ocrd 3.0.0b3__py3-none-any.whl → 3.0.0b5__py3-none-any.whl

ocrd/cli/bashlib.py

@@ -76,10 +76,10 @@ def bashlib_constants(name):
 @click.option('--ocrd-tool', help="path to ocrd-tool.json of processor to feed", default=None)
 @click.option('--executable', help="name of processor executable in ocrd-tool.json", default=None)
 @click.option('-m', '--mets', help="METS to process", default=DEFAULT_METS_BASENAME)
-@click.option('-w', '--working-dir', help="Working Directory")
+@click.option('-U', '--mets-server-url', help='TCP host URI or UDS path of METS server', default=None)
+@click.option('-d', '--working-dir', help="Working Directory")
 @click.option('-I', '--input-file-grp', help='File group(s) used as input.', default=None)
 @click.option('-O', '--output-file-grp', help='File group(s) used as output.', default=None)
-# repeat some other processor options for convenience (will be ignored here)
 @click.option('-g', '--page-id', help="ID(s) of the pages to process")
 @click.option('--overwrite', is_flag=True, default=False, help="Remove output pages/images if they already exist\n"
               "(with '--page-id', remove only those).\n"
@@ -126,9 +126,10 @@ def bashlib_input_files(ocrd_tool, executable, **kwargs):
             def executable(self):
                 # needed for ocrd_tool lookup
                 return executable
+        processor_class = FullBashlibProcessor
     else:
         # we have no true metadata file, so fill in just to make it work
-        class FullBashlibProcessor(BashlibProcessor):
+        class UnknownBashlibProcessor(BashlibProcessor):
             @property
             def ocrd_tool(self):
                 # needed to satisfy the validator
@@ -142,5 +143,6 @@ def bashlib_input_files(ocrd_tool, executable, **kwargs):
             def version(self):
                 # needed to satisfy the validator and wrapper
                 return '1.0'
+        processor_class = UnknownBashlibProcessor
 
-    ocrd_cli_wrap_processor(FullBashlibProcessor, **kwargs)
+    ocrd_cli_wrap_processor(processor_class, **kwargs)

ocrd/cli/ocrd_tool.py

@@ -125,7 +125,7 @@ def ocrd_tool_tool_list_resources(ctx):
 @click.argument('res_name')
 @pass_ocrd_tool
 def ocrd_tool_tool_resolve_resource(ctx, res_name):
-    ctx.processor(None).resolve_resource(res_name)
+    print(ctx.processor(None).resolve_resource(res_name))
 
 @ocrd_tool_tool.command('show-resource', help="Dump a tool's file resource")
 @click.argument('res_name')

ocrd/cli/validate.py

@@ -102,16 +102,19 @@ def validate_page(page, **kwargs):
 @validate_cli.command('tasks')
 @click.option('--workspace', nargs=1, required=False, help='Workspace directory these tasks are to be run. If omitted, only validate syntax')
 @click.option('-M', '--mets-basename', nargs=1, default=DEFAULT_METS_BASENAME, help='Basename of the METS file, used in conjunction with --workspace')
+@click.option('-U', '--mets-server-url', help='TCP host URI or UDS path of METS server')
 @click.option('--overwrite', is_flag=True, default=False, help='When checking against a concrete workspace, simulate overwriting output or page range.')
 @click.option('-g', '--page-id', help="ID(s) of the pages to process")
 @click.argument('tasks', nargs=-1, required=True)
-def validate_process(tasks, workspace, mets_basename, overwrite, page_id):
+def validate_process(tasks, workspace, mets_basename, mets_server_url, overwrite, page_id):
     '''
     Validate a sequence of tasks passable to `ocrd process`
     '''
     if workspace:
-        _inform_of_result(validate_tasks([ProcessorTask.parse(t) for t in tasks],
-            Workspace(Resolver(), directory=workspace, mets_basename=mets_basename), page_id=page_id, overwrite=overwrite))
+        _inform_of_result(validate_tasks(
+            [ProcessorTask.parse(t) for t in tasks],
+            Workspace(Resolver(), directory=workspace, mets_basename=mets_basename, mets_server_url=mets_server_url),
+            page_id=page_id, overwrite=overwrite))
     else:
         for t in [ProcessorTask.parse(t) for t in tasks]:
             _inform_of_result(t.validate())

ocrd/cli/workspace.py

@@ -36,6 +36,17 @@ class WorkspaceCtx():
                 = self.resolver.resolve_mets_arguments(directory, mets_url, mets_basename, mets_server_url)
         self.automatic_backup = automatic_backup
 
