ocrd 3.3.2__py3-none-any.whl → 3.4.1__py3-none-any.whl

ocrd/cli/__init__.py

@@ -67,6 +67,12 @@ Variables:
 \b
 {config.describe('OCRD_EXISTING_OUTPUT', wrap_text=False)}
 \b
+{config.describe('OCRD_MAX_MISSING_OUTPUTS')}
+\b
+{config.describe('OCRD_MAX_PARALLEL_PAGES')}
+\b
+{config.describe('OCRD_PROCESSING_PAGE_TIMEOUT')}
+\b
 {config.describe('OCRD_METS_CACHING')}
 \b
 {config.describe('OCRD_MAX_PROCESSOR_CACHE')}

ocrd/cli/ocrd_tool.py

@@ -18,7 +18,8 @@ from ocrd.processor import Processor
 from ocrd_utils import (
     set_json_key_value_overrides,
     parse_json_string_or_file,
-    parse_json_string_with_comments as loads
+    parse_json_string_with_comments as loads,
+    get_moduledir
 )
 from ocrd_validators import ParameterValidator, OcrdToolValidator
 
@@ -104,6 +105,13 @@ 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):
+    print(dumps({tool_name: get_moduledir(tool_name)
+                 for tool_name in ctx.json['tools']},
+                indent=True))
+
 # ----------------------------------------------------------------------
 # ocrd ocrd-tool tool
 # ----------------------------------------------------------------------

ocrd/cli/workspace.py

@@ -474,11 +474,11 @@ def workspace_find(ctx, file_grp, mimetype, page_id, file_id, output_field, incl
                 if wait:
                     time.sleep(wait)
             if undo_download and f.url and f.local_filename:
-                f.local_filename = None
                 modified_mets = True
                 if not keep_files:
                     ctx.log.debug("rm %s [cwd=%s]", f.local_filename, workspace.directory)
                     unlink(f.local_filename)
+                f.local_filename = None
             ret_entry = [f.ID if field == 'pageId' else str(getattr(f, field)) or '' for field in output_field]
             ret.append(ret_entry)
     if modified_mets:

ocrd/decorators/ocrd_cli_options.py

@@ -56,7 +56,7 @@ def ocrd_cli_options(f):
         # subcommands. So we have to work around that by creating a
         # pseudo-subcommand handled in ocrd_cli_wrap_processor
         argument('subcommand', nargs=1, required=False,
-                 type=click.Choice([AgentType.PROCESSING_WORKER, AgentType.PROCESSOR_SERVER])),
+                 type=click.Choice(list(map(str, AgentType)))),
     ]
     for param in params:
         param(f)

ocrd/mets_server.py

@@ -46,7 +46,7 @@ class OcrdFileModel(BaseModel):
     ):
         return OcrdFileModel(
             file_grp=file_grp, file_id=file_id, page_id=page_id, mimetype=mimetype, url=url,
-            local_filename=str(local_filename)
+            local_filename=str(local_filename) if local_filename else None
         )
 
 
@@ -314,7 +314,7 @@ class ClientSideOcrdMets:
         else:
             r = self.session.request("POST", self.url, json=MpxReq.add_file(self.ws_dir_path, kwargs))
             if not r.ok:
-                raise RuntimeError(f"Failed to add file ({str(data)}): {r.json()[errors]}")
+                raise RuntimeError(f"Failed to add file ({str(data)}): {r.json()}")
 
         return ClientSideOcrdFile(
             None, fileGrp=file_grp,

ocrd/processor/base.py

@@ -29,8 +29,7 @@ from frozendict import frozendict
 # this is where the fixes came from:
 from loky import Future, ProcessPoolExecutor
 import multiprocessing as mp
-from threading import Timer
-from _thread import interrupt_main
+from multiprocessing.pool import ThreadPool
 
