ocrd 3.3.1__tar.gz → 3.4.0__tar.gz

{ocrd-3.3.1/src/ocrd.egg-info → ocrd-3.4.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.3.1
+Version: 3.4.0
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

ocrd-3.4.0/VERSION

@@ -0,0 +1 @@
+3.4.0

{ocrd-3.3.1 → ocrd-3.4.0}/src/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-3.3.1 → ocrd-3.4.0}/src/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
         )
 
 
@@ -131,7 +131,7 @@ class ClientSideOcrdMets:
     def __init__(self, url, workspace_path: Optional[str] = None):
         self.protocol = "tcp" if url.startswith("http://") else "uds"
         self.log = getLogger(f"ocrd.models.ocrd_mets.client.{url}")
-        self.url = url if self.protocol == "tcp" else f'http+unix://{url.replace("/", "%2F")}'
+        self.url = url if self.protocol == "tcp" else f'http+unix://{url.replace("/", "%2F").replace(".", "%2E")}'
         self.ws_dir_path = workspace_path if workspace_path else None
 
         if self.protocol == "tcp" and "tcp_mets" in self.url:
@@ -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-3.3.1 → ocrd-3.4.0}/src/ocrd/resolver.py

@@ -308,4 +308,8 @@ class Resolver():
                     if not is_file_in_directory(directory, mets_url):
                         raise ValueError("--mets '%s' has a directory part inconsistent with --directory '%s'" % (mets_url, directory))
 
+        if mets_server_url and not mets_server_url.startswith('http://'):
+            # UDS socket
+            mets_server_url = str(Path(mets_server_url).resolve())
+
         return str(Path(directory).resolve()), str(mets_url), str(mets_basename), mets_server_url

{ocrd-3.3.1 → ocrd-3.4.0}/src/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),

{ocrd-3.3.1 → ocrd-3.4.0/src/ocrd.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.3.1
+Version: 3.4.0
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

{ocrd-3.3.1 → ocrd-3.4.0}/src/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-3.3.1 → ocrd-3.4.0}/src/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-3.3.1 → ocrd-3.4.0}/src/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-3.3.1 → ocrd-3.4.0}/tests/test_mets_server.py

@@ -181,7 +181,7 @@ def test_mets_server_str(start_mets_server):
     assert str(f) == '<ClientSideOcrdFile fileGrp=OCR-D-IMG, ID=FILE_0001_IMAGE, mimetype=image/tiff, url=---, local_filename=OCR-D-IMG/FILE_0001_IMAGE.tif]/>'
     a = workspace_server.mets.agents[0]
     assert str(a) == '<ClientSideOcrdAgent [type=OTHER, othertype=SOFTWARE, role=CREATOR, otherrole=---, name=DFG-Koordinierungsprojekt zur Weiterentwicklung von Verfahren der Optical Character Recognition (OCR-D)]/>'
-    assert str(workspace_server.mets) == '<ClientSideOcrdMets[url=%s]>' % ('http+unix://%2Ftmp%2Focrd-mets-server.sock' if mets_server_url == TRANSPORTS[0] else TRANSPORTS[1])
+    assert str(workspace_server.mets) == '<ClientSideOcrdMets[url=%s]>' % ('http+unix://%2Ftmp%2Focrd-mets-server%2Esock' if mets_server_url == TRANSPORTS[0] else TRANSPORTS[1])
 
 def test_mets_test_unimplemented(start_mets_server):
     _, workspace_server = start_mets_server

ocrd-3.3.1/VERSION

@@ -1 +0,0 @@
-3.3.1

{ocrd-3.3.1 → ocrd-3.4.0}/src/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: