ocrd 3.8.1__py3-none-any.whl → 3.9.1__py3-none-any.whl

ocrd/processor/base.py

@@ -824,51 +824,59 @@ class Processor():
         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
-            output_file_id = input_files[input_pos].ID
-        output_file = next(self.workspace.mets.find_files(ID=output_file_id), None)
-        if output_file and config.OCRD_EXISTING_OUTPUT != 'OVERWRITE':
-            # short-cut avoiding useless computation:
-            raise FileExistsError(
-                f"A file with ID=={output_file_id} already exists {output_file} and neither force nor ignore are set"
-            )
-        result = self.process_page_pcgts(*input_pcgts, page_id=page_id)
-        for image_result in result.images:
-            image_file_id = f'{output_file_id}_{image_result.file_id_suffix}'
-            image_file_path = join(self.output_file_grp, f'{image_file_id}.png')
-            if isinstance(image_result.alternative_image, PageType):
-                # special case: not an alternative image, but replacing the original image
-                # (this is needed by certain processors when the original's coordinate system
-                #  cannot or must not be kept)
-                image_result.alternative_image.set_imageFilename(image_file_path)
-                image_result.alternative_image.set_imageWidth(image_result.pil.width)
-                image_result.alternative_image.set_imageHeight(image_result.pil.height)
-            elif isinstance(image_result.alternative_image, AlternativeImageType):
-                image_result.alternative_image.set_filename(image_file_path)
-            elif image_result.alternative_image is None:
-                pass  # do not reference in PAGE result
-            else:
-                raise ValueError(f"process_page_pcgts returned an OcrdPageResultImage of unknown type "
-                                 f"{type(image_result.alternative_image)}")
-            self.workspace.save_image_file(
-                image_result.pil,
-                image_file_id,
-                self.output_file_grp,
+        output_file_grps = self.output_file_grp.split(',')
+        output_file_ids = [make_file_id(input_files[input_pos], output_file_grp)
+                           if input_files[input_pos].fileGrp != output_file_grp else
+                           # input=output fileGrp: re-use ID exactly
+                           input_files[input_pos].ID
+                           for output_file_grp in output_file_grps]
+        if config.OCRD_EXISTING_OUTPUT != 'OVERWRITE':
+            for output_file_id in output_file_ids:
+                if output_file := next(self.workspace.mets.find_files(ID=output_file_id), None):
+                    # short-cut avoiding useless computation:
+                    raise FileExistsError(
+                        f"A file with ID=={output_file_id} already exists {output_file}"
+                        " and OCRD_EXISTING_OUTPUT != OVERWRITE"
+                    )
+        results = self.process_page_pcgts(*input_pcgts, page_id=page_id)
+        if len(results) > len(output_file_grps):
+            self._base_logger.error(f"processor returned {len(results) - len(output_file_grps)} "
+                                    f"more results than specified output fileGrps for page {page_id}")
+        for result, output_file_id, output_file_grp in zip(results, output_file_ids, output_file_grps):
+            for image_result in result.images:
+                image_file_id = f'{output_file_id}_{image_result.file_id_suffix}'
+                image_file_path = join(output_file_grp, f'{image_file_id}.png')
+                if isinstance(image_result.alternative_image, PageType):
+                    # special case: not an alternative image, but replacing the original image
+                    # (this is needed by certain processors when the original's coordinate system
+                    #  cannot or must not be kept, e.g. dewarping)
+                    image_result.alternative_image.set_imageFilename(image_file_path)
+                    image_result.alternative_image.set_imageWidth(image_result.pil.width)
+                    image_result.alternative_image.set_imageHeight(image_result.pil.height)
+                elif isinstance(image_result.alternative_image, AlternativeImageType):
+                    image_result.alternative_image.set_filename(image_file_path)
+                elif image_result.alternative_image is None:
+                    pass  # do not reference in PAGE result
+                else:
+                    raise ValueError(f"process_page_pcgts returned an OcrdPageResultImage of unknown type "
+                                     f"{type(image_result.alternative_image)}")
+                self.workspace.save_image_file(
+                    image_result.pil,
+                    image_file_id,
+                    output_file_grp,
+                    page_id=page_id,
+                    file_path=image_file_path,
+                )
+            result.pcgts.set_pcGtsId(output_file_id)
+            self.add_metadata(result.pcgts)
+            self.workspace.add_file(
+                file_id=output_file_id,
+                file_grp=output_file_grp,
                 page_id=page_id,
-                file_path=image_file_path,
+                local_filename=os.path.join(output_file_grp, output_file_id + '.xml'),
+                mimetype=MIMETYPE_PAGE,
+                content=to_xml(result.pcgts),
             )
-        result.pcgts.set_pcGtsId(output_file_id)
-        self.add_metadata(result.pcgts)
-        self.workspace.add_file(
-            file_id=output_file_id,
-            file_grp=self.output_file_grp,
-            page_id=page_id,
-            local_filename=os.path.join(self.output_file_grp, output_file_id + '.xml'),
-            mimetype=MIMETYPE_PAGE,
-            content=to_xml(result.pcgts),
-        )
 