 from click import wrap_text
 from deprecated import deprecated
@@ -783,11 +782,16 @@ class Processor():
         page_id = input_files[input_pos].pageId
         self._base_logger.info("processing page %s", page_id)
         for i, input_file in enumerate(input_files):
+            grp = self.input_file_grp.split(',')[i]
             if input_file is None:
-                grp = self.input_file_grp.split(',')[i]
                 self._base_logger.debug(f"ignoring missing file for input fileGrp {grp} for page {page_id}")
                 continue
             assert isinstance(input_file, get_args(OcrdFileType))
+            if not input_file.local_filename:
+                self._base_logger.error(f'No local file exists for page {page_id} in file group {grp}')
+                if config.OCRD_MISSING_INPUT == 'ABORT':
+                    raise MissingInputFile(grp, page_id, input_file.mimetype)
+                continue
             self._base_logger.debug(f"parsing file {input_file.ID} for page {page_id}")
             try:
                 page_ = page_from_file(input_file)
@@ -796,6 +800,9 @@ class Processor():
             except ValueError as err:
                 # not PAGE and not an image to generate PAGE for
                 self._base_logger.error(f"non-PAGE input for page {page_id}: {err}")
+        if not any(input_pcgts):
+            self._base_logger.warning(f'skipping page {page_id}')
+            return
         output_file_id = make_file_id(input_files[input_pos], self.output_file_grp)
         if input_files[input_pos].fileGrp == self.output_file_grp:
             # input=output fileGrp: re-use ID exactly
@@ -1107,7 +1114,11 @@ class Processor():
             self._base_logger.critical(f"Could not find any files for selected pageId {self.page_id}.\n"
                                        f"compare '{self.page_id}' with the output of 'orcd workspace list-page'.")
         ifts = []
-        for page, ifiles in pages.items():
+        # use physical page order
+        for page in self.workspace.mets.physical_pages:
+            if page not in pages:
+                continue
+            ifiles = pages[page]
             for i, ifg in enumerate(ifgs):
                 if not ifiles[i]:
                     # could be from non-unique with on_error=skip or from true gap
@@ -1150,18 +1161,15 @@ def _page_worker(timeout, *input_files):
     """
     page_id = next((file.pageId for file in input_files
                     if hasattr(file, 'pageId')), "")
-    if timeout > 0:
-        timer = Timer(timeout, interrupt_main)
-        timer.start()
+    pool = ThreadPool(processes=1)
     try:
-        _page_worker_processor.process_page_file(*input_files)
+        #_page_worker_processor.process_page_file(*input_files)
+        async_result = pool.apply_async(_page_worker_processor.process_page_file, input_files)
+        async_result.get(timeout or None)
         _page_worker_processor.logger.debug("page worker completed for page %s", page_id)
-    except KeyboardInterrupt:
+    except mp.TimeoutError:
         _page_worker_processor.logger.debug("page worker timed out for page %s", page_id)
-        raise TimeoutError()
-    finally:
-        if timeout > 0:
-            timer.cancel()
+        raise
 
 def generate_processor_help(ocrd_tool, processor_instance=None, subcommand=None):
     """Generate a string describing the full CLI of this processor including params.

ocrd/workspace.py

@@ -5,7 +5,7 @@ from shutil import copyfileobj
 from re import sub
 from tempfile import NamedTemporaryFile
 from contextlib import contextmanager
-from typing import Optional, Union
+from typing import Optional, Union, Callable
 
 from cv2 import COLOR_GRAY2BGR, COLOR_RGB2BGR, cvtColor
 from PIL import Image
@@ -457,6 +457,20 @@ class Workspace():
             with atomic_write(self.mets_target) as f:
                 f.write(self.mets.to_xml(xmllint=True).decode('utf-8'))
 
