revert d96b532 until OCR-D/spec#121 is unresolved

kba · Jun 8, 2020 · b5aaa3d · b5aaa3d
1 parent 1411cf1
commit b5aaa3d
Show file tree

Hide file tree

Showing 3 changed files with 20 additions and 14 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -15,7 +15,7 @@ Changed:
     output file group is targeted but for pages that aren't in that group, #471
   * `ocrd_cli_wrap_processor`: Take `page_id` into account when doing `WorkspaceValidator.check_file_grp`
   * `run_cli` accepts an `overwrite` parameter to pass on to processor calls, #471
-  * `Task.validate`: set implicit input/output groups from ocrd-tool.json, #471
+  * <del>`Task.validate`: set implicit input/output groups from ocrd-tool.json, #471</del> blocked by OCR-D/spec#121
   * `ocrd process`: support --overwrite and pass on to processor calls, #471
 
 Added:

diff --git a/ocrd/ocrd/task_sequence.py b/ocrd/ocrd/task_sequence.py
@@ -57,21 +57,24 @@ def validate(self):
         parameters = {}
         if self.parameter_path:
             parameters = parse_json_string_or_file(self.parameter_path)
-        # make implicit input/output groups explicit by defaulting to what is
-        # provided in ocrd-tool.json
-        # TODO adapt once OCR-D/spec#121 lands
-        actual_output_grps = [*self.ocrd_tool_json['output_file_grp']]
-        for i, grp in enumerate(self.output_file_grps):
-            actual_output_grps[i] = grp
-        self.output_file_grps = actual_output_grps
-        actual_input_grps = [*self.ocrd_tool_json['input_file_grp']]
-        for i, grp in enumerate(self.input_file_grps):
-            actual_input_grps[i] = grp
-        self.input_file_grps = actual_input_grps
+        # TODO uncomment and adapt once OCR-D/spec#121 lands
+        # # make implicit input/output groups explicit by defaulting to what is
+        # # provided in ocrd-tool.json
+        # actual_output_grps = [*self.ocrd_tool_json['output_file_grp']]
+        # for i, grp in enumerate(self.output_file_grps):
+            # actual_output_grps[i] = grp
+        # self.output_file_grps = actual_output_grps
+        # actual_input_grps = [*self.ocrd_tool_json['input_file_grp']]
+        # for i, grp in enumerate(self.input_file_grps):
+            # actual_input_grps[i] = grp
+        # self.input_file_grps = actual_input_grps
         param_validator = ParameterValidator(self.ocrd_tool_json)
         report = param_validator.validate(parameters)
         if not report.is_valid:
             raise Exception(report.errors)
+        # TODO remove once OCR-D/spec#121 lands
+        if 'output_file_grp' in self.ocrd_tool_json and not self.output_file_grps:
+            raise Exception("Processor requires output_file_grp but none was provided.")
         return report
 
     def __str__(self):

diff --git a/tests/test_task_sequence.py b/tests/test_task_sequence.py
@@ -79,8 +79,11 @@ def test_parse_no_in(self):
     def test_parse_implicit_after_validate(self):
         task = ProcessorTask.parse('%s -I IN -O OUT -p \'{"param1": true}\'' % SAMPLE_NAME_REQUIRED_PARAM)
         task.validate()
-        self.assertEqual(task.input_file_grps, ['IN', 'SECOND_IN'])
-        self.assertEqual(task.output_file_grps, ['OUT', 'SECOND_OUT'])
+        # TODO uncomment and adapt once OCR-D/spec#121 lands
+        # self.assertEqual(task.input_file_grps, ['IN', 'SECOND_IN'])
+        # self.assertEqual(task.output_file_grps, ['OUT', 'SECOND_OUT'])
+        self.assertEqual(task.input_file_grps, ['IN'])
+        self.assertEqual(task.output_file_grps, ['OUT'])
 
     def test_parse_unknown(self):
         with self.assertRaisesRegex(Exception, 'Failed parsing task description'):