ocrd 3.0.0b7__py3-none-any.whl → 3.0.1__py3-none-any.whl

ocrd/cli/__init__.py

@@ -16,7 +16,7 @@ def command_with_replaced_help(*replacements):
 
     class CommandWithReplacedHelp(click.Command):
         def get_help(self, ctx):
-            newhelp = super().get_help(ctx)
+            newhelp : str = super().get_help(ctx)
             for replacement in replacements:
                 newhelp = re.sub(*replacement, newhelp)
             # print(newhelp)
@@ -83,6 +83,8 @@ Variables:
 \b
 {config.describe('OCRD_NETWORK_RABBITMQ_CLIENT_CONNECT_ATTEMPTS')}
 \b
+{config.describe('OCRD_NETWORK_RABBITMQ_HEARTBEAT')}
+\b
 {config.describe('OCRD_PROFILE_FILE')}
 \b
 {config.describe('OCRD_PROFILE', wrap_text=False)}

ocrd/decorators/__init__.py

@@ -48,6 +48,9 @@ def ocrd_cli_wrap_processor(
     # ocrd_network params end #
     **kwargs
 ):
+    # init logging handlers so no imported libs can preempt ours
+    initLogging()
+
     # FIXME: remove workspace arg entirely
     processor = processorClass(None)
     if not sys.argv[1:]:
@@ -89,8 +92,6 @@ def ocrd_cli_wrap_processor(
         # Used for checking/starting network agents for the WebAPI architecture
         check_and_run_network_agent(processorClass, subcommand, address, database, queue)
 
-    # from here: single-run processing context
-    initLogging()
     if 'parameter' in kwargs:
         # Disambiguate parameter file/literal, and resolve file
         def resolve(name):

ocrd/mets_server.py

@@ -1,8 +1,10 @@
 """
 # METS server functionality
 """
+import os
 import re
 from os import _exit, chmod
+import signal
 from typing import Dict, Optional, Union, List, Tuple
 from time import sleep
 from pathlib import Path
@@ -155,13 +157,13 @@ class ClientSideOcrdMets:
         Request writing the changes to the file system
         """
         if not self.multiplexing_mode:
-            self.session.request("PUT", url=self.url)
+            return self.session.request("PUT", url=self.url).text
         else:
-            self.session.request(
+            return self.session.request(
                 "POST",
                 self.url,
                 json=MpxReq.save(self.ws_dir_path)
-            )
+            ).json()["text"]
 
     def stop(self):
         """
@@ -169,14 +171,13 @@ class ClientSideOcrdMets:
         """
         try:
             if not self.multiplexing_mode:
-                self.session.request("DELETE", self.url)
-                return
+                return self.session.request("DELETE", self.url).text
             else:
-                self.session.request(
+                return self.session.request(
                     "POST",
                     self.url,
                     json=MpxReq.stop(self.ws_dir_path)
-                )
+                ).json()["text"]
         except ConnectionError:
             # Expected because we exit the process without returning
             pass
@@ -323,7 +324,7 @@ class ClientSideOcrdMets:
 
 
 class MpxReq:
-    """This class wrapps the request bodies needed for the tcp forwarding
+    """This class wraps the request bodies needed for the tcp forwarding
 
     For every mets-server-call like find_files or workspace_path a special request_body is
     needed to call `MetsServerProxy.forward_tcp_request`. These are created by this functions.
@@ -346,12 +347,12 @@ class MpxReq:
     @staticmethod
     def save(ws_dir_path: str) -> Dict:
         return MpxReq.__args_wrapper(
-            ws_dir_path, method_type="PUT", response_type="empty", request_url="", request_data={})
+            ws_dir_path, method_type="PUT", response_type="text", request_url="", request_data={})
 
     @staticmethod
     def stop(ws_dir_path: str) -> Dict:
         return MpxReq.__args_wrapper(
-            ws_dir_path, method_type="DELETE", response_type="empty", request_url="", request_data={})
+            ws_dir_path, method_type="DELETE", response_type="text", request_url="", request_data={})
 
     @staticmethod
     def reload(ws_dir_path: str) -> Dict:
@@ -428,18 +429,24 @@ class OcrdMetsServer:
 
     @staticmethod
     def kill_process(mets_server_pid: int):
-        subprocess_run(args=["kill", "-s", "SIGINT", f"{mets_server_pid}"], shell=False, universal_newlines=True)
+        os.kill(mets_server_pid, signal.SIGINT)
+        sleep(3)
+        try:
+            os.kill(mets_server_pid, signal.SIGKILL)
+        except ProcessLookupError as e:
+            pass
 
     def shutdown(self):
+        pid = os.getpid()
+        self.log.info(f"Shutdown method of mets server[{pid}] invoked, sending SIGTERM signal.")
+        os.kill(pid, signal.SIGTERM)
         if self.is_uds:
             if Path(self.url).exists():
-                self.log.debug(f'UDS socket {self.url} still exists, removing it')
+                self.log.warning(f"Due to a server shutdown, removing the existing UDS socket file: {self.url}")
                 Path(self.url).unlink()
-        # os._exit because uvicorn catches SystemExit raised by sys.exit
-        _exit(0)
 
     def startup(self):
-        self.log.info("Starting up METS server")
+        self.log.info(f"Configuring the Mets Server")
 
         workspace = self.workspace
 
@@ -465,32 +472,49 @@ class OcrdMetsServer:
             """
             Write current changes to the file system
             """
-            return workspace.save_mets()
+            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__}")
+            return response
 
