ocrd 3.0.2__tar.gz → 3.0.4__tar.gz

{ocrd-3.0.2/src/ocrd.egg-info → ocrd-3.0.4}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.1
 Name: ocrd
-Version: 3.0.2
+Version: 3.0.4
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

ocrd-3.0.4/VERSION

@@ -0,0 +1 @@
+3.0.4

{ocrd-3.0.2 → ocrd-3.0.4}/src/ocrd/mets_server.py

@@ -474,7 +474,7 @@ class OcrdMetsServer:
             """
             workspace.save_mets()
             response = Response(content="The Mets Server is writing changes to disk.", media_type='text/plain')
-            self.log.info(f"PUT / -> {response.__dict__}")
+            self.log.debug(f"PUT / -> {response.__dict__}")
             return response
 
         @app.delete(path='/')
@@ -485,7 +485,7 @@ class OcrdMetsServer:
             workspace.save_mets()
             response = Response(content="The Mets Server will shut down soon...", media_type='text/plain')
             self.shutdown()
-            self.log.info(f"DELETE / -> {response.__dict__}")
+            self.log.debug(f"DELETE / -> {response.__dict__}")
             return response
 
         @app.post(path='/reload')
@@ -495,25 +495,25 @@ class OcrdMetsServer:
             """
             workspace.reload_mets()
             response = Response(content=f"Reloaded from {workspace.directory}", media_type='text/plain')
-            self.log.info(f"POST /reload -> {response.__dict__}")
+            self.log.debug(f"POST /reload -> {response.__dict__}")
             return response
 
         @app.get(path='/unique_identifier', response_model=str)
         async def unique_identifier():
             response = Response(content=workspace.mets.unique_identifier, media_type='text/plain')
-            self.log.info(f"GET /unique_identifier -> {response.__dict__}")
+            self.log.debug(f"GET /unique_identifier -> {response.__dict__}")
             return response
 
         @app.get(path='/workspace_path', response_model=str)
         async def workspace_path():
             response = Response(content=workspace.directory, media_type="text/plain")
-            self.log.info(f"GET /workspace_path -> {response.__dict__}")
+            self.log.debug(f"GET /workspace_path -> {response.__dict__}")
             return response
 
         @app.get(path='/physical_pages', response_model=OcrdPageListModel)
         async def physical_pages():
             response = {'physical_pages': workspace.mets.physical_pages}
-            self.log.info(f"GET /physical_pages -> {response}")
+            self.log.debug(f"GET /physical_pages -> {response}")
             return response
 
         @app.get(path='/physical_pages', response_model=OcrdPageListModel)
@@ -523,13 +523,13 @@ class OcrdMetsServer:
         @app.get(path='/file_groups', response_model=OcrdFileGroupListModel)
         async def file_groups():
             response = {'file_groups': workspace.mets.file_groups}
-            self.log.info(f"GET /file_groups -> {response}")
+            self.log.debug(f"GET /file_groups -> {response}")
             return response
 
         @app.get(path='/agent', response_model=OcrdAgentListModel)
         async def agents():
             response = OcrdAgentListModel.create(workspace.mets.agents)
-            self.log.info(f"GET /agent -> {response.__dict__}")
+            self.log.debug(f"GET /agent -> {response.__dict__}")
             return response
 
         @app.post(path='/agent', response_model=OcrdAgentModel)
@@ -538,7 +538,7 @@ class OcrdMetsServer:
             kwargs['_type'] = kwargs.pop('type')
             workspace.mets.add_agent(**kwargs)
             response = agent
-            self.log.info(f"POST /agent -> {response.__dict__}")
+            self.log.debug(f"POST /agent -> {response.__dict__}")
             return response
 
         @app.get(path="/file", response_model=OcrdFileListModel)
@@ -557,14 +557,14 @@ class OcrdMetsServer:
                 fileGrp=file_grp, ID=file_id, pageId=page_id, mimetype=mimetype, local_filename=local_filename, url=url
             )
             response = OcrdFileListModel.create(found)
-            self.log.info(f"GET /file -> {response.__dict__}")
+            self.log.debug(f"GET /file -> {response.__dict__}")
             return response
 
         @app.post(path='/file', response_model=OcrdFileModel)
         async def add_file(
             file_grp: str = Form(),
             file_id: str = Form(),
-            page_id: Optional[str] = Form(),
+            page_id: Optional[str] = Form(None),
             mimetype: str = Form(),
             url: Optional[str] = Form(None),
             local_filename: Optional[str] = Form(None),
@@ -582,7 +582,7 @@ class OcrdMetsServer:
             kwargs = file_resource.dict()
             workspace.add_file(**kwargs, force=force)
             response = file_resource
-            self.log.info(f"POST /file -> {response.__dict__}")
+            self.log.debug(f"POST /file -> {response.__dict__}")
             return response
 
