ocrd 3.6.0__py3-none-any.whl → 3.7.0__py3-none-any.whl

ocrd/cli/__init__.py

@@ -32,12 +32,11 @@ from ..decorators import ocrd_loglevel
 from .ocrd_tool import ocrd_tool_cli
 from .workspace import workspace_cli
 from .process import process_cli
-from .bashlib import bashlib_cli
 from .validate import validate_cli
 from .resmgr import resmgr_cli
 from .zip import zip_cli
-from .log import log_cli
 from .network import network_cli
+from .bashlib import bashlib_cli
 
 
 __all__ = ['cli']
@@ -117,9 +116,8 @@ def cli(**kwargs):  # pylint: disable=unused-argument
 cli.add_command(ocrd_tool_cli)
 cli.add_command(workspace_cli)
 cli.add_command(process_cli)
-cli.add_command(bashlib_cli)
 cli.add_command(zip_cli)
 cli.add_command(validate_cli)
-cli.add_command(log_cli)
 cli.add_command(resmgr_cli)
 cli.add_command(network_cli)
+cli.add_command(bashlib_cli)

ocrd/cli/bashlib.py

@@ -6,24 +6,14 @@ OCR-D CLI: bash library
     :nested: full
 
 """
-from __future__ import print_function
-import sys
-import click
 
+# WARNING: bashlib processors have been deprecated as of v3 of the OCR-D/core API
+#          and will be removed in v3.7.0. We retain the `ocrd bashlib` CLI only
+#          to not break the `ocrd bashlib filename` command, which is used in CD
+#          scripts to get the `share` directory of the core installation. 
+
+import click
 from ocrd.constants import BASHLIB_FILENAME
-import ocrd.constants
-import ocrd_utils.constants
-from ocrd_utils.constants import DEFAULT_METS_BASENAME
-import ocrd_models.constants
-import ocrd_validators.constants
-from ocrd.decorators import (
-    parameter_option,
-    parameter_override_option,
-    ocrd_loglevel,
-    ocrd_cli_wrap_processor
-)
-from ocrd_utils import make_file_id
-from ocrd.processor import Processor
 
 # ----------------------------------------------------------------------
 # ocrd bashlib
@@ -50,104 +40,3 @@ def bashlib_filename():
     """
     print(BASHLIB_FILENAME)
 
-
-@bashlib_cli.command('constants')
-@click.argument('name')
-def bashlib_constants(name):
-    """
-    Query constants from ocrd_utils and ocrd_models
-    """
-    all_constants = {}
-    for src in [ocrd.constants, ocrd_utils.constants, ocrd_models.constants, ocrd_validators.constants]:
-        for k in src.__all__:
-            all_constants[k] = src.__dict__[k]
-    if name in ['*', 'KEYS', '__all__']:
-        print(sorted(all_constants.keys()))
-        sys.exit(0)
-    if name not in all_constants:
-        print("ERROR: name '%s' is not a known constant" % name, file=sys.stderr)
-        sys.exit(1)
-    val = all_constants[name]
-    if isinstance(val, dict):
-        # make this bash-friendly (show initialization for associative array)
-        for key in val:
-            print("[%s]=%s" % (key, val[key]), end=' ')
-    else:
-        print(val)
-
-
-@bashlib_cli.command('input-files')
-@click.option('--ocrd-tool', help="path to ocrd-tool.json of processor to feed", default=None)
-@click.option('--executable', help="name of processor executable in ocrd-tool.json", default=None)
-@click.option('-m', '--mets', help="METS to process", default=DEFAULT_METS_BASENAME)
-@click.option('-U', '--mets-server-url', help='TCP host URI or UDS path of METS server', default=None)
-@click.option('-d', '--working-dir', help="Working Directory")
-@click.option('-I', '--input-file-grp', help='File group(s) used as input.', default=None)
-@click.option('-O', '--output-file-grp', help='File group(s) used as output.', default=None)
-@click.option('-g', '--page-id', help="ID(s) of the pages to process")
-@click.option('--overwrite', is_flag=True, default=False, help="Remove output pages/images if they already exist\n"
-              "(with '--page-id', remove only those).\n"
-              "Short-hand for OCRD_EXISTING_OUTPUT=OVERWRITE")
-@click.option('--debug', is_flag=True, default=False, help="Abort on any errors with full stack trace.\n"
-              "Short-hand for OCRD_MISSING_OUTPUT=ABORT")
-@parameter_option
-@parameter_override_option
-@ocrd_loglevel
-def bashlib_input_files(ocrd_tool, executable, **kwargs):
-    """
-    List input files for processing
-
-    Instantiate a processor and workspace from the given processing options.
-    Then loop through the input files of the input fileGrp, and for each one,
-    print its `url`, `ID`, `mimetype` and `pageId`, as well as its recommended
-    `outputFileId` (from ``make_file_id``).
-
-    (The printing format is one associative array initializer per line.)
-    """
-    class BashlibProcessor(Processor):
-        # go half way of the normal run_processor / process_workspace call tree
-        # by just delegating to process_workspace, overriding process_page_file
-        # to ensure all input files exist locally (without persisting them in the METS)
-        # and print what needs to be acted on in bash-friendly way
-        def process_page_file(self, *input_files):
-            for field in ['url', 'local_filename', 'ID', 'mimetype', 'pageId']:
-                # make this bash-friendly (show initialization for associative array)
-                if len(input_files) > 1:
-                    # single quotes allow us to preserve the list value inside the alist
-                    value = ' '.join(str(getattr(res, field)) for res in input_files)
-                else:
-                    value = str(getattr(input_files[0], field))
-                print(f"[{field}]='{value}'", end=' ')
-            output_file_id = make_file_id(input_files[0], kwargs['output_file_grp'])
-            print(f"[outputFileId]='{output_file_id}'")
-    if ocrd_tool and executable:
-        class FullBashlibProcessor(BashlibProcessor):
-            @property
-            def metadata_location(self):
-                # needed for metadata loading and validation mechanism
-                return ocrd_tool
-
-            @property
-            def executable(self):
-                # needed for ocrd_tool lookup
-                return executable
-        processor_class = FullBashlibProcessor
-    else:
-        # we have no true metadata file, so fill in just to make it work
-        class UnknownBashlibProcessor(BashlibProcessor):
-            @property
-            def ocrd_tool(self):
-                # needed to satisfy the validator
-                return {'executable': '',
-                        # required now
-                        'input_file_grp_cardinality': 1,
-                        'output_file_grp_cardinality': 1,
-                        'steps': ['']}
-
-            @property
-            def version(self):
-                # needed to satisfy the validator and wrapper
-                return '1.0'
-        processor_class = UnknownBashlibProcessor
-
-    ocrd_cli_wrap_processor(processor_class, **kwargs)