+    def _apply_mets_file(self, filename_or_url: str, fun: Callable):
+        if not filename_or_url:
+            # avoid "finding" just any file
+            raise ValueError("requires non-empty filename or URL")
+        with pushd_popd(self.directory):
+            if Path(filename_or_url).exists():
+                return fun(filename_or_url)
+            if image_file := next(self.mets.find_files(local_filename=str(filename_or_url)), None):
+                return fun(image_file.local_filename)
+            if image_file := next(self.mets.find_files(url=str(filename_or_url)), None):
+                return fun(self.download_file(image_file).local_filename)
+            with download_temporary_file(filename_or_url) as f:
+                return fun(f.name)
+
     def resolve_image_exif(self, image_url):
         """
         Get the EXIF metadata about an image URL as :py:class:`ocrd_models.ocrd_exif.OcrdExif`
@@ -467,19 +481,7 @@ class Workspace():
         Returns:
             :py:class:`ocrd_models.ocrd_exif.OcrdExif`
         """
-        if not image_url:
-            # avoid "finding" just any file
-            raise ValueError(f"'image_url' must be a non-empty string, not '{image_url}' ({type(image_url)})")
-        try:
-            f = next(self.mets.find_files(local_filename=str(image_url)))
-            return exif_from_filename(f.local_filename)
-        except StopIteration:
-            try:
-                f = next(self.mets.find_files(url=str(image_url)))
-                return exif_from_filename(self.download_file(f).local_filename)
-            except StopIteration:
-                with download_temporary_file(image_url) as f:
-                    return exif_from_filename(f.name)
+        return self._apply_mets_file(image_url, exif_from_filename)
 
     @deprecated(version='1.0.0', reason="Use workspace.image_from_page and workspace.image_from_segment")
     def resolve_image_as_pil(self, image_url, coords=None):
@@ -498,22 +500,9 @@ class Workspace():
         return self._resolve_image_as_pil(image_url, coords)
 
     def _resolve_image_as_pil(self, image_url, coords=None):
-        if not image_url:
-            # avoid "finding" just any file
-            raise Exception("Cannot resolve empty image path")
         log = getLogger('ocrd.workspace._resolve_image_as_pil')
-        with pushd_popd(self.directory):
-            try:
-                f = next(self.mets.find_files(local_filename=str(image_url)))
-                pil_image = Image.open(f.local_filename)
-            except StopIteration:
-                try:
-                    f = next(self.mets.find_files(url=str(image_url)))
-                    pil_image = Image.open(self.download_file(f).local_filename)
-                except StopIteration:
-                    with download_temporary_file(image_url) as f:
-                        pil_image = Image.open(f.name)
-            pil_image.load() # alloc and give up the FD
+        pil_image = self._apply_mets_file(image_url, Image.open)
+        pil_image.load() # alloc and give up the FD
 
         # Pillow does not properly support higher color depths
         # (e.g. 16-bit or 32-bit or floating point grayscale),
@@ -788,16 +777,14 @@ class Workspace():
             raise Exception('Found no AlternativeImage that satisfies all requirements ' +
                             'filename="%s" in page "%s"' % (
                                 filename, page_id))
-        if not all(feature in page_coords['features']
-                   for feature in feature_selector.split(',') if feature):
-            raise Exception('Found no AlternativeImage that satisfies all requirements ' +
-                            'selector="%s" in page "%s"' % (
-                                feature_selector, page_id))
-        if any(feature in page_coords['features']
-               for feature in feature_filter.split(',') if feature):
-            raise Exception('Found no AlternativeImage that satisfies all requirements ' +
-                            'filter="%s" in page "%s"' % (
-                                feature_filter, page_id))
+        if (not all(feature in page_coords['features']
+                    for feature in feature_selector.split(',') if feature) or
+            any(feature in page_coords['features']
+                for feature in feature_filter.split(',') if feature)):
+            raise Exception('Found no AlternativeImage that satisfies all requirements' +
+                            ' selector="%s"' % feature_selector +
+                            ' filter="%s"' % feature_filter +
+                            ' in page "%s"' % page_id)
         # ensure DPI will be set in image meta-data again
         if 'DPI' in page_coords:
             dpi = page_coords['DPI']
@@ -1049,16 +1036,14 @@ class Workspace():
             raise Exception('Found no AlternativeImage that satisfies all requirements ' +
                             'filename="%s" in segment "%s"' % (
                                 filename, segment.id))
-        if not all(feature in segment_coords['features']
-                   for feature in feature_selector.split(',') if feature):
+        if (not all(feature in segment_coords['features']
+                    for feature in feature_selector.split(',') if feature) or
+            any(feature in segment_coords['features']
+                for feature in feature_filter.split(',') if feature)):
             raise Exception('Found no AlternativeImage that satisfies all requirements' +
-                            'selector="%s" in segment "%s"' % (
-                                feature_selector, segment.id))
-        if any(feature in segment_coords['features']
-               for feature in feature_filter.split(',') if feature):
-            raise Exception('Found no AlternativeImage that satisfies all requirements ' +
-                            'filter="%s" in segment "%s"' % (
-                                feature_filter, segment.id))
+                            ' selector="%s"' % feature_selector +
+                            ' filter="%s"' % feature_filter +
+                            ' in segment "%s"' % segment.id)
         # ensure DPI will be set in image meta-data again
         if 'DPI' in segment_coords:
             dpi = segment_coords['DPI']