         # ------------- #

{ocrd-3.0.2 → ocrd-3.0.4}/src/ocrd/processor/base.py

@@ -534,15 +534,20 @@ class Processor():
                     # forward messages from log queue (in subprocesses) to all root handlers
                     log_listener = logging.handlers.QueueListener(log_queue, *logging.root.handlers, respect_handler_level=True)
                     log_listener.start()
+                tasks = None
                 try:
                     self._base_logger.debug("started executor %s with %d workers", str(executor), max_workers or 1)
                     tasks = self.process_workspace_submit_tasks(executor, max_seconds)
                     stats = self.process_workspace_handle_tasks(tasks)
                 finally:
                     executor.shutdown(kill_workers=True, wait=False)
+                    self._base_logger.debug("stopped executor %s after %d tasks", str(executor), len(tasks) if tasks else -1)
                     if max_workers > 1:
-                        log_listener.stop()
-                        del log_listener
+                        # can cause deadlock:
+                        #log_listener.stop()
+                        # not much better:
+                        #log_listener.enqueue_sentinel()
+                        pass
 
             except NotImplementedError:
                 # fall back to deprecated method
@@ -670,11 +675,12 @@ class Processor():
                 nr_succeeded += 1
             # else skipped - already exists
         nr_errors = dict(nr_errors)
+        nr_all = nr_succeeded + nr_failed
         if nr_failed > 0:
-            nr_all = nr_succeeded + nr_failed
             if config.OCRD_MAX_MISSING_OUTPUTS > 0 and nr_failed / nr_all > config.OCRD_MAX_MISSING_OUTPUTS:
                 raise Exception(f"too many failures with {reason} output ({nr_failed} of {nr_all}, {str(nr_errors)})")
             self._base_logger.warning("%s %d of %d pages due to %s", reason, nr_failed, nr_all, str(nr_errors))
+        self._base_logger.debug("succeeded %d, missed %d of %d pages due to %s", nr_succeeded, nr_failed, nr_all, str(nr_errors))
         return nr_succeeded, nr_failed, nr_errors, len(tasks)
 
     def process_workspace_handle_page_task(self, page_id : str, input_files : List[Optional[OcrdFileType]], task : TFuture) -> Union[bool, Exception]:

{ocrd-3.0.2 → ocrd-3.0.4}/src/ocrd/processor/helpers.py

@@ -56,13 +56,19 @@ def run_processor(
     - :py:attr:`output_file_grp`
     - :py:attr:`parameter` (after applying any :py:attr:`parameter_override` settings)
 
-    Warning: Avoid setting the `instance_caching` flag to True. It may have unexpected side effects.
-    This flag is used for an experimental feature we would like to adopt in future.
-
     Run the processor on the workspace (creating output files in the filesystem).
 
     Finally, write back the workspace (updating the METS in the filesystem).
 
+    If :py:attr:`instance_caching` is True, then processor instances (for the same set
+    of :py:attr:`parameter` values) will be cached internally. Thus, these objects (and
+    all their memory resources, like loaded models) get re-used instead of re-instantiated
+    when a match occurs - as long as the program is being run. They only get deleted (and
+    their resources freed) when as many as :py:data:`~ocrd_utils.config.OCRD_MAX_PROCESSOR_CACHE`
+    instances have already been cached while this particular parameter set was re-used
+    least frequently. (See :py:class:`~ocrd_network.ProcessingWorker` and
+    :py:class:`~ocrd_network.ProcessorServer` for use-cases.)
+
     Args:
         processorClass (object): Python class of the module processor.
     """
@@ -137,7 +143,7 @@ def run_processor(
         t1_cpu,
         processor.input_file_grp or '',
         processor.output_file_grp or '',
-        json.dumps(processor.parameter) or '',
+        json.dumps(dict(processor.parameter or {})),
         processor.page_id or ''
     )
     workspace.mets.add_agent(
@@ -148,7 +154,7 @@ def run_processor(
         otherrole=otherrole,
         notes=[({'option': 'input-file-grp'}, processor.input_file_grp or ''),
                ({'option': 'output-file-grp'}, processor.output_file_grp or ''),
-               ({'option': 'parameter'}, json.dumps(processor.parameter or '')),
+               ({'option': 'parameter'}, json.dumps(dict(processor.parameter or {}))),
                ({'option': 'page-id'}, processor.page_id or '')]
     )
     workspace.save_mets()

