ocrd 3.5.1__py3-none-any.whl → 3.7.0__py3-none-any.whl

ocrd/cli/__init__.py

@@ -10,13 +10,14 @@ import click
 
 from ocrd_utils import config
 
+
 # pylint: disable=wrong-import-position
 
 def command_with_replaced_help(*replacements):
 
     class CommandWithReplacedHelp(click.Command):
         def get_help(self, ctx):
-            newhelp : str = super().get_help(ctx)
+            newhelp: str = super().get_help(ctx)
             for replacement in replacements:
                 newhelp = re.sub(*replacement, newhelp)
             # print(newhelp)
@@ -24,18 +25,18 @@ def command_with_replaced_help(*replacements):
 
     return CommandWithReplacedHelp
 
+
 # pylint: enable=wrong-import-position
 
 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']
@@ -102,20 +103,21 @@ Variables:
 {config.describe('OCRD_LOGGING_DEBUG')}
 """
 
+
 @click.group(epilog=_epilog)
 @click.version_option(package_name='ocrd')
 @ocrd_loglevel
-def cli(**kwargs): # pylint: disable=unused-argument
+def cli(**kwargs):  # pylint: disable=unused-argument
     """
     Entry-point of multi-purpose CLI for OCR-D
     """
 
+
 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

@@ -6,29 +6,20 @@ 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
 # ----------------------------------------------------------------------
 
+
 @click.group('bashlib')
 def bashlib_cli():
     """
@@ -39,6 +30,7 @@ def bashlib_cli():
 # ocrd bashlib filename
 # ----------------------------------------------------------------------
 
+
 @bashlib_cli.command('filename')
 def bashlib_filename():
     """
@@ -48,101 +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/cli/network.py

@@ -12,7 +12,6 @@ from ocrd_network.cli import (
     client_cli,
     processing_server_cli,
     processing_worker_cli,
-    processor_server_cli,
 )
 
 
@@ -27,4 +26,3 @@ def network_cli():
 network_cli.add_command(client_cli)
 network_cli.add_command(processing_server_cli)
 network_cli.add_command(processing_worker_cli)
-network_cli.add_command(processor_server_cli)

ocrd/cli/ocrd_tool.py

@@ -23,6 +23,7 @@ from ocrd_utils import (
 )
 from ocrd_validators import ParameterValidator, OcrdToolValidator
 
+
 class OcrdToolCtx():
 
     def __init__(self, filename):
@@ -36,25 +37,31 @@ class OcrdToolCtx():
 
         class BashProcessor(Processor):
             @property
-            def metadata(inner_self): # pylint: disable=no-self-argument,arguments-renamed
+            def metadata(inner_self):  # pylint: disable=no-self-argument,arguments-renamed
                 return self.json
+
             @property
-            def executable(inner_self): # pylint: disable=no-self-argument,arguments-renamed
+            def executable(inner_self):  # pylint: disable=no-self-argument,arguments-renamed
                 return self.tool_name
+
             @property
-            def moduledir(inner_self): # pylint: disable=no-self-argument,arguments-renamed
+            def moduledir(inner_self):  # pylint: disable=no-self-argument,arguments-renamed
                 return os.path.dirname(self.filename)
+
             # set docstrings to empty
             __doc__ = None
             # HACK: override the module-level docstring, too
             getmodule(OcrdToolCtx).__doc__ = None
-            def process(inner_self): # pylint: disable=no-self-argument,arguments-renamed
+
+            def process(inner_self):  # pylint: disable=no-self-argument,arguments-renamed
                 return super()
 
         self.processor = BashProcessor
 
+
 pass_ocrd_tool = click.make_pass_decorator(OcrdToolCtx)
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool
 # ----------------------------------------------------------------------
@@ -65,6 +72,7 @@ pass_ocrd_tool = click.make_pass_decorator(OcrdToolCtx)
 def ocrd_tool_cli(ctx, json_file):
     ctx.obj = OcrdToolCtx(json_file)
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool version
 # ----------------------------------------------------------------------
@@ -74,6 +82,7 @@ def ocrd_tool_cli(ctx, json_file):
 def ocrd_tool_version(ctx):
     print(ctx.json['version'])
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool validate
 # ----------------------------------------------------------------------
@@ -86,6 +95,7 @@ def ocrd_tool_validate(ctx):
     if not report.is_valid:
         return 128
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool list-tools
 # ----------------------------------------------------------------------
@@ -96,6 +106,7 @@ def ocrd_tool_list(ctx):
     for tool in ctx.json['tools']:
         print(tool)
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool dump-tools
 # ----------------------------------------------------------------------
@@ -105,6 +116,7 @@ def ocrd_tool_list(ctx):
 def ocrd_tool_dump(ctx):
     print(dumps(ctx.json['tools'], indent=True))
 
