ocrd 3.11.0__py3-none-any.whl → 3.12.1__py3-none-any.whl

ocrd/cli/workspace.py

@@ -6,7 +6,7 @@ OCR-D CLI: workspace management
     :nested: full
 """
 from os import rmdir, unlink
-from os.path import dirname, relpath, normpath, exists, join, isabs, isdir
+from os.path import abspath, dirname, relpath, normpath, exists, join, isabs, isdir
 from pathlib import Path
 from json import loads, dumps
 import sys
@@ -47,7 +47,7 @@ class WorkspaceCtx():
     def workspace(self):
         return Workspace(
             self.resolver,
-            directory=self.directory,
+            self.directory,
             mets_basename=self.mets_basename,
             automatic_backup=self.automatic_backup,
             mets_server_url=self.mets_server_url,
@@ -85,7 +85,7 @@ def workspace_cli(ctx, directory, mets, mets_basename, mets_server_url, backup):
     initLogging()
     ctx.obj = WorkspaceCtx(
         directory,
-        mets_url=mets,
+        mets,
         mets_basename=mets_basename,
         mets_server_url=mets_server_url,
         automatic_backup=backup
@@ -254,22 +254,26 @@ def workspace_add_file(ctx, file_grp, file_id, mimetype, page_id, ignore, check_
     log.debug("Adding '%s'", fname)
     local_filename = None
     if not (fname.startswith('http://') or fname.startswith('https://')):
-        if not fname.startswith(ctx.directory):
-            if not isabs(fname) and exists(join(ctx.directory, fname)):
+        assert isabs(ctx.directory)
+        # try to resolve relative to workspace or CWD
+        if not isabs(fname):
+            if exists(join(ctx.directory, fname)):
                 fname = join(ctx.directory, fname)
             else:
-                log.debug("File '%s' is not in workspace, copying", fname)
-                try:
-                    fname = ctx.resolver.download_to_directory(ctx.directory, fname, subdir=file_grp)
-                except FileNotFoundError:
-                    if check_file_exists:
-                        log.error("File '%s' does not exist, halt execution!" % fname)
-                        sys.exit(1)
-        if check_file_exists and not exists(fname):
-            log.error("File '%s' does not exist, halt execution!" % fname)
-            sys.exit(1)
+                fname = abspath(fname)
         if fname.startswith(ctx.directory):
+            if check_file_exists and not exists(fname):
+                log.error("File '%s' does not exist, halt execution!" % fname)
+                sys.exit(1)
             fname = relpath(fname, ctx.directory)
+        else:
+            log.debug("File '%s' is not in workspace, copying", fname)
+            try:
+                fname = ctx.resolver.download_to_directory(ctx.directory, fname, subdir=file_grp)
+            except FileNotFoundError:
+                if check_file_exists:
+                    log.error("File '%s' does not exist, halt execution!" % fname)
+                    sys.exit(1)
         local_filename = fname
 
     if not page_id:
@@ -299,8 +303,8 @@ def workspace_add_file(ctx, file_grp, file_id, mimetype, page_id, ignore, check_
 @click.option('-g', '--page-id', help="physical page ID of the file", required=False)
 @click.option('-i', '--file-id', help="ID of the file. If not provided, derive from fileGrp and filename", required=False)
 @click.option('-u', '--url', help="Remote URL of the file", required=False)
-@click.option('-l', '--local-filename', help="Local filesystem path in the workspace directory "
-              "(copied from source file if different)", required=False)
+@click.option('-l', '--local-filename', help="Relative filesystem path in the workspace directory "
+              "(copied from source path if different)", required=False)
 @click.option('-G', '--file-grp', help="File group USE of the file", required=True)
 @click.option('-n', '--dry-run', help="Don't actually do anything to the METS or filesystem, just preview",
               default=False, is_flag=True)
@@ -414,38 +418,45 @@ def workspace_cli_bulk_add(ctx, regex, mimetype, page_id, file_id, url, local_fi
                 raise ValueError(f"OcrdFile attribute '{param_name}' unset ({file_dict})")
             for group_name in group_dict:
                 file_dict[param_name] = file_dict[param_name].replace('{{ %s }}' % group_name, group_dict[group_name])
+            if '{{ ' in file_dict[param_name]:
+                log.warning("Probably failed to match a capture group in %s: '%s'", param_name, file_dict[param_name])
 