ocrd/mets_server.py

@@ -258,12 +258,12 @@ class ClientSideOcrdMets:
 
     def add_agent(self, **kwargs):
         if not self.multiplexing_mode:
-            return self.session.request("POST", f"{self.url}/agent", json=OcrdAgentModel.create(**kwargs).dict())
+            return self.session.request("POST", f"{self.url}/agent", json=OcrdAgentModel.create(**kwargs).model_dump())
         else:
             self.session.request(
                 "POST",
                 self.url,
-                json=MpxReq.add_agent(self.ws_dir_path, OcrdAgentModel.create(**kwargs).dict())
+                json=MpxReq.add_agent(self.ws_dir_path, OcrdAgentModel.create(**kwargs).model_dump())
             ).json()
             return OcrdAgentModel.create(**kwargs)
 
@@ -305,7 +305,7 @@ class ClientSideOcrdMets:
             mimetype=mimetype, url=url, local_filename=local_filename
         )
         # add force+ignore
-        kwargs = {**kwargs, **data.dict()}
+        kwargs = {**kwargs, **data.model_dump()}
 
         if not self.multiplexing_mode:
             r = self.session.request("POST", f"{self.url}/file", data=kwargs)
@@ -530,7 +530,7 @@ class OcrdMetsServer:
 
         @app.post(path='/agent', response_model=OcrdAgentModel)
         async def add_agent(agent: OcrdAgentModel):
-            kwargs = agent.dict()
+            kwargs = agent.model_dump()
             kwargs['_type'] = kwargs.pop('type')
             workspace.mets.add_agent(**kwargs)
             response = agent
@@ -575,7 +575,7 @@ class OcrdMetsServer:
                 local_filename=local_filename
             )
             # Add to workspace
-            kwargs = file_resource.dict()
+            kwargs = file_resource.model_dump()
             workspace.add_file(**kwargs, force=force)
             response = file_resource
             self.log.debug(f"POST /file -> {response.__dict__}")

{ocrd-3.6.0.dist-info → ocrd-3.7.0.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.6.0
+Version: 3.7.0
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0
@@ -35,7 +35,7 @@ Requires-Dist: opencv-python-headless
 Requires-Dist: paramiko
 Requires-Dist: pika>=1.2.0
 Requires-Dist: Pillow>=7.2.0
-Requires-Dist: pydantic==1.*
+Requires-Dist: pydantic>=2.0.0
 Requires-Dist: python-magic
 Requires-Dist: python-multipart
 Requires-Dist: pyyaml
@@ -220,12 +220,6 @@ Also contains the command line tool `ocrd`.
 
 See [README for `ocrd`](./README_ocrd.md) for further information.
 
-## bash library
-
-Builds a bash script that can be sourced by other bash scripts to create OCRD-compliant CLI.
-
-See [README for `bashlib`](./README_bashlib.md) for further information.
-
 ## Testing
 
 Download assets (`make assets`)