+
 @ocrd_tool_cli.command('dump-module-dirs', help="Dump module directory of each tool")
 @pass_ocrd_tool
 def ocrd_tool_dump_module_dirs(ctx):
@@ -112,6 +124,7 @@ def ocrd_tool_dump_module_dirs(ctx):
                  for tool_name in ctx.json['tools']},
                 indent=True))
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool
 # ----------------------------------------------------------------------
@@ -124,6 +137,7 @@ def ocrd_tool_tool(ctx, tool_name):
         raise Exception("No such tool: %s" % tool_name)
     ctx.tool_name = tool_name
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool description
 # ----------------------------------------------------------------------
@@ -133,29 +147,34 @@ def ocrd_tool_tool(ctx, tool_name):
 def ocrd_tool_tool_description(ctx):
     print(ctx.json['tools'][ctx.tool_name]['description'])
 
+
 @ocrd_tool_tool.command('list-resources', help="List tool's file resources")
 @pass_ocrd_tool
 def ocrd_tool_tool_list_resources(ctx):
     ctx.processor(None).list_resources()
 
+
 @ocrd_tool_tool.command('resolve-resource', help="Get a tool's file resource full path name")
 @click.argument('res_name')
 @pass_ocrd_tool
 def ocrd_tool_tool_resolve_resource(ctx, res_name):
     print(ctx.processor(None).resolve_resource(res_name))
 
+
 @ocrd_tool_tool.command('show-resource', help="Dump a tool's file resource")
 @click.argument('res_name')
 @pass_ocrd_tool
 def ocrd_tool_tool_show_resource(ctx, res_name):
     ctx.processor(None).show_resource(res_name)
 
+
 @ocrd_tool_tool.command('help', help="Generate help for processors")
 @click.argument('subcommand', required=False)
 @pass_ocrd_tool
 def ocrd_tool_tool_params_help(ctx, subcommand):
     ctx.processor(None).show_help(subcommand=subcommand)
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool categories
 # ----------------------------------------------------------------------
@@ -165,6 +184,7 @@ def ocrd_tool_tool_params_help(ctx, subcommand):
 def ocrd_tool_tool_categories(ctx):
     print('\n'.join(ctx.json['tools'][ctx.tool_name]['categories']))
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool steps
 # ----------------------------------------------------------------------
@@ -174,6 +194,7 @@ def ocrd_tool_tool_categories(ctx):
 def ocrd_tool_tool_steps(ctx):
     print('\n'.join(ctx.json['tools'][ctx.tool_name]['steps']))
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool dump
 # ----------------------------------------------------------------------
@@ -183,6 +204,7 @@ def ocrd_tool_tool_steps(ctx):
 def ocrd_tool_tool_dump(ctx):
     print(dumps(ctx.json['tools'][ctx.tool_name], indent=True))
 
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool parse-params
 # ----------------------------------------------------------------------

ocrd/cli/process.py

@@ -13,6 +13,7 @@ from ocrd.task_sequence import run_tasks
 
 from ..decorators import ocrd_loglevel
 
+
 # ----------------------------------------------------------------------
 # ocrd process
 # ----------------------------------------------------------------------

ocrd/cli/resmgr.py

@@ -18,7 +18,6 @@ from ocrd_utils import (
     getLogger,
     get_moduledir,
     get_ocrd_tool_json,
-    resource_filename,
     initLogging,
     RESOURCE_LOCATIONS,
 )

ocrd/cli/validate.py

@@ -17,12 +17,13 @@ from ocrd.task_sequence import ProcessorTask, validate_tasks
 from ocrd_utils import initLogging, parse_json_string_or_file, DEFAULT_METS_BASENAME
 from ocrd_validators import (
     OcrdToolValidator,
-    OcrdZipValidator,
+    # OcrdZipValidator,
     PageValidator,
     ParameterValidator,
-    WorkspaceValidator,
+    # WorkspaceValidator,
 )
 
+
 def _inform_of_result(report):
     if not report.is_valid:
         print(report.to_xml())
@@ -36,6 +37,7 @@ def validate_cli():
     """
     initLogging()
 
+
 @validate_cli.command('tool-json')
 @click.argument('ocrd_tool', required=False, nargs=1)
 def validate_ocrd_tool(ocrd_tool):
@@ -48,6 +50,7 @@ def validate_ocrd_tool(ocrd_tool):
         ocrd_tool = loads(f.read())
     _inform_of_result(OcrdToolValidator.validate(ocrd_tool))
 
+
 @validate_cli.command('parameters')
 @click.argument('ocrd_tool')
 @click.argument('executable')
@@ -60,24 +63,31 @@ def validate_parameters(ocrd_tool, executable, param_json):
         ocrd_tool = loads(f.read())
     _inform_of_result(ParameterValidator(ocrd_tool['tools'][executable]).validate(parse_json_string_or_file(param_json)))
 
