crop: create AlternativeImage when successful (to prevent inconsisten…

…t annotation in case AlternativeImage was already present)

ocrd_tesserocr/crop.py

@@ -10,7 +10,7 @@
 from ocrd_models.ocrd_page import (
     MetadataItemType,
     LabelsType, LabelType,
-    CoordsType,
+    CoordsType, AlternativeImageType,
     to_xml
 )
 from ocrd_models.ocrd_page_generateds import BorderType
@@ -19,11 +19,12 @@
 from .config import TESSDATA_PREFIX, OCRD_TOOL
 from .common import (
     bbox_from_points, points_from_bbox,
-    bbox_from_xywh
+    bbox_from_xywh, save_image_file
 )
 
 TOOL = 'ocrd-tesserocr-crop'
 LOG = getLogger('processor.TesserocrCrop')
+FILEGRP_IMG = 'OCR-D-IMG-CROP'
 
 class TesserocrCrop(Processor):
 
@@ -150,7 +151,19 @@ def process(self):
                     LOG.debug("Padded page border: %i:%i,%i:%i", min_x, max_x, min_y, max_y)
                     border = BorderType(Coords=CoordsType(
                         points_from_bbox(min_x, min_y, max_x, max_y)))
+                    # update PAGE (annotate border):
                     page.set_Border(border)
+                    # update METS (add the image file):
+                    page_image = page_image.crop(
+                        box=(min_x, min_y, max_x, max_y))
+                    file_id = input_file.ID.replace(self.input_file_grp, FILEGRP_IMG)
+                    file_path = save_image_file(self.workspace, page_image,
+                                                file_id,
+                                                page_id=page_id,
+                                                file_grp=FILEGRP_IMG)
+                    # update PAGE (reference the image file):
+                    page.add_AlternativeImage(AlternativeImageType(
+                        filename=file_path, comments="cropped"))
                 else:
                     LOG.error("Cannot find valid extent for page '%s'", page_id)