{ocrd-3.6.0.dist-info → ocrd-3.7.0.dist-info}/RECORD

@@ -1,7 +1,6 @@
 ocrd/__init__.py,sha256=ZswMVmlqFhAEIzMR3my6IKPq9XLH21aDPC_m_8Jh4dA,1076
 ocrd/constants.py,sha256=6dn3mG54WqHsKInmLZp4kJjNqqPtBoFoSuLUuRbOps0,740
-ocrd/lib.bash,sha256=Eu-_Eb3uUNEoa3GmSDyUm_4FmmSJctglPOWz8EcBJ2c,10395
-ocrd/mets_server.py,sha256=eXIbSip6gYi5RyJZlriIrR9lUGpJjL0kXq3UDZPeqVs,22274
+ocrd/mets_server.py,sha256=LbZ0U2_o0W7cWO639U7E816dXabro8-8yHGX0quvHn4,22304
 ocrd/ocrd-all-tool.json,sha256=EYXmMzP68p3KzL8nUZ16TCX2chQzKkAeISvuXqI_yIw,2094
 ocrd/resolver.py,sha256=7uwHRxaK8YMdKHe_a2dfrcNwL6UhQRJRVBrIX7GST7Q,15443
 ocrd/resource_list.yml,sha256=82-PiqkZnka1kTj3MQqNn4wXWKHHtoFchsQuetWuqFs,2633
@@ -10,9 +9,8 @@ ocrd/task_sequence.py,sha256=r4e4iaP9AXzTL2xQZpfYnHuFXty5pE-ym3gIyUz1aJc,7180
 ocrd/workspace.py,sha256=UL_gX0KA-MmpayBl9KGYTfcl-1Canj8S991G9RHhu70,65216
 ocrd/workspace_backup.py,sha256=aUOnYeJ-nWu-Zve27B0cYd9ZtBkmQX4F4Wim2UcrR5I,3624
 ocrd/workspace_bagger.py,sha256=4viSQoWteW0V4B_blB6asJXd4-qniGGJyCPfKnrsyrY,12054
-ocrd/cli/__init__.py,sha256=klgd85WNdLzqivbkEbt5cMOFvEoMoEE2KnT_oKAwTBs,2888
-ocrd/cli/bashlib.py,sha256=RLp6ejgbxLjGfNmeniVTuzLtrGeJrGJAZYc9JPT7U68,5998
-ocrd/cli/log.py,sha256=jRgxdoJS14OPXI9LpovPqB3JcxlcGZH7QOde6-TmykQ,1566
+ocrd/cli/__init__.py,sha256=-n2jpGBZs_OMpI31E7CljGVdoFxDhgCAYwibcl_vp1Q,2838
+ocrd/cli/bashlib.py,sha256=sEpTKbqM5DEo6838Ki5aFU8QsokA2SqQ841gcBu7M5M,1148
 ocrd/cli/network.py,sha256=iQ0AhQRGvIFyJY9RBArUiA_wuz7IfNKvU4L8KpVggnY,530
 ocrd/cli/ocrd_tool.py,sha256=kB3Y3tj7Fpz6Ts4KgVlznhXpAx8gCDvJTnO39j8SGL4,7679
 ocrd/cli/process.py,sha256=yfhBSYmuY5k2AccKwiNvG9hCDx1coYyWjq9BBwYaL3Y,1234
@@ -67,10 +65,10 @@ ocrd_network/cli/client.py,sha256=H5fiJhBqbFn4_B2p3V20GejGTIYO-mNglh3y5nzUGhs,10
 ocrd_network/cli/processing_server.py,sha256=NsuI0f9h4KDwe39YugmHo5cJ_29chcLLQ7DThKfPO7s,770
 ocrd_network/cli/processing_worker.py,sha256=ZuaCkbKV_WKJV7cGOjZ6RLrjjppymnwNCiznFMlclAg,1897
 ocrd_network/models/__init__.py,sha256=eVYMZaktzlyHKx-zI7GLYyRlZd3Vi_lNgsqSSFwqb6U,475
-ocrd_network/models/job.py,sha256=6NxcNIUHMS9Ft5UGWegngB6uNUOnE_6nOnRGKFbjI6A,4243
-ocrd_network/models/messages.py,sha256=XnyLMX77NchgmtKJRtqtBFsk_sCR4OGEuWm_d3uDkj8,657
+ocrd_network/models/job.py,sha256=9bwp8DFoRH96WnRpkDV3XRfXCBiupzK6WXjqPsTcvLg,4440
+ocrd_network/models/messages.py,sha256=OUDTjUiaATStsSAHCEDilUhBSruPsjpBtIBsllqN2Z0,672
 ocrd_network/models/workflow.py,sha256=GL8q7RX9fGdXG3iVyJpCeLXbWa-2qI_SIxqhzxs9VK8,189
-ocrd_network/models/workspace.py,sha256=7kjCTY0ixqcyJP9eHnptkrJnPTCD3zFNfhApZz9w6OU,1568
+ocrd_network/models/workspace.py,sha256=rZcBWNlQOZX2KukP79IDRrXJvZ-H5pPH3WpPuw72HBM,1596
 ocrd_network/rabbitmq_utils/__init__.py,sha256=XLIqZhfin4I4m80G9B__UcP45Lz10_mEpMYLXGOByUk,741
 ocrd_network/rabbitmq_utils/connector.py,sha256=N6mzjIf5FkVIno3FI1AksZY4F5jMUAm8baay0nXZx8w,11343
 ocrd_network/rabbitmq_utils/constants.py,sha256=Zu_dKJASfrgnIvEZZlFX9uDR9y6w7zy0KhW7gP7wHDE,1063
@@ -91,9 +89,9 @@ ocrd_utils/constants.py,sha256=ImbG1d8t2MW3uuFi-mN6aY90Zn74liAKZBKlfuKN86w,3278
 ocrd_utils/deprecate.py,sha256=luAqGWUSF-9DHmTd2lDiQoQPA5SrJazdoDPQYQ6A7Z4,1029
 ocrd_utils/image.py,sha256=tG5WnNtrrvGjm2-r6NVs1Jm7z8fee3MuLKotAD6C2RU,24818
 ocrd_utils/introspect.py,sha256=LPhgcUuoicQcURDCWlCpSdbfVyxID5vmQPXJ9vzuYV0,1977
-ocrd_utils/logging.py,sha256=Kj_z92pXbzWmc0jMJ299Pup9hfcnzJH8ltvI9w7STZc,7824
+ocrd_utils/logging.py,sha256=-cCi_9kIzmLUixfnDcx2jq9IQuwMqrU-71RJhKOQilQ,7929
 ocrd_utils/ocrd_logging.conf,sha256=JlWmA_5vg6HnjPGjTC4mA5vFHqmnEinwllSTiOw5CCo,3473
-ocrd_utils/os.py,sha256=GstXB4i3kDBy7PXU-TaPYV4BI-lkqd_cYKl6uDkVMkw,9829
+ocrd_utils/os.py,sha256=QEOramsUmBDzZxslPMZhfTviPr7EnopXKEgNc5zwwTs,9817
 ocrd_utils/str.py,sha256=4P0MdX0LCTqDTnsi_y5wNOBXW_TuTFANF7NYRXjo4x0,10136
 ocrd_validators/__init__.py,sha256=ZFc-UqRVBk9o1YesZFmr9lOepttNJ_NKx1Zdb7g_YsU,972
 ocrd_validators/bagit-profile.yml,sha256=sdQJlSi7TOn1E9WYMOZ1shewJ-i_nPaKmsAFkh28TGY,1011
@@ -117,9 +115,9 @@ ocrd_validators/xlink.xsd,sha256=8fW7YAMWXN2PbB_MMvj9H5ZeFoEBDzuYBtlGC8_6ijw,318
 ocrd_validators/xsd_mets_validator.py,sha256=YgiuNtwNDtn3LuvdFFscnmsGREF_wQ4wtA76yE2Iljw,469
 ocrd_validators/xsd_page_validator.py,sha256=ggt-nmaz-DDyAPwm3ZMVvtChuV2BJ2ZEEbWpePL9vTk,469
 ocrd_validators/xsd_validator.py,sha256=ahJo_oVvTK_JB0Cu4CkMC8l_gbzsyW91AxGtelMjqrg,2115
-ocrd-3.6.0.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-ocrd-3.6.0.dist-info/METADATA,sha256=Qc1Nap_yo-Y3_8FYaVIvGXV79q4NKrCZs6h4BEKXIg4,10523
-ocrd-3.6.0.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
-ocrd-3.6.0.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
-ocrd-3.6.0.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.6.0.dist-info/RECORD,,
+ocrd-3.7.0.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.7.0.dist-info/METADATA,sha256=yCPxM32p95ODwi46-eVJxTjxCheoiuxOZujpKYmEzIA,10340
+ocrd-3.7.0.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
+ocrd-3.7.0.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
+ocrd-3.7.0.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.7.0.dist-info/RECORD,,

