ocrd 3.0.0b3__tar.gz → 3.0.0b4__tar.gz

{ocrd-3.0.0b3/src/ocrd.egg-info → ocrd-3.0.0b4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.0.0b3
+Version: 3.0.0b4
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

ocrd-3.0.0b4/VERSION

@@ -0,0 +1 @@
+3.0.0b4

{ocrd-3.0.0b3 → ocrd-3.0.0b4}/src/ocrd/processor/base.py

@@ -166,11 +166,14 @@ class Processor():
 
         (Override if ``ocrd-tool.json`` is not distributed with the Python package.)
         """
-        # XXX HACK
-        module_tokens = self.__module__.split('.')
-        if module_tokens[0] == 'src':
-            module_tokens.pop(0)
-        return resource_filename(module_tokens[0], self.metadata_filename)
+        module = inspect.getmodule(self)
+        module_tokens = module.__package__.split('.')
+        # for namespace packages, we cannot just use the first token
+        for i in range(len(module_tokens)):
+            prefix = '.'.join(module_tokens[:i + 1])
+            if sys.modules[prefix].__spec__.has_location:
+                return resource_filename(prefix, self.metadata_filename)
+        raise Exception("cannot find top-level module prefix for %s", module.__package__)
 
     @cached_property
     def metadata_rawdict(self) -> dict:
@@ -455,17 +458,17 @@ class Processor():
                 nr_copied = 0
 
                 # set up multithreading
-                if self.max_workers <= 0:
-                    max_workers = max(0, config.OCRD_MAX_PARALLEL_PAGES)
-                else:
-                    max_workers = max(0, min(config.OCRD_MAX_PARALLEL_PAGES, self.max_workers))
+                max_workers = max(0, config.OCRD_MAX_PARALLEL_PAGES)
+                if self.max_workers > 0 and self.max_workers < config.OCRD_MAX_PARALLEL_PAGES:
+                    self._base_logger.info("limiting number of threads from %d to %d", max_workers, self.max_workers)
+                    max_workers = self.max_workers
                 if max_workers > 1:
                     assert isinstance(workspace.mets, ClientSideOcrdMets), \
                         "OCRD_MAX_PARALLEL_PAGES>1 requires also using --mets-server-url"
-                if self.max_page_seconds <= 0:
-                    max_seconds = max(0, config.OCRD_PROCESSING_PAGE_TIMEOUT)
-                else:
-                    max_seconds = max(0, min(config.OCRD_PROCESSING_PAGE_TIMEOUT, self.max_page_seconds))
+                max_seconds = max(0, config.OCRD_PROCESSING_PAGE_TIMEOUT)
+                if self.max_page_seconds > 0 and self.max_page_seconds < config.OCRD_PROCESSING_PAGE_TIMEOUT:
+                    self._base_logger.info("limiting page timeout from %d to %d sec", max_seconds, self.max_page_seconds)
+                    max_seconds = self.max_page_seconds
                 executor = ThreadPoolExecutor(
                     max_workers=max_workers or 1,
                     thread_name_prefix=f"pagetask.{workspace.mets.unique_identifier}"

{ocrd-3.0.0b3 → ocrd-3.0.0b4}/src/ocrd/workspace.py

@@ -121,7 +121,10 @@ class Workspace():
         """
         Reload METS from the filesystem.
         """
-        self.mets = OcrdMets(filename=self.mets_target)
+        if self.is_remote:
+            self.mets.reload()
+        else:
+            self.mets = OcrdMets(filename=self.mets_target)
 
     @deprecated_alias(pageId="page_id")
     @deprecated_alias(ID="file_id")

{ocrd-3.0.0b3 → ocrd-3.0.0b4/src/ocrd.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.0.0b3
+Version: 3.0.0b4
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

ocrd-3.0.0b3/VERSION

@@ -1 +0,0 @@
-3.0.0b3