+    def workspace(self):
+        return Workspace(
+            self.resolver,
+            directory=self.directory,
+            mets_basename=self.mets_basename,
+            automatic_backup=self.automatic_backup,
+            mets_server_url=self.mets_server_url,
+        )
+    def backup_manager(self):
+        return WorkspaceBackupManager(self.workspace())
+
 
 pass_workspace = click.make_pass_decorator(WorkspaceCtx)
 
@@ -138,6 +149,7 @@ def workspace_clone(ctx, clobber_mets, download, file_grp, file_id, page_id, mim
         LOG.warning(DeprecationWarning("Use 'ocrd workspace --directory DIR clone' instead of argument 'WORKSPACE_DIR' ('%s')" % workspace_dir))
         ctx.directory = workspace_dir
 
+    assert not ctx.mets_server_url
     workspace = ctx.resolver.workspace_from_url(
         mets_url,
         dst_dir=ctx.directory,
@@ -173,10 +185,11 @@ def workspace_init(ctx, clobber_mets, directory):
     if directory:
         LOG.warning(DeprecationWarning("Use 'ocrd workspace --directory DIR init' instead of argument 'DIRECTORY' ('%s')" % directory))
         ctx.directory = directory
+    assert not ctx.mets_server_url
     workspace = ctx.resolver.workspace_from_nothing(
         directory=ctx.directory,
         mets_basename=ctx.mets_basename,
-        clobber_mets=clobber_mets
+        clobber_mets=clobber_mets,
     )
     workspace.save_mets()
     print(workspace.directory)
@@ -200,13 +213,7 @@ def workspace_add_file(ctx, file_grp, file_id, mimetype, page_id, ignore, check_
     Add a file or http(s) URL FNAME to METS in a workspace.
     If FNAME is not an http(s) URL and is not a workspace-local existing file, try to copy to workspace.
     """
-    workspace = Workspace(
-        ctx.resolver,
-        directory=ctx.directory,
-        mets_basename=ctx.mets_basename,
-        automatic_backup=ctx.automatic_backup,
-        mets_server_url=ctx.mets_server_url,
-    )
+    workspace = ctx.workspace()
 
     log = getLogger('ocrd.cli.workspace.add')
     if not mimetype:
@@ -313,13 +320,7 @@ def workspace_cli_bulk_add(ctx, regex, mimetype, page_id, file_id, url, local_fi
 
     """
     log = getLogger('ocrd.cli.workspace.bulk-add') # pylint: disable=redefined-outer-name
-    workspace = Workspace(
-        ctx.resolver,
-        directory=ctx.directory,
-        mets_basename=ctx.mets_basename,
-        automatic_backup=ctx.automatic_backup,
-        mets_server_url=ctx.mets_server_url,
-    )
+    workspace = ctx.workspace()
 
     try:
         pat = re.compile(regex)
@@ -455,12 +456,7 @@ def workspace_find(ctx, file_grp, mimetype, page_id, file_id, output_field, incl
     output_field = [snake_to_camel.get(x, x) for x in output_field]
     modified_mets = False
     ret = []
-    workspace = Workspace(
-        ctx.resolver,
-        directory=ctx.directory,
-        mets_basename=ctx.mets_basename,
-        mets_server_url=ctx.mets_server_url,
-    )
+    workspace = ctx.workspace()
     with pushd_popd(workspace.directory):
         for f in workspace.find_files(
                 file_id=file_id,
@@ -510,7 +506,7 @@ def workspace_remove_file(ctx, id, force, keep_file):  # pylint: disable=redefin
     (If any ``ID`` starts with ``//``, then its remainder
      will be interpreted as a regular expression.)
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+    workspace = ctx.workspace()
     for i in id:
         workspace.remove_file(i, force=force, keep_file=keep_file)
     workspace.save_mets()
@@ -528,7 +524,7 @@ def rename_group(ctx, old, new):
     """
     Rename fileGrp (USE attribute ``NEW`` to ``OLD``).
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
+    workspace = ctx.workspace()
     workspace.rename_file_group(old, new)
     workspace.save_mets()
 
