MAINT: Move from PyPDF2 to pypdf

Author: MartinThoma

.isort.cfg

@@ -1,7 +1,7 @@
 # pre-commit run --all-files
 repos:
 -   repo: https://github.com/pre-commit/pre-commit-hooks
-    rev: v4.1.0
+    rev: v4.4.0
     hooks:
     -   id: check-ast
     -   id: check-byte-order-marker
@@ -17,25 +17,22 @@ repos:
         exclude: results/
     -   id: check-added-large-files
         args: ['--maxkb=1000']
--   repo: https://gitlab.com/pycqa/flake8
-    rev: 3.9.2
+-   repo: https://github.com/PyCQA/flake8
+    rev: 6.0.0
     hooks:
     -   id: flake8
         args: ["--ignore", "E,W,F"]
 # -   repo: https://github.com/pre-commit/mirrors-mypy
 #     rev: v0.942
 #     hooks:
 #     -   id: mypy
--   repo: https://github.com/asottile/seed-isort-config
-    rev: v2.2.0
-    hooks:
-    -   id: seed-isort-config
--   repo: https://github.com/pre-commit/mirrors-isort
-    rev: v5.10.1
+-   repo: https://github.com/pycqa/isort
+    rev: 5.11.4
     hooks:
     -   id: isort
+        name: isort (python)
 -   repo: https://github.com/psf/black
-    rev: 22.3.0
+    rev: 22.12.0
     hooks:
     -   id: black
 # -   repo: https://github.com/asottile/pyupgrade

.pre-commit-config.yaml

@@ -19,7 +19,7 @@
 import numpy as np
 import pdfminer
 import pdfplumber
-import PyPDF2
+import pypdf
 import pypdfium2 as pdfium
 import requests
 import tika
@@ -135,9 +135,9 @@ def pymupdf_get_text(data: bytes) -> str:
     return text
 
 
-def pypdf2_get_text(data: bytes) -> str:
+def pypdf_get_text(data: bytes) -> str:
     text = ""
-    reader = PyPDF2.PdfFileReader(BytesIO(data))
+    reader = pypdf.PdfReader(BytesIO(data))
     for page in reader.pages:
         text += page.extract_text() + "\n"
     return text
@@ -153,11 +153,11 @@ def pdfium_get_text(data: bytes) -> str:
     return text
 
 
-def pypdf2_watermarking(watermark_data: bytes, data: bytes) -> bytes:
-    watermark_pdf = PyPDF2.PdfReader(BytesIO(watermark_data))
+def pypdf_watermarking(watermark_data: bytes, data: bytes) -> bytes:
+    watermark_pdf = pypdf.PdfReader(BytesIO(watermark_data))
     watermark_page = watermark_pdf.pages[0]
-    reader = PyPDF2.PdfReader(BytesIO(data))
-    writer = PyPDF2.PdfWriter()
+    reader = pypdf.PdfReader(BytesIO(data))
+    writer = pypdf.PdfWriter()
     for page in reader.pages:
         page.merge_page(watermark_page)
         writer.add_page(page)
@@ -167,15 +167,15 @@ def pypdf2_watermarking(watermark_data: bytes, data: bytes) -> bytes:
         return bytes_stream.read()
 
 
-def pypdf2_image_extraction(data: bytes) -> List[Tuple[str, bytes]]:
+def pypdf_image_extraction(data: bytes) -> List[Tuple[str, bytes]]:
     images = []
     try:
-        reader = PyPDF2.PdfReader(BytesIO(data))
+        reader = pypdf.PdfReader(BytesIO(data))
         for page in reader.pages:
             for image in page.images:
                 images.append((image.name, image.data))
     except Exception as exc:
-        print(f"PyPDF2 Image extraction failure: {exc}")
+        print(f"pypdf Image extraction failure: {exc}")
     return images
 
 
@@ -444,7 +444,7 @@ def write_benchmark_report(
             if len([el for el in image_extraction_times[name] if el is not None]) > 0
         ]
         print(names)
-        print(image_extraction_times["pypdf2"])
+        print(image_extraction_times["pypdf"])
         averages = [np.mean(image_extraction_times[name]) for name in names]
         sort_order = np.argsort([avg for avg in averages])
         for place, index in enumerate(sort_order, start=1):
@@ -553,16 +553,16 @@ def get_text_extraction_score(doc: Document, library_name: str):
             license="Apache v2",
             last_release_date="2020-03-21",
         ),
-        "pypdf2": Library(
-            "PyPDF2",
-            "pypdf2",
-            "https://pypi.org/project/PyPDF2/",
-            pypdf2_get_text,
-            version=PyPDF2.__version__,
-            watermarking_function=pypdf2_watermarking,
+        "pypdf": Library(
+            "pypdf",
+            "pypdf",
+            "https://pypi.org/project/pypdf/",
+            pypdf_get_text,
+            version=pypdf.__version__,
+            watermarking_function=pypdf_watermarking,
             license="BSD 3-Clause",
             last_release_date="2022-09-25",
-            image_extraction_function=pypdf2_image_extraction,
+            image_extraction_function=pypdf_image_extraction,
         ),
         "pdfminer": Library(
             "pdfminer.six",

