Convert to pixel data

dangerzone/common.py

@@ -1,6 +1,7 @@
 import sys
 import os
 import inspect
+import tempfile
 
 
 class Common(object):
@@ -9,7 +10,9 @@ class Common(object):
     """
 
     def __init__(self):
-        pass
+        # Temporary directory to store pixel data
+        self.tmpdir = tempfile.TemporaryDirectory()
+        print(f"tmpdir is: {self.tmpdir.name}")
 
     def get_resource_path(self, filename):
         if getattr(sys, "dangerzone_dev", False):

dangerzone/main_window.py

@@ -1,6 +1,6 @@
 from PyQt5 import QtCore, QtGui, QtWidgets
 
-from .tasks import PullImageTask, BuildContainerTask
+from .tasks import PullImageTask, BuildContainerTask, ConvertToPixels
 
 
 class MainWindow(QtWidgets.QMainWindow):
@@ -40,11 +40,11 @@ class MainWindow(QtWidgets.QMainWindow):
         central_widget.setLayout(layout)
         self.setCentralWidget(central_widget)
 
-        self.tasks = [PullImageTask, BuildContainerTask]
+        self.tasks = [PullImageTask, BuildContainerTask, ConvertToPixels]
 
     def start(self, filename):
         print(f"Input document: {filename}")
-        self.document_filename = filename
+        self.common.document_filename = filename
         self.show()
 
         self.next_task()
@@ -52,12 +52,16 @@ class MainWindow(QtWidgets.QMainWindow):
     def next_task(self):
         if len(self.tasks) == 0:
             print("Tasks finished")
+            self.task_label.setText("Tasks finished")
             return
 
+        self.task_details.setText("")
+
         self.current_task = self.tasks.pop(0)(self.common)
         self.current_task.update_label.connect(self.update_label)
         self.current_task.update_details.connect(self.update_details)
-        self.current_task.thread_finished.connect(self.next_task)
+        self.current_task.task_finished.connect(self.next_task)
+        self.current_task.task_failed.connect(self.task_failed)
         self.current_task.start()
 
     def update_label(self, s):
@@ -66,10 +70,16 @@ class MainWindow(QtWidgets.QMainWindow):
     def update_details(self, s):
         self.task_details.setText(s)
 
+    def task_failed(self, err):
+        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}"
+        )
+
     def scroll_to_bottom(self, minimum, maximum):
         self.details_scrollarea.verticalScrollBar().setValue(maximum)
 
     def closeEvent(self, e):
-        print("closing")
         e.accept()
         self.app.quit()

dangerzone/tasks.py

@@ -1,10 +1,13 @@
 import subprocess
 import time
+import tempfile
+import os
 from PyQt5 import QtCore, QtWidgets, QtGui
 
 
 class TaskBase(QtCore.QThread):
-    thread_finished = QtCore.pyqtSignal()
+    task_finished = QtCore.pyqtSignal()
+    task_failed = QtCore.pyqtSignal(str)
     update_label = QtCore.pyqtSignal(str)
     update_details = QtCore.pyqtSignal(str)
 
@@ -34,6 +37,8 @@ class TaskBase(QtCore.QThread):
             output += p.stdout.read()
             self.update_details.emit(output)
 
+        return output
+
 
 class PullImageTask(TaskBase):
     def __init__(self, common):
@@ -45,7 +50,7 @@ class PullImageTask(TaskBase):
         self.update_details.emit("")
         args = ["podman", "pull", "ubuntu:18.04"]
         self.execute_podman(args, watch="stderr")
-        self.thread_finished.emit()
+        self.task_finished.emit()
 