ocrd_network/models/job.py

@@ -13,7 +13,7 @@ class PYJobInput(BaseModel):
     workspace_id: Optional[str] = None
     description: Optional[str] = None
     input_file_grps: List[str]
-    output_file_grps: Optional[List[str]]
+    output_file_grps: Optional[List[str]] = None
     page_id: Optional[str] = None
     parameters: dict = {}  # Always set to empty dict when None, otherwise it fails ocr-d-validation
     result_queue_name: Optional[str] = None
@@ -23,8 +23,8 @@ class PYJobInput(BaseModel):
     # If set, specifies a list of job ids this job depends on
     depends_on: Optional[List[str]] = None
 
-    class Config:
-        schema_extra = {
+    model_config = {
+        'json_schema_extra': {
             'example': {
                 'path_to_mets': '/path/to/mets.xml',
                 'description': 'The description of this execution',
@@ -34,6 +34,7 @@ class PYJobInput(BaseModel):
                 'parameters': {}
             }
         }
+    }
 
 
 class PYJobOutput(BaseModel):
@@ -42,12 +43,12 @@ class PYJobOutput(BaseModel):
     job_id: str
     processor_name: str
     state: JobState = JobState.unset
-    path_to_mets: Optional[str]
-    workspace_id: Optional[str]
+    path_to_mets: Optional[str] = None
+    workspace_id: Optional[str] = None
     input_file_grps: List[str]
-    output_file_grps: Optional[List[str]]
+    output_file_grps: Optional[List[str]] = None
     page_id: Optional[str] = None
-    log_file_path: Optional[str]
+    log_file_path: Optional[str] = None
 
