ocrd 3.0.2__py3-none-any.whl → 3.0.4__py3-none-any.whl

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/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/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.dist-info → ocrd-3.0.4.dist-info}/METADATA

@@ -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
@@ -24,15 +24,13 @@ Requires-Dist: Flask
 Requires-Dist: frozendict>=2.3.4
 Requires-Dist: gdown
 Requires-Dist: httpx>=0.22.0
-Requires-Dist: importlib_metadata; python_version < "3.8"
-Requires-Dist: importlib_resources; python_version < "3.10"
 Requires-Dist: jsonschema>=4
 Requires-Dist: loky
 Requires-Dist: lxml
 Requires-Dist: memory-profiler>=0.58.0
 Requires-Dist: numpy
 Requires-Dist: ocrd-fork-bagit>=1.8.1.post2
-Requires-Dist: ocrd-fork-bagit_profile>=1.3.0.post1
+Requires-Dist: ocrd-fork-bagit-profile>=1.3.0.post1
 Requires-Dist: opencv-python-headless
 Requires-Dist: paramiko
 Requires-Dist: pika>=1.2.0
@@ -42,10 +40,12 @@ Requires-Dist: python-magic
 Requires-Dist: python-multipart
 Requires-Dist: pyyaml
 Requires-Dist: requests
-Requires-Dist: requests_unixsocket2
+Requires-Dist: requests-unixsocket2
 Requires-Dist: shapely
 Requires-Dist: uvicorn
 Requires-Dist: uvicorn>=0.17.6
+Requires-Dist: importlib-resources; python_version < "3.10"
+Requires-Dist: importlib-metadata; python_version < "3.8"
 
 # OCR-D/core
 

{ocrd-3.0.2.dist-info → ocrd-3.0.4.dist-info}/RECORD

@@ -1,7 +1,7 @@
 ocrd/__init__.py,sha256=ZswMVmlqFhAEIzMR3my6IKPq9XLH21aDPC_m_8Jh4dA,1076
 ocrd/constants.py,sha256=6dn3mG54WqHsKInmLZp4kJjNqqPtBoFoSuLUuRbOps0,740
 ocrd/lib.bash,sha256=le6XqAOEacdjP3JNSlPkxwRH1y0oVjNQM2tX5d6QFO4,10901
-ocrd/mets_server.py,sha256=wiOvSSYZMicjQC958dD2i7uHMwfZCaDZQd7E5l50ayU,22436
+ocrd/mets_server.py,sha256=EL6CMtVjmVPs9--3Vn-JyhU2VDMNw-6QN_XWRKFp6uk,22451
 ocrd/ocrd-all-tool.json,sha256=9bX2VYnUwhTAzAvKaoT77BFzbgBGgyIt7qBqARpwWNc,586
 ocrd/resolver.py,sha256=Ba9ALQbTXz6_mla4VqN9tAfHoj6aKuNJAU4tIDnjcHE,14952
 ocrd/resource_list.yml,sha256=82-PiqkZnka1kTj3MQqNn4wXWKHHtoFchsQuetWuqFs,2633
@@ -26,8 +26,8 @@ ocrd/decorators/mets_find_options.py,sha256=d4oATKMP6bFQHNqOK6nLqgUiWF2FYdkPvzkT
 ocrd/decorators/ocrd_cli_options.py,sha256=lIvtE8re1VmpHm45u71ltE0QJS8nyd28HhLC7zGSvlo,2691
 ocrd/decorators/parameter_option.py,sha256=n8hYw7XVTd3i3tvpK8F1Jx_CqRp6EGF9qJVH95yj92Q,1076
 ocrd/processor/__init__.py,sha256=39ymNwYRdc-b_OJzzKmWCvo2ga3KdsGSYDHE1Hzkn_w,274
-ocrd/processor/base.py,sha256=L396ON0p77D8r34CESJ2SIYspTZ6mYFqaBq6DYg6ywM,59315
-ocrd/processor/helpers.py,sha256=8ngrqAJ01BSoSJNsIoK_YfA8QdryM5y0MqaqA9f7ELM,10483
+ocrd/processor/base.py,sha256=wAu3d5E2Vt-rWBejzgZOOnxuI0njgShuLfRZPxyuUGk,59731
+ocrd/processor/helpers.py,sha256=gIc6PdvOS1sR0UkYlrdZopImAXxXglDBNpgNZGWHO7Y,10987
 ocrd/processor/ocrd_page_result.py,sha256=eDkpyVHcpaBzTHXiGrcNk9PP9Xr-XZru2w_uoX_ZeNA,510
 ocrd/processor/builtin/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ocrd/processor/builtin/dummy_processor.py,sha256=a-4kKJ1JeXQuBIyyN8w2R3s7ov-wAfyEdEz3nxrf0sU,3479
@@ -43,7 +43,7 @@ ocrd_models/ocrd_exif.py,sha256=wRSprHxCy9LCXw41Fi9kp-CbFc5NFX9ZFIFNszB41qk,4585
 ocrd_models/ocrd_file.py,sha256=7lyHezuNnl2FEYV1lV35-QTCrgYAL-3wO2ulFUNq2Ak,9717
 ocrd_models/ocrd_mets.py,sha256=cRBEnvRNmRUhcrKouqygTQqYA-XdgyA-FNLwt86V0vg,42878
 ocrd_models/ocrd_page.py,sha256=TTCnvpKGyZx1dqH8LnDiVVVPjU6emWGVLO_4o9rQHtw,6233
-ocrd_models/ocrd_page_generateds.py,sha256=vjXPob5UMxYxPI6un8SYTL673Nhph0E2WEZexBgbvY4,841794
+ocrd_models/ocrd_page_generateds.py,sha256=IWoN3V-v3C4JgyPaFh9OQC87ob__wUP1Q6ELBxhLA1w,841794
 ocrd_models/ocrd_xml_base.py,sha256=OW57mXLlwm1nH8CNefvXmwLRws9KL9zSrb-3vH--mX8,1641
 ocrd_models/report.py,sha256=luZxvzAAQyGYOlRNSJQUIUIANG81iGmBW5ag-uXxKCA,2026
 ocrd_models/utils.py,sha256=0_WHf5NEn1WC8MKJc6X_RK8gW-70Z09_mslkKOj7uF8,2369
@@ -120,9 +120,9 @@ ocrd_validators/xlink.xsd,sha256=8fW7YAMWXN2PbB_MMvj9H5ZeFoEBDzuYBtlGC8_6ijw,318
 ocrd_validators/xsd_mets_validator.py,sha256=4GWfLyqkmca0x7osDuXuExYuM0HWVrKoqn0S35sFhHU,467
 ocrd_validators/xsd_page_validator.py,sha256=BNz_9u-Ek4UCeyZu3KxSQoolfW9lvuaSR9nIu1XXxeE,467
 ocrd_validators/xsd_validator.py,sha256=6HrVAf6SzCvfUIuQdIzz9bOq4V-zhyii9yrUPoK2Uvo,2094
-ocrd-3.0.2.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-ocrd-3.0.2.dist-info/METADATA,sha256=e6sO3z4kPE8qe8Ne-aPLX5Tl03M_NcrORBnlAQodLDE,10442
-ocrd-3.0.2.dist-info/WHEEL,sha256=In9FTNxeP60KnTkGw7wk6mJPYd_dQSjEZmXdBdMCI-8,91
-ocrd-3.0.2.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
-ocrd-3.0.2.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.0.2.dist-info/RECORD,,
+ocrd-3.0.4.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.0.4.dist-info/METADATA,sha256=BRwaS739FiiA7T2QsmeVSBSddJu_FyGyh7Jzxy2D1_g,10442
+ocrd-3.0.4.dist-info/WHEEL,sha256=P9jw-gEje8ByB7_hXoICnHtVCrEwMQh-630tKvQWehc,91
+ocrd-3.0.4.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
+ocrd-3.0.4.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.0.4.dist-info/RECORD,,

{ocrd-3.0.2.dist-info → ocrd-3.0.4.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (75.8.0)
+Generator: setuptools (75.3.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

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