Canon cr3 previews (#1958)

* Extract THMB and PRVW images from Canon CR3 file

* Added test for Canon CR3 preview extraction.

Added test data Canon-R6-pruned.CR3 (first 492016 bytes of https://raw.pixls.us/getfile.php/4659/nice/Canon%20-%20Canon%20EOS%20R6%20-%203:2.CR3).

See https://github.com/Exiv2/exiv2/issues/1893

* Fixed format specifier

* Update src/bmffimage.cpp

Co-authored-by: Miloš Komarčević <4973094+kmilos@users.noreply.github.com>

* Update src/bmffimage.cpp

Co-authored-by: Miloš Komarčević <4973094+kmilos@users.noreply.github.com>

* retrigger checks

Co-authored-by: Miloš Komarčević <4973094+kmilos@users.noreply.github.com>

include/exiv2/bmffimage.hpp

@@ -106,6 +106,28 @@ namespace Exiv2
         void parseXmp(uint64_t length,uint64_t start);
         //@}
 
+        //@{
+        /*!
+        @brief Parse a Canon PRVW or THMB box and add an entry to the set
+            of native previews.
+        @param data Buffer containing the box
+        @param out Logging stream
+        @param bTrace Controls logging
+        @param width_offset Index of image width field in data
+        @param height_offset Index of image height field in data
+        @param size_offset Index of image size field in data
+        @param relative_position Location of the start of image data in the file,
+            relative to the current file position indicator.
+        */
+        void parseCr3Preview(DataBuf &data,
+                             std::ostream &out,
+                             bool bTrace,
+                             uint16_t width_offset,
+                             uint16_t height_offset,
+                             uint32_t size_offset,
+                             uint16_t relative_position);
+        //@}
+
         //! @name Manipulators
         //@{
         void readMetadata() override /* override */;

src/bmffimage.cpp

@@ -75,7 +75,9 @@
 #define TAG_cmt4 0x434D5434 /**< "CMT4" gpsID */
 #define TAG_colr 0x636f6c72 /**< "colr" */
 #define TAG_exif 0x45786966 /**< "Exif" Used by JXL*/
-#define TAG_xml  0x786d6c20 /**< "xml"  Used by JXL*/
+#define TAG_xml  0x786d6c20 /**< "xml "  Used by JXL*/
+#define TAG_thmb 0x54484d42 /**< "THMB" Canon thumbnail */
+#define TAG_prvw 0x50525657 /**< "PRVW" Canon preview image */
 
 // *****************************************************************************
 // class member definitions
@@ -429,7 +431,12 @@ namespace Exiv2
                     out << " uuidName " << name << std::endl;
                     bLF = false;
                 }
-                if (name == "cano") {
+                if (name == "cano" || name == "canp" ) {
+                    if (name == "canp") {
+                        // based on
+                        // https://github.com/lclevy/canon_cr3/blob/7be75d6/parse_cr3.py#L271
+                        io_->seek(8, BasicIo::cur);
+                    }
                     while (io_->tell() < box_end) {
                         io_->seek(boxHandler(out,option,box_end,depth + 1), BasicIo::beg);
                     }
@@ -456,10 +463,16 @@ namespace Exiv2
             case TAG_xml:
                 parseXmp(box_length,io_->tell());
                 break;
+            case TAG_thmb:
+                parseCr3Preview(data, out, bTrace, 4, 6, 8, 16);
+                break;
+            case TAG_prvw:
+                parseCr3Preview(data, out, bTrace, 6, 8, 12, 16);
+                break;
 
             default: break ; /* do nothing */
         }
-        if ( bLF&& bTrace) out << std::endl;
+        if (bLF && bTrace) out << std::endl;
 
         // return address of next box
         return box_end;
@@ -540,6 +553,36 @@ namespace Exiv2
         }
     }
 
