Re-attempt 10 times, some tweaks to OCR, pre-filter card pool HEAD main

1 files changed, 137 insertions, 106 deletions

@@ -55,6 +55,24 @@ void wizard::set_status_callback(status_callback_type callback) {
 }
 
 Magick::Image wizard::run() {
+  tesseract_ptr api{new tesseract::TessBaseAPI()};
+
+  if (api->Init(nullptr, "eng", tesseract::OEM_TESSERACT_LSTM_COMBINED)) {
+    throw std::runtime_error("Could not initialize tesseract");
+  }
+
+  std::string allChars;
+  for (char ch : cards_.getCharacters()) {
+    allChars.push_back(ch);
+    if (std::isalpha(ch)) {
+      allChars.push_back(std::toupper(ch));
+    }
+  }
+
+  if (!api->SetVariable("tessedit_char_whitelist", allChars.c_str())) {
+    throw std::runtime_error("Could not set tesseract allowlist");
+  }
+
   std::string text = text_;  //"what the heck, it's just some gay guy";
   // getline(std::cin, text);
   if (text.back() == '\r') {
@@ -68,104 +86,109 @@ Magick::Image wizard::run() {
 
   std::string canonText = hatkirby::lowercase(text);
   designer des(canonText, cards_.getTitles());
-  std::list<usage> res = des.generate(rng_);
 
   Magick::Image endImage;
-  bool firstSlice = false;
-  int ui = 0;
-
-  for (const usage& u : res) {
-    const card& theCard = cards_.getCard(u.cardId);
-    const std::string& cardName = theCard.name;
-
-    std::cout << cardName.substr(0, u.strIndex) << "["
-              << cardName.substr(u.strIndex, u.strLen) << "]"
-              << cardName.substr(u.strIndex + u.strLen) << std::endl;
-
-    std::cout << "Downloading image..." << std::endl;
-    if (status_callback_) {
-      status_callback_("Downloading image...");
-    }
-
-    Magick::Image cardImg = images_.get(theCard.uuid, theCard.imageUri);
-
-    std::cout << "Reading text..." << std::endl;
-    if (status_callback_) {
-      status_callback_("Reading text...");
-    }
-
-    Magick::Image titleImg = cardImg;
-    titleImg.magick("TIFF");
-
-    // titleImg.threshold(MaxRGB / 2);
-    titleImg.write("pre.tif");
-
-    Magick::Geometry margin;
-
-    if (theCard.frame == card_frame::m2015) {
-      margin = Magick::Geometry{595, 46, 57, 54};
-    } else if (theCard.frame == card_frame::modern) {
-      margin = Magick::Geometry{581, 50, 63, 57};
-    }
-
-    titleImg.crop(margin);
-    titleImg.zoom({margin.width() * 5, margin.height() * 5});
-
-    // titleImg.quantizeColorSpace(Magick::GRAYColorspace);
-    // titleImg.quantizeColors(2);
-    // titleImg.quantize();
-    titleImg.backgroundColor("white");
-    titleImg.matte(false);
-    titleImg.resolutionUnits(Magick::PixelsPerInchResolution);
-    titleImg.density({300, 300});
-    titleImg.type(Magick::GrayscaleType);
+  bool success = false;
+  for (int attempt = 0; attempt < 10; attempt++) {
+    endImage = {};
+
+    std::list<usage> res = des.generate(rng_);
+
+    bool firstSlice = false;
+    int ui = 0;
+
+    bool failure = false;
+    for (const usage& u : res) {
+      const card& theCard = cards_.getCard(u.cardId);
+      const std::string& cardName = theCard.name;
+
+      std::cout << cardName.substr(0, u.strIndex) << "["
+                << cardName.substr(u.strIndex, u.strLen) << "]"
+                << cardName.substr(u.strIndex + u.strLen) << std::endl;
+
+      std::cout << "Downloading image..." << std::endl;
+      if (status_callback_) {
+        std::ostringstream status_stream;
+        status_stream << "Processing card " << (ui + 1) << "/" << res.size()
+                      << " (attempt " << (attempt + 1) << ")...";
+        status_callback_(status_stream.str());
+      }
 
-    titleImg.write("title.tif");
+      Magick::Image cardImg = images_.get(theCard.uuid, theCard.imageUri);
 
-    Magick::Blob titleBlob;
-    titleImg.write(&titleBlob);
+      std::cout << "Reading text..." << std::endl;
 
-    pix_ptr titlePix{
-        pixReadMemTiff(reinterpret_cast<const unsigned char*>(titleBlob.data()),
-                       titleBlob.length(), 0)};
+      Magick::Image titleImg = cardImg;
+      titleImg.magick("TIFF");
 
-    tesseract_ptr api{new tesseract::TessBaseAPI()};
+      // titleImg.threshold(MaxRGB / 2);
+      titleImg.write("pre.tif");
 
-    if (api->Init(nullptr, "eng")) {
-      throw std::runtime_error("Could not initialize tesseract");
-    }
+      Magick::Geometry margin;
 