 
 class DBProcessorJob(Document):
@@ -55,22 +56,22 @@ class DBProcessorJob(Document):
     """
     job_id: str
     processor_name: str
-    path_to_mets: Optional[str]
-    workspace_id: Optional[str]
-    description: Optional[str]
+    path_to_mets: Optional[str] = None
+    workspace_id: Optional[str] = None
+    description: Optional[str] = None
     state: JobState = JobState.unset
     input_file_grps: List[str]
-    output_file_grps: Optional[List[str]]
-    page_id: Optional[str]
-    parameters: Optional[dict]
-    depends_on: Optional[List[str]]
-    result_queue_name: Optional[str]
-    callback_url: Optional[str]
-    internal_callback_url: Optional[str]
-    start_time: Optional[datetime]
-    end_time: Optional[datetime]
-    exec_time: Optional[str]
-    log_file_path: Optional[str]
+    output_file_grps: Optional[List[str]] = None
+    page_id: Optional[str] = None
+    parameters: Optional[dict] = None
+    depends_on: Optional[List[str]] = None
+    result_queue_name: Optional[str] = None
+    callback_url: Optional[str] = None
+    internal_callback_url: Optional[str] = None
+    start_time: Optional[datetime] = None
+    end_time: Optional[datetime] = None
+    exec_time: Optional[str] = None
+    log_file_path: Optional[str] = None
 
     class Settings:
         use_enum_values = True
@@ -99,9 +100,9 @@ class PYWorkflowJobOutput(BaseModel):
     page_id: str
     page_wise: bool = False
     job_id: str
-    path_to_mets: Optional[str]
-    workspace_id: Optional[str]
-    description: Optional[str]
+    path_to_mets: Optional[str] = None
+    workspace_id: Optional[str] = None
+    description: Optional[str] = None
 
 
 class DBWorkflowJob(Document):
@@ -114,10 +115,10 @@ class DBWorkflowJob(Document):
     # key: page_id
     # value: List of and processing job ids sorted in dependency order
     processing_job_ids: Dict
-    path_to_mets: Optional[str]
-    workspace_id: Optional[str]
-    description: Optional[str]
-    workflow_callback_url: Optional[str]
+    path_to_mets: Optional[str] = None
+    workspace_id: Optional[str] = None
+    description: Optional[str] = None
+    workflow_callback_url: Optional[str] = None
 
     class Settings:
         use_enum_values = True

ocrd_network/models/messages.py

@@ -11,8 +11,8 @@ class PYResultMessage(BaseModel):
     path_to_mets: Optional[str] = None
     workspace_id: Optional[str] = None
 
-    class Config:
-        schema_extra = {
+    model_config = {
+        "json_schema_extra": {
             "example": {
                 "job_id": "d8e36726-ed28-5476-b83c-bc31d2eecf1f",
                 "state": JobState.success,
@@ -20,3 +20,4 @@ class PYResultMessage(BaseModel):
                 "workspace_id": "c7f25615-fc17-4365-a74d-ad20e1ddbd0e"
             }
         }
+    }

ocrd_network/models/workspace.py

@@ -25,10 +25,10 @@ class DBWorkspace(Document):
     workspace_mets_path: str
     ocrd_identifier: str
     bagit_profile_identifier: str
-    ocrd_base_version_checksum: Optional[str]
-    ocrd_mets: Optional[str]
-    bag_info_adds: Optional[dict]
-    mets_server_url: Optional[str]
+    ocrd_base_version_checksum: Optional[str] = None
+    ocrd_mets: Optional[str] = None
+    bag_info_adds: Optional[dict] = None
+    mets_server_url: Optional[str] = None
     deleted: bool = False
 
     class Settings:

ocrd_utils/logging.py

@@ -75,6 +75,9 @@ _ocrdLevel2pythonLevel = {
 
 
 def tf_disable_interactive_logs():
+    """
+    Disable the interactive logging of tf/keras and set the log level to error or higher
+    """
     try:
         from os import environ  # pylint: disable=import-outside-toplevel
         # This env variable must be set before importing from Keras

ocrd_utils/os.py

@@ -171,7 +171,7 @@ def list_all_resources(executable, moduled=None, xdg_data_home=None):
                    # files over directories; but we want data files only
                    # todo: more code and cache exclusion patterns!
                    ['*.py', '*.py[cod]', '*~', 'ocrd-tool.json',
-                    'environment.pickle', 'resource_list.yml', 'lib.bash']):
+                    'environment.pickle', 'resource_list.yml']):
                 continue
             candidates.append(resource)
     # recurse once

