Add more samples - especially one book (#7)

Author: MartinThoma

.github/workflows/ci.in

@@ -1 +1,2 @@
 pydantic
+PyPDF2

.github/workflows/ci.txt

@@ -2,9 +2,11 @@
 # This file is autogenerated by pip-compile with python 3.10
 # To update, run:
 #
-#    pip-compile ci.in
+#    pip-compile .github/workflows/ci.in
 #
 pydantic==1.9.0
-    # via -r ci.in
+    # via -r .github/workflows/ci.in
+pypdf2==1.27.9
+    # via -r .github/workflows/ci.in
 typing-extensions==4.1.1
     # via pydantic

.github/workflows/file-size-check.yaml

@@ -25,8 +25,8 @@ jobs:
       uses: tj-actions/[email protected]
     - name: List all changed files
       run: |
-        # MAXSIZE is 1 MB
-        MAXSIZE=1000000
+        # MAXSIZE is 10 MB
+        MAXSIZE=10000000
         for FILENAME in ${{ steps.changed-files.outputs.all_changed_files }}; do
           FILESIZE=$(stat -c%s "$FILENAME")
           echo "Size of $FILENAME = $FILESIZE bytes."

.github/workflows/json_consistency.py

@@ -3,21 +3,22 @@
 import os
 import sys
 from pathlib import Path
-from typing import List
+from typing import List, Optional
 
 from pydantic import BaseModel, NonNegativeInt
+from PyPDF2 import PdfFileReader
 
 
 class PdfEntry(BaseModel):
     path: str
     producer: str
     pages: NonNegativeInt
     creation_date: datetime.datetime
-    images: NonNegativeInt
+    images: Optional[NonNegativeInt]
 
 
 class MainPdfFile(BaseModel):
-    data: list[PdfEntry]
+    data: List[PdfEntry]
 
 
 def main():
@@ -34,6 +35,7 @@ def main():
             seen_failure = True
         else:
             print(f"✅ Found {entry.path}")
+            check_meta(entry)
 
     # Are all files registered?
     pdf_paths = Path(".").glob("**/*.pdf")
@@ -46,5 +48,47 @@ def main():
         sys.exit(1)
 
 
+def pdf_to_datetime(date_str):
+    if not date_str.startswith("D:"):
+        print(f"❌ ERROR: Invalid date: {date_str}")
+    date_str = date_str[2:]
+    if len(date_str) < 14:
+        print(f"❌ ERROR: Invalid date: {date_str}")
+    return datetime.datetime(
+        int(date_str[0:4]),  # year
+        int(date_str[4:6]), # month
+        int(date_str[6:8]),  # day
+        int(date_str[8:10]),  # hour
+        int(date_str[10:12]),  # minute
+        int(date_str[12:14]),  # second
+    )
+
+
+def check_meta(entry: PdfEntry):
+    reader = PdfFileReader(entry.path)
+    if reader.isEncrypted:
+        return
+    info = reader.getDocumentInfo()
+    if info.get("/Producer") != entry.producer:
+        print(
+            f"❌ ERROR: Producer mismatch: {entry.producer} vs {info.get('/Producer')}"
+        )
+
+    pdf_date = pdf_to_datetime(info.get("/CreationDate")).isoformat()
+    entry_date = entry.creation_date.isoformat()[:19]
+    if pdf_date != entry_date:
+        print(
+            f"❌ ERROR: Creation date mismatch: {entry_date} vs {pdf_date}"
+        )
+    # if entry.images is not None:
+    #     if info.get("/XObject") is None:
+    #         if entry.images > 0:
+    #             print(f"❌ ERROR: No XObject, but {entry.images} expected")
+    #     elif len(info.get("/XObject")) != entry.images:
+    #         print(
+    #             f"❌ ERROR: XObject count mismatch: {entry.images} vs {len(info.get('/XObject'))}"
+    #         )
+
+
 if __name__ == "__main__":
     main()

007-imagemagick-images/README.md

@@ -0,0 +1,8 @@
+Generated via [imagemagick](https://legacy.imagemagick.org/Usage/formats/#ps):
+
+```bash
+convert -quality 100 * imagemagick-images.pdf
+convert -compress none smile.png imagemagick-ASCII85Decode.pdf
+convert -compress lzw smile.png imagemagick-lzw.pdf
+convert -alpha off -monochrome -compress fax smile.png imagemagick-CCITTFaxDecode.pdf
+```