{ocrd-3.3.2.dist-info → ocrd-3.4.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.3.2
+Version: 3.4.1
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

{ocrd-3.3.2.dist-info → ocrd-3.4.1.dist-info}/RECORD

@@ -1,32 +1,32 @@
 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=cplsjQ5OMsBRDPLLbN-6qr3VETg6xOX21tdJOxwgCWk,22471
+ocrd/mets_server.py,sha256=o01N5vAdPeu-xCgngcMRQm0bzBOhn_IFMV0AbeXDu9g,22491
 ocrd/ocrd-all-tool.json,sha256=EYXmMzP68p3KzL8nUZ16TCX2chQzKkAeISvuXqI_yIw,2094
 ocrd/resolver.py,sha256=A7BrZlUGrfJye-etaEuT-fdJFgvQcCxWovjufT-WmRY,15119
 ocrd/resource_list.yml,sha256=82-PiqkZnka1kTj3MQqNn4wXWKHHtoFchsQuetWuqFs,2633
 ocrd/resource_manager.py,sha256=kIWDoKxWH4IJE1gcoTcCRQjYjieCqiQclyuyF6Y9b8A,16813
 ocrd/task_sequence.py,sha256=spiaUQaMM7M8WdBDoQGmLuTPm7tOugYXD6rcJ2UXzxw,6991
-ocrd/workspace.py,sha256=-j3X83K0f4vtd5jwfu6_R53RJ2R8gt1HYpyrv8YP2bg,65661
+ocrd/workspace.py,sha256=eLuGSJtOh3y2miKgcF8219YH1RkAaEi-qwXHarz8O8k,64916
 ocrd/workspace_backup.py,sha256=iab_JjZ_mMP-G8NIUk4PZmfpNlQuGRoqc3NbTSSew1w,3621
 ocrd/workspace_bagger.py,sha256=yU8H3xR5WmQKvgQewac71ie-DUWcfLnMS01D55zsEHQ,11971
-ocrd/cli/__init__.py,sha256=-BiwIakeCkWx0Jd2yX9_ahfdV4VKz_5yqGEJ_2zKakQ,2734
+ocrd/cli/__init__.py,sha256=LpQb8ne1nzAq2j52lGWDTZlBCmrLwUsz17PTwJkWNcU,2884
 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
-ocrd/cli/ocrd_tool.py,sha256=EyD5VdLm2WTzQnR-hZKpn-D4-dsWr2PIE5IoY1O3mfE,7357
+ocrd/cli/ocrd_tool.py,sha256=MLTqbtBCw8wBSScv2S4_xp5Jyz5fwD7BzUJxI9R4TBw,7653
 ocrd/cli/process.py,sha256=8KD0i7LT01H9u5CC1vktYMEVpS67da_rp_09_EOECmw,1233
 ocrd/cli/resmgr.py,sha256=mk8KZweC_7ENAFnC6FvFf7Zv_W1wqJTmk0EMd9XSvf4,10132
 ocrd/cli/validate.py,sha256=nvageDaHCETcE71X5lu7i_4JKpgo9MrvJKinVPLYUTI,5727
-ocrd/cli/workspace.py,sha256=bsp6YXEgwABIUFbSENmxV1c4oxRwc2L-BpeDPlYfhHE,40501
+ocrd/cli/workspace.py,sha256=t40r3tnzz0VivhmT0HSeA5-2xdvUleRYvniqLIBGoWs,40501
 ocrd/cli/zip.py,sha256=MMJLw3OXWiJVfVtrdJcBkbB8vA1IzSautluazZRuCQ0,5910
 ocrd/decorators/__init__.py,sha256=n2Lb1WLXGlvPrhNTSGZYRqugpa__MZSWV546EmQnTtc,7678
 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=lIvtE8re1VmpHm45u71ltE0QJS8nyd28HhLC7zGSvlo,2691
+ocrd/decorators/ocrd_cli_options.py,sha256=psS7u42mXTOWIXQd9kcrgW7kDnFURHbmZ0946aqBz3A,2659
 ocrd/decorators/parameter_option.py,sha256=n8hYw7XVTd3i3tvpK8F1Jx_CqRp6EGF9qJVH95yj92Q,1076
 ocrd/processor/__init__.py,sha256=39ymNwYRdc-b_OJzzKmWCvo2ga3KdsGSYDHE1Hzkn_w,274
-ocrd/processor/base.py,sha256=eQzrwPcGxyo0y7jlAF2g5jOabB-jq_LuJfTn87Qxjwk,59929
+ocrd/processor/base.py,sha256=_h0V5FevEPLb1q0zGtShuKXRj_tOWhD0M7_ufn34MPc,60476
 ocrd/processor/helpers.py,sha256=WFdC5zeB8F7T0FkpJwfTqWsSPNRtBCBUmFLgixw-rYs,10999
 ocrd/processor/ocrd_page_result.py,sha256=eDkpyVHcpaBzTHXiGrcNk9PP9Xr-XZru2w_uoX_ZeNA,510
 ocrd/processor/builtin/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
@@ -51,9 +51,9 @@ ocrd_models/xpath_functions.py,sha256=AwR8tHf56-mmIksnw_GeOQ760sxNHqK92T7z9OfsEE
