Finish making convert to safe PDF work

dangerzone/common.py

@@ -11,8 +11,10 @@ class Common(object):
 
     def __init__(self):
         # Temporary directory to store pixel data
-        self.tmpdir = tempfile.TemporaryDirectory()
+        self.pixel_dir = tempfile.TemporaryDirectory()
-        print(f"tmpdir is: {self.tmpdir.name}")
+        self.safe_dir = tempfile.TemporaryDirectory()
+        print(f"pixel_dir is: {self.pixel_dir.name}")
+        print(f"safe_dir is: {self.safe_dir.name}")
 
     def get_resource_path(self, filename):
         if getattr(sys, "dangerzone_dev", False):

dangerzone/main_window.py

@@ -1,6 +1,8 @@
+import shutil
+import os
 from PyQt5 import QtCore, QtGui, QtWidgets
 
-from .tasks import PullImageTask, BuildContainerTask, ConvertToPixels
+from .tasks import PullImageTask, BuildContainerTask, ConvertToPixels, ConvertToPDF
 
 
 class MainWindow(QtWidgets.QMainWindow):
@@ -40,7 +42,7 @@ class MainWindow(QtWidgets.QMainWindow):
         central_widget.setLayout(layout)
         self.setCentralWidget(central_widget)
 
-        self.tasks = [PullImageTask, BuildContainerTask, ConvertToPixels]
+        self.tasks = [PullImageTask, BuildContainerTask, ConvertToPixels, ConvertToPDF]
 
     def start(self, filename):
         print(f"Input document: {filename}")
@@ -51,8 +53,7 @@ class MainWindow(QtWidgets.QMainWindow):
 
     def next_task(self):
         if len(self.tasks) == 0:
-            print("Tasks finished")
+            self.save_safe_pdf()
-            self.task_label.setText("Tasks finished")
             return
 
         self.task_details.setText("")
@@ -74,9 +75,31 @@ class MainWindow(QtWidgets.QMainWindow):
         self.task_label.setText("Task failed :(")
         self.task_details.setWordWrap(True)
         self.task_details.setText(
-            f"Temporary directory: {self.common.tmpdir.name}\n\n{err}"
+            f"Directory with pixel data: {self.common.pixel_dir.name}\n\n{err}"
         )
 
+    def save_safe_pdf(self):
+        suggested_filename = (
+            f"{os.path.splitext(self.common.document_filename)[0]}-safe.pdf"
+        )
+
+        filename = QtWidgets.QFileDialog.getSaveFileName(
+            self, "Save safe PDF", suggested_filename, filter="Documents (*.pdf)"
+        )
+        if filename[0] == "":
+            print("Save file dialog canceled")
+        else:
+            source_filename = f"{self.common.safe_dir.name}/safe-output-compressed.pdf"
+            dest_filename = filename[0]
+            shutil.move(source_filename, dest_filename)
+
+            # Clean up
+            self.common.pixel_dir.cleanup()
+            self.common.safe_dir.cleanup()
+
+            # Quit
+            self.app.quit()
+
     def scroll_to_bottom(self, minimum, maximum):
         self.details_scrollarea.verticalScrollBar().setValue(maximum)
 

dangerzone/tasks.py

@@ -34,6 +34,9 @@ class TaskBase(QtCore.QThread):
                 output += line
                 self.update_details.emit(output)
 
+            if watch == "stdout":
+                output += p.stderr.read()
+            else:
                 output += p.stdout.read()
             self.update_details.emit(output)
 
@@ -86,9 +89,9 @@ class ConvertToPixels(TaskBase):
             "-v",
             f"{self.common.document_filename}:/tmp/input_file",
             "-v",
-            f"{self.common.tmpdir.name}:/dangerzone",
+            f"{self.common.pixel_dir.name}:/dangerzone",
             "dangerzone",
-            "/tmp/convert_to_pixels",
+            "convert_to_pixels",
         ]
         output = self.execute_podman(args)
 
@@ -112,7 +115,7 @@ class ConvertToPixels(TaskBase):
                 f"page-{i}.height",
             ]
         expected_filenames.sort()
-        actual_filenames = os.listdir(self.common.tmpdir.name)
+        actual_filenames = os.listdir(self.common.pixel_dir.name)
         actual_filenames.sort()
 
         if expected_filenames != actual_filenames:
@@ -123,9 +126,9 @@ class ConvertToPixels(TaskBase):
 