@@ -549,7 +545,7 @@ def remove_group(ctx, group, recursive, force, keep_files):
     (If any ``GROUP`` starts with ``//``, then its remainder
      will be interpreted as a regular expression.)
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
+    workspace = ctx.workspace()
     for g in group:
         workspace.remove_file_group(g, recursive=recursive, force=force, keep_files=keep_files)
     workspace.save_mets()
@@ -571,7 +567,7 @@ def prune_files(ctx, file_grp, mimetype, page_id, file_id):
     (If any ``FILTER`` starts with ``//``, then its remainder
      will be interpreted as a regular expression.)
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+    workspace = ctx.workspace()
     with pushd_popd(workspace.directory):
         for f in workspace.find_files(
             file_id=file_id,
@@ -608,8 +604,7 @@ def clean(ctx, dry_run, directories, path_glob):
     If no PATH_GLOB are specified, then all files and directories
     may match.
     """
-    log = getLogger('ocrd.cli.workspace.clean')
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+    workspace = ctx.workspace()
     allowed_files = [normpath(f.local_filename) for f in workspace.find_files(local_only=True)]
     allowed_files.append(relpath(workspace.mets_target, start=workspace.directory))
     allowed_dirs = set(dirname(path) for path in allowed_files)
@@ -627,7 +622,7 @@ def clean(ctx, dry_run, directories, path_glob):
             if normpath(path) in allowed_files:
                 continue
             if dry_run:
-                log.info('unlink(%s)' % path)
+                ctx.log.info('unlink(%s)' % path)
             else:
                 unlink(path)
         if not directories:
@@ -637,7 +632,7 @@ def clean(ctx, dry_run, directories, path_glob):
             if normpath(path) in allowed_dirs:
                 continue
             if dry_run:
-                log.info('rmdir(%s)' % path)
+                ctx.log.info('rmdir(%s)' % path)
             else:
                 rmdir(path)
 
@@ -651,7 +646,7 @@ def list_groups(ctx):
     """
     List fileGrp USE attributes
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
+    workspace = ctx.workspace()
     print("\n".join(workspace.mets.file_groups))
 
 # ----------------------------------------------------------------------
@@ -677,7 +672,7 @@ def list_pages(ctx, output_field, output_format, chunk_number, chunk_index, page
     (If any ``FILTER`` starts with ``//``, then its remainder
      will be interpreted as a regular expression.)
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
+    workspace = ctx.workspace()
     find_kwargs = {}
     if page_id_range and 'ID' in output_field:
         find_kwargs['pageId'] = page_id_range
@@ -724,7 +719,7 @@ def get_id(ctx):
     """
     Get METS id if any
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename)
+    workspace = ctx.workspace()
     ID = workspace.mets.unique_identifier
     if ID:
         print(ID)
@@ -744,7 +739,7 @@ def set_id(ctx, id):   # pylint: disable=redefined-builtin
 
     Otherwise will create a new <mods:identifier type="purl">{{ ID }}</mods:identifier>.
     """
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+    workspace = ctx.workspace()
     workspace.mets.unique_identifier = id
     workspace.save_mets()
 
@@ -767,7 +762,7 @@ def update_page(ctx, attr_value_pairs, order, orderlabel, contentids, page_id):
     if contentids:
         update_kwargs['CONTENTIDS'] = contentids
     try:
-        workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+        workspace = ctx.workspace()
         workspace.mets.update_physical_page_attributes(page_id, **update_kwargs)
         workspace.save_mets()
     except Exception as err:
@@ -805,7 +800,7 @@ def merge(ctx, overwrite, force, copy_files, filegrp_mapping, fileid_mapping, pa
     mets_path = Path(mets_path)
     if filegrp_mapping:
         filegrp_mapping = loads(filegrp_mapping)
-    workspace = Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup)
+    workspace = ctx.workspace()
     other_workspace = Workspace(ctx.resolver, directory=str(mets_path.parent), mets_basename=str(mets_path.name))
     workspace.merge(
         other_workspace,
@@ -829,11 +824,12 @@ def merge(ctx, overwrite, force, copy_files, filegrp_mapping, fileid_mapping, pa
 # ----------------------------------------------------------------------
 
 @workspace_cli.group('backup')
-@click.pass_context
+@pass_workspace
 def workspace_backup_cli(ctx): # pylint: disable=unused-argument
     """
     Backing and restoring workspaces - dev edition
     """
+    assert not ctx.mets_server_url, "Workspace backups currently not interoperable with METS Server"
 
 @workspace_backup_cli.command('add')
 @pass_workspace
@@ -841,7 +837,7 @@ def workspace_backup_add(ctx):
     """
     Create a new backup
     """
-    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
+    backup_manager = ctx.backup_manager()
     backup_manager.add()
 
 @workspace_backup_cli.command('list')
@@ -850,7 +846,7 @@ def workspace_backup_list(ctx):
     """
     List backups
     """
-    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
+    backup_manager = ctx.backup_manager()
     for b in backup_manager.list():
         print(b)
 