         @app.delete(path='/')
-        async def stop():
+        def stop():
             """
             Stop the mets server
             """
-            getLogger('ocrd.models.ocrd_mets').info(f'Shutting down METS Server {self.url}')
             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__}")
+            return response
 
         @app.post(path='/reload')
-        async def workspace_reload_mets():
+        def workspace_reload_mets():
             """
             Reload mets file from the file system
             """
             workspace.reload_mets()
-            return Response(content=f'Reloaded from {workspace.directory}', media_type="text/plain")
+            response = Response(content=f"Reloaded from {workspace.directory}", media_type='text/plain')
+            self.log.info(f"POST /reload -> {response.__dict__}")
+            return response
 
         @app.get(path='/unique_identifier', response_model=str)
         async def unique_identifier():
-            return Response(content=workspace.mets.unique_identifier, media_type='text/plain')
+            response = Response(content=workspace.mets.unique_identifier, media_type='text/plain')
+            self.log.info(f"GET /unique_identifier -> {response.__dict__}")
+            return response
 
         @app.get(path='/workspace_path', response_model=str)
         async def workspace_path():
-            return Response(content=workspace.directory, media_type="text/plain")
+            response = Response(content=workspace.directory, media_type="text/plain")
+            self.log.info(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}")
+            return response
 
         @app.get(path='/physical_pages', response_model=OcrdPageListModel)
         async def physical_pages():
@@ -498,18 +522,24 @@ class OcrdMetsServer:
 
         @app.get(path='/file_groups', response_model=OcrdFileGroupListModel)
         async def file_groups():
-            return {'file_groups': workspace.mets.file_groups}
+            response = {'file_groups': workspace.mets.file_groups}
+            self.log.info(f"GET /file_groups -> {response}")
+            return response
 
         @app.get(path='/agent', response_model=OcrdAgentListModel)
         async def agents():
-            return OcrdAgentListModel.create(workspace.mets.agents)
+            response = OcrdAgentListModel.create(workspace.mets.agents)
+            self.log.info(f"GET /agent -> {response.__dict__}")
+            return response
 
         @app.post(path='/agent', response_model=OcrdAgentModel)
         async def add_agent(agent: OcrdAgentModel):
             kwargs = agent.dict()
             kwargs['_type'] = kwargs.pop('type')
             workspace.mets.add_agent(**kwargs)
-            return agent
+            response = agent
+            self.log.info(f"POST /agent -> {response.__dict__}")
+            return response
 
         @app.get(path="/file", response_model=OcrdFileListModel)
         async def find_files(
@@ -526,7 +556,9 @@ class OcrdMetsServer:
             found = workspace.mets.find_all_files(
                 fileGrp=file_grp, ID=file_id, pageId=page_id, mimetype=mimetype, local_filename=local_filename, url=url
             )
-            return OcrdFileListModel.create(found)
+            response = OcrdFileListModel.create(found)
+            self.log.info(f"GET /file -> {response.__dict__}")
+            return response
 
         @app.post(path='/file', response_model=OcrdFileModel)
         async def add_file(
@@ -549,7 +581,9 @@ class OcrdMetsServer:
             # Add to workspace
             kwargs = file_resource.dict()
             workspace.add_file(**kwargs, force=force)
-            return file_resource
+            response = file_resource
+            self.log.info(f"POST /file -> {response.__dict__}")
+            return response
 
         # ------------- #
 
@@ -557,9 +591,6 @@ class OcrdMetsServer:
             # Create socket and change to world-readable and -writable to avoid permission errors
             self.log.debug(f"chmod 0o677 {self.url}")
             server = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
-            if Path(self.url).exists() and not is_socket_in_use(self.url):
-                # remove leftover unused socket which blocks startup
-                Path(self.url).unlink()
             server.bind(self.url)  # creates the socket file
             atexit.register(self.shutdown)
             server.close()
@@ -571,16 +602,5 @@ class OcrdMetsServer:
         uvicorn_kwargs['log_config'] = None
         uvicorn_kwargs['access_log'] = False
 