ocrd/cli/log.py

@@ -1,56 +0,0 @@
-"""
-OCR-D CLI: Logging
-
-.. click:: ocrd.cli.log:log_cli
-    :prog: ocrd log
-    :nested: full
-"""
-import click
-from ocrd_utils import initLogging, getLogger, getLevelName
-
-
-class LogCtx():
-
-    def __init__(self, name):
-        self.name = name
-
-    def log(self, lvl, *args, **kwargs):
-        logger = getLogger(self.name)
-        logger.log(getLevelName(lvl), *args, **kwargs)
-
-
-pass_log = click.make_pass_decorator(LogCtx)
-
-
-@click.group("log")
-@click.option('-n', '--name', envvar='OCRD_TOOL_NAME', default='log_cli', metavar='LOGGER_NAME',
-              help='Name of the logger', show_default=True)
-@click.pass_context
-def log_cli(ctx, name, *args, **kwargs):
-    """
-    Logging
-
-    Logger name will be ocrd.OCRD_TOOL_NAME where OCRD_TOOL_NAME is normally
-    (when using bashlib) the name of the processor.
-    """
-    initLogging()
-    ctx.obj = LogCtx('ocrd.' + name)
-
-
-def _bind_log_command(lvl):
-    @click.argument('msgs', nargs=-1)
-    @pass_log
-    def _log_wrapper(ctx, msgs):
-        if not msgs:
-            ctx.log(lvl.upper(), '')
-        elif len(msgs) == 1 and msgs[0] == '-':
-            for stdin_line in click.get_text_stream('stdin'):
-                ctx.log(lvl.upper(), stdin_line.rstrip('\n'))
-        else:
-            msg = list(msgs) if '%s' in msgs[0] else ' '.join([x.replace('%', '%%') for x in msgs])
-            ctx.log(lvl.upper(), msg)
-    return _log_wrapper
-
-
-for _lvl in ['trace', 'debug', 'info', 'warning', 'error', 'critical']:
-    log_cli.command(_lvl, help="Log a %s message" % _lvl.upper())(_bind_log_command(_lvl))