+
 @validate_cli.command('page')
 @click.argument('page', required=True, nargs=1)
-@click.option('--page-textequiv-consistency', help="How strict to check PAGE multi-level textequiv consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict')
-@click.option('--page-textequiv-strategy', help="Strategy to determine the correct textequiv", type=click.Choice(['first']), default='first')
+@click.option('--page-textequiv-consistency', help="How strict to check PAGE multi-level textequiv consistency",
+              type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict')
+@click.option('--page-textequiv-strategy', help="Strategy to determine the correct textequiv",
+              type=click.Choice(['first']), default='first')
 @click.option('--check-baseline', help="Whether Baseline must be fully within TextLine/Coords", is_flag=True, default=False)
-@click.option('--check-coords', help="Whether *Region/TextLine/Word/Glyph must each be fully contained within Border/*Region/TextLine/Word, resp.", is_flag=True, default=False)
+@click.option('--check-coords', help="Whether *Region/TextLine/Word/Glyph must each be fully contained "
+              "within Border/*Region/TextLine/Word, resp.", is_flag=True, default=False)
 def validate_page(page, **kwargs):
     '''
     Validate PAGE against OCR-D conventions
     '''
     _inform_of_result(PageValidator.validate(filename=page, **kwargs))
 
+
 #  @validate_cli.command('zip')
 #  @click.argument('src', type=click.Path(dir_okay=True, readable=True, resolve_path=True), required=True)
 #  @click.option('-Z', '--skip-unzip', help="Treat SRC as a directory not a ZIP", is_flag=True, default=False)
 #  @click.option('-B', '--skip-bag', help="Whether to skip all checks of manifests and files", is_flag=True, default=False)
-#  @click.option('-C', '--skip-checksums', help="Whether to omit checksum checks but still check basic BagIt conformance", is_flag=True, default=False)
-#  @click.option('-D', '--skip-delete', help="Whether to skip deleting the unpacked OCRD-ZIP dir after valdiation", is_flag=True, default=False)
+#  @click.option('-C', '--skip-checksums', help="Whether to omit checksum checks but still check basic BagIt conformance",
+#                is_flag=True, default=False)
+#  @click.option('-D', '--skip-delete', help="Whether to skip deleting the unpacked OCRD-ZIP dir after valdiation",
+#                is_flag=True, default=False)
 #  @click.option('-j', '--processes', help="Number of parallel processes", type=int, default=1)
 #  def validate(src, **kwargs):
 #      """
@@ -87,11 +97,16 @@ def validate_page(page, **kwargs):
 #      """
 #      _inform_of_result(OcrdZipValidator(Resolver(), src).validate(**kwargs))
 
+
 #  @validate_cli.command('workspace')
 #  @click.option('-a', '--download', is_flag=True, help="Download all files")
-#  @click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(['imagefilename', 'dimension', 'mets_unique_identifier', 'mets_file_group_names', 'mets_files', 'pixel_density', 'page', 'url']))
-#  @click.option('--page-textequiv-consistency', '--page-strictness', help="How strict to check PAGE multi-level textequiv consistency", type=click.Choice(['strict', 'lax', 'fix', 'off']), default='strict')
-#  @click.option('--page-coordinate-consistency', help="How fierce to check PAGE multi-level coordinate consistency", type=click.Choice(['poly', 'baseline', 'both', 'off']), default='poly')
+#  @click.option('-s', '--skip', help="Tests to skip", default=[], multiple=True, type=click.Choice(
+#                ['imagefilename', 'dimension', 'mets_unique_identifier', 'mets_file_group_names', 'mets_files',
+#                'pixel_density', 'page', 'url']))
+#  @click.option('--page-textequiv-consistency', '--page-strictness', type=click.Choice(['strict', 'lax', 'fix', 'off']),
+#                help="How strict to check PAGE multi-level textequiv consistency", default='strict')
+#  @click.option('--page-coordinate-consistency', help="How fierce to check PAGE multi-level coordinate consistency",
+#                type=click.Choice(['poly', 'baseline', 'both', 'off']), default='poly')
 #  @click.argument('mets_url')
 #  def validate_workspace(mets_url, **kwargs):
 #      '''
@@ -99,11 +114,15 @@ def validate_page(page, **kwargs):
 #      '''
 #      _inform_of_result(WorkspaceValidator.validate(Resolver(), mets_url, **kwargs))
 
+
 @validate_cli.command('tasks')
-@click.option('--workspace', nargs=1, required=False, help='Workspace directory these tasks are to be run. If omitted, only validate syntax')
-@click.option('-M', '--mets-basename', nargs=1, default=DEFAULT_METS_BASENAME, help='Basename of the METS file, used in conjunction with --workspace')
+@click.option('--workspace', nargs=1, required=False, help='Workspace directory these tasks are to be run. '
+              'If omitted, only validate syntax')
+@click.option('-M', '--mets-basename', nargs=1, default=DEFAULT_METS_BASENAME,
+              help='Basename of the METS file, used in conjunction with --workspace')
 @click.option('-U', '--mets-server-url', help='TCP host URI or UDS path of METS server')
-@click.option('--overwrite', is_flag=True, default=False, help='When checking against a concrete workspace, simulate overwriting output or page range.')
+@click.option('--overwrite', is_flag=True, default=False,
+              help='When checking against a concrete workspace, simulate overwriting output or page range.')
 @click.option('-g', '--page-id', help="ID(s) of the pages to process")
 @click.argument('tasks', nargs=-1, required=True)
 def validate_process(tasks, workspace, mets_basename, mets_server_url, overwrite, page_id):