     def process_page_pcgts(self, *input_pcgts: Optional[OcrdPage], page_id: Optional[str] = None) -> OcrdPageResult:
         """

ocrd/processor/ocrd_page_result.py

@@ -1,4 +1,5 @@
 from dataclasses import dataclass, field
+import copy
 from typing import List, Union, Optional
 from ocrd_models.ocrd_page import OcrdPage
 from PIL.Image import Image
@@ -8,12 +9,85 @@ from ocrd_models.ocrd_page_generateds import AlternativeImageType, PageType
 
 @dataclass
 class OcrdPageResultImage():
+    """
+    Encapsulates a single ``AlternativeImage`` reference to be persisted
+    as image file to the :py:class:`ocrd.Workspace`.
+    """
     pil: Image
+    """
+    image data to be saved
+    """
     file_id_suffix: str
+    """
+    a suffix to append to the file name when saving
+    (something like ``.IMG`` according to OCR-D
+    conventions for PAGE-XML)
+    """
     alternative_image: Optional[Union[AlternativeImageType, PageType]]
+    """
+    the ``AlternativeImage`` instance that references this image;
+    to be amended with the actual (final) ``@filename`` when saving
+
+    alternatively, can be a ``Page`` instance: in that case,
+    amend its ``@imageFilename`` (i.e. replace the original image
+    of the PAGE-XML)
+    """
 
 
 @dataclass
 class OcrdPageResult():
+    """
+    Encapsulates the return type of :py:func:`ocrd.Processor.process_page_pcgts`,
+    i.e. an instance of :py:class:`ocrd_models.ocrd_page.OcrdPage` and an
+    accompanying list of :py:class:`OcrdPageResultImage` that contain all
+    image files referenced via ``AlternativeImage`` to be persisted into the
+    :py:class:`ocrd.Workspace` along with the PAGE-XML itself.
+    """
     pcgts: OcrdPage
     images: List[OcrdPageResultImage] = field(default_factory=list)
+
+class OcrdPageResultVariadicListWrapper():
+    """
+    Proxy object for :py:class:`ocrd.SingleOcrdPageResult` allowing
+    list semantics (i.e. multi-valued return from
+    :py:func:`ocrd.Processor.process_page_pcgts`) without changing
+    the API introduced in version 3.0.
+
+    Everything but list access will yield the old (singular valued)
+    semantics.
+    """
+    def __init__(
+            self,
+            pcgts: OcrdPage,
+            *args):
+        self._results = [SingleOcrdPageResult(pcgts)] + [
+            SingleOcrdPageResult(arg) for arg in args]
+
+    def __getitem__(self, key):
+        return self._results[key]
+
+    def __contains__(self, key):
+        return key in self._results
+
+    def __len__(self):
+        return len(self._results)
+
+    def __iter__(self):
+        return iter(self._results)
+
+    def __repr__(self):
+        return repr(self._results)
+
+    # allow copy() without infinite recursion
+    def __copy__(self):
+        return OcrdPageResultVariadicListWrapper(*copy.copy(self._results))
+
+    # allow deepcopy() without infinite recursion
+    def __deepcopy__(self, memo):
+        return OcrdPageResultVariadicListWrapper(*copy.deepcopy(self._results))
+
+    # delegate to all members of first result
+    def __getattr__(self, name):
+        return getattr(self._results[0], name)
+
+SingleOcrdPageResult, OcrdPageResult = OcrdPageResult, OcrdPageResultVariadicListWrapper

{ocrd-3.8.1.dist-info → ocrd-3.9.1.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.8.1
+Version: 3.9.1
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

