auto rename pdf files

Author: longendu

aiutil/__init__.py

@@ -1,3 +1,3 @@
 """A utils Python package for data scientists."""
 
-__version__ = "0.86.1"
+__version__ = "0.87.0"

aiutil/pdf.py

@@ -1,7 +1,13 @@
 """Manipulating PDFs."""
 
+import datetime
+from pathlib import Path
+import re
 from typing import Iterable
 from pypdf import PdfWriter, PdfReader
+import pdfplumber
+
+FMT = "%Y%m%d"
 
 
 def extract_pages(file: str, subfiles: dict[str, int | Iterable[int]]) -> None:
@@ -42,3 +48,45 @@ def _extract_pages(
         writer.add_page(reader.pages[index])
     with open(output, "wb") as fout:
         writer.write(fout)
+
+
+def extract_text_first_page(path: str | Path) -> str:
+    """Extract the text of the first page of a PDF file.
+
+    :param path: The path of the PDF file.
+    :return: The text of the first page.
+    """
+    with pdfplumber.open(path) as pdf:
+        page = pdf.pages[0]
+        return page.extract_text()
+
+
+def _rename_puget_sound_energy(path: Path, text_first_page: str) -> Path:
+    m = re.search(r"Issued: (\w+ \d{1,2}, \d{4})", text_first_page)
+    date = datetime.datetime.strptime(m.group(1), "%B %d, %Y").strftime(FMT)
+    path_new = path.with_name(f"pse_{date}.pdf")
+    path.rename(path_new)
+    return path_new
+
+
+def _rename_bellevue_water(path: Path, text_first_page: str) -> Path:
+    m = re.search(r"Bill Date: (\d{1,2}/\d{1,2}/\d{4})", text_first_page)
+    date = datetime.datetime.strptime(m.group(1), "%m/%d/%Y").strftime(FMT)
+    path_new = path.with_name(f"bellevue_water_{date}.pdf")
+    path.rename(path_new)
+    return path_new
+
+
+def rename_auto(path: str | Path) -> Path:
+    """Rename a PDF file automatically based on its content.
+
+    :param path: The path of the PDF file.
+    :return: The path of the renamed PDF file.
+    """
+    if isinstance(path, str):
+        path = Path(path)
+    text = extract_text_first_page(path)
+    if "Puget Sound Energy" in text:
+        return _rename_puget_sound_energy(path, text)
+    if "MyUtilityBill.bellevuewa.gov" in text:
+        return _rename_bellevue_water(path, text)

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "aiutil"
-version = "0.86.1"
+version = "0.87.0"
 description = "A utils Python package for data scientists."
 authors = [{ name = "Benjamin Du", email = "[email protected]" }]
 requires-python = ">=3.10,<3.14"
@@ -30,6 +30,7 @@ dependencies = [
     "paramiko>=3.2.0",
     "nbformat>=5.10.4",
     "nbconvert>=7.16.6",
+    "pdfplumber>=0.11.7",
 ]
 
 [project.optional-dependencies]