         # Where to copy from
         if src_path_option:
-            src_path = src_path_option
             for group_name in group_dict:
-                src_path = src_path.replace('{{ %s }}' % group_name, group_dict[group_name])
-            srcpath = Path(src_path)
+                src_path_option = src_path_option.replace('{{ %s }}' % group_name, group_dict[group_name])
+            if '{{ ' in src_path_option:
+                log.warning("Probably failed to match a capture group in source path: '%s'", src_path_option)
+            src_path = Path(src_path_option)
         else:
-            srcpath = file_path
+            src_path = file_path
 
         # derive --file-id from filename if not --file-id not explicitly set
         if not file_id:
-            id_field = srcpath.stem if file_path != srcpath else file_path.stem
+            id_field = src_path.stem if file_path != src_path else file_path.stem
             file_dict['file_id'] = safe_filename('%s_%s' % (file_dict['file_grp'], id_field))
         if not mimetype:
             try:
-                file_dict['mimetype'] = EXT_TO_MIME[srcpath.suffix]
+                file_dict['mimetype'] = EXT_TO_MIME[src_path.suffix]
             except KeyError:
                 log.error("Cannot guess MIME type from extension '%s' for '%s'. "
-                          "Set --mimetype explicitly" % (srcpath.suffix, srcpath))
+                          "Set --mimetype explicitly" % (src_path.suffix, src_path))
 
         # copy files if src != url
+        src_path = src_path.resolve().absolute()
         if local_filename_is_src:
-            file_dict['local_filename'] = srcpath
-        else:
-            destpath = Path(workspace.directory, file_dict['local_filename'])
-            if srcpath != destpath and not destpath.exists():
-                log.info("cp '%s' '%s'", srcpath, destpath)
-                if not dry_run:
-                    if not destpath.parent.is_dir():
-                        destpath.parent.mkdir()
-                    destpath.write_bytes(srcpath.read_bytes())
+            if str(src_path).startswith(workspace.directory):
+                file_dict['local_filename'] = src_path.relative_to(workspace.directory)
+            else:
+                file_dict['local_filename'] = src_path.name # will be copied-in below
+
+        dest_path = Path(workspace.directory, file_dict['local_filename'])
+        if src_path != dest_path and not dest_path.exists():
+            log.info("cp '%s' '%s'", src_path, dest_path)
+            if not dry_run:
+                if not dest_path.parent.is_dir():
+                    dest_path.parent.mkdir()
+                dest_path.write_bytes(src_path.read_bytes())
 
         # Add to workspace (or not)
         fileGrp = file_dict.pop('file_grp')
@@ -465,8 +476,7 @@ def workspace_cli_bulk_add(ctx, regex, mimetype, page_id, file_id, url, local_fi
 @workspace_cli.command('find')
 @mets_find_options
 @click.option('-k', '--output-field', help="Output field. Repeat for multiple fields, will be joined with tab",
-              default=['local_filename'],
-              show_default=True,
+              default=['local_filename'], show_default=True,
               multiple=True,
               type=click.Choice([
                   'url',
@@ -869,7 +879,7 @@ def merge(ctx, overwrite, force, copy_files, filegrp_mapping, fileid_mapping, pa
     the same semantics as in ``ocrd workspace find``, see ``ocrd workspace find --help``
     for an explanation.
     """
-    mets_path = Path(mets_path)
+    mets_path = Path(mets_path).absolute()
     if filegrp_mapping:
         filegrp_mapping = loads(filegrp_mapping)
     assert not ctx.mets_server_url, \

ocrd/mets_server.py

@@ -418,7 +418,8 @@ class OcrdMetsServer:
     def create_process(mets_server_url: str, ws_dir_path: str, log_file: str) -> int:
         sub_process = Popen(
             args=["ocrd", "workspace", "-U", f"{mets_server_url}", "-d", f"{ws_dir_path}", "server", "start"],
-            stdout=open(file=log_file, mode="w"), stderr=open(file=log_file, mode="a"), cwd=ws_dir_path,
+            stdout=open(file=log_file, mode="w"), stderr=open(file=log_file, mode="a"),
+            #cwd=ws_dir_path, # rs: if relative, this will cause wrong path resolution in the subprocess
             shell=False, universal_newlines=True, start_new_session=True
         )
         # Wait for the mets server to start
@@ -446,7 +447,7 @@ class OcrdMetsServer:
                 Path(self.url).unlink()
 
     def startup(self):
-        self.log.info("Configuring the METS Server")
+        self.log.info("Configuring the METS Server for %s", self.workspace)
 
         workspace = self.workspace
 

ocrd/processor/base.py

@@ -539,7 +539,7 @@ class Processor():
                     self._base_logger.info("limiting page timeout from %d to %d sec", max_seconds, self.max_page_seconds)
                     max_seconds = self.max_page_seconds
 