@@ -862,7 +858,7 @@ def workspace_backup_restore(ctx, choose_first, bak):
     """
     Restore backup BAK
     """
-    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
+    backup_manager = ctx.backup_manager()
     backup_manager.restore(bak, choose_first)
 
 @workspace_backup_cli.command('undo')
@@ -871,7 +867,7 @@ def workspace_backup_undo(ctx):
     """
     Restore the last backup
     """
-    backup_manager = WorkspaceBackupManager(Workspace(ctx.resolver, directory=ctx.directory, mets_basename=ctx.mets_basename, automatic_backup=ctx.automatic_backup))
+    backup_manager = ctx.backup_manager()
     backup_manager.undo()
 
 
@@ -888,15 +884,24 @@ def workspace_serve_cli(ctx): # pylint: disable=unused-argument
 @workspace_serve_cli.command('stop')
 @pass_workspace
 def workspace_serve_stop(ctx): # pylint: disable=unused-argument
-    """Stop the METS server"""
-    workspace = Workspace(
-        ctx.resolver,
-        directory=ctx.directory,
-        mets_basename=ctx.mets_basename,
-        mets_server_url=ctx.mets_server_url,
-    )
+    """Stop the METS server (saving changes to disk)"""
+    workspace = ctx.workspace()
     workspace.mets.stop()
 
+@workspace_serve_cli.command('reload')
+@pass_workspace
+def workspace_serve_reload(ctx): # pylint: disable=unused-argument
+    """Reload the METS server from disk"""
+    workspace = ctx.workspace()
+    workspace.mets.reload()
+
+@workspace_serve_cli.command('save')
+@pass_workspace
+def workspace_serve_save(ctx): # pylint: disable=unused-argument
+    """Save the METS changes to disk"""
+    workspace = ctx.workspace()
+    workspace.mets.save()
+
 @workspace_serve_cli.command('start')
 @pass_workspace
 def workspace_serve_start(ctx): # pylint: disable=unused-argument

ocrd/decorators/ocrd_cli_options.py

@@ -43,6 +43,7 @@ def ocrd_cli_options(f):
         option('--address', type=ServerAddressParamType()),
         option('--queue', type=QueueServerParamType()),
         option('--database', type=DatabaseParamType()),
+        option('-R', '--resolve-resource'),
         option('-C', '--show-resource'),
         option('-L', '--list-resources', is_flag=True, default=False),
         option('-J', '--dump-json', is_flag=True, default=False),

ocrd/lib.bash

@@ -27,8 +27,8 @@ ocrd__log () {
 ## Ensure minimum version
 # ht https://stackoverflow.com/posts/4025065
 ocrd__minversion () {
-    local minversion_raw="$1"
     set -e
+    local minversion_raw="$1"
     local version_raw=$(ocrd --version|sed 's/ocrd, version //')
     local version_mmp=$(echo "$version_raw" | grep -Eo '([0-9]+\.?){3}')
     local version_prerelease_suffix="${version_raw#$version_mmp}"
@@ -123,6 +123,7 @@ ocrd__usage () {
 ## declare -A ocrd__argv=()
 ## ```
 ocrd__parse_argv () {
+    set -e
 
     # if [[ -n "$ZSH_VERSION" ]];then
     #     print -r -- ${+ocrd__argv} ${(t)ocrd__argv}
@@ -135,11 +136,16 @@ ocrd__parse_argv () {
         ocrd__raise "Must set \$params (declare -A params)"
     fi
 
+    if ! declare -p "params_json" >/dev/null 2>/dev/null ;then
+        ocrd__raise "Must set \$params_json (declare params_json)"
+    fi
+
     if [[ $# = 0 ]];then
         ocrd__usage
         exit 1
     fi
 
+    ocrd__argv[debug]=false
     ocrd__argv[overwrite]=false
     ocrd__argv[profile]=false
     ocrd__argv[profile_file]=
@@ -170,6 +176,7 @@ ocrd__parse_argv () {
             -w|--working-dir) ocrd__argv[working_dir]=$(realpath "$2") ; shift ;;
             -m|--mets) ocrd__argv[mets_file]=$(realpath "$2") ; shift ;;
             -U|--mets-server-url) ocrd__argv[mets_server_url]="$2" ; shift ;;
+            --debug) ocrd__argv[debug]=true ;;
             --overwrite) ocrd__argv[overwrite]=true ;;
             --profile) ocrd__argv[profile]=true ;;
             --profile-file) ocrd__argv[profile_file]=$(realpath "$2") ; shift ;;
@@ -242,17 +249,6 @@ ocrd__parse_argv () {
         trap showtime DEBUG
     fi
 