-    api->SetImage(titlePix.get());
-    api->Recognize(nullptr);
+      if (theCard.frame == card_frame::m2015) {
+        margin = Magick::Geometry{595, 50, 57, 54};
+      } else if (theCard.frame == card_frame::modern) {
+        margin = Magick::Geometry{581, 50, 63, 57};
+      }
 
-    tesseract::ResultIterator* ri = api->GetIterator();
-    tesseract::PageIteratorLevel level = tesseract::RIL_TEXTLINE;
-    bool foundName = false;
-    size_t extraChars = 0;
+      titleImg.crop(margin);
+      titleImg.zoom({margin.width() * 5, margin.height() * 5});
+
+      // titleImg.quantizeColorSpace(Magick::GRAYColorspace);
+      // titleImg.quantizeColors(2);
+      // titleImg.quantize();
+      titleImg.backgroundColor("white");
+      titleImg.matte(false);
+      titleImg.resolutionUnits(Magick::PixelsPerInchResolution);
+      titleImg.density({300, 300});
+      titleImg.type(Magick::GrayscaleType);
+      titleImg.gamma(3);
+      // titleImg.level(0.2, 0.8);
+
+      titleImg.write("title.tif");
+
+      Magick::Blob titleBlob;
+      titleImg.write(&titleBlob);
+
+      pix_ptr titlePix{pixReadMemTiff(
+          reinterpret_cast<const unsigned char*>(titleBlob.data()),
+          titleBlob.length(), 0)};
+
+      api->SetImage(titlePix.get());
+      api->Recognize(nullptr);
+
+      tesseract::ResultIterator* ri = api->GetIterator();
+      tesseract::PageIteratorLevel level = tesseract::RIL_TEXTLINE;
+      bool foundName = false;
+      size_t extraChars = 0;
+
+      if (ri) {
+        do {
+          const char* line = ri->GetUTF8Text(level);
+
+          if (line) {
+            std::string lineStr(line);
+
+            size_t leadin = hatkirby::lowercase(lineStr).find(
+                hatkirby::lowercase((cardName)));
+            if (leadin != std::string::npos) {
+              foundName = true;
+              extraChars = leadin;
+
+              break;
+            } /* else {
+                std::cout << "WRONG: " << lineStr << std::endl;
+              }*/
+          }
+        } while (ri->Next(level));
+      }
 
-    if (ri) {
-      do {
-        const char* line = ri->GetUTF8Text(level);
-
-        if (line) {
-          std::string lineStr(line);
-
-          size_t leadin = hatkirby::lowercase(lineStr).find(
-              hatkirby::lowercase((cardName)));
-          // if (leadin != std::string::npos)
-          {
-            foundName = true;
-            // extraChars = leadin;
-
-            break;
-          } /* else {
-             std::cout << "WRONG: " << lineStr << std::endl;
-           }*/
-        }
-      } while (ri->Next(level));
-    }
+      if (!foundName) {
+        failure = true;
+        break;
+      }
 
-    if (foundName) {
       level = tesseract::RIL_SYMBOL;
 
       for (int i = 0; i < extraChars; i++) {
@@ -233,32 +256,40 @@ Magick::Image wizard::run() {
 
       cardImg.magick("PNG");
       cardImg.write("slice.png");
-    } else {
-      std::cout << "Didn't find name" << std::endl;
-    }
 
-    if (!firstSlice) {
-      firstSlice = true;
-      endImage = cardImg;
+      if (!firstSlice) {
+        firstSlice = true;
+        endImage = cardImg;
 
-      if (theCard.frame == card_frame::m2015) {
-        // endImage.extent({endImage.columns(), endImage.rows() + 6},
-        // Magick::SouthGravity);
-      }
-    } else {
-      int xoff = endImage.columns();
+        if (theCard.frame == card_frame::m2015) {
+          // endImage.extent({endImage.columns(), endImage.rows() + 6},
+          // Magick::SouthGravity);
+        }
+      } else {
+        int xoff = endImage.columns();
 
-      endImage.backgroundColor("black");
+        endImage.backgroundColor("black");
 
-      endImage.extent({endImage.columns() + cardImg.columns(), cardImg.rows()},
-                      Magick::WestGravity);
+        endImage.extent(
+            {endImage.columns() + cardImg.columns(), cardImg.rows()},
+            Magick::WestGravity);
 
-      endImage.composite(cardImg, xoff,
-                         (theCard.frame == card_frame::m2015) ? 6 : 0);
+        endImage.composite(cardImg, xoff,
+                           (theCard.frame == card_frame::m2015) ? 6 : 0);
+      }
+
+      // break;
+      ui++;
     }
 
-    // break;
-    ui++;
+    if (!failure) {
+      success = true;
+      break;
+    }
+  }
+
+  if (!success) {
+    throw std::domain_error("Could not generate card");
   }
 
   endImage.magick("PNG");