-                if isinstance(workspace.mets, ClientSideOcrdMets):
+                if max_workers > 1:
                     executor_cls = ProcessPoolExecutor
                     log_queue = mp.get_context('fork').Queue()
                 else:
@@ -553,7 +553,7 @@ class Processor():
                     initializer=_page_worker_set_ctxt,
                     initargs=(self, log_queue),
                 )
-                if isinstance(workspace.mets, ClientSideOcrdMets):
+                if max_workers > 1:
                     assert executor.active # ensure pre-forking
                     # forward messages from log queue (in subprocesses) to all root handlers
                     log_listener = logging.handlers.QueueListener(log_queue, *logging.root.handlers,

{ocrd-3.11.0.dist-info → ocrd-3.12.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: ocrd
-Version: 3.11.0
+Version: 3.12.1
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

{ocrd-3.11.0.dist-info → ocrd-3.12.1.dist-info}/RECORD

@@ -1,6 +1,6 @@
 ocrd/__init__.py,sha256=ZswMVmlqFhAEIzMR3my6IKPq9XLH21aDPC_m_8Jh4dA,1076
 ocrd/constants.py,sha256=REPY-y28MMsrTWBNB4oOsvX3W06Xr2fvtv9wuWH9oAI,633
-ocrd/mets_server.py,sha256=LbZ0U2_o0W7cWO639U7E816dXabro8-8yHGX0quvHn4,22304
+ocrd/mets_server.py,sha256=No9UJnD14etKrfGaH_ZKgeaMR2-mPAD1mOJ1j8lwzlU,22415
 ocrd/ocrd-all-tool.json,sha256=qVTZq8cJtKQuEgtqYQVkpYQPz4BegO09b7TT-LcfsJs,3134
 ocrd/resolver.py,sha256=7uwHRxaK8YMdKHe_a2dfrcNwL6UhQRJRVBrIX7GST7Q,15443
 ocrd/resource_manager.py,sha256=2wo3JSCYE1oA0VgI8H901IsC-fnx6vRJ5qSMFgYNorE,20664
@@ -15,7 +15,7 @@ ocrd/cli/ocrd_tool.py,sha256=kB3Y3tj7Fpz6Ts4KgVlznhXpAx8gCDvJTnO39j8SGL4,7679
 ocrd/cli/process.py,sha256=yfhBSYmuY5k2AccKwiNvG9hCDx1coYyWjq9BBwYaL3Y,1234
 ocrd/cli/resmgr.py,sha256=7hRRi8EryQwakRdZgguee3ercA5_T48BKGWWfgAVfzM,8072
 ocrd/cli/validate.py,sha256=P8jrzAnoU-5TUjLNA7s_ZMY2Krw5Y-SVIZPhdOk25cw,5931
-ocrd/cli/workspace.py,sha256=0UzKN7vvD0n5wwxldzLHOlikDDIyiBiV1PuTOKCnnnE,41279
+ocrd/cli/workspace.py,sha256=_QPRyDW0F8SPM8ClHNmtzOou_bHD9XzZleB89cI7-e0,41884
 ocrd/cli/zip.py,sha256=3HMUbVsPTK3SRuF5oZnCZLjoqXJK-AYpA-rMqenY858,5965
 ocrd/decorators/__init__.py,sha256=gU7WDp2re8BF8DU4PVRRqgQpp7TR5wMOnoafcbMAjws,6324
 ocrd/decorators/loglevel_option.py,sha256=QHVZ_72rkt2hqGSP9vuIXUYo5YhBxLsldUsaVVpb_YY,802
@@ -23,7 +23,7 @@ ocrd/decorators/mets_find_options.py,sha256=8fiSdk-415o6-iBPB2T9He_v52qE8cTj3cCn
 ocrd/decorators/ocrd_cli_options.py,sha256=Bemkq3V3QkOI3nNqGzphaNW7gjU9vNN-M5F2DvxvioM,2479
 ocrd/decorators/parameter_option.py,sha256=TnCIcV9L5oAnI1Ew2TyFzo5FAwiIzWl2pn8oaD9jfEU,1056
 ocrd/processor/__init__.py,sha256=39ymNwYRdc-b_OJzzKmWCvo2ga3KdsGSYDHE1Hzkn_w,274
-ocrd/processor/base.py,sha256=0SbFhLYUC8VMPZdCd_Y329514IuOz1X1jrCbuuX8Kwg,62074
+ocrd/processor/base.py,sha256=oTrQAEJtbIqNwS9xk7qGuwcYdh9hy5MFMsNk98if4og,62012
 ocrd/processor/helpers.py,sha256=4lR_QvZsxvh7f8_uK9YzdHP5-hvFU4qqYM_Cu_k41KI,10937
 ocrd/processor/ocrd_page_result.py,sha256=hHV1TlKhKFN848cUCqR31v2R3HH4HEoeyGXqUc2DLkY,2945
 ocrd/processor/builtin/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -38,7 +38,7 @@ ocrd/processor/builtin/param_command_transkribus-to-prima.json,sha256=AvPNNS5uBm
 ocrd/processor/builtin/shell_processor.py,sha256=000qP9wGfoJWtovFXaNYaDfKw_9VNEg7keulYoe0GyM,5943
 ocrd/processor/builtin/dummy/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ocrd/processor/builtin/dummy/ocrd-tool.json,sha256=t_M3HABw7k_Ufi1L9Mr4t3LSCRnu0HH8-fvEs3u2PQY,3487