{ocrd-3.8.1.dist-info → ocrd-3.9.1.dist-info}/RECORD

@@ -23,9 +23,9 @@ 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=DxBsRn8VLsfNvc9_2BU0KxUv4t9XtHSSu9uiabxn8Nk,59850
+ocrd/processor/base.py,sha256=yHwxd4ZkHLPuFgqQmOeDhMWAdCnHY_ptOjiSWj-FZqI,60600
 ocrd/processor/helpers.py,sha256=4lR_QvZsxvh7f8_uK9YzdHP5-hvFU4qqYM_Cu_k41KI,10937
-ocrd/processor/ocrd_page_result.py,sha256=qo9pGV4r9S5--NAq5clIJOfs4b1vavoDOTbDqAEAAKA,507
+ocrd/processor/ocrd_page_result.py,sha256=hHV1TlKhKFN848cUCqR31v2R3HH4HEoeyGXqUc2DLkY,2945
 ocrd/processor/builtin/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ocrd/processor/builtin/dummy_processor.py,sha256=SmMRtN0w88kBU24654ThT-yf84SFsFW4BOcmwsDDWdc,3533
 ocrd/processor/builtin/filter_processor.py,sha256=9mbMq_XTJa8wrlbNdf46GUMNdjedz-enxafsCrnNhEo,4295
@@ -46,16 +46,16 @@ ocrd_models/ocrd_agent.py,sha256=Nm0XDNCmWZ8O3xsXaY-WmEghttXmh90UKmAObCL99IY,561
 ocrd_models/ocrd_exif.py,sha256=HSLPn_WBDRIlMtKNYilLHm8WjX-b14HgnqT_KfzjS_0,4680
 ocrd_models/ocrd_file.py,sha256=9-mfDb91RVy3p9rKryl-C39P4Of6Rb8OZBuxAee4VrI,9723
 ocrd_models/ocrd_mets.py,sha256=lz9mlDq9A9UmZDoN8lh5XRnBzdAtLLZywDZSbyZPS84,50905
-ocrd_models/ocrd_page.py,sha256=SRInM4HgcfDzbJH4ZO5B53ST2lAO7pgTXluwi0yzkf0,6211
-ocrd_models/ocrd_page_generateds.py,sha256=IWoN3V-v3C4JgyPaFh9OQC87ob__wUP1Q6ELBxhLA1w,841794
+ocrd_models/ocrd_page.py,sha256=Hed1PJ4JWSkTVj7mVOWDaJqtZ9Fc9czzOfPr6flFohw,6818
+ocrd_models/ocrd_page_generateds.py,sha256=hBIhOs_slXdQza_zokBfjjsrimX76h9I_6prRWbgVAk,911920
 ocrd_models/ocrd_xml_base.py,sha256=iOnDl2zBNhN-Q4moLWiFkSqXvfRzxE5wbp5Tjsu1W6A,1642
 ocrd_models/report.py,sha256=CX-t9ZDi2VmAy8M1Azsh83UsvE_f5pMeEC7tPaA-ztU,2021
 ocrd_models/utils.py,sha256=A-H11ZJ65ZjH4DPK9s_Yz6JtA9fbTQ2jY-__9s7Hrg8,2320
 ocrd_models/xpath_functions.py,sha256=VM2f9hl8ja4NrDOEQRSYdx7GewwAxfoyGMDjqjgA_7g,1439
 ocrd_network/__init__.py,sha256=NWlSgXi7z45ow37AmITxfCB1d-L39rO8ttyxNJ-z8G0,376
-ocrd_network/client.py,sha256=pL-g79cQgulXyGYgLOh--oxl1hZEMu48PTbuvMW1jIE,3007
-ocrd_network/client_utils.py,sha256=Ne1a0fteb-TBuc0EAD6X_fh2RAl4hmPt2oluhpB28iU,5371
-ocrd_network/constants.py,sha256=XyRYjFO38yIBD6s1wsA-z6V16tBmbUw4LXlFkj-tQC8,1943
+ocrd_network/client.py,sha256=hi13uDUYC5t7xHtZEUYwNBAZOvovWaScfCtFSORVg7Q,3224
+ocrd_network/client_utils.py,sha256=d5UE0MdDJxsYxIQemKcoUuALOiPJ8Cew8bjgsg9d71w,5709
+ocrd_network/constants.py,sha256=mUjpkZDYPdRZmOeC0jyzQkuLuWrODLFzlrAHkguKWGg,1942
 ocrd_network/database.py,sha256=-SddvaMLKn0pjdONyvWmjxfPJd6viedAIp6Lj1sU1Zs,10705
 ocrd_network/logging_utils.py,sha256=hXwS46FzY_HTh92DgnxTuARxj8C18bOBmFKVrvBlUgc,2409
 ocrd_network/param_validators.py,sha256=Jl1VwiPPKJ50k-xEHLdvW-1QDOkJHCiMz4k9Ipqm-Uc,1489
@@ -68,7 +68,7 @@ ocrd_network/server_utils.py,sha256=Lxby62gHvrSbHgpWXvyZGdsWajp2TFzyxjHdMZWBESk,
 ocrd_network/tcp_to_uds_mets_proxy.py,sha256=yRW-O6ihd31gf7xqQBIBb_ZQQgqisMyOdRI216ehq_A,3160
 ocrd_network/utils.py,sha256=yE-nV_sv171tPp7weIFOxYw6HJlxvGBmrS8b1rIHS7c,6760
 ocrd_network/cli/__init__.py,sha256=VBjjXcn-2O5gerqE6UdNfS-EkVFEVPQFHylsn8F9kfY,317
-ocrd_network/cli/client.py,sha256=H5fiJhBqbFn4_B2p3V20GejGTIYO-mNglh3y5nzUGhs,10350
+ocrd_network/cli/client.py,sha256=WoLt1NZAOtHeECegUBcop8K2_D0S8khrLjFZhV_38ww,10551