ocrd/lib.bash

@@ -1,310 +0,0 @@
-((BASH_VERSINFO<4 || BASH_VERSINFO==4 && BASH_VERSINFO[1]<4)) && \
-    echo >&2 "bash $BASH_VERSION is too old. Please install bash 4.4 or newer." && \
-    exit 1
-
-## ### `ocrd__raise`
-##
-## Raise an error and exit.
-ocrd__raise () {
-    echo >&2 "ERROR: $1"; exit 127
-}
-
-## ### `ocrd__log`
-##
-## Delegate logging to `ocrd log`
-ocrd__log () {
-    local log_level="${ocrd__argv[log_level]:-}"
-    if [[ -n "$log_level" ]];then
-        ocrd -l "$log_level" log "$@"
-    else
-        ocrd log "$@"
-    fi
-}
-
-
-## ### `ocrd__minversion`
-##
-## Ensure minimum version
-# ht https://stackoverflow.com/posts/4025065
-ocrd__minversion () {
-    set -e
-    local minversion_raw="$1"
-    local version_raw=$(ocrd --version|sed 's/ocrd, version //')
-    local version_mmp=$(echo "$version_raw" | grep -Eo '([0-9]+\.?){3}')
-    local version_prerelease_suffix="${version_raw#$version_mmp}"
-    if [[ -z $version_prerelease_suffix ]];then
-      version_prerelease_suffix=0
-    fi
-    local minversion_mmp=$(echo "$minversion_raw" | grep -Eo '([0-9]+\.?){3}')
-    local minversion_prerelease_suffix="${minversion_raw#$minversion_mmp}"
-    if [[ -z $minversion_prerelease_suffix ]];then
-      minversion_prerelease_suffix=0
-    fi
-    local IFS='.'
-    version=($version_mmp)
-    minversion=($minversion_mmp)
-    # MAJOR > MAJOR
-    if (( ${version[0]} > ${minversion[0]} ));then
-        return
-    # MAJOR == MAJOR
-    elif (( ${version[0]} == ${minversion[0]} ));then
-        # MINOR > MINOR
-        if (( ${version[1]} > ${minversion[1]} ));then
-            return
-        # MINOR == MINOR
-        elif (( ${version[1]} == ${minversion[1]} ));then
-            # PATCH > PATCH
-            if (( ${version[2]} > ${minversion[2]} ));then
-                return
-            elif (( ${version[2]} == ${minversion[2]}));then
-              # Match prerelease suffix like a1, b1 alphabetically
-              if [ "$version_prerelease_suffix" = "$minversion_prerelease_suffix" -o "$version_prerelease_suffix" \> "$minversion_prerelease_suffix" ]; then
-                return
-              fi
-            fi
-        fi
-    fi
-    ocrd__raise "ocrd/core is too old ($version_raw < $minversion_raw). Please update OCR-D/core"
-}
-
-## ### `ocrd__dumpjson`
-##
-## Output ocrd-tool.json.
-##
-## Requires `$OCRD_TOOL_JSON` and `$OCRD_TOOL_NAME` to be set:
-##
-## ```sh
-## export OCRD_TOOL_JSON=/path/to/ocrd-tool.json
-## export OCRD_TOOL_NAME=ocrd-foo-bar
-## ```
-##
-ocrd__dumpjson () {
-    ocrd ocrd-tool "$OCRD_TOOL_JSON" tool "$OCRD_TOOL_NAME" dump
-}
-
-##
-## Output file resource path.
-##
-ocrd__resolve_resource () {
-    ocrd ocrd-tool "$OCRD_TOOL_JSON" tool "$OCRD_TOOL_NAME" resolve-resource "$1"
-}
-
-##
-## Output file resource content.
-##
-ocrd__show_resource () {
-    ocrd ocrd-tool "$OCRD_TOOL_JSON" tool "$OCRD_TOOL_NAME" show-resource "$1"
-}
-
-##
-## Output file resources names.
-##
-ocrd__list_resources () {
-    ocrd ocrd-tool "$OCRD_TOOL_JSON" tool "$OCRD_TOOL_NAME" list-resources
-}
-
-## ### `ocrd__usage`
-##
-## Print usage
-##
-ocrd__usage () {
-    declare -a _args=(ocrd-tool "$OCRD_TOOL_JSON" tool "$OCRD_TOOL_NAME" help)
-    if [ -v ocrd__subcommand ];then
-        _args+=($ocrd__subcommand)
-    fi
-    ocrd ${_args[@]}
-}
-
-## ### `ocrd__parse_argv`
-##
-## Expects an associative array ("hash"/"dict") `ocrd__argv` to be defined:
-##
-## ```sh
-## declare -A ocrd__argv=()
-## ```
-ocrd__parse_argv () {
-    set -e
-
-    # if [[ -n "$ZSH_VERSION" ]];then
-    #     print -r -- ${+ocrd__argv} ${(t)ocrd__argv}
-    # fi
-    if ! declare -p "ocrd__argv" >/dev/null 2>/dev/null ;then
-        ocrd__raise "Must set \$ocrd__argv (declare -A ocrd__argv)"
-    fi
-
-    if ! declare -p "params" >/dev/null 2>/dev/null ;then
-        ocrd__raise "Must set \$params (declare -A params)"
-    fi
-
-    if ! declare -p "params_json" >/dev/null 2>/dev/null ;then
-        ocrd__raise "Must set \$params_json (declare params_json)"
-    fi
-
-    if [[ $# = 0 ]];then
-        ocrd__usage
-        exit 1
-    fi
-
-    ocrd__argv[debug]=false
-    ocrd__argv[overwrite]=false
-    ocrd__argv[profile]=false
-    ocrd__argv[profile_file]=
-    ocrd__argv[mets_server_url]=
-    ocrd__argv[mets_file]="$PWD/mets.xml"
-
-    local __parameters=()
-    local __parameter_overrides=()
-
-    if [[ $1 == 'worker' || $1 == 'server' ]];then
-         ocrd__subcommand="$1" ; shift ;
-    fi
-
-    while [[ "${1:-}" = -* ]];do
-        case "$1" in
-            -l|--log-level) ocrd__argv[log_level]=$2 ; shift ;;
-            --log-filename) exec 2> "$2" ; shift ;;
-            -h|--help|--usage) ocrd__usage; exit ;;
-            -J|--dump-json) ocrd__dumpjson; exit ;;
-            -D|--dump-module-dir) echo $(dirname "$OCRD_TOOL_JSON"); exit ;;
-            -C|--show-resource) ocrd__show_resource "$2"; exit ;;
-            -L|--list-resources) ocrd__list_resources; exit ;;
-            -p|--parameter)  __parameters+=(-p "$(ocrd__resolve_resource "$2" 2>/dev/null || echo "$2")") ; shift ;;
-            -P|--parameter-override) __parameter_overrides+=(-P "$2" "$3") ; shift ; shift ;;
-            -g|--page-id) ocrd__argv[page_id]=$2 ; shift ;;
-            -O|--output-file-grp) ocrd__argv[output_file_grp]=$2 ; shift ;;