-ocrd-3.11.0.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.12.1.dist-info/licenses/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
 ocrd_modelfactory/__init__.py,sha256=sjAwPwDzetvPHdV6nPquHtMdFUBYRmo1P-VKER9YCWM,4404
 ocrd_models/__init__.py,sha256=A0aj0mOraNb-xfiUueACdoaqISnp0qH-F49nTJg2vCs,380
 ocrd_models/constants.py,sha256=R7-jOGabFd8HP0qRWfTMk0RcUmdwN-jhmDVbUW_QfU4,6961
@@ -56,12 +56,12 @@ ocrd_models/xpath_functions.py,sha256=VM2f9hl8ja4NrDOEQRSYdx7GewwAxfoyGMDjqjgA_7
 ocrd_network/__init__.py,sha256=NWlSgXi7z45ow37AmITxfCB1d-L39rO8ttyxNJ-z8G0,376
 ocrd_network/client.py,sha256=hi13uDUYC5t7xHtZEUYwNBAZOvovWaScfCtFSORVg7Q,3224
 ocrd_network/client_utils.py,sha256=d5UE0MdDJxsYxIQemKcoUuALOiPJ8Cew8bjgsg9d71w,5709
-ocrd_network/constants.py,sha256=mUjpkZDYPdRZmOeC0jyzQkuLuWrODLFzlrAHkguKWGg,1942
+ocrd_network/constants.py,sha256=lre5j0JAHAzZQMWPOJnVnSG79KIT5RmEtJ9PmHUNR-c,1771
 ocrd_network/database.py,sha256=p2vaFgVzhA1l8CTY9vTMT6mOjzHSJK1tE1dkPrUUcp8,10670
 ocrd_network/logging_utils.py,sha256=hXwS46FzY_HTh92DgnxTuARxj8C18bOBmFKVrvBlUgc,2409
 ocrd_network/param_validators.py,sha256=Jl1VwiPPKJ50k-xEHLdvW-1QDOkJHCiMz4k9Ipqm-Uc,1489
 ocrd_network/process_helpers.py,sha256=t2qltUpRefzLwdSGsiUEOGYO4Pz2OH7arpgjmCAeXMU,3086
-ocrd_network/processing_server.py,sha256=z21DvRleEeo0hkpc1-2z0jLKuf5WSipL95MVEns8eJE,38457
+ocrd_network/processing_server.py,sha256=vHKDW_ZPVzCJpC34nnO7_L8zeRq7kH1eO_DqpPfmKEw,38453
 ocrd_network/processing_worker.py,sha256=5AtvIhfcePzltKj4SElh7Aj9zlUOEiMVPTjtXuFSbT8,12659
 ocrd_network/resource_manager_server.py,sha256=Ihz2g9uhkPSJee9GL7485DFC4cORhro4JQI6QzHoUA4,7255
 ocrd_network/server_cache.py,sha256=orfAMw3LwUnduRHFAB6MpfoORTDoPV4ntSdAcQHBOyI,13148
@@ -91,7 +91,7 @@ ocrd_network/runtime_data/connection_clients.py,sha256=HKf_aSfwg11JeH6qiQXnqxbnv
 ocrd_network/runtime_data/deployer.py,sha256=j3tcauURZtu7MKcEIE9B5eMCMSYMbxhB8LmtK72Zk1c,5314
 ocrd_network/runtime_data/hosts.py,sha256=P1bLh1NjgL-ajgP-VhGCACvy6rgJ5nZhGKsHaldzatk,8921
 ocrd_network/runtime_data/network_agents.py,sha256=wwN7IJei4UdlyfhjvdmB5TB4O0Gn8icSkArJe-suvAY,4523
-ocrd_network/runtime_data/network_services.py,sha256=5aH3RNGCi1fBuSdRp_Xz0MzyD_FmnvPnaBYAiYY3gp4,7891
+ocrd_network/runtime_data/network_services.py,sha256=HDEs2ZAtZEQFEnWUWa2UP5l3IH0ek19PWmtSEObzC-8,7706
 ocrd_utils/__init__.py,sha256=Cl_lrZxjXuTZ_me4I_lpaFNTpSdacWhQetOtHdrkUsU,6057
 ocrd_utils/config.py,sha256=Oe8JBGb8r4z274XNWcdMV-GApzxmAYO8hHmbAV5bXf8,12609
 ocrd_utils/constants.py,sha256=6lqMLeJdkFBlvGVmGjcExWbRKzNU6QT0kADBb5BkcBc,3464
@@ -124,8 +124,8 @@ 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-3.11.0.dist-info/METADATA,sha256=pgkMu0pKPHEcGCCS63Dmxo9fnfWfUI9owWzouzCfECw,11479
-ocrd-3.11.0.dist-info/WHEEL,sha256=_zCd3N1l69ArxyTb8rzEoP9TpbYXkqRFSNOD5OuxnTs,91
-ocrd-3.11.0.dist-info/entry_points.txt,sha256=CI-NoDR1BYmsuAsJmPAn4NrN9guzdedHGUbC8QSmdGs,266
-ocrd-3.11.0.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.11.0.dist-info/RECORD,,
+ocrd-3.12.1.dist-info/METADATA,sha256=XQvJMEg3Aw7kMcV7go_4BTD32c4M024v4DWffD3-YHo,11479
+ocrd-3.12.1.dist-info/WHEEL,sha256=wUyA8OaulRlbfwMtmQsvNngGrxQHAvkKcvRmdizlJi0,92
+ocrd-3.12.1.dist-info/entry_points.txt,sha256=CI-NoDR1BYmsuAsJmPAn4NrN9guzdedHGUbC8QSmdGs,266
+ocrd-3.12.1.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.12.1.dist-info/RECORD,,