 ocrd_network/cli/processing_server.py,sha256=NsuI0f9h4KDwe39YugmHo5cJ_29chcLLQ7DThKfPO7s,770
 ocrd_network/cli/processing_worker.py,sha256=ZuaCkbKV_WKJV7cGOjZ6RLrjjppymnwNCiznFMlclAg,1897
 ocrd_network/cli/resmgr_server.py,sha256=sc0VX_RehTbg8Qp7ht_DvVqsrdL5b9Zw3bBgWcAD13A,826
@@ -81,7 +81,7 @@ ocrd_network/rabbitmq_utils/__init__.py,sha256=XLIqZhfin4I4m80G9B__UcP45Lz10_mEp
 ocrd_network/rabbitmq_utils/connector.py,sha256=N6mzjIf5FkVIno3FI1AksZY4F5jMUAm8baay0nXZx8w,11343
 ocrd_network/rabbitmq_utils/constants.py,sha256=Zu_dKJASfrgnIvEZZlFX9uDR9y6w7zy0KhW7gP7wHDE,1063
 ocrd_network/rabbitmq_utils/consumer.py,sha256=3WeryDmo0dSD9U0eLODbDElscvhEYjNeCBIewQHYfws,2488
-ocrd_network/rabbitmq_utils/helpers.py,sha256=5G0wrBlDtmCItzp-fMZjYr1oeaqDUcii5qeyjXI-ilM,5372
+ocrd_network/rabbitmq_utils/helpers.py,sha256=gbP9Ks4c_ksMln-VQ7GCND6ok_lttm6wW-R7Wszo5qA,5374
 ocrd_network/rabbitmq_utils/ocrd_messages.py,sha256=wwzfMWbXmOFo_nd32_XySCso91_Ul-aGm_GhGncNxD4,4419
 ocrd_network/rabbitmq_utils/publisher.py,sha256=mw4XQQhRE1xUQVgEUseyG845iIgVO-9GdGwNH6nUFms,2433
 ocrd_network/runtime_data/__init__.py,sha256=PnWuuagElbkTzGtPWQEk5wlFtDxqT7B48S0Zrgt8H68,320
@@ -123,9 +123,9 @@ 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-3.8.1.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-ocrd-3.8.1.dist-info/METADATA,sha256=cn6Wts1L_Gu7xs9Eachncz78IqE5YZS4N5FoJ8xZU7w,11396
-ocrd-3.8.1.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
-ocrd-3.8.1.dist-info/entry_points.txt,sha256=CI-NoDR1BYmsuAsJmPAn4NrN9guzdedHGUbC8QSmdGs,266
-ocrd-3.8.1.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.8.1.dist-info/RECORD,,
+ocrd-3.9.1.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.9.1.dist-info/METADATA,sha256=89_vM8lCOcs1eMAHyEO6YunESQbjiWG6jZNkWsrBYV0,11396
+ocrd-3.9.1.dist-info/WHEEL,sha256=iAkIy5fosb7FzIOwONchHf19Qu7_1wCWyFNR5gu9nU0,91
+ocrd-3.9.1.dist-info/entry_points.txt,sha256=CI-NoDR1BYmsuAsJmPAn4NrN9guzdedHGUbC8QSmdGs,266
+ocrd-3.9.1.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.9.1.dist-info/RECORD,,

ocrd_models/ocrd_page.py

@@ -2,6 +2,7 @@
 API to PAGE-XML, generated with generateDS from XML schema.
 """
 from io import StringIO
+import copy
 from typing import Dict, Union, Any
 from lxml import etree as ET
 from elementpath import XPath2Parser, XPathContext
@@ -212,6 +213,25 @@ class OcrdPage():
         self.xpath_context = XPathContext(self.etree)
         self.xpath = lambda expression: self.xpath_parser.parse(expression).get_results(self.xpath_context)
 
+    # allow copy() without infinite recursion
+    def __copy__(self):
+        return OcrdPage(
+            copy.copy(self._pcgts),
+            copy.copy(self.etree),
+            copy.copy(self.mapping),
+            copy.copy(self.revmap),
+        )
+
+    # allow deepcopy() without infinite recursion
+    def __deepcopy__(self, memo):
+        return OcrdPage(
+            copy.deepcopy(self._pcgts, memo),
+            copy.deepcopy(self.etree, memo),
+            copy.deepcopy(self.mapping, memo),
+            copy.deepcopy(self.revmap, memo),
+        )
+
+    # delegate to all members of ._pcgts
     def __getattr__(self, name):
         return getattr(self._pcgts, name)