Revert 4fa05b9. Make a note.

src/ccstruct/imagedata.cpp

@@ -47,6 +47,9 @@ ImageData::ImageData(bool vertical, Image pix) : page_number_(0), vertical_text_
   SetPix(pix);
 }
 ImageData::~ImageData() {
+#ifdef TESSERACT_IMAGEDATA_AS_PIX
+  internal_pix_.destroy();
+#endif
 }
 
 // Builds and returns an ImageData from the basic data. Note that imagedata,
@@ -177,12 +180,28 @@ bool ImageData::SkipDeSerialize(TFile *fp) {
 // In case of missing PNG support in Leptonica use PNM format,
 // which requires more memory.
 void ImageData::SetPix(Image pix) {
+#ifdef TESSERACT_IMAGEDATA_AS_PIX
+  internal_pix_ = pix;
+#else
   SetPixInternal(pix, &image_data_);
+#endif
 }
 
 // Returns the Pix image for *this. Must be pixDestroyed after use.
 Image ImageData::GetPix() const {
+#ifdef TESSERACT_IMAGEDATA_AS_PIX
+#  ifdef GRAPHICS_DISABLED
+  /* The only caller of this is the scaling functions to prescale the
+   * source. Thus we can just return a new pointer to the same data. */
+  return pixClone(internal_pix_);
+#  else
+  /* pixCopy always does an actual copy, so the caller can modify the
+   * changed data. */
+  return pixCopy(NULL, internal_pix_);
+#  endif
+#else
   return GetPixInternal(image_data_);
+#endif
 }
 
 // Gets anything and everything with a non-nullptr pointer, prescaled to a
@@ -303,6 +322,7 @@ void ImageData::AddBoxes(const std::vector<TBOX> &boxes, const std::vector<std::
   }
 }
 
+#ifndef TESSERACT_IMAGEDATA_AS_PIX
 // Saves the given Pix as a PNG-encoded string and destroys it.
 // In case of missing PNG support in Leptonica use PNM format,
 // which requires more memory.
@@ -331,6 +351,7 @@ Image ImageData::GetPixInternal(const std::vector<char> &image_data) {
   }
   return pix;
 }
+#endif
 
 // Parses the text string as a box file and adds any discovered boxes that
 // match the page number. Returns false on error.

src/ccstruct/imagedata.h

@@ -152,6 +152,11 @@ class TESS_API ImageData {
 private:
   std::string imagefilename_; // File to read image from.
   int32_t page_number_;  // Page number if multi-page tif or -1.
+  // see https://github.com/tesseract-ocr/tesseract/pull/2965
+  // EP: reconsider for tess6.0/opencv
+#ifdef TESSERACT_IMAGEDATA_AS_PIX
+  Image internal_pix_;
+#endif
   std::vector<char> image_data_;  // PNG/PNM file data.
   std::string language_;          // Language code for image.
   std::string transcription_;     // UTF-8 ground truth of image.