{ocrd-3.11.0.dist-info → ocrd-3.12.1.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (80.9.0)
+Generator: setuptools (80.10.2)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

ocrd_network/constants.py

@@ -2,8 +2,6 @@ from enum import Enum
 
 DOCKER_IMAGE_MONGO_DB = "mongo"
 DOCKER_IMAGE_RABBIT_MQ = "rabbitmq:4.2-management"
-# These feature flags are required by default to use the newer version
-DOCKER_RABBIT_MQ_FEATURES = "quorum_queue,implicit_default_bindings,classic_mirrored_queue_version"
 
 NETWORK_PROTOCOLS = ["http://", "https://"]
 OCRD_ALL_TOOL_JSON = "ocrd-all-tool.json"

ocrd_network/processing_server.py

@@ -388,7 +388,7 @@ class ProcessingServer(FastAPI):
         self.cache_processing_requests.update_request_counter(workspace_key=workspace_key, by_value=0)
 
         # This check is done to return early in case a workspace_id is provided
-        # but the abs mets path cannot be queried from the DB
+        # but the mets path cannot be queried from the DB
         request_mets_path = await validate_and_return_mets_path(self.log, data)
 
         page_ids = expand_page_ids(data.page_id)

ocrd_network/runtime_data/network_services.py

@@ -2,7 +2,7 @@ from __future__ import annotations
 from logging import Logger
 from typing import Any, Dict, List, Optional, Union
 
-from ..constants import DOCKER_IMAGE_MONGO_DB, DOCKER_IMAGE_RABBIT_MQ, DOCKER_RABBIT_MQ_FEATURES
+from ..constants import DOCKER_IMAGE_MONGO_DB, DOCKER_IMAGE_RABBIT_MQ
 from ..database import verify_mongodb_available
 from ..rabbitmq_utils import verify_rabbitmq_available
 from .connection_clients import create_docker_client
@@ -136,9 +136,7 @@ class DataRabbitMQ(DataNetworkService):
             env = [
                 # The default credentials to be used by the processing workers
                 f"RABBITMQ_DEFAULT_USER={rmq_user}",
-                f"RABBITMQ_DEFAULT_PASS={rmq_password}",
-                # These feature flags are required by default to use the newer version
-                f"RABBITMQ_FEATURE_FLAGS={DOCKER_RABBIT_MQ_FEATURES}"
+                f"RABBITMQ_DEFAULT_PASS={rmq_password}"
             ]
         if not ports_mapping:
             # 5672, 5671 - used by AMQP 0-9-1 and AMQP 1.0 clients without and with TLS