Handle image and line regions in output formats ALTO, hOCR and text

Tested-by: Merlijn Wajer <merlijn@archive.org>
Signed-off-by: Stefan Weil <sw@weilnetz.de>

src/api/altorenderer.cpp

@@ -13,9 +13,11 @@
 // See the License for the specific language governing permissions and
 // limitations under the License.
 
+#include "errcode.h" // for ASSERT_HOST
 #ifdef _WIN32
-#  include "host.h" // windows.h for MultiByteToWideChar, ...
+#  include "host.h"  // windows.h for MultiByteToWideChar, ...
 #endif
+#include "tprintf.h" // for tprintf
 
 #include <tesseract/baseapi.h>
 #include <tesseract/renderer.h>
@@ -174,6 +176,36 @@ char *TessBaseAPI::GetAltoText(ETEXT_DESC *monitor, int page_number) {
       continue;
     }
 
+    int left, top, right, bottom;
+    auto block_type = res_it->BlockType();
+
+    switch (block_type) {
+      case PT_FLOWING_IMAGE:
+      case PT_HEADING_IMAGE:
+      case PT_PULLOUT_IMAGE: {
+        // Handle all kinds of images.
+        // TODO: optionally add TYPE, for example TYPE="photo".
+        alto_str << "\t\t\t\t<Illustration ID=\"cblock_" << bcnt++ << "\"";
+        AddBoxToAlto(res_it, RIL_BLOCK, alto_str);
+        alto_str << "</Illustration>\n";
+        res_it->Next(RIL_BLOCK);
+        continue;
+      }
+      case PT_HORZ_LINE:
+      case PT_VERT_LINE:
+        // Handle horizontal and vertical lines.
+        alto_str << "\t\t\t\t<GraphicalElement ID=\"cblock_" << bcnt++ << "\"";
+        AddBoxToAlto(res_it, RIL_BLOCK, alto_str);
+        alto_str << "</GraphicalElement >\n";
+        res_it->Next(RIL_BLOCK);
+        continue;
+      case PT_NOISE:
+        tprintf("TODO: Please report image which triggers the noise case.\n");
+        ASSERT_HOST(false);
+      default:
+        break;
+    }
+
     if (res_it->IsAtBeginningOf(RIL_BLOCK)) {
       alto_str << "\t\t\t\t<ComposedBlock ID=\"cblock_" << bcnt << "\"";
       AddBoxToAlto(res_it, RIL_BLOCK, alto_str);
@@ -200,7 +232,6 @@ char *TessBaseAPI::GetAltoText(ETEXT_DESC *monitor, int page_number) {
     bool last_word_in_tblock = res_it->IsAtFinalElement(RIL_PARA, RIL_WORD);
     bool last_word_in_cblock = res_it->IsAtFinalElement(RIL_BLOCK, RIL_WORD);
 
-    int left, top, right, bottom;
     res_it->BoundingBox(RIL_WORD, &left, &top, &right, &bottom);
 
     do {

src/api/baseapi.cpp

@@ -1371,6 +1371,22 @@ char *TessBaseAPI::GetUTF8Text() {
     if (it->Empty(RIL_PARA)) {
       continue;
     }
+    auto block_type = it->BlockType();
+    switch (block_type) {
+      case PT_FLOWING_IMAGE:
+      case PT_HEADING_IMAGE:
+      case PT_PULLOUT_IMAGE:
+      case PT_HORZ_LINE:
+      case PT_VERT_LINE:
+        // Ignore images and lines for text output.
+        continue;
+      case PT_NOISE:
+        tprintf("TODO: Please report image which triggers the noise case.\n");
+        ASSERT_HOST(false);
+      default:
+        break;
+    }
+
     const std::unique_ptr<const char[]> para_text(it->GetUTF8Text(RIL_PARA));
     text += para_text.get();
   } while (it->Next(RIL_PARA));

src/api/hocrrenderer.cpp

@@ -189,6 +189,36 @@ char *TessBaseAPI::GetHOCRText(ETEXT_DESC *monitor, int page_number) {
 
   std::unique_ptr<ResultIterator> res_it(GetIterator());
   while (!res_it->Empty(RIL_BLOCK)) {
+    int left, top, right, bottom;
+    auto block_type = res_it->BlockType();
+    switch (block_type) {
+      case PT_FLOWING_IMAGE:
+      case PT_HEADING_IMAGE:
+      case PT_PULLOUT_IMAGE: {
+        // Handle all kinds of images.
+        res_it.get()->BoundingBox(RIL_TEXTLINE, &left, &top, &right, &bottom);
+        hocr_str << "   <div class='ocr_photo' id='block_" << page_id << '_'
+                 << bcnt++ << "' title=\"bbox " << left << " " << top << " "
+                 << right << " " << bottom << "\"></div>\n";
+        res_it->Next(RIL_BLOCK);
+        continue;
+      }
+      case PT_HORZ_LINE:
+      case PT_VERT_LINE:
+        // Handle horizontal and vertical lines.
+        res_it.get()->BoundingBox(RIL_TEXTLINE, &left, &top, &right, &bottom);
+        hocr_str << "   <div class='ocr_separator' id='block_" << page_id << '_'
+                 << bcnt++ << "' title=\"bbox " << left << " " << top << " "
+                 << right << " " << bottom << "\"></div>\n";
+        res_it->Next(RIL_BLOCK);
+        continue;
+      case PT_NOISE:
+        tprintf("TODO: Please report image which triggers the noise case.\n");
+        ASSERT_HOST(false);
+      default:
+        break;
+    }
+
     if (res_it->Empty(RIL_WORD)) {
       res_it->Next(RIL_WORD);
       continue;
@@ -218,7 +248,7 @@ char *TessBaseAPI::GetHOCRText(ETEXT_DESC *monitor, int page_number) {
     }
     if (res_it->IsAtBeginningOf(RIL_TEXTLINE)) {
       hocr_str << "\n     <span class='";
-      switch (res_it->BlockType()) {
+      switch (block_type) {
         case PT_HEADING_TEXT:
           hocr_str << "ocr_header";
           break;
@@ -228,6 +258,11 @@ char *TessBaseAPI::GetHOCRText(ETEXT_DESC *monitor, int page_number) {
         case PT_CAPTION_TEXT:
           hocr_str << "ocr_caption";
           break;
+        case PT_FLOWING_IMAGE:
+        case PT_HEADING_IMAGE:
+        case PT_PULLOUT_IMAGE:
+          ASSERT_HOST(false);
+          break;
         default:
           hocr_str << "ocr_line";
       }
@@ -248,12 +283,10 @@ char *TessBaseAPI::GetHOCRText(ETEXT_DESC *monitor, int page_number) {
     hocr_str << "\n      <span class='ocrx_word'"
              << " id='"
              << "word_" << page_id << "_" << wcnt << "'";
-    int left, top, right, bottom;
     bool bold, italic, underlined, monospace, serif, smallcaps;
     int pointsize, font_id;
-    const char *font_name;
     res_it->BoundingBox(RIL_WORD, &left, &top, &right, &bottom);
-    font_name =
+    const char *font_name =
         res_it->WordFontAttributes(&bold, &italic, &underlined, &monospace,
                                    &serif, &smallcaps, &pointsize, &font_id);
     hocr_str << " title='bbox " << left << " " << top << " " << right << " "