-        self.log.debug("Starting uvicorn")
+        self.log.info("Starting the uvicorn Mets Server")
         uvicorn.run(app, **uvicorn_kwargs)
-
-
-def is_socket_in_use(socket_path):
-    if Path(socket_path).exists():
-        client = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
-        try:
-            client.connect(socket_path)
-        except OSError:
-            return False
-        client.close()
-        return True

ocrd/processor/base.py

@@ -130,7 +130,8 @@ class DummyExecutor:
     def __init__(self, initializer=None, initargs=(), **kwargs):
         initializer(*initargs)
     def shutdown(self, **kwargs):
-        pass
+        # allow gc to catch processor instance (unless cached)
+        _page_worker_set_ctxt(None, None)
     def submit(self, fn, *args, **kwargs) -> DummyFuture:
         return DummyFuture(fn, *args, **kwargs)
 
@@ -372,7 +373,7 @@ class Processor():
                 deprecated(version='3.0', reason='process() should be replaced with process_page_pcgts() or process_page_file() or process_workspace()')(getattr(self, 'process')))
 
     def __del__(self):
-        self._base_logger.debug("shutting down")
+        self._base_logger.debug("shutting down %s in %s", repr(self), mp.current_process().name)
         self.shutdown()
 
     def show_help(self, subcommand=None):
@@ -505,7 +506,7 @@ class Processor():
                 # set up multitasking
                 max_workers = max(0, config.OCRD_MAX_PARALLEL_PAGES)
                 if self.max_workers > 0 and self.max_workers < config.OCRD_MAX_PARALLEL_PAGES:
-                    self._base_logger.info("limiting number of threads from %d to %d", max_workers, self.max_workers)
+                    self._base_logger.info("limiting number of workers from %d to %d", max_workers, self.max_workers)
                     max_workers = self.max_workers
                 if max_workers > 1:
                     assert isinstance(workspace.mets, ClientSideOcrdMets), \
@@ -518,12 +519,9 @@ class Processor():
                 if max_workers > 1:
                     executor_cls = ProcessPoolExecutor
                     log_queue = mp.Queue()
-                    # 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)
                 else:
                     executor_cls = DummyExecutor
                     log_queue = None
-                    log_listener = None
                 executor = executor_cls(
                     max_workers=max_workers or 1,
                     # only forking method avoids pickling
@@ -533,6 +531,8 @@ class Processor():
                     initargs=(self, log_queue),
                 )
                 if max_workers > 1:
+                    # 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()
                 try:
                     self._base_logger.debug("started executor %s with %d workers", str(executor), max_workers or 1)
@@ -542,6 +542,7 @@ class Processor():
                     executor.shutdown(kill_workers=True, wait=False)
                     if max_workers > 1:
                         log_listener.stop()
+                        del log_listener
 
             except NotImplementedError:
                 # fall back to deprecated method

ocrd/processor/builtin/dummy/ocrd-tool.json

@@ -16,6 +16,26 @@
           "description": "Whether to actually copy files (true) or just create PAGE-XML as a side effect (false)"
         }
       }
+    },
+    "ocrd-filter": {
+      "executable": "ocrd-filter",
+      "description": "Bare-bones processor can be dynamically configured to remove segments based on XPath queries",
+      "steps": ["recognition/post-correction"],
+      "categories": ["Quality assurance"],
+      "input_file_grp_cardinality": 1,
+      "output_file_grp_cardinality": 1,
+      "parameters": {
+        "select": {
+          "type": "string",
+          "default": "//*[ends-with(local-name(),'Region')]",
+          "description": "Which segments to select for removal. An XPath 2.0 query expression (path and optional predicates), with 'pc' as namespace prefix for PAGE-XML and our extension functions (see help text). Only selection of segment hierarchy elements is allowed (so e.g. `*` would be equivalent to `pc:NoiseRegion|pc:LineDrawingRegion|pc:AdvertRegion|pc:ImageRegion|pc:ChartRegion|pc:MusicRegion|pc:GraphicRegion|pc:UnknownRegion|pc:CustomRegion|pc:SeparatorRegion|pc:MathsRegion|pc:TextRegion|pc:MapRegion|pc:ChemRegion|pc:TableRegion|pc:TextLine|pc:Word|pc:Glyph`, but `pc:MetadataItem` or `pc:Border` or `pc:Coords` would not match).\nFor example, to remove words or glyphs with low text confidence, select '(pc:Word|pc:Glyph)[pc:TextEquiv/@conf < 0.7]'. Or low layout confidence, '*[pc:Coords/@conf < 0.7]'.\nTo remove high pixel-to-character rate, select '*[pc:pixelarea(.) div string-length(pc:textequiv(.)) > 10000]'."
+        },
+        "plot": {
+          "type": "boolean",
+          "default": false,
+          "description": "Whether to extract an image for each filtered segment and write to the output fileGrp."
+        }
+      }
     }
   }
 }

