
from tqdm.auto import tqdm
|
| 33 |
|
| 34 |
|
|
|
|
| 35 |
def simple_rename(filepath, target_ext=".txt"):
|
| 36 |
_fp = Path(filepath)
|
| 37 |
basename = _fp.stem
|
| 38 |
return f"OCR_{basename}_{target_ext}"
|
| 39 |
|
| 40 |
+
|
| 41 |
def rm_local_text_files(name_contains="RESULT_"):
|
| 42 |
"""
|
| 43 |
rm_local_text_files - remove local text files
|
|
|
|
| 45 |
Args:
|
| 46 |
name_contains (str, optional): [description]. Defaults to "OCR_".
|
| 47 |
"""
|
| 48 |
+
files = [
|
| 49 |
+
f
|
| 50 |
+
for f in Path.cwd().iterdir()
|
| 51 |
+
if f.is_file() and f.suffix == ".txt" and name_contains in f.name
|
| 52 |
+
]
|
| 53 |
logging.info(f"removing {len(files)} text files")
|
| 54 |
for f in files:
|
| 55 |
os.remove(f)
|
| 56 |
logging.info("done")
|
| 57 |
|
| 58 |
+
|
| 59 |
def corr(
|
| 60 |
s: str,
|
| 61 |
add_space_when_numerics=False,
|