README.md

@@ -292,76 +292,76 @@
                 "read": 0.24839401245117188
             }
         },
-        "pypdf2": {
+        "pypdf": {
             "1601.03642": {
-                "image_extraction": 0.0035588741302490234,
-                "read": 0.1545581817626953,
-                "watermark": 0.4532604217529297
+                "image_extraction": 0.003720998764038086,
+                "read": 0.16930723190307617,
+                "watermark": 0.40448474884033203
             },
             "1602.06541": {
-                "image_extraction": 0.2223806381225586,
-                "read": 0.5582091808319092,
-                "watermark": 1.2958548069000244
+                "image_extraction": 0.21662640571594238,
+                "read": 0.5400035381317139,
+                "watermark": 1.1498773097991943
             },
             "1707.09725": {
-                "image_extraction": 3.600585699081421,
-                "read": 2.0959312915802,
-                "watermark": 5.662652492523193
+                "image_extraction": 3.632526397705078,
+                "read": 1.9727542400360107,
+                "watermark": 5.599521636962891
             },
             "2201.00021": {
-                "image_extraction": 0.899993896484375,
-                "read": 0.5941448211669922,
-                "watermark": 0.8802309036254883
+                "image_extraction": 0.9102218151092529,
+                "read": 0.561309814453125,
+                "watermark": 1.030742883682251
             },
             "2201.00022": {
-                "image_extraction": 0.5003249645233154,
-                "read": 0.2769145965576172,
-                "watermark": 0.9552063941955566
+                "image_extraction": 0.49163293838500977,
+                "read": 0.35618138313293457,
+                "watermark": 0.7521810531616211
             },
             "2201.00029": {
-                "image_extraction": 0.003554821014404297,
-                "read": 0.38430094718933105,
-                "watermark": 0.09974980354309082
+                "image_extraction": 0.0033931732177734375,
+                "read": 0.41935181617736816,
+                "watermark": 0.08749532699584961
             },
             "2201.00037": {
-                "image_extraction": 0.010822772979736328,
-                "read": 0.957435131072998,
-                "watermark": 1.8120055198669434
+                "image_extraction": 0.010039329528808594,
+                "read": 1.0310795307159424,
+                "watermark": 1.6997003555297852
             },
             "2201.00069": {
-                "image_extraction": 5.672004461288452,
-                "read": 0.35195350646972656,
-                "watermark": 1.1070070266723633
+                "image_extraction": 5.814162731170654,
+                "read": 0.4030001163482666,
+                "watermark": 0.9838786125183105
             },
             "2201.00151": {
-                "image_extraction": 0.003854990005493164,
-                "read": 5.403608083724976,
-                "watermark": 13.4715895652771
+                "image_extraction": 0.003250598907470703,
+                "read": 5.729317903518677,
+                "watermark": 12.279357194900513
             },
             "2201.00178": {
-                "image_extraction": 0.7386953830718994,
-                "read": 0.443634033203125,
-                "watermark": 1.1078128814697266
+                "image_extraction": 0.7503547668457031,
+                "read": 0.48607897758483887,
+                "watermark": 1.0085842609405518
             },
             "2201.00200": {
-                "image_extraction": 0.004326581954956055,
-                "read": 0.5157289505004883,
-                "watermark": 0.4838368892669678
+                "image_extraction": 0.0038993358612060547,
+                "read": 0.5258574485778809,
+                "watermark": 0.4673922061920166
             },
             "2201.00201": {
-                "image_extraction": 0.7092070579528809,
-                "read": 0.27915358543395996,
-                "watermark": 0.793565034866333
+                "image_extraction": 0.7103419303894043,
+                "read": 0.30803465843200684,
+                "watermark": 0.7354044914245605
             },
             "2201.00214": {
-                "image_extraction": 0.4208667278289795,
-                "read": 19.764751195907593,
-                "watermark": 45.63583517074585
+                "image_extraction": 0.3845233917236328,
+                "read": 18.78579568862915,
+                "watermark": 43.98939323425293
             },
             "GeoTopo-book": {
-                "image_extraction": 1.4834551811218262,
-                "read": 4.467116117477417,
-                "watermark": 11.597773551940918
+                "image_extraction": 1.5176374912261963,
+                "read": 5.871168851852417,
+                "watermark": 10.380576372146606
             }
         },
         "tika": {
@@ -506,17 +506,17 @@
             "2201.00214": 0.9784680209521233,
             "GeoTopo-book": 0.9658770842721947
         },
-        "pypdf2": {
+        "pypdf": {
             "1601.03642": 0.9897718873016692,
-            "1602.06541": 0.9802022141888839,
+            "1602.06541": 0.9802517774812367,
             "1707.09725": 0.942859219351859,
             "2201.00021": 0.9665893615444081,
             "2201.00022": 0.9730826246391511,
             "2201.00029": 0.9766931724705237,
-            "2201.00037": 0.9406599334671981,
+            "2201.00037": 0.9402707115289559,
             "2201.00069": 0.9644961587230552,
-            "2201.00151": 0.9393172305985578,
-            "2201.00178": 0.9316599563831572,
+            "2201.00151": 0.9395715828027728,
+            "2201.00178": 0.9316599563831571,
             "2201.00200": 0.974868026918479,
             "2201.00201": 0.9830162051681347,
             "2201.00214": 0.9761955366631243,

benchmark.py

@@ -0,0 +1,2 @@
+[tool.isort]
+profile = "black"