scionoftech · February 17, 2025 04:27 · Jun 14, 2019 · Jun 14, 2019
diff --git a/pdf2excel.py b/pdf2excel.py
@@ -22,7 +22,7 @@ def to_excel(path, output_path):
 
 @click.command()
 @click.option('--input_path', '-ip', help='input file path')
-@click.option('--output_path', '-op', help='output file path')
+@click.option('--output_path', '-op', help='output folder path')
 def convert(input_path, output_path):
     """This program generates xlsx files from pdf."""
     to_excel(input_path, output_path)

diff --git a/pdf2excel.py b/pdf2excel.py
@@ -0,0 +1,32 @@
+import click
+from pathlib import Path
+import pdfplumber
+import pandas as pd
+from pandas import ExcelWriter
+
+
+def to_excel(path, output_path):
+    with pdfplumber.open(path) as pdf:
+        data = []
+        for i in range(len(pdf.pages)):
+            page = pdf.pages[i]
+            # print(first_page.chars[0])
+            data += page.extract_table()
+            # print(data)
+        df = pd.DataFrame(data)
+        filename = Path(path).name.split('.')
+        writer = ExcelWriter(output_path + os.sep + filename[0] + '.xlsx')
+        df.to_excel(writer, 'Sheet1', index=False)
+        writer.save()
+
+
+@click.command()
+@click.option('--input_path', '-ip', help='input file path')
+@click.option('--output_path', '-op', help='output file path')
+def convert(input_path, output_path):
+    """This program generates xlsx files from pdf."""
+    to_excel(input_path, output_path)
+
+
+if __name__ == '__main__':
+    convert()