From b1d17b605088e118a8799fd027a892310e00d29e Mon Sep 17 00:00:00 2001
From: Hongzhuo Liang <lianghongzhuo@126.com>
Date: Fri, 27 Jan 2023 17:57:04 +0100
Subject: [PATCH] support pikepdf instead of pypdf2

---
 krop/mainwindow.py |  1 -
 krop/pdfcropper.py | 79 ++++++++++++++--------------------------------
 2 files changed, 24 insertions(+), 56 deletions(-)

diff --git a/krop/mainwindow.py b/krop/mainwindow.py
index e8adadf..fd1ae32 100644
--- a/krop/mainwindow.py
+++ b/krop/mainwindow.py
@@ -413,7 +413,6 @@ def slotKrop(self):
             pdf = PdfFile()
             pdf.loadFromFile(inputFileName)
             cropper = PdfCropper()
-            cropper.copyDocumentRoot(pdf)
             for nr in pages:
                 c = self.viewer.cropValues(nr)
                 cropper.addPageCropped(pdf, nr, c, alwaysinclude, rotation)
diff --git a/krop/pdfcropper.py b/krop/pdfcropper.py
index db30646..42500de 100644
--- a/krop/pdfcropper.py
+++ b/krop/pdfcropper.py
@@ -15,23 +15,8 @@
 
 import copy
 import sys
+from pikepdf import Pdf
 
-# Unless specified otherwise, use PyPDF2 instead of pyPdf if available.
-usepypdf2 = '--no-PyPDF2' not in sys.argv
-if usepypdf2:
-    try:
-        from PyPDF2 import PdfFileReader, PdfFileWriter
-    except ImportError:
-        usepypdf2 = False
-if not usepypdf2:
-    try:
-        from pyPdf import PdfFileReader, PdfFileWriter
-    except ImportError:
-        _msg = "Please install PyPDF2 (or its predecessor pyPdf) first."\
-            "\n\tOn recent versions of Ubuntu, the following should do the trick:"\
-            "\n\tsudo apt-get install python-pypdf2"\
-            "\n\t(or, if using python3) sudo apt-get install python3-pypdf2"
-        raise RuntimeError(_msg)
 
 class PdfEncryptedError(Exception):
     pass
@@ -55,8 +40,6 @@ def writeToFile(self, filename):
         stream.close()
     def addPageCropped(self, pdffile, pagenumber, croplist, rotate=0):
         pass
-    def copyDocumentRoot(self, pdffile):
-        pass
 
 
 class PyPdfFile(AbstractPdfFile):
@@ -64,23 +47,16 @@ class PyPdfFile(AbstractPdfFile):
     def __init__(self):
         self.reader = None
     def loadFromStream(self, stream):
-        if usepypdf2:
-            self.reader = PdfFileReader(stream, strict=False)
-        else:
-            self.reader = PdfFileReader(stream)
-        if self.reader.isEncrypted:
-            try:
-                if not self.reader.decrypt(''):
-                    raise PdfEncryptedError
-            except:
-                raise PdfEncryptedError
-    def getPage(self, nr):
-        page = self.reader.getPage(nr-1)
+        self.reader = Pdf.open(stream)
+        if self.reader.is_encrypted:
+            raise PdfEncryptedError
+
 
 class PyPdfCropper(AbstractPdfCropper):
     """Implementation of PdfCropper using pyPdf"""
     def __init__(self):
-        self.output = PdfFileWriter()
+        self.pdf = Pdf.new()
+
     def writeToStream(self, stream):
         # For certain large pdf files, PdfFileWriter.write() causes the error:
         #  maximum recursion depth exceeded while calling a Python object
@@ -88,37 +64,31 @@ def writeToStream(self, stream):
         # We therefore temporarily increase the recursion limit.
         old_reclimit = sys.getrecursionlimit()
         sys.setrecursionlimit(10000)
-        self.output.write(stream)
+        self.pdf.save(stream)
         sys.setrecursionlimit(old_reclimit)
+
     def addPageCropped(self, pdffile, pagenumber, croplist, alwaysinclude, rotate=0):
-        page = pdffile.reader.getPage(pagenumber)
+        page = pdffile.reader.pages[pagenumber]
         if not croplist and alwaysinclude:
-            self.output.addPage(page)
+            self.pdf.pages.append(page)
         for c in croplist:
-            newpage = copy.copy(page)
-            self.cropPage(newpage, c, rotate)
-            self.output.addPage(newpage)
-    def cropPage(self, page, crop, rotate):
+            new_box = self.getCropPageParm(page, c)
+            # Update the various PDF boxes
+            new_page = copy.copy(page)
+            new_page.mediabox = new_box
+            new_page.cropbox = new_box
+            new_page.trimbox = new_box
+            if rotate != 0:
+                new_page.rotate(rotate, True)
+            self.pdf.pages.append(new_page)
+
+    def getCropPageParm(self, page, crop):
         # Note that the coordinate system is up-side down compared with Qt.
-        x0, y0 = page.cropBox.lowerLeft
-        x1, y1 = page.cropBox.upperRight
+        x0, y0, x1, y1 = page.cropbox
         x0, y0, x1, y1 = float(x0), float(y0), float(x1), float(y1)
         x0, x1 = x0+crop[0]*(x1-x0), x1-crop[2]*(x1-x0)
         y0, y1 = y0+crop[3]*(y1-y0), y1-crop[1]*(y1-y0)
-        # Update the various PDF boxes
-        for box in (page.artBox, page.bleedBox, page.cropBox, page.mediaBox, page.trimBox):
-            box.lowerLeft = (x0, y0)
-            box.upperRight = (x1, y1)
-        if rotate != 0:
-            page.rotateClockwise(rotate)
-
-    def copyDocumentRoot(self, pdffile):
-        # Sounds promising in PyPDF2 (see PdfFileWriter.cloneDocumentFromReader),
-        # but doesn't seem to produce a readable PDF:
-        # self.output.cloneReaderDocumentRoot(pdffile.reader)
-        # Instead, this copies at least the named destinations for links:
-        for dest in pdffile.reader.namedDestinations.values():
-            self.output.addNamedDestinationObject(dest)
+        return [x0, y0, x1, y1]
 
 
 def optimizePdfGhostscript(oldfilename, newfilename):
@@ -128,4 +98,3 @@ def optimizePdfGhostscript(oldfilename, newfilename):
 
 PdfFile = PyPdfFile
 PdfCropper = PyPdfCropper
-