OCR-D · bertsky · Jan 10, 2023 · Apr 19, 2024 · Jun 6, 2024 · Jun 6, 2024
diff --git a/src/ocrd/decorators/__init__.py b/src/ocrd/decorators/__init__.py
@@ -1,3 +1,4 @@
+from functools import partial
 import sys
 
 from ocrd_utils import (
@@ -8,6 +9,8 @@
     getLogger,
     parse_json_string_with_comments,
     set_json_key_value_overrides,
+    parse_json_string_or_file,
+    list_resource_candidates,
 )
 from ocrd_validators import WorkspaceValidator
 from ocrd_network import ProcessingWorker, ProcessorServer, AgentType
@@ -71,6 +74,18 @@ def ocrd_cli_wrap_processor(
 
     LOG = getLogger('ocrd.cli_wrap_processor')
     # LOG.info('kwargs=%s' % kwargs)
+    if 'parameter' in kwargs:
+        # Disambiguate parameter file/literal, and resolve file
+        disposable = processorClass(workspace=None)
+        # we cannot use Processor.resolve_resource() directly
+        # because it exits if unsuccessful
+        resolve = partial(list_resource_candidates,
+                          disposable.ocrd_tool['executable'],
+                          moduled=disposable.moduledir)
+        kwargs['parameter'] = parse_json_string_or_file(*list(kwargs['parameter']),
+                                                        resolve_preset_file=resolve)
+    else:
+        kwargs['parameter'] = dict()
     # Merge parameter overrides and parameters
     if 'parameter_override' in kwargs:
         set_json_key_value_overrides(kwargs['parameter'], *kwargs['parameter_override'])

diff --git a/src/ocrd/decorators/parameter_option.py b/src/ocrd/decorators/parameter_option.py
@@ -1,5 +1,5 @@
 from click import option
-from ocrd_utils import parse_json_string_or_file
+#from ocrd_utils import parse_json_string_or_file
 
 __all__ = ['parameter_option', 'parameter_override_option']
 
@@ -11,7 +11,9 @@ def _handle_param_option(ctx, param, value):
                                 help="Parameters, either JSON string or path to JSON file",
                                 multiple=True,
                                 default=['{}'],
-                                callback=_handle_param_option)
+                                )
+                                # now handled in ocrd_cli_wrap_processor to resolve processor preset files
+                                # callback=_handle_param_option)
 
 parameter_override_option = option('-P', '--parameter-override',
                                 help="Parameter override",

diff --git a/src/ocrd_utils/str.py b/src/ocrd_utils/str.py
@@ -2,6 +2,7 @@
 Utility functions for strings, paths and URL.
 """
 
+from os.path import exists
 import re
 import json
 from typing import List, Union
@@ -159,7 +160,7 @@ def parse_json_string_with_comments(val):
     jsonstr = re.sub(r'^\s*#.*$', '', val, flags=re.MULTILINE)
     return json.loads(jsonstr)
 
-def parse_json_string_or_file(*values):    # pylint: disable=unused-argument
+def parse_json_string_or_file(*values, resolve_preset_file=None):    # pylint: disable=unused-argument
     """
     Parse a string as either the path to a JSON object or a literal JSON object.
 
@@ -173,7 +174,12 @@ def parse_json_string_or_file(*values):    # pylint: disable=unused-argument
             continue
         try:
             try:
-                with open(value, 'r') as f:
+                path = value
+                if callable(resolve_preset_file):
+                    for candidate in resolve_preset_file(value):
+                        if exists(candidate):
+                            path = candidate
+                with open(path, 'r') as f:
                     value_parsed = parse_json_string_with_comments(f.read())
             except (FileNotFoundError, OSError):
                 value_parsed = parse_json_string_with_comments(value.strip())

diff --git a/tests/processor/test_processor.py b/tests/processor/test_processor.py
@@ -1,14 +1,16 @@
 import json
 
 from tempfile import TemporaryDirectory
-from os.path import join
+from pathlib import Path
+from os import environ
 from tests.base import CapturingTestCase as TestCase, assets, main # pylint: disable=import-error, no-name-in-module
 from tests.data import DummyProcessor, DummyProcessorWithRequiredParameters, DummyProcessorWithOutput, IncompleteProcessor
 
 from ocrd_utils import MIMETYPE_PAGE, pushd_popd, initLogging, disableLogging
 from ocrd.resolver import Resolver
 from ocrd.processor.base import Processor, run_processor, run_cli
 
+from unittest import mock
 import pytest
 
 class TestProcessor(TestCase):
@@ -48,8 +50,8 @@ def test_with_mets_url_input_files(self):
         assert [f.mimetype for f in processor.input_files] == [MIMETYPE_PAGE, MIMETYPE_PAGE]
 
     def test_parameter(self):
-        with TemporaryDirectory() as tempdir:
-            jsonpath = join(tempdir, 'params.json')
+        with TemporaryDirectory():
+            jsonpath = Path('params.json').name
             with open(jsonpath, 'w') as f:
                 f.write('{"baz": "quux"}')
             with open(jsonpath, 'r') as f:
@@ -73,6 +75,30 @@ def test_params_missing_required(self):
         with self.assertRaisesRegex(Exception, 'is a required property'):
             DummyProcessorWithRequiredParameters(workspace=self.workspace)
 
+    def test_params_preset_resolve(self):
+        with pushd_popd(tempdir=True) as tempdir:
+            with mock.patch.dict(environ, {'XDG_DATA_HOME': str(tempdir)}):
+                path = Path(tempdir) / 'ocrd-resources' / 'ocrd-dummy'
+                path.mkdir(parents=True)
+                path = str(path / 'preset.json')
+                with open(path, 'w') as out:
+                    # it would be nicer to test some existing processor which does take params
+                    out.write('{}')
+                # FIXME: we cannot directly use self.workspace, but need a temporary copy
+                assert 0 == run_cli("ocrd-dummy",
+                                    resolver=Resolver(),
+                                    mets_url=self.workspace.mets_target,
+                                    input_file_grp="OCR-D-IMG",
+                                    output_file_grp="DUMMY",
+                                    parameter=path)
+                assert 0 == run_cli("ocrd-dummy",
+                                    resolver=Resolver(),
+                                    mets_url=self.workspace.mets_target,
+                                    input_file_grp="OCR-D-IMG",
+                                    output_file_grp="DUMMY",
+                                    parameter='preset.json',
+                                    overwrite=True)
+
     def test_params(self):
         proc = Processor(workspace=self.workspace)
         self.assertEqual(proc.parameter, {})

diff --git a/tests/test_decorators.py b/tests/test_decorators.py
@@ -4,6 +4,7 @@
 from click.testing import CliRunner
 from tempfile import TemporaryDirectory
 from os.path import join, exists
+import pytest
 
 from tests.base import CapturingTestCase as TestCase, assets, main, copy_of_directory # pylint: disable=import-error, no-name-in-module
 from tests.data import DummyProcessor
@@ -105,6 +106,7 @@ def test_processor_run(self):
                 exit_code, out, err = self.invoke_cli(cli_dummy_processor, ['-p', '{"baz": "forty-two"}', '--mets', 'mets.xml', *DEFAULT_IN_OUT])
                 assert not exit_code
 
+    @pytest.mark.skip(reason="now deferred to ocrd_cli_wrap_processor (to resolve preset resources)")
     def test_param_merging(self):
         json1 = '{"foo": 23, "bar": 100}'
         json2 = '{"foo": 42}'