ocrd/processor/builtin/dummy_processor.py

@@ -13,9 +13,6 @@ from ocrd_utils import (
     make_file_id,
     MIME_TO_EXT,
     MIMETYPE_PAGE,
-    parse_json_string_with_comments,
-    resource_string,
-    config
 )
 from ocrd_modelfactory import page_from_file
 

ocrd/processor/builtin/filter_processor.py

@@ -0,0 +1,108 @@
+# pylint: disable=missing-module-docstring,invalid-name
+from typing import Optional
+
+from lxml import etree
+import click
+
+from ocrd import Processor, OcrdPageResult, OcrdPageResultImage
+from ocrd.decorators import ocrd_cli_options, ocrd_cli_wrap_processor
+from ocrd_models import OcrdPage
+
+_SEGTYPES = [
+    "NoiseRegion",
+    "LineDrawingRegion",
+    "AdvertRegion",
+    "ImageRegion",
+    "ChartRegion",
+    "MusicRegion",
+    "GraphicRegion",
+    "UnknownRegion",
+    "CustomRegion",
+    "SeparatorRegion",
+    "MathsRegion",
+    "TextRegion",
+    "MapRegion",
+    "ChemRegion",
+    "TableRegion",
+    "TextLine",
+    "Word",
+    "Glyph"
+]
+
+class FilterProcessor(Processor):
+    def process_page_pcgts(self, *input_pcgts: Optional[OcrdPage], page_id: Optional[str] = None) -> OcrdPageResult:
+        """
+        Remove PAGE segment hierarchy elements based on flexible selection criteria.
+
+        Open and deserialise PAGE input file, then iterate over the segment hierarchy
+        down to the level required for ``select`` (which could be multiple levels at once).
+
+        Remove any segments matching XPath query ``select`` from that hierarchy (and from
+        the `ReadingOrder` if it is a region type).
+
+        \b
+        Besides full XPath 2.0 syntax, this supports extra predicates:
+        - `pc:pixelarea()` for the number of pixels of the bounding box (or sum area on node sets),
+        - `pc:textequiv()` for the first TextEquiv unicode string (or concatenated string on node sets).
+
+        If ``plot`` is `true`, then extract and write an image file for all removed segments
+        to the output fileGrp (without reference to the PAGE).
+
+        Produce a new PAGE output file by serialising the resulting hierarchy.
+        """
+        pcgts = input_pcgts[0]
+        result = OcrdPageResult(pcgts)
+        nodes = pcgts.xpath(self.parameter['select'])
+        # get PAGE objects from matching etree nodes
+        # but allow only hierarchy segments
+        segments = [segment for segment in map(pcgts.revmap.get, nodes)
+                    if segment.__class__.__name__.replace('Type', '') in _SEGTYPES]
+        if not(len(segments)):
+            self.logger.info("no matches")
+            return result
+        rodict = pcgts.get_Page().get_ReadingOrderGroups()
+        if self.parameter['plot']:
+            page_image, page_coords, _ = self.workspace.image_from_page(pcgts.get_Page(), page_id)
+        for segment in segments:
+            segtype = segment.original_tagname_
+            self.logger.info("matched %s segment %s", segtype, segment.id)
+            parent = segment.parent_object_
+            partype = parent.__class__.__name__.replace('Type', '')
+            if partype == 'Page':
+                getattr(parent, 'get_' + segtype)().remove(segment)
+            elif partype.endswith('Region'):
+                if segtype.endswith('Region'):
+                    getattr(parent, 'get_' + segtype)().remove(segment)
+                else:
+                    parent.TextLine.remove(segment)
+            elif partype == 'TextLine':
+                parent.Word.remove(segment)
+            elif partype == 'Word':
+                parent.Glyph.remove(segment)
+            else:
+                raise Exception(f"unexpected type ({partype}) of parent for matched segment ({segtype})")
+            segment.parent_object_ = None
+            if segtype.endswith('Region') and segment.id in rodict:
+                # remove from ReadingOrder as well
+                roelem = rodict[segment.id]
+                rorefs = getattr(roelem.parent_object_, roelem.__class__.__name__.replace('Type', ''))
+                rorefs.remove(roelem)
+                roelem.parent_object_ = None
+                del rodict[segment.id]
+            if self.parameter['plot']:
+                segment_image, _ = self.workspace.image_from_segment(segment, page_image, page_coords)
+                result.images.append(OcrdPageResultImage(segment_image, segment.id + '.IMG', None))
+        return result
+
+    @property
+    def metadata_filename(self):
+        return 'processor/builtin/dummy/ocrd-tool.json'
+
+    @property
+    def executable(self):
+        return 'ocrd-filter'
+
+@click.command()
+@ocrd_cli_options
+def cli(*args, **kwargs):
+    return ocrd_cli_wrap_processor(FilterProcessor, *args, **kwargs)

ocrd/resource_manager.py

@@ -23,6 +23,10 @@ yaml.constructor.SafeConstructor.yaml_constructors['tag:yaml.org,2002:timestamp'
 
