Don't optimize non-8-bit images

Also add test cases for additional coverage on image optimization.

Don't optimize non-8-bit images
Also add test cases for additional coverage on image optimization.
Jay Berkenbilt
1 parent eb49e07c
Showing 9 changed files with 80 additions and 5 deletions
ChangeLog
TODO
manual/qpdf-manual.xml
qpdf/qpdf.cc
qpdf/qpdf.testcov
qpdf/qtest/qpdf.test
qpdf/qtest/qpdf/optimize-images-unsupported-json.out
qpdf/qtest/qpdf/optimize-images-unsupported.out
qpdf/qtest/qpdf/unsupported-optimization.pdf
 2019-01-31  Jay Berkenbilt  <ejb@ql.org>
+	* Bug fix: do better pre-checks on images before optimizing;
+	refuse to optimize images that can't be converted to JPEG because
+	of colorspace or depth.
+
 	* Add new options --externalize-inline-images, which converts
 	inline images larger than a specified size to regular images, and
 	--ii-min-bytes, which tweaks that size.
-Now
-===
-
-* Deal with compiler warnings
-
 Soon
 ====
@@ -4456,6 +4456,13 @@ print &quot;\n&quot;;
         </listitem>
         <listitem>
          <para>
+           When optimizing images, detect and refuse to optimize
+           images that can't be converted to JPEG because of bit depth
+           or color space.
+         </para>
+        </listitem>
+        <listitem>
+         <para>
           Linearization and page manipulation APIs now detect and
           recover from files that have duplicate Page objects in the
           pages tree.
@@ -3816,6 +3816,18 @@ ImageOptimizer::makePipeline(std::string const&amp; description, Pipeline* next)
         }
         return result;
     }
+    QPDFObjectHandle components_obj = dict.getKey("/BitsPerComponent");
+    if (! (components_obj.isInteger() && (components_obj.getIntValue() == 8)))
+    {
+        QTC::TC("qpdf", "qpdf image optimize bits per component");
+        if (o.verbose && (! description.empty()))
+        {
+            std::cout << whoami << ": " << description
+                      << ": not optimizing because image has other than"
+                      << " 8 bits per component" << std::endl;
+        }
+        return result;
+    }
     // Files have been seen in the wild whose width and height are
     // floating point, which is goofy, but we can deal with it.
     JDIMENSION w = static_cast<JDIMENSION>(
@@ -3844,6 +3856,7 @@ ImageOptimizer::makePipeline(std::string const&amp; description, Pipeline* next)
     }
     else
     {
+        QTC::TC("qpdf", "qpdf image optimize colorspace");
         if (o.verbose && (! description.empty()))
         {
             std::cout << whoami << ": " << description
@@ -437,3 +437,5 @@ QPDFTokenizer inline image at EOF the old way 0
 QPDFTokenizer found EI after more than one try 0
 QPDFPageObjectHelper externalize inline image 0
 QPDFPageObjectHelper keep inline image 0
+qpdf image optimize colorspace 0
+qpdf image optimize bits per component 0
@@ -2088,6 +2088,8 @@ my @image_opt = (
      '--oi-min-width=0 --oi-min-height=0 --oi-min-area=0 --ii-min-bytes=0'],
     ['large-inline-image', 'inline-images-keep-some', ''],
     ['large-inline-image', 'inline-images-keep-all', '--keep-inline-images'],
+    ['unsupported-optimization', 'unsupported',
+     '--oi-min-width=0 --oi-min-height=0 --oi-min-area=0'],
     );
 $n_tests += 2 * scalar(@image_opt);
+{
+  "pages": [
+    {
+      "contents": [
+        "4 0 R"
+      ],
+      "images": [
+        {
+          "bitspercomponent": 1,
+          "colorspace": "/DeviceGray",
+          "decodeparms": [
+            null
+          ],
+          "filter": [
+            "/FlateDecode"
+          ],
+          "filterable": true,
+          "height": 200,
+          "name": "/Im1",
+          "object": "6 0 R",
+          "width": 100
+        },
+        {
+          "bitspercomponent": 8,
+          "colorspace": "/XeviceGray",
+          "decodeparms": [
+            null
+          ],
+          "filter": [
+            "/FlateDecode"
+          ],
+          "filterable": true,
+          "height": 200,
+          "name": "/Im2",
+          "object": "7 0 R",
+          "width": 200
+        }
+      ],
+      "label": null,
+      "object": "3 0 R",
+      "outlines": [],
+      "pageposfrom1": 1
+    }
+  ],
+  "parameters": {
+    "decodelevel": "generalized"
+  },
+  "version": 1
+}
+qpdf: image /Im1 on page 1: not optimizing because image has other than 8 bits per component
+qpdf: image /Im2 on page 1: not optimizing because qpdf can't optimize images with this colorspace
+qpdf: wrote file a.pdf