-    # check fileGrps
-    local _valopts=( --workspace "${ocrd__argv[working_dir]}" --mets-basename "$(basename ${ocrd__argv[mets_file]})" )
-    if [[ ${ocrd__argv[overwrite]} = true ]]; then
-        _valopts+=( --overwrite )
-    fi
-    if [[ -n "${ocrd__argv[page_id]:-}" ]]; then
-        _valopts+=( --page-id "${ocrd__argv[page_id]}" )
-    fi
-    _valopts+=( "${OCRD_TOOL_NAME#ocrd-} -I ${ocrd__argv[input_file_grp]} -O ${ocrd__argv[output_file_grp]} ${__parameters[*]@Q} ${__parameter_overrides[*]@Q}" )
-    ocrd validate tasks "${_valopts[@]}" || exit $?
-
     # check parameters
     local params_parsed retval
     params_parsed="$(ocrd ocrd-tool "$OCRD_TOOL_JSON" tool $OCRD_TOOL_NAME parse-params "${__parameters[@]}" "${__parameter_overrides[@]}")" || {
@@ -261,10 +257,12 @@ ocrd__parse_argv () {
 $params_parsed"
     }
     eval "$params_parsed"
+    params_json="$(ocrd ocrd-tool "$OCRD_TOOL_JSON" tool $OCRD_TOOL_NAME parse-params --json "${__parameters[@]}" "${__parameter_overrides[@]}")"
 
 }
 
 ocrd__wrap () {
+    set -e
 
     declare -gx OCRD_TOOL_JSON="$1"
     declare -gx OCRD_TOOL_NAME="$2"
@@ -272,6 +270,7 @@ ocrd__wrap () {
     shift
     declare -Agx params
     params=()
+    declare -g params_json
     declare -Agx ocrd__argv
     ocrd__argv=()
 
@@ -293,22 +292,26 @@ ocrd__wrap () {
 
     ocrd__parse_argv "$@"
 
-    i=0
-    declare -ag ocrd__files=()
-    while read line; do
-        eval declare -Ag "ocrd__file$i=( $line )"
-        eval "ocrd__files[$i]=ocrd__file$i"
-        let ++i
-    done < <(ocrd bashlib input-files \
+    declare -ag ocrd__files
+    IFS=$'\n'
+    ocrd__files=( $(ocrd bashlib input-files \
                   --ocrd-tool $OCRD_TOOL_JSON \
                   --executable $OCRD_TOOL_NAME \
+                  $(if [[ ${ocrd__argv[debug]} = true ]]; then echo --debug; fi) \
+                  $(if [[ ${ocrd__argv[overwrite]} = true ]]; then echo --overwrite; fi) \
                   -m "${ocrd__argv[mets_file]}" \
+                  -d "${ocrd__argv[working_dir]}" \
+                  ${ocrd__argv[mets_server_url]:+-U} ${ocrd__argv[mets_server_url]:-} \
+                  -p "$params_json" \
                   -I "${ocrd__argv[input_file_grp]}" \
                   -O "${ocrd__argv[output_file_grp]}" \
-                  ${ocrd__argv[page_id]:+-g} ${ocrd__argv[page_id]:-})
+                  ${ocrd__argv[page_id]:+-g} ${ocrd__argv[page_id]:-}) )
+    IFS=$' \t\n'
 }
 
 ## usage: pageId=$(ocrd__input_file 3 pageId)
 ocrd__input_file() {
-    eval echo "\${${ocrd__files[$1]}[$2]}"
+    declare -A input_file
+    eval input_file=( "${ocrd__files[$1]}" )
+    eval echo "${input_file[$2]}"
 }

ocrd/mets_server.py

@@ -88,6 +88,14 @@ class OcrdFileGroupListModel(BaseModel):
         return OcrdFileGroupListModel(file_groups=file_groups)
 
 
+class OcrdPageListModel(BaseModel):
+    physical_pages: List[str] = Field()
+
+    @staticmethod
+    def create(physical_pages: List[str]):
+        return OcrdPageListModel(physical_pages=physical_pages)
+
+
 class OcrdAgentListModel(BaseModel):
     agents: List[OcrdAgentModel] = Field()
 
@@ -210,6 +218,17 @@ class ClientSideOcrdMets:
             ).json()["text"]
             return self.ws_dir_path
 