 # pylint: enable=wrong-import-position
 
+# pylint: enable=wrong-import-position
+
+# pylint: enable=wrong-import-position
+
 from ocrd_validators import OcrdResourceListValidator
 from ocrd_utils import getLogger, directory_size, get_moduledir, guess_media_type, config
 from ocrd_utils.os import get_processor_resource_types, list_all_resources, pushd_popd, get_ocrd_tool_json

{ocrd-3.0.0b7.dist-info → ocrd-3.0.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.0.0b7
+Version: 3.0.1
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0
@@ -17,6 +17,7 @@ Requires-Dist: click>=7
 Requires-Dist: cryptography<43.0.0
 Requires-Dist: Deprecated==1.2.0
 Requires-Dist: docker
+Requires-Dist: elementpath
 Requires-Dist: fastapi>=0.78.0
 Requires-Dist: filetype
 Requires-Dist: Flask

{ocrd-3.0.0b7.dist-info → ocrd-3.0.1.dist-info}/RECORD

@@ -1,16 +1,16 @@
 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=U62eih1_O_N0StunVFkEustFs2PlrcMzccraj6_QRk4,21295
+ocrd/mets_server.py,sha256=wiOvSSYZMicjQC958dD2i7uHMwfZCaDZQd7E5l50ayU,22436
 ocrd/ocrd-all-tool.json,sha256=9bX2VYnUwhTAzAvKaoT77BFzbgBGgyIt7qBqARpwWNc,586
 ocrd/resolver.py,sha256=Ba9ALQbTXz6_mla4VqN9tAfHoj6aKuNJAU4tIDnjcHE,14952
 ocrd/resource_list.yml,sha256=82-PiqkZnka1kTj3MQqNn4wXWKHHtoFchsQuetWuqFs,2633
-ocrd/resource_manager.py,sha256=8BMVKJq8J56hugi8vtGn9Ffuk7oRkbs197aG74aKbCY,16733
+ocrd/resource_manager.py,sha256=kIWDoKxWH4IJE1gcoTcCRQjYjieCqiQclyuyF6Y9b8A,16813
 ocrd/task_sequence.py,sha256=spiaUQaMM7M8WdBDoQGmLuTPm7tOugYXD6rcJ2UXzxw,6991
 ocrd/workspace.py,sha256=cedqK7es2i2nwQCiUiVyWk3j4-nH7bsi6TF7v8siTio,65794
 ocrd/workspace_backup.py,sha256=iab_JjZ_mMP-G8NIUk4PZmfpNlQuGRoqc3NbTSSew1w,3621
 ocrd/workspace_bagger.py,sha256=yU8H3xR5WmQKvgQewac71ie-DUWcfLnMS01D55zsEHQ,11971
-ocrd/cli/__init__.py,sha256=lNR6wMf7JhQ8Jf33tUkowJr0mB3423OMY0_6dkMRLvU,2672
+ocrd/cli/__init__.py,sha256=-BiwIakeCkWx0Jd2yX9_ahfdV4VKz_5yqGEJ_2zKakQ,2734
 ocrd/cli/bashlib.py,sha256=ypFBM3-IULz_IEBx0Y04eGt9VbQWwEWm4ujm9g_hPWY,6009
 ocrd/cli/log.py,sha256=6_FrVmTKIIVNUaNLkuOJx8pvPhensHMuayJ0PA7T-XA,1562
 ocrd/cli/network.py,sha256=oWBHFEURxfUdb_t-F4svP_ri7o5mqBoNQnLZLbsZLTA,602
@@ -20,50 +20,51 @@ ocrd/cli/resmgr.py,sha256=bTE-MpF7RRCHhgAbknqZUFHgHScIK6FR3S4h4DEAets,10080
 ocrd/cli/validate.py,sha256=nvageDaHCETcE71X5lu7i_4JKpgo9MrvJKinVPLYUTI,5727
 ocrd/cli/workspace.py,sha256=KTbSzIUrba5WoYETvM9ElRZVsDUHCGVvjoFgBGZS2nU,40468
 ocrd/cli/zip.py,sha256=MMJLw3OXWiJVfVtrdJcBkbB8vA1IzSautluazZRuCQ0,5910
-ocrd/decorators/__init__.py,sha256=NWqZUTSBve6yIBgTN3miGmz4w1IFefLhG92qXx5ADtw,7620
+ocrd/decorators/__init__.py,sha256=PyXX7vxdWkRHixas9dWUtyO3YLczcly8ZEpfZDSMVp8,7639
 ocrd/decorators/loglevel_option.py,sha256=tgipROEu3t4hkwWvFssd80k2SbTBwBIC4WNE6Gc-XAg,798