 
 class BuildContainerTask(TaskBase):
@@ -59,4 +64,85 @@ class BuildContainerTask(TaskBase):
         self.update_details.emit("")
         args = ["podman", "build", "-t", "dangerzone", container_path]
         self.execute_podman(args)
-        self.thread_finished.emit()
+        self.task_finished.emit()
+
+
+class ConvertToPixels(TaskBase):
+    def __init__(self, common):
+        super(ConvertToPixels, self).__init__()
+        self.common = common
+
+        self.max_image_width = 10000
+        self.max_image_height = 10000
+        self.max_image_size = self.max_image_width * self.max_image_height * 3
+
+    def run(self):
+        self.update_label.emit("Converting document to pixels")
+        args = [
+            "podman",
+            "run",
+            "--network",
+            "none",
+            "-v",
+            f"{self.common.document_filename}:/tmp/input_file",
+            "-v",
+            f"{self.common.tmpdir.name}:/dangerzone",
+            "dangerzone",
+            "/tmp/convert_to_pixels",
+        ]
+        output = self.execute_podman(args)
+
+        # How many pages was that?
+        num_pages = None
+        for line in output.split("\n"):
+            if line.startswith("Document has "):
+                num_pages = line.split(" ")[2]
+                break
+        if not num_pages or not num_pages.isdigit() or int(num_pages) <= 0:
+            self.task_failed.emit("Invalid number of pages returned")
+            return
+        num_pages = int(num_pages)
+
+        # Make sure we have the files we expect
+        expected_filenames = []
+        for i in range(1, num_pages + 1):
+            expected_filenames += [
+                f"page-{i}.rgb",
+                f"page-{i}.width",
+                f"page-{i}.height",
+            ]
+        expected_filenames.sort()
+        actual_filenames = os.listdir(self.common.tmpdir.name)
+        actual_filenames.sort()
+
+        if expected_filenames != actual_filenames:
+            self.task_failed.emit(
+                f"We expected these files:\n{expected_filenames}\n\nBut we got these files:\n{actual_filenames}"
+            )
+            return
+
+        # 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:
+                w_str = f.read().strip()
+            with open(f"{self.common.tmpdir.name}/page-{i}.height") as f:
+                h_str = f.read().strip()
+            w = int(w_str)
+            h = int(h_str)
+            if (
+                not w_str.isdigit()
+                or not h_str.isdigit()
+                or w <= 0
+                or w > self.max_image_width
+                or h <= 0
+                or h > self.max_image_height
+            ):
+                self.task_failed.emit(f"Page {i} has invalid geometry")
+                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:
+                self.task_failed.emit(f"Page {i} has an invalid RGB file size")
+                return
+
+        self.task_finished.emit()

share/container/Containerfile

@@ -1,8 +1,6 @@
 FROM ubuntu:18.04
 
 RUN apt-get update && \
-    apt-get install -y poppler-utils imagemagick
+    apt-get install -y file poppler-utils imagemagick
-
-RUN useradd -ms /bin/bash user
-USER user:user
 
+COPY convert_to_pixels /tmp/convert_to_pixels

share/container/convert_to_pixels

@@ -0,0 +1,38 @@
+#!/bin/bash
+
+IMG_DEPTH=8
+
+echo "Separating document into pages"
+
+pdfseparate /tmp/input_file /tmp/page-%d.pdf
+
+NUM_PAGES=$(find /tmp/page-*.pdf |wc -l)
+echo "Document has $NUM_PAGES pages"
+echo
+
+for FILENAME in $(find /tmp/page-*.pdf); do
+    FILENAME_BASE=${FILENAME%.pdf}
+    PAGE=$(basename $FILENAME_BASE |cut -d"-" -f2)
+
+    echo "Converting page $PAGE to pixels"
+
+    # Convert to png
+    pdftocairo "$FILENAME" -png -singlefile "$FILENAME_BASE"
+    
+    # Get the width and height
+    IMG_WIDTH=$(identify -format "%w" "$FILENAME_BASE.png")
+    IMG_HEIGHT=$(identify -format "%h" "$FILENAME_BASE.png")
+    echo $IMG_WIDTH > $FILENAME_BASE.width
+    echo $IMG_HEIGHT > $FILENAME_BASE.height
+    
+    # Convert to rgb
+    convert "$FILENAME_BASE.png" -depth $IMG_DEPTH rgb:"$FILENAME_BASE.rgb"
+
+    # Delete the png
+    rm "$FILENAME_BASE.png"
+
+    # Move files needed for the next step to the mounted volume
+    mv "$FILENAME_BASE.rgb" /dangerzone
+    mv "$FILENAME_BASE.width" /dangerzone
+    mv "$FILENAME_BASE.height" /dangerzone
+done