 ocrd_network/__init__.py,sha256=gMejC614J5PPGgXDKBiQS0jt-Jx8qOrLbWH7zt8x8Gs,374
 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/constants.py,sha256=AAcE6zZQNcNp2oqPD6oIgoVLSs4IHTkg8AS92WCQ6Xo,1968
 ocrd_network/database.py,sha256=fcft7vdRDoR7vmPL1xNYTIeOg5DwRPcggwYDYxLy5ik,10706
-ocrd_network/logging_utils.py,sha256=bO9TQqBXw9CIZEKp8tHXbeE2NuJWMiaQDHzS05b4ajo,2153
+ocrd_network/logging_utils.py,sha256=ijWpM8B943Jx6F0NeK3ggni0198UYjM5NCkYpARLk_E,2472
 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=qBiYk4wgTLqhHvbmDWu_F626BfSfyvkoCD-i0ZwsBSE,42109
@@ -89,14 +89,14 @@ ocrd_network/runtime_data/hosts.py,sha256=ml19ptzH4TFofyJR-Qp_Mn3sZUFbWoNe__rRXZ
 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=1_8j4kpKK5gxhFKObbBEzmq4JazTtKEkGe14Ch_tpw0,11796
+ocrd_utils/config.py,sha256=PGRnqrDT7lWJCd9ezoAEBJ5CyVJfFnvCEPTLjtKduOQ,12143
 ocrd_utils/constants.py,sha256=ImbG1d8t2MW3uuFi-mN6aY90Zn74liAKZBKlfuKN86w,3278
 ocrd_utils/deprecate.py,sha256=4i50sZsA3Eevqn5D-SL5yGf9KEZfGCV4A5Anzn1GRMs,1026
 ocrd_utils/image.py,sha256=zNNX1cnRy6yvrxx8mnYQiqWraAh5-i4a1AOfCCg4SmI,24781
 ocrd_utils/introspect.py,sha256=gfBlmeEFuRmRUSgdSK0jOxRpYqDRXl2IAE6gv2MZ6as,1977
-ocrd_utils/logging.py,sha256=sHNfih9kBfvKsHdLqMK_ew9Pld1GsRyYlrZHIySujnw,7313
+ocrd_utils/logging.py,sha256=XYTL7DxUvdX4V56jhAYH6PkhjMFOmaa0kf_XkhSTTe0,7816
 ocrd_utils/ocrd_logging.conf,sha256=JlWmA_5vg6HnjPGjTC4mA5vFHqmnEinwllSTiOw5CCo,3473
-ocrd_utils/os.py,sha256=acRRdDBI8L6BK0Mf773yKEzwdpZSFRBJEKB2crL4EjU,9865
+ocrd_utils/os.py,sha256=tMjikpVXJ8sCgYBOrgjgT3vlR2Pok39nSKysYc6mUQ4,9863
 ocrd_utils/str.py,sha256=cRgqYILDGOAqWr0qrCrV52I3y4wvpwDVtnBGEUjXNS4,10116
 ocrd_validators/__init__.py,sha256=ZFc-UqRVBk9o1YesZFmr9lOepttNJ_NKx1Zdb7g_YsU,972
 ocrd_validators/bagit-profile.yml,sha256=sdQJlSi7TOn1E9WYMOZ1shewJ-i_nPaKmsAFkh28TGY,1011
@@ -120,9 +120,9 @@ ocrd_validators/xlink.xsd,sha256=8fW7YAMWXN2PbB_MMvj9H5ZeFoEBDzuYBtlGC8_6ijw,318
 ocrd_validators/xsd_mets_validator.py,sha256=4GWfLyqkmca0x7osDuXuExYuM0HWVrKoqn0S35sFhHU,467
 ocrd_validators/xsd_page_validator.py,sha256=BNz_9u-Ek4UCeyZu3KxSQoolfW9lvuaSR9nIu1XXxeE,467
 ocrd_validators/xsd_validator.py,sha256=6HrVAf6SzCvfUIuQdIzz9bOq4V-zhyii9yrUPoK2Uvo,2094
-ocrd-3.3.2.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-ocrd-3.3.2.dist-info/METADATA,sha256=ovylNGJP1F1XfiV0cgYS1MXdkJr0YkTWJp9GOrvFYvk,10442
-ocrd-3.3.2.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
-ocrd-3.3.2.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
-ocrd-3.3.2.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.3.2.dist-info/RECORD,,
+ocrd-3.4.1.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.4.1.dist-info/METADATA,sha256=bNMeKlfpm41iZkct1OWhmw-aPP1GbuvsB1wXksYq7_Q,10442
+ocrd-3.4.1.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
+ocrd-3.4.1.dist-info/entry_points.txt,sha256=4hcJ2LkK_OlIabHnKgFit35Ap7b5Lz1Gb4hzkxV0Kiw,152
+ocrd-3.4.1.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.4.1.dist-info/RECORD,,