 ocrd/decorators/mets_find_options.py,sha256=d4oATKMP6bFQHNqOK6nLqgUiWF2FYdkPvzkTVRMYpKo,635
 ocrd/decorators/ocrd_cli_options.py,sha256=hr2EugwAY_-GJ7F7g77Od9o9eAqhfLBHSpfmCql2OCU,2665
 ocrd/decorators/parameter_option.py,sha256=n8hYw7XVTd3i3tvpK8F1Jx_CqRp6EGF9qJVH95yj92Q,1076
 ocrd/processor/__init__.py,sha256=39ymNwYRdc-b_OJzzKmWCvo2ga3KdsGSYDHE1Hzkn_w,274
-ocrd/processor/base.py,sha256=8kFrYLd1cSHBaBolHjXdIVHwmV8muKgBCBrAYw7kWTQ,59154
-ocrd/processor/concurrent.py,sha256=IMMBFGDGqfpgm7Rp6J_dnXVckyBKntwARwcKDnxadHQ,38217
+ocrd/processor/base.py,sha256=rVTQeUpZR_rBDh44Q7Xtl5TGcEdyBhMTDfpAgx4eLPg,59295
 ocrd/processor/helpers.py,sha256=8ngrqAJ01BSoSJNsIoK_YfA8QdryM5y0MqaqA9f7ELM,10483
 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=iWiw_jJXOqwr7-hFjdkmTCCo1xGr6MLGOshx81PTu-8,3548
+ocrd/processor/builtin/dummy_processor.py,sha256=a-4kKJ1JeXQuBIyyN8w2R3s7ov-wAfyEdEz3nxrf0sU,3479
+ocrd/processor/builtin/filter_processor.py,sha256=nDnXjo2tDCjodURb8VlB4VJtRwWGx261jH7AG91edSk,4317
 ocrd/processor/builtin/dummy/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ocrd/processor/builtin/dummy/ocrd-tool.json,sha256=aTA2FZRsRsrkbTctkazFeRu4xsTF6yCdeY07cMzOyt4,677
-ocrd_modelfactory/__init__.py,sha256=0baYSJXrOCTCguHkE6hBeqpGNVUe3aZUocv64A-DMDk,4094
+ocrd/processor/builtin/dummy/ocrd-tool.json,sha256=VoI37paWiUyMkTN5Qqau8R1Clmw24-HcZu4wjy1Br9Y,2311
+ocrd_modelfactory/__init__.py,sha256=NyJT1uSvmeEwibRFOkh0AEoVnYfP0mzxU--pP23B-TQ,4404
 ocrd_models/__init__.py,sha256=A0aj0mOraNb-xfiUueACdoaqISnp0qH-F49nTJg2vCs,380
 ocrd_models/constants.py,sha256=fI6Qz4OPOm6UBLQ_P2dlpjcwB0XFJZ7AgxxKqgc75X0,2724
 ocrd_models/mets-empty.xml,sha256=dFixfbxSXrgjZx9BfdIKWHX-khNmp7dNYaFe2qQSwCY,1203
 ocrd_models/ocrd_agent.py,sha256=E9OtDhz9UfKb6ou2qvsuCL9NlO1V6zMb0s8nVq8dVos,5609
-ocrd_models/ocrd_exif.py,sha256=5BRLjvB6jg36V68i8jvVnT2SSNnpqLbhLsaMuP51Scw,4583
+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=sVIvvMeBT8eZnOfW0DTjQUNyu62-llz0v_Ga5Xo-tUM,5393
-ocrd_models/ocrd_page_generateds.py,sha256=wfx3vESMAi08rl6-16zNVJe4E3B6APIvL6RCr1roAzg,774092
+ocrd_models/ocrd_page.py,sha256=TTCnvpKGyZx1dqH8LnDiVVVPjU6emWGVLO_4o9rQHtw,6233
+ocrd_models/ocrd_page_generateds.py,sha256=vjXPob5UMxYxPI6un8SYTL673Nhph0E2WEZexBgbvY4,841794
 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
+ocrd_models/xpath_functions.py,sha256=AwR8tHf56-mmIksnw_GeOQ760sxNHqK92T7z9OfsEEs,1437
 ocrd_network/__init__.py,sha256=gMejC614J5PPGgXDKBiQS0jt-Jx8qOrLbWH7zt8x8Gs,374
-ocrd_network/client.py,sha256=j9PJ9QHI6qJ_qAxVlpTk5mLfq0Xemj_Br-0SX-09U5M,2834
-ocrd_network/client_utils.py,sha256=JDBzDOGY2G_bgS8fQ677CMHG9DWSvIuWTlEFzdsXr_Q,4527
+ocrd_network/client.py,sha256=rzqtamZ8krRRy-QTO-AeWH8Lr3HhRiQe2R1-Lovd40g,3020
+ocrd_network/client_utils.py,sha256=VVZMNBgGznh41exZ78S48X3DDwHcWTuOq-LNdxjRvak,5002