+    @property
+    def physical_pages(self) -> List[str]:
+        if not self.multiplexing_mode:
+            return self.session.request("GET", f"{self.url}/physical_pages").json()["physical_pages"]
+        else:
+            return self.session.request(
+                "POST",
+                self.url,
+                json=MpxReq.physical_pages(self.ws_dir_path)
+            ).json()["physical_pages"]
+
     @property
     def file_groups(self):
         if not self.multiplexing_mode:
@@ -284,15 +303,17 @@ class ClientSideOcrdMets:
             file_id=ID, page_id=pageId,
             mimetype=mimetype, url=url, local_filename=local_filename
         )
+        # add force+ignore
+        kwargs = {**kwargs, **data.dict()}
 
         if not self.multiplexing_mode:
-            r = self.session.request("POST", f"{self.url}/file", data=data.dict())
-            if not r:
-                raise RuntimeError("Add file failed. Please check provided parameters")
+            r = self.session.request("POST", f"{self.url}/file", data=kwargs)
+            if not r.ok:
+                raise RuntimeError(f"Failed to add file ({str(data)}): {r.json()}")
         else:
-            r = self.session.request("POST", self.url, json=MpxReq.add_file(self.ws_dir_path, data.dict()))
-            if "error" in r:
-                raise RuntimeError(f"Add file failed: Msg: {r['error']}")
+            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]}")
 
         return ClientSideOcrdFile(
             None, fileGrp=file_grp,
@@ -347,6 +368,11 @@ class MpxReq:
         return MpxReq.__args_wrapper(
             ws_dir_path, method_type="GET", response_type="text", request_url="workspace_path", request_data={})
 
+    @staticmethod
+    def physical_pages(ws_dir_path: str) -> Dict:
+        return MpxReq.__args_wrapper(
+            ws_dir_path, method_type="GET", response_type="dict", request_url="physical_pages", request_data={})
+
     @staticmethod
     def file_groups(ws_dir_path: str) -> Dict:
         return MpxReq.__args_wrapper(
@@ -466,6 +492,10 @@ class OcrdMetsServer:
         async def workspace_path():
             return Response(content=workspace.directory, media_type="text/plain")
 
+        @app.get(path='/physical_pages', response_model=OcrdPageListModel)
+        async def physical_pages():
+            return {'physical_pages': workspace.mets.physical_pages}
+
         @app.get(path='/file_groups', response_model=OcrdFileGroupListModel)
         async def file_groups():
             return {'file_groups': workspace.mets.file_groups}
@@ -505,7 +535,8 @@ class OcrdMetsServer:
             page_id: Optional[str] = Form(),
             mimetype: str = Form(),
             url: Optional[str] = Form(None),
-            local_filename: Optional[str] = Form(None)
+            local_filename: Optional[str] = Form(None),
+            force: bool = Form(False),
         ):
             """
             Add a file
@@ -517,7 +548,7 @@ class OcrdMetsServer:
             )
             # Add to workspace
             kwargs = file_resource.dict()
-            workspace.add_file(**kwargs)
+            workspace.add_file(**kwargs, force=force)
             return file_resource
 
         # ------------- #

ocrd/processor/base.py

@@ -16,7 +16,7 @@ import json
 import os
 from os import getcwd
 from pathlib import Path
-from typing import List, Optional, Union, get_args
+from typing import Any, List, Optional, Union, get_args
 import sys
 import inspect
 import tarfile
@@ -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:
@@ -336,7 +339,7 @@ class Processor():
         self._finalizer = weakref.finalize(self, self.shutdown)
         # workaround for deprecated#72 (@deprecated decorator does not work for subclasses):
         setattr(self, 'process',
-                deprecated(version='3.0', reason='process() should be replaced with process_page() and process_workspace()')(getattr(self, 'process')))
+                deprecated(version='3.0', reason='process() should be replaced with process_page_pcgts() or process_page_file() or process_workspace()')(getattr(self, 'process')))
 