-            -I|--input-file-grp) ocrd__argv[input_file_grp]=$2 ; shift ;;
-            -w|--working-dir) ocrd__argv[working_dir]=$(realpath "$2") ; shift ;;
-            -m|--mets) ocrd__argv[mets_file]=$(realpath "$2") ; shift ;;
-            -U|--mets-server-url) ocrd__argv[mets_server_url]="$2" ; shift ;;
-            --debug) ocrd__argv[debug]=true ;;
-            --overwrite) ocrd__argv[overwrite]=true ;;
-            --profile) ocrd__argv[profile]=true ;;
-            --profile-file) ocrd__argv[profile_file]=$(realpath "$2") ; shift ;;
-            -V|--version) ocrd ocrd-tool "$OCRD_TOOL_JSON" version; exit ;;
-            --queue) ocrd__worker_queue="$2" ; shift ;;
-            --database) ocrd__worker_database="$2" ; shift ;;
-            *) ocrd__raise "Unknown option '$1'" ;;
-        esac
-        shift
-    done
-
-    if [ -v ocrd__worker_queue -o -v ocrd__worker_database -o -v ocrd__subcommand ]; then
-        if ! [ -v ocrd__subcommand ] ; then
-            ocrd__raise "Provide subcommand 'worker' for Processing Worker"
-        elif ! [ -v ocrd__worker_database ]; then
-            ocrd__raise "For the Processing Worker --database is required"
-        elif ! [ -v ocrd__worker_queue ]; then
-            ocrd__raise "For the Processing Worker --queue is required"
-        fi
-        if [ ${ocrd__subcommand} = "worker" ]; then
-            ocrd network processing-worker $OCRD_TOOL_NAME --queue "${ocrd__worker_queue}" --database "${ocrd__worker_database}"
-        else
-            ocrd__raise "subcommand must be 'worker' not '${ocrd__subcommand}'"
-        fi
-        exit
-    fi
-
-    if [[ ! -e "${ocrd__argv[mets_file]}" ]]; then
-        ocrd__raise "METS file '${ocrd__argv[mets_file]}' not found"
-    fi
-
-    if [[ ! -d "${ocrd__argv[working_dir]:=$(dirname "${ocrd__argv[mets_file]}")}" ]]; then
-        ocrd__raise "workdir '${ocrd__argv[working_dir]}' not a directory. Use -w/--working-dir to set correctly"
-    fi
-
-    if [[ ! "${ocrd__argv[log_level]:=INFO}" =~ OFF|ERROR|WARN|INFO|DEBUG|TRACE ]]; then
-        ocrd__raise "log level '${ocrd__argv[log_level]}' is invalid"
-    fi
-
-    if [[ -z "${ocrd__argv[input_file_grp]:=}" ]]; then
-        ocrd__raise "Provide --input-file-grp/-I explicitly!"
-    fi
-
-    if [[ -z "${ocrd__argv[output_file_grp]:=}" ]]; then
-        ocrd__raise "Provide --output-file-grp/-O explicitly!"
-    fi
-
-    # enable profiling (to be extended/acted upon by caller)
-    if [[ ${ocrd__argv[profile]} = true ]]; then
-        if [[ -n "${ocrd__argv[profile_file]}" ]]; then
-            exec 3> "${ocrd__argv[profile_file]}"
-        else
-            exec 3>&2
-        fi
-        BASH_XTRACEFD=3
-        # just the builtin tracer (without timing):
-        #set -x
-        # our own (including timing):
-        DEPTH=+++++++++++
-        shopt -s extdebug
-        showtime() { date "+${DEPTH:0:$BASH_SUBSHELL+1} %H:%M:%S $BASH_COMMAND" >&3; }
-        declare +t showtime # no trace here
-        trap showtime DEBUG
-    fi
-
-    # check parameters
-    local params_parsed retval
-    params_parsed="$(ocrd ocrd-tool "$OCRD_TOOL_JSON" tool $OCRD_TOOL_NAME parse-params "${__parameters[@]}" "${__parameter_overrides[@]}")" || {
-        retval=$?
-        ocrd__raise "Failed to parse parameters (retval $retval):
-$params_parsed"
-    }
-    eval "$params_parsed"
-    params_json="$(ocrd ocrd-tool "$OCRD_TOOL_JSON" tool $OCRD_TOOL_NAME parse-params --json "${__parameters[@]}" "${__parameter_overrides[@]}")"
-
-}
-
-ocrd__wrap () {
-    set -e
-
-    declare -gx OCRD_TOOL_JSON="$1"
-    declare -gx OCRD_TOOL_NAME="$2"
-    shift
-    shift
-    declare -Agx params
-    params=()
-    declare -g params_json
-    declare -Agx ocrd__argv
-    ocrd__argv=()
-
-    if ! which "ocrd" >/dev/null 2>/dev/null;then
-        ocrd__raise "ocrd not in \$PATH"
-    fi
-
-    if ! declare -p "OCRD_TOOL_JSON" >/dev/null 2>/dev/null;then
-        ocrd__raise "Must set \$OCRD_TOOL_JSON"
-    elif [[ ! -r "$OCRD_TOOL_JSON" ]];then
-        ocrd__raise "Cannot read \$OCRD_TOOL_JSON: '$OCRD_TOOL_JSON'"
-    fi
-
-   if [[ -z "$OCRD_TOOL_NAME" ]];then
-        ocrd__raise "Must set \$OCRD_TOOL_NAME"
-    elif ! ocrd ocrd-tool "$OCRD_TOOL_JSON" list-tools|grep -q "$OCRD_TOOL_NAME";then
-        ocrd__raise "No such command \$OCRD_TOOL_NAME: $OCRD_TOOL_NAME"
-    fi
-
-    ocrd__parse_argv "$@"
-
-    declare -ag ocrd__files
-    IFS=$'\n'
-    ocrd__files=( $(ocrd bashlib input-files \
-                  --ocrd-tool $OCRD_TOOL_JSON \
-                  --executable $OCRD_TOOL_NAME \
-                  $(if [[ ${ocrd__argv[debug]} = true ]]; then echo --debug; fi) \
-                  $(if [[ ${ocrd__argv[overwrite]} = true ]]; then echo --overwrite; fi) \
-                  -m "${ocrd__argv[mets_file]}" \
-                  -d "${ocrd__argv[working_dir]}" \
-                  ${ocrd__argv[mets_server_url]:+-U} ${ocrd__argv[mets_server_url]:-} \
-                  -p "$params_json" \
-                  -I "${ocrd__argv[input_file_grp]}" \
-                  -O "${ocrd__argv[output_file_grp]}" \
-                  ${ocrd__argv[page_id]:+-g} ${ocrd__argv[page_id]:-}) )
-    IFS=$' \t\n'
-}
-
-## usage: pageId=$(ocrd__input_file 3 pageId)
-ocrd__input_file() {
-    declare -A input_file
-    eval input_file=( "${ocrd__files[$1]}" )
-    eval echo "${input_file[$2]}"
-}