 ocrd_network/constants.py,sha256=IeNtcU6dqonDE3Zw83_61auhS8X2b8wsjAxYg1zvK-M,1902
 ocrd_network/database.py,sha256=fcft7vdRDoR7vmPL1xNYTIeOg5DwRPcggwYDYxLy5ik,10706
 ocrd_network/logging_utils.py,sha256=bO9TQqBXw9CIZEKp8tHXbeE2NuJWMiaQDHzS05b4ajo,2153
 ocrd_network/param_validators.py,sha256=Jl1VwiPPKJ50k-xEHLdvW-1QDOkJHCiMz4k9Ipqm-Uc,1489
 ocrd_network/process_helpers.py,sha256=KpkInXsa5bgrxvTOouyMJ0NgJhaz0J9Gjs5sZHBcH64,2373
-ocrd_network/processing_server.py,sha256=8EFQIcjiQ-AEc4goaggxa-hYWhJW9fMzR768ZIX8vZE,41139
-ocrd_network/processing_worker.py,sha256=VV_pC5ljtjPpMY89MElAMZ-YUJdKk7IPugxFfDy8viA,12396
-ocrd_network/processor_server.py,sha256=RVX0MSzzayDdqHg7Kga7L4vOsjmQcoRYObh6Naxz-U4,9858
-ocrd_network/server_cache.py,sha256=GKX91SkVahgAk0nGWT0S-0qWp7a9ubxncjwwLTKs7JY,12829
-ocrd_network/server_utils.py,sha256=NBSZvLD7o43fMtSkqR70ougW-IWCW9J71GHqZIBeD0Q,10982
-ocrd_network/tcp_to_uds_mets_proxy.py,sha256=pNEl5jH09M-zuUIx25nog7IZ-fkXabi3724kQ_3iMTs,2894
-ocrd_network/utils.py,sha256=Vm8EiBBSKv5fAlkcb6nB8-duFN633ZIRAcPKfk7iyZY,6366
+ocrd_network/processing_server.py,sha256=qBiYk4wgTLqhHvbmDWu_F626BfSfyvkoCD-i0ZwsBSE,42109
+ocrd_network/processing_worker.py,sha256=fhIvmDQAYOkHYtUs5IB8Jk2lOKUTIBk3DskAsFloijA,12591
+ocrd_network/processor_server.py,sha256=2CD9TlinXk6x1jFjP5VWOXgJe8lAQdxc9zjZuVy3EOw,9931
+ocrd_network/server_cache.py,sha256=LpvJ-_Lbaeo4M3t8rZDdm9DAErZr8lDlma6pYc0m7aQ,13149
+ocrd_network/server_utils.py,sha256=Uge5F2VagPAEpcyU_Qf8AiecObIGXE0ilD8DaK7bTdE,12222
+ocrd_network/tcp_to_uds_mets_proxy.py,sha256=yRW-O6ihd31gf7xqQBIBb_ZQQgqisMyOdRI216ehq_A,3160
+ocrd_network/utils.py,sha256=XzPXeSPCVjWLQM540PCpxfJ5hqjJ85_OQBjnf9HlDtE,6759
 ocrd_network/cli/__init__.py,sha256=F7YVqxw-9glz6-ghG0Kp5XXeV1-rL1emVSXLCWxdTF0,306
-ocrd_network/cli/client.py,sha256=4bpsDRwm3WVFHA6ITM-VKAFx-JQ6ldbBKFVYleWhiW8,7515
+ocrd_network/cli/client.py,sha256=XYWbeSiPK4BQXuyTq_FTOXEKljXVLkukWfx07aKbthY,8424
 ocrd_network/cli/processing_server.py,sha256=rAci6RsHlZ0c87GuLdfdCQCiGNcDEu4NEEQiwKJqVUo,796
 ocrd_network/cli/processing_worker.py,sha256=ZuaCkbKV_WKJV7cGOjZ6RLrjjppymnwNCiznFMlclAg,1897
 ocrd_network/cli/processor_server.py,sha256=Vto7UekFo_g83aHqwDmhge9bhPzk0b7O-L46dSfIpJc,1259
@@ -74,7 +75,7 @@ ocrd_network/models/ocrd_tool.py,sha256=WhxSwDyEXtF03Cu8u2tLZcYM0tCacL4PX1GveAxn
 ocrd_network/models/workflow.py,sha256=GL8q7RX9fGdXG3iVyJpCeLXbWa-2qI_SIxqhzxs9VK8,189
 ocrd_network/models/workspace.py,sha256=42G8RoS0wJ902LHyM09NIaqHoRdmP-Oib3d5I07sTBQ,1579
 ocrd_network/rabbitmq_utils/__init__.py,sha256=8MRawAiSpZ9IQUBcLAS7sYOzNC9sI11eZOLk7k92_ZQ,691
-ocrd_network/rabbitmq_utils/connector.py,sha256=tLYayGlGRrqLkQChhL5FyUyF0MQTi4w-nWKxz_IuX-0,11342
+ocrd_network/rabbitmq_utils/connector.py,sha256=N6mzjIf5FkVIno3FI1AksZY4F5jMUAm8baay0nXZx8w,11343