     def show_help(self, subcommand=None):
         """
@@ -355,6 +358,7 @@ class Processor():
         """
         Verify that :py:attr:`input_file_grp` and :py:attr:`output_file_grp` fulfill the processor's requirements.
         """
+        # verify input and output file groups in parameters
         assert self.input_file_grp is not None
         assert self.output_file_grp is not None
         input_file_grps = self.input_file_grp.split(',')
@@ -371,12 +375,23 @@ class Processor():
                     assert len(grps) >= minimum, msg % (len(grps), str(spec))
                 if maximum > 0:
                     assert len(grps) <= maximum, msg % (len(grps), str(spec))
-        assert_file_grp_cardinality(input_file_grps, self.ocrd_tool['input_file_grp_cardinality'],
-                                    "Unexpected number of input file groups %d vs %s")
-        assert_file_grp_cardinality(output_file_grps, self.ocrd_tool['output_file_grp_cardinality'],
-                                    "Unexpected number of output file groups %d vs %s")
+        # FIXME: enforce unconditionally as soon as grace period for deprecation is over
+        if 'input_file_grp_cardinality' in self.ocrd_tool:
+            assert_file_grp_cardinality(input_file_grps, self.ocrd_tool['input_file_grp_cardinality'],
+                                        "Unexpected number of input file groups %d vs %s")
+        if 'output_file_grp_cardinality' in self.ocrd_tool:
+            assert_file_grp_cardinality(output_file_grps, self.ocrd_tool['output_file_grp_cardinality'],
+                                        "Unexpected number of output file groups %d vs %s")
+        # verify input and output file groups in METS
         for input_file_grp in input_file_grps:
-            assert input_file_grp in self.workspace.mets.file_groups
+            assert input_file_grp in self.workspace.mets.file_groups, \
+                f"input fileGrp {input_file_grp} does not exist in workspace {self.workspace}"
+        for output_file_grp in output_file_grps:
+            assert output_file_grp not in self.workspace.mets.file_groups \
+                or config.OCRD_EXISTING_OUTPUT in ['OVERWRITE', 'SKIP'] \
+                or not any(self.workspace.mets.find_files(
+                    pageId=self.page_id, fileGrp=output_file_grp)), \
+                    f"output fileGrp {output_file_grp} already exists in workspace {self.workspace}"
         # keep this for backwards compatibility:
         return True
 
@@ -455,17 +470,17 @@ 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}"
@@ -543,7 +558,11 @@ class Processor():
 
             except NotImplementedError:
                 # fall back to deprecated method
-                self.process()
+                try:
+                    self.process()
+                except Exception as err:
+                    # suppress the NotImplementedError context
+                    raise err from None
 
     def _copy_page_file(self, input_file : OcrdFileType) -> None:
         """
@@ -571,7 +590,6 @@ class Processor():
             local_filename=os.path.join(self.output_file_grp, output_file_id + '.xml'),
             mimetype=MIMETYPE_PAGE,
             content=to_xml(input_pcgts),
-            force=config.OCRD_EXISTING_OUTPUT == 'OVERWRITE',
         )
 
     def process_page_file(self, *input_files : Optional[OcrdFileType]) -> None:
@@ -613,6 +631,8 @@ class Processor():
                 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)}")
@@ -622,7 +642,6 @@ class Processor():
                 self.output_file_grp,
                 page_id=page_id,
                 file_path=image_file_path,
-                force=config.OCRD_EXISTING_OUTPUT == 'OVERWRITE',
             )
         result.pcgts.set_pcGtsId(output_file_id)
         self.add_metadata(result.pcgts)
@@ -633,7 +652,6 @@ class Processor():
             local_filename=os.path.join(self.output_file_grp, output_file_id + '.xml'),
             mimetype=MIMETYPE_PAGE,
             content=to_xml(result.pcgts),
-            force=config.OCRD_EXISTING_OUTPUT == 'OVERWRITE',
         )
 
     def process_page_pcgts(self, *input_pcgts : Optional[OcrdPage], page_id : Optional[str] = None) -> OcrdPageResult:

ocrd/processor/builtin/dummy_processor.py

@@ -47,7 +47,6 @@ class DummyProcessor(Processor):
                     mimetype=input_file.mimetype,
                     local_filename=local_filename,
                     content=f.read(),
-                    force=config.OCRD_EXISTING_OUTPUT == 'OVERWRITE',
                 )
             file_id = file_id + '_PAGE'
             pcgts = page_from_file(output_file)
@@ -62,7 +61,6 @@ class DummyProcessor(Processor):
                                     local_filename=join(self.output_file_grp, file_id + '.xml'),
                                     mimetype=MIMETYPE_PAGE,
                                     content=to_xml(pcgts),
-                                    force=config.OCRD_EXISTING_OUTPUT == 'OVERWRITE',
             )
         else:
             if self.parameter['copy_files']:

ocrd/processor/helpers.py

@@ -89,7 +89,7 @@ def run_processor(
 
     ocrd_tool = processor.ocrd_tool
     name = '%s v%s' % (ocrd_tool['executable'], processor.version)
-    otherrole = ocrd_tool['steps'][0]
+    otherrole = ocrd_tool.get('steps', [''])[0]
     logProfile = getLogger('ocrd.process.profile')
     log.debug("Processor instance %s (%s doing %s)", processor, name, otherrole)
     t0_wall = perf_counter()
@@ -234,10 +234,10 @@ def get_cached_processor(parameter: dict, processor_class):
 def get_processor(
         processor_class,
         parameter: Optional[dict] = None,
-        workspace: Workspace = None,
-        page_id: str = None,
-        input_file_grp: List[str] = None,
-        output_file_grp: List[str] = None,
+        workspace: Optional[Workspace] = None,
+        page_id: Optional[str] = None,
+        input_file_grp: Optional[List[str]] = None,
+        output_file_grp: Optional[List[str]] = None,
         instance_caching: bool = False,
 ):
     if processor_class:
@@ -258,6 +258,7 @@ def get_processor(
         else:
             # avoid passing workspace already (deprecated chdir behaviour)
             processor = processor_class(None, parameter=parameter)
+        assert processor
         # set current processing parameters
         processor.workspace = workspace
         processor.page_id = page_id

ocrd/processor/ocrd_page_result.py

@@ -1,5 +1,5 @@
 from dataclasses import dataclass, field
-from typing import List, Union
+from typing import List, Union, Optional
 from ocrd_models.ocrd_page import OcrdPage
 from PIL.Image import Image
 
@@ -9,7 +9,7 @@ from ocrd_models.ocrd_page_generateds import AlternativeImageType, PageType
 class OcrdPageResultImage():
     pil : Image
     file_id_suffix : str
-    alternative_image : Union[AlternativeImageType, PageType]
+    alternative_image : Optional[Union[AlternativeImageType, PageType]]
 
 @dataclass
 class OcrdPageResult():

ocrd/workspace.py

@@ -19,6 +19,7 @@ from ocrd_models.ocrd_page import parse, BorderType, to_xml
 from ocrd_modelfactory import exif_from_filename, page_from_file
 from ocrd_utils import (
     atomic_write,
+    config,
     getLogger,
     image_from_polygon,
     coordinates_of_segment,
@@ -121,7 +122,10 @@ 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")
@@ -424,6 +428,8 @@ class Workspace():
             kwargs["pageId"] = kwargs.pop("page_id")
             if "file_id" in kwargs:
                 kwargs["ID"] = kwargs.pop("file_id")
+            if config.OCRD_EXISTING_OUTPUT == 'OVERWRITE':
+                kwargs["force"] = True
 
             ret = self.mets.add_file(file_grp, **kwargs)
 

{ocrd-3.0.0b3.dist-info → ocrd-3.0.0b5.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: ocrd
-Version: 3.0.0b3
+Version: 3.0.0b5
 Summary: OCR-D framework
 Author-email: Konstantin Baierer <unixprog@gmail.com>
 License: Apache License 2.0

{ocrd-3.0.0b3.dist-info → ocrd-3.0.0b5.dist-info}/RECORD

@@ -1,36 +1,36 @@
 ocrd/__init__.py,sha256=ZswMVmlqFhAEIzMR3my6IKPq9XLH21aDPC_m_8Jh4dA,1076
 ocrd/constants.py,sha256=6dn3mG54WqHsKInmLZp4kJjNqqPtBoFoSuLUuRbOps0,740
-ocrd/lib.bash,sha256=dlOQd36OVsibwEPXg7i-e_JmIc-3uH-SIwGcE8OJ3Cc,10744
-ocrd/mets_server.py,sha256=SrCnCmWBuyEauHrNF0jLZO_Wi8WoPCMWxhec2wE5Y3w,20160
+ocrd/lib.bash,sha256=le6XqAOEacdjP3JNSlPkxwRH1y0oVjNQM2tX5d6QFO4,10901
+ocrd/mets_server.py,sha256=U62eih1_O_N0StunVFkEustFs2PlrcMzccraj6_QRk4,21295
 ocrd/ocrd-all-tool.json,sha256=9bX2VYnUwhTAzAvKaoT77BFzbgBGgyIt7qBqARpwWNc,586
 ocrd/resolver.py,sha256=Ba9ALQbTXz6_mla4VqN9tAfHoj6aKuNJAU4tIDnjcHE,14952
 ocrd/resource_list.yml,sha256=82-PiqkZnka1kTj3MQqNn4wXWKHHtoFchsQuetWuqFs,2633
 ocrd/resource_manager.py,sha256=8BMVKJq8J56hugi8vtGn9Ffuk7oRkbs197aG74aKbCY,16733
 ocrd/task_sequence.py,sha256=spiaUQaMM7M8WdBDoQGmLuTPm7tOugYXD6rcJ2UXzxw,6991
-ocrd/workspace.py,sha256=4s0qscEosS7rQ0jfn1qJeT9B3eC31YippAX-RUjXghA