+    void BmffImage::parseCr3Preview(DataBuf &data,
+                                    std::ostream& out,
+                                    bool bTrace,
+                                    uint16_t width_offset,
+                                    uint16_t height_offset,
+                                    uint32_t size_offset,
+                                    uint16_t relative_position)
+    {
+        // Derived from https://github.com/lclevy/canon_cr3
+        NativePreview nativePreview;
+        long here = io_->tell();
+        enforce(here >= 0 &&
+                here <= std::numeric_limits<long>::max() - relative_position,
+                kerCorruptedMetadata);
+        nativePreview.position_ = here + relative_position;
+        nativePreview.width_ =  data.read_uint16(width_offset, endian_);
+        nativePreview.height_ = data.read_uint16(height_offset, endian_);
+        nativePreview.size_ = data.read_uint32(size_offset, endian_);
+        nativePreview.filter_ = "";
+        nativePreview.mimeType_ = "image/jpeg";
+        nativePreviews_.push_back(nativePreview);
+
+        if (bTrace) {
+            out << Internal::stringFormat("width,height,size = %u,%u,%u",
+                                          nativePreview.width_,
+                                          nativePreview.height_,
+                                          nativePreview.size_);
+        }
+    }
+
     void BmffImage::setComment(const std::string& /*comment*/)
     {
         // bmff files are read-only

tests/bugfixes/github/test_issue_1893.py

@@ -0,0 +1,65 @@
+# -*- coding: utf-8 -*-
+
+import system_tests
+import unittest
+from tempfile import TemporaryDirectory
+import shutil
+import hashlib
+import os
+
+# test needs system_tests.BT.vv['enable_bmff']=1
+bSkip=system_tests.BT.verbose_version().get('enable_bmff')!='1'
+
+if bSkip:
+    raise unittest.SkipTest('*** requires enable_bmff=1 ***')
+
+file_basename = 'Canon-R6-pruned.CR3'
+previews_expected = (
+    ('Canon-R6-pruned-preview1.jpg', 'a182ef12ac883309b4dfc66b87eac1891286d3ae'),
+    ('Canon-R6-pruned-preview2.jpg', '524a07f1797854e349ae140e2682ba37147fa6b2')
+)
+
+class issue_1893_cr3_preview(metaclass=system_tests.CaseMeta):
+    """
+    Check that THMB and PRVW images are extracted from Canon CR3 files
+    """
+    url = "https://github.com/Exiv2/exiv2/issues/1893"
+    filename = "$data_path/" + file_basename
+    commands=[] # see setUp()
+
+    if bSkip:
+        retval=[]
+        stdin=[]
+        stderr=[]
+        stdout=[]
+        print("*** test skipped.  requires enable_bmff=1***")
+    else:
+        retval = [ 0, 0]
+        stderr = [ "",""]
+        stdin  = [ "", ""]
+        stdout = ["""Preview 1: image/jpeg, 160x120 pixels, 16005 bytes
+Preview 2: image/jpeg, 1620x1080 pixels, 389450 bytes
+""", ""]
+
+    def post_tests_hook(self):
+        if self.commands:
+            for j, sha1 in previews_expected:
+                p = os.path.join(self.preview_image_tmp_dir.name, j)
+                self.assertTrue(os.path.isfile(p))
+                h = hashlib.sha1(open(p, 'rb').read()).hexdigest()
+                self.assertEqual(h, sha1)
+
+    def setUp(self):
+        if bSkip:
+            return
+        # Avoid polluting the test data directory with extracted previews
+        self.preview_image_tmp_dir = TemporaryDirectory()
+        shutil.copy(self.expand_variables(self.filename),
+                    self.preview_image_tmp_dir.name)
+        p = os.path.join(
+            self.preview_image_tmp_dir.name,
+            file_basename)
+        self.commands = [
+            self.expand_variables("$exiv2 -pp ") + p,
+            self.expand_variables("$exiv2 -ep ") + p
+            ]