 ocrd_network/rabbitmq_utils/constants.py,sha256=Zu_dKJASfrgnIvEZZlFX9uDR9y6w7zy0KhW7gP7wHDE,1063
 ocrd_network/rabbitmq_utils/consumer.py,sha256=3WeryDmo0dSD9U0eLODbDElscvhEYjNeCBIewQHYfws,2488
 ocrd_network/rabbitmq_utils/helpers.py,sha256=y8FTC1ml_IBNcFo14GgCNtNRxYDotQn7U14HmTkv6h0,4874
@@ -83,12 +84,12 @@ ocrd_network/rabbitmq_utils/publisher.py,sha256=mw4XQQhRE1xUQVgEUseyG845iIgVO-9G
 ocrd_network/runtime_data/__init__.py,sha256=3jYkmT4mxMUcpbDaSw7Ld0KTedGEx_5vUQPDjwUyJZc,367
 ocrd_network/runtime_data/config_parser.py,sha256=Vr0FbsqmsoiuhDgZ7KFdeFZj9JvUulcOS2PCRFQQNHY,2364
 ocrd_network/runtime_data/connection_clients.py,sha256=DZyAvkNyMaIddGJs56s2pMP_fK-XWAtICxk1cjvkWYM,4207
-ocrd_network/runtime_data/deployer.py,sha256=W7SHn6XBC5fIaTJIMnHOMxEyp7dVqvHY4_cepaloMhQ,8315
+ocrd_network/runtime_data/deployer.py,sha256=LkDUG0uJf_V4SteiOM3EWwhKtdANCjmAOEAJJDshN30,9111
 ocrd_network/runtime_data/hosts.py,sha256=ml19ptzH4TFofyJR-Qp_Mn3sZUFbWoNe__rRXZSj_WE,12185
 ocrd_network/runtime_data/network_agents.py,sha256=5p_zKLqECBIHLw-Ya6eKcKSZcUM4ESiipEIphVxHBEA,5192
 ocrd_network/runtime_data/network_services.py,sha256=xrPpFUU_Pa-XzGe2FEt5RmO17xqykIUmTr_9g6S7XSs,7892
 ocrd_utils/__init__.py,sha256=U_zAQJwxg_aJ4CR84CKMNAUP6Cob8Er8Ikj42JmnUKo,5977
-ocrd_utils/config.py,sha256=Rkqv5wWEmlDDD0l1IWo9TPgn5ppPnHPRH9FfkMST29E,11117
+ocrd_utils/config.py,sha256=smjUAGK5n0iKZCs4dZAtrZurelcaavlFqhIUJWNMOi0,11796
 ocrd_utils/constants.py,sha256=ImbG1d8t2MW3uuFi-mN6aY90Zn74liAKZBKlfuKN86w,3278
 ocrd_utils/deprecate.py,sha256=4i50sZsA3Eevqn5D-SL5yGf9KEZfGCV4A5Anzn1GRMs,1026
 ocrd_utils/image.py,sha256=zNNX1cnRy6yvrxx8mnYQiqWraAh5-i4a1AOfCCg4SmI,24781
@@ -119,9 +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-3.0.0b7.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-ocrd-3.0.0b7.dist-info/METADATA,sha256=lc6oZVb9SPl97qvspPigal4yb_1DBBRZwE3GQUTAK_o,10417
-ocrd-3.0.0b7.dist-info/WHEEL,sha256=P9jw-gEje8ByB7_hXoICnHtVCrEwMQh-630tKvQWehc,91
-ocrd-3.0.0b7.dist-info/entry_points.txt,sha256=tV_gAdO8cbnOjS0GmKfJKbN60xBAV2DQRX6hEjleSjE,94
-ocrd-3.0.0b7.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.0.0b7.dist-info/RECORD,,
+ocrd-3.0.1.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.0.1.dist-info/METADATA,sha256=y6mdBSjKmkTMSJ7F1LAvjCWeagTBRAeiYaka9_Z2Djc,10442
+ocrd-3.0.1.dist-info/WHEEL,sha256=P9jw-gEje8ByB7_hXoICnHtVCrEwMQh-630tKvQWehc,91
+ocrd-3.0.1.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
+ocrd-3.0.1.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.0.1.dist-info/RECORD,,

{ocrd-3.0.0b7.dist-info → ocrd-3.0.1.dist-info}/entry_points.txt

@@ -1,3 +1,4 @@
 [console_scripts]
 ocrd = ocrd.cli:cli
 ocrd-dummy = ocrd.processor.builtin.dummy_processor:cli
+ocrd-filter = ocrd.processor.builtin.filter_processor:cli