         # Make sure the files are the correct sizes
         for i in range(1, num_pages + 1):
-            with open(f"{self.common.tmpdir.name}/page-{i}.width") as f:
+            with open(f"{self.common.pixel_dir.name}/page-{i}.width") as f:
                 w_str = f.read().strip()
-            with open(f"{self.common.tmpdir.name}/page-{i}.height") as f:
+            with open(f"{self.common.pixel_dir.name}/page-{i}.height") as f:
                 h_str = f.read().strip()
             w = int(w_str)
             h = int(h_str)
@@ -141,8 +144,34 @@ class ConvertToPixels(TaskBase):
                 return
 
             # Make sure the RGB file is the correct size
-            if os.path.getsize(f"{self.common.tmpdir.name}/page-{i}.rgb") != w * h * 3:
+            if (
+                os.path.getsize(f"{self.common.pixel_dir.name}/page-{i}.rgb")
+                != w * h * 3
+            ):
                 self.task_failed.emit(f"Page {i} has an invalid RGB file size")
                 return
 
         self.task_finished.emit()
+
+
+class ConvertToPDF(TaskBase):
+    def __init__(self, common):
+        super(ConvertToPDF, self).__init__()
+        self.common = common
+
+    def run(self):
+        self.update_label.emit("Converting pixels to safe PDF")
+        args = [
+            "podman",
+            "run",
+            "--network",
+            "none",
+            "-v",
+            f"{self.common.pixel_dir.name}:/dangerzone",
+            "-v",
+            f"{self.common.safe_dir.name}:/safezone",
+            "dangerzone",
+            "convert_to_pdf",
+        ]
+        self.execute_podman(args)
+        self.task_finished.emit()

share/container/Containerfile

@@ -1,6 +1,10 @@
 FROM ubuntu:18.04
 
 RUN apt-get update && \
-    apt-get install -y file poppler-utils imagemagick
+    apt-get install -y file poppler-utils imagemagick ghostscript
 
-COPY convert_to_pixels /tmp/convert_to_pixels
+# Fix imagemagick policy to allow writing PDFs
+RUN sed -i '/rights="none" pattern="PDF"/c\<policy domain="coder" rights="read|write" pattern="PDF" />' /etc/ImageMagick-6/policy.xml
+
+COPY convert_to_pixels /usr/local/bin/convert_to_pixels
+COPY convert_to_pdf /usr/local/bin/convert_to_pdf

share/container/convert_to_pdf

@@ -0,0 +1,55 @@
+#!/bin/bash
+
+die() {
+    echo "$1" >&2
+    exit 1
+}
+
+IMG_DEPTH=8
+NUM_PAGES=$(find /dangerzone/page-*.rgb |wc -l)
+
+echo "Document has $NUM_PAGES pages"
+echo
+
+# Convert rgb files to png files
+for RGB_FILENAME in $(find /dangerzone/page-*.rgb); do
+    FILENAME_BASE=${RGB_FILENAME%.rgb}
+    PAGE=$(basename $FILENAME_BASE |cut -d"-" -f2)
+
+    WIDTH_FILENAME=${FILENAME_BASE}.width
+    HEIGHT_FILENAME=${FILENAME_BASE}.height
+    PNG_FILENAME=/tmp/$(basename $FILENAME_BASE).png
+    PDF_FILENAME=/tmp/$(basename $FILENAME_BASE).pdf
+
+    echo "Converting page $PAGE from pixels to PNG"
+    
+    IMG_WIDTH=$(cat $WIDTH_FILENAME)
+    IMG_HEIGHT=$(cat $HEIGHT_FILENAME)
+
+    CONVERT_MSGS=$(convert -size "${IMG_WIDTH}x${IMG_HEIGHT}" -depth ${IMG_DEPTH} rgb:"$RGB_FILENAME" png:"$PNG_FILENAME" 2>&1)
+    if [ $? -ne 0 ]; then
+        die "Page $PAGE conversion failed: $CONVERT_MSGS"
+    fi
+
+    echo "Converting page $PAGE from PNG to PDF"
+    CONVERT_MSGS=$(convert "$PNG_FILENAME" "$PDF_FILENAME" 2>&1)
+    if [ $? -ne 0 ]; then
+        die "Page $PAGE conversion failed: $CONVERT_MSGS"
+    fi
+done
+
+echo "Merging $NUM_PAGES pages into a single PDF"
+
+# Put PDF filenames into an array
+declare -a args
+for PAGE in $(seq 1 $NUM_PAGES); do
+    args+=("/tmp/page-$PAGE.pdf")
+done
+args+=("/safezone/safe-output.pdf")
+
+# Merge them
+pdfunite "${args[@]}"
+
+# Compress
+echo "Compressing PDF"
+ps2pdf /safezone/safe-output.pdf /safezone/safe-output-compressed.pdf