ocrd_network/constants.py

@@ -11,12 +11,16 @@ OCRD_ALL_TOOL_JSON = "ocrd-all-tool.json"
 SERVER_ALL_PAGES_PLACEHOLDER = "all_pages"
 
 
-class AgentType(str, Enum):
+class StrEnum(str, Enum):
+    def __str__(self):
+        return self.value
+
+class AgentType(StrEnum):
     PROCESSING_WORKER = "worker"
     PROCESSOR_SERVER = "server"
 
 
-class DeployType(str, Enum):
+class DeployType(StrEnum):
     # Deployed by the Processing Server config file
     DOCKER = "docker"
     NATIVE = "native"
@@ -26,7 +30,7 @@ class DeployType(str, Enum):
 
 
 # TODO: Make the states uppercase
-class JobState(str, Enum):
+class JobState(StrEnum):
     # The processing job is cached inside the Processing Server requests cache
     cached = "CACHED"
     # The processing job was cancelled due to failed dependencies
@@ -43,7 +47,7 @@ class JobState(str, Enum):
     unset = "UNSET"
 
 
-class NetworkLoggingDirs(str, Enum):
+class NetworkLoggingDirs(StrEnum):
     METS_SERVERS = "mets_servers"
     PROCESSING_JOBS = "processing_jobs"
     PROCESSING_SERVERS = "processing_servers"
@@ -51,7 +55,7 @@ class NetworkLoggingDirs(str, Enum):
     PROCESSOR_SERVERS = "processor_servers"
 