{ocrd-3.0.2 → ocrd-3.0.4/src/ocrd.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
-Metadata-Version: 2.2
+Metadata-Version: 2.1
 Name: ocrd
-Version: 3.0.2
+Version: 3.0.4
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

{ocrd-3.0.2 → ocrd-3.0.4}/src/ocrd_models/ocrd_page_generateds.py

@@ -2,7 +2,7 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Sun Sep 15 21:49:27 2024 by generateDS.py version 2.44.1.
+# Generated Mon Feb 17 10:32:54 2025 by generateDS.py version 2.44.1.
 # Python 3.8.17+ (heads/3.8-dirty:1663f8ba84, Aug 15 2023, 18:13:01)  [GCC 8.3.0]
 #
 # Command line options:
@@ -7112,7 +7112,7 @@ class OrderedGroupIndexedType(GeneratedsSuper):
             else:
                 cleaned.append(entry)
         for entry in cleaned:
-            entry.export(outfile, level, entry.ns_prefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
+            entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
 # end class OrderedGroupIndexedType
 
 
@@ -8075,7 +8075,7 @@ class OrderedGroupType(GeneratedsSuper):
             else:
                 cleaned.append(entry)
         for entry in cleaned:
-            entry.export(outfile, level, entry.ns_prefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
+            entry.export(outfile, level, namespaceprefix_, namespacedef_='', name_=entry.__class__.__name__[:-4], pretty_print=pretty_print)
 # end class OrderedGroupType
 
 

{ocrd-3.0.2 → ocrd-3.0.4}/tests/test_mets_server.py

@@ -240,7 +240,7 @@ def test_find_all_files(start_mets_server : Tuple[str, Workspace]):
     assert len(mets.find_all_files(mimetype='//application/.*')) == 22, '22 application/.*'
     assert len(mets.find_all_files(mimetype=MIMETYPE_PAGE)) == 20, '20 ' + MIMETYPE_PAGE
     assert len(mets.find_all_files(local_filename='OCR-D-IMG/FILE_0005_IMAGE.tif')) == 1, '1 FILE xlink:href="OCR-D-IMG/FILE_0005_IMAGE.tif"'
-    assert len(mets.find_all_files(url='https://github.com/OCR-D/assets/raw/master/data/SBB0000F29300010000/00000001_DESKEW.tif')) == 1, '1 URL xlink:href="https://github.com/OCR-D/assets/raw/master/data/SBB0000F29300010000/00000001_DESKEW.tif"'
+    assert len(mets.find_all_files(url='https://github.com/OCR-D/assets/raw/master/data/SBB0000F29300010000/data/00000001_DESKEW.tif')) == 1, '1 URL xlink:href="https://github.com/OCR-D/assets/raw/master/data/SBB0000F29300010000/data/00000001_DESKEW.tif"'
     assert len(mets.find_all_files(pageId='PHYS_0001..PHYS_0005')) == 35, '35 files for page "PHYS_0001..PHYS_0005"'
     assert len(mets.find_all_files(pageId='//PHYS_000(1|2)')) == 34, '34 files in PHYS_001 and PHYS_0002'
     assert len(mets.find_all_files(pageId='//PHYS_0001,//PHYS_0005')) == 18, '18 files in PHYS_001 and PHYS_0005 (two regexes)'

{ocrd-3.0.2 → ocrd-3.0.4}/tests/test_workspace.py

@@ -224,7 +224,7 @@ def test_superfluous_copies_in_ws_dir(tmp_path):
     https://github.com/OCR-D/core/issues/227
     """
     # arrange
-    src_path = assets.path_to('SBB0000F29300010000/data/mets_one_file.xml')
+    src_path = assets.path_to('sample_bagit-with-fetch/data/PPN595930174.xml')
     dst_path = join(tmp_path, 'mets.xml')
     copyfile(src_path, dst_path)
     ws1 = Workspace(Resolver(), tmp_path)
@@ -233,12 +233,12 @@ def test_superfluous_copies_in_ws_dir(tmp_path):
     assert count_files(tmp_path) == 1
 
     # act
-    for file in ws1.mets.find_all_files():
-        ws1.download_file(file)
+    file1 = next(ws1.mets.find_files(fileGrp='DEFAULT'))
+    ws1.download_file(file1)
 
     # assert
     assert count_files(tmp_path) == 2
-    assert exists(join(tmp_path, 'OCR-D-IMG/FILE_0005_IMAGE.tif'))
+    assert exists(join(tmp_path, 'DEFAULT/FILE_0000_DEFAULT.jpg'))
 
 
 @pytest.fixture(name='sbb_data_tmp')

ocrd-3.0.2/VERSION

@@ -1 +0,0 @@
-3.0.2