ocrd-dummy: make copying optional and disable by default, #803

OCR-D · Mar 7, 2022 · 4097791 · 4097791
1 parent decb06f
commit 4097791
Show file tree

Hide file tree

Showing 4 changed files with 30 additions and 18 deletions.
diff --git a/ocrd/ocrd/processor/builtin/dummy/ocrd-tool.json b/ocrd/ocrd/processor/builtin/dummy/ocrd-tool.json
@@ -1,8 +1,15 @@
 {
   "executable": "ocrd-dummy",
-  "description": "Bare-bones processor that copies file from input group to output group",
+  "description": "Bare-bones processor creates PAGE-XML and optionally copies file from input group to output group",
   "steps": ["preprocessing/optimization"],
   "categories": ["Image preprocessing"],
   "input_file_grp": "DUMMY_INPUT",
-  "output_file_grp": "DUMMY_OUTPUT"
+  "output_file_grp": "DUMMY_OUTPUT",
+  "parameters": {
+    "copy_files": {
+      "type": "boolean",
+      "default": false,
+      "description": "Whether to actually copy files (true) or just create PAGE-XML as a side effect (false)"
+    }
+  }
 }
diff --git a/ocrd/ocrd/processor/builtin/dummy_processor.py b/ocrd/ocrd/processor/builtin/dummy_processor.py
@@ -21,13 +21,14 @@
 
 class DummyProcessor(Processor):
     """
-    Bare-bones processor that only copies mets:file from input group to output group.
+    Bare-bones processor creates PAGE-XML and optionally copies file from input group to output group
     """
 
     def process(self):
         LOG = getLogger('ocrd.dummy')
         assert_file_grp_cardinality(self.input_file_grp, 1)
         assert_file_grp_cardinality(self.output_file_grp, 1)
+        copy_files = self.parameter['copy_files']
         for input_file in self.input_files:
             input_file = self.workspace.download_file(input_file)
             file_id = make_file_id(input_file, self.output_file_grp)
@@ -39,24 +40,26 @@ def process(self):
             LOG.info("cp %s %s # %s -> %s", input_file.url, local_filename, input_file.ID, file_id)
             if input_file.mimetype == MIMETYPE_PAGE:
                 # Source file is PAGE-XML: Write out in-memory PcGtsType
-                self.workspace.add_file(
-                    ID=file_id,
-                    file_grp=self.output_file_grp,
-                    pageId=input_file.pageId,
-                    mimetype=input_file.mimetype,
-                    local_filename=local_filename,
-                    content=to_xml(pcgts).encode('utf-8'))
-            else:
-                # Source file is not PAGE-XML: Copy byte-by-byte
-                with open(input_file.local_filename, 'rb') as f:
-                    content = f.read()
+                if copy_files:
                     self.workspace.add_file(
                         ID=file_id,
                         file_grp=self.output_file_grp,
                         pageId=input_file.pageId,
                         mimetype=input_file.mimetype,
                         local_filename=local_filename,
-                        content=content)
+                        content=to_xml(pcgts).encode('utf-8'))
+            else:
+                # Source file is not PAGE-XML: Copy byte-by-byte
+                if copy_files:
+                    with open(input_file.local_filename, 'rb') as f:
+                        content = f.read()
+                        self.workspace.add_file(
+                            ID=file_id,
+                            file_grp=self.output_file_grp,
+                            pageId=input_file.pageId,
+                            mimetype=input_file.mimetype,
+                            local_filename=local_filename,
+                            content=content)
                 if input_file.mimetype.startswith('image/'):
                     # write out the PAGE-XML representation for this image
                     page_file_id = file_id + '_PAGE'
@@ -76,7 +79,7 @@ def process(self):
 
     def __init__(self, *args, **kwargs):
         kwargs['ocrd_tool'] = OCRD_TOOL
-        kwargs['version'] = '0.0.2'
+        kwargs['version'] = '0.0.3'
         super(DummyProcessor, self).__init__(*args, **kwargs)
 
 @click.command()

diff --git a/tests/processor/test_ocrd_dummy.py b/tests/processor/test_ocrd_dummy.py
@@ -21,6 +21,7 @@ def test_copies_ok(self):
                 DummyProcessor,
                 input_file_grp='OCR-D-IMG',
                 output_file_grp='OUTPUT',
+                parameter={'copy_files': True},
                 workspace=workspace
             )
             output_files = workspace.mets.find_all_files(fileGrp='OUTPUT')
@@ -38,6 +39,7 @@ def test_copies_ok(self):
                 DummyProcessor,
                 input_file_grp='OUTPUT',
                 output_file_grp='OUTPUT2',
+                parameter={'copy_files': True},
                 workspace=workspace
             )
             output2_files = workspace.mets.find_all_files(fileGrp='OUTPUT2')

diff --git a/tests/test_task_sequence.py b/tests/test_task_sequence.py
@@ -142,8 +142,8 @@ def test_task_run(self):
                 ws.save_mets()
                 files_before = len(ws.mets.find_all_files())
                 run_tasks('mets.xml', 'DEBUG', None, [
-                    "dummy -I OCR-D-IMG -O GRP1",
-                    "dummy -I GRP1 -O GRP2",
+                    "dummy -I OCR-D-IMG -O GRP1 -P copy_files true",
+                    "dummy -I GRP1 -O GRP2 -P copy_files true",
                 ])
                 ws.reload_mets()
                 # step 1: 2 images in OCR-D-IMG -> 2 images 2 PAGEXML in GRP1