 
-class ServerApiTags(str, Enum):
+class ServerApiTags(StrEnum):
     ADMIN = "admin"
     DISCOVERY = "discovery"
     PROCESSING = "processing"

ocrd_network/logging_utils.py

@@ -9,11 +9,22 @@ def configure_file_handler_with_formatter(logger: Logger, log_file: Path, mode:
     file_handler = FileHandler(filename=log_file, mode=mode)
     file_handler.setFormatter(Formatter(LOG_FORMAT))
     logger.addHandler(file_handler)
+    try:
+        log_file.chmod(0o666)
+    except PermissionError:
+        # if the file exists the permissions are supposed to already fit
+        pass
 
 
 def get_root_logging_dir(module_name: NetworkLoggingDirs) -> Path:
     module_log_dir = Path(config.OCRD_NETWORK_LOGS_ROOT_DIR, module_name.value)
-    module_log_dir.mkdir(parents=True, exist_ok=True)
+    try:
+        module_log_dir.mkdir(parents=True, exist_ok=True)
+        module_log_dir.chmod(0o777)
+    except PermissionError:
+        # if the folder exists the permissions are supposed to already fit
+        pass
+
     return module_log_dir
 
 

ocrd_utils/config.py

@@ -277,13 +277,23 @@ config.add(name="OCRD_NETWORK_SOCKETS_ROOT_DIR",
            description="The root directory where all mets server related socket files are created",
            parser=lambda val: Path(val),
            default=(True, Path(gettempdir(), "ocrd_network_sockets")))
-config.OCRD_NETWORK_SOCKETS_ROOT_DIR.mkdir(parents=True, exist_ok=True)
+config.OCRD_NETWORK_SOCKETS_ROOT_DIR.mkdir(mode=0o777, parents=True, exist_ok=True)
+try:
+    config.OCRD_NETWORK_SOCKETS_ROOT_DIR.chmod(0o777)
+except PermissionError:
+    # if the folder exists the permissions are supposed to already fit
+    pass
 
 config.add(name="OCRD_NETWORK_LOGS_ROOT_DIR",
            description="The root directory where all ocrd_network related file logs are stored",
            parser=lambda val: Path(val),
            default=(True, Path(gettempdir(), "ocrd_network_logs")))
-config.OCRD_NETWORK_LOGS_ROOT_DIR.mkdir(parents=True, exist_ok=True)
+config.OCRD_NETWORK_LOGS_ROOT_DIR.mkdir(mode=0o777, parents=True, exist_ok=True)
+try:
+    config.OCRD_NETWORK_LOGS_ROOT_DIR.chmod(0o777)
+except PermissionError:
+    # if the folder exists the permissions are supposed to already fit
+    pass
 
 config.add("HOME",
     description="Directory to look for `ocrd_logging.conf`, fallback for unset XDG variables.",

ocrd_utils/logging.py

@@ -32,6 +32,7 @@ import logging
 import logging.config
 from pathlib import Path
 import sys
+from os import chmod
 
 from .constants import LOG_FORMAT, LOG_TIMEFMT
 from .config import config
@@ -166,6 +167,15 @@ def initLogging(builtin_only=False, force_reinit=False, silent=not config.OCRD_L
         if not silent:
             print(f"[LOGGING] Picked up logging config at {config_file}", file=sys.stderr)
         logging.config.fileConfig(config_file)
+        # Set permission of processing-server logfile to 666 to prevent possible permission erros while logging
+        try:
+            network_logger = logging.getLogger("ocrd_network")
+            for handler in network_logger.handlers:
+                if isinstance(handler, logging.FileHandler):
+                    chmod(handler.baseFilename, 0o666)
+        except PermissionError:
+            # if the file exists the permissions are supposed to already fit
+            pass
     else:
         if not silent:
             print("[LOGGING] Initializing logging with built-in defaults", file=sys.stderr)

ocrd_utils/os.py

@@ -254,7 +254,7 @@ def guess_media_type(input_file : str, fallback : str = None, application_xml :
     if mimetype is None:
         mimetype = EXT_TO_MIME.get(''.join(Path(input_file).suffixes), fallback)
     if mimetype is None:
-        raise ValueError("Could not determine MIME type of input_file must")
+        raise ValueError(f"Could not determine MIME type of {input_file}")
     if mimetype == 'application/xml':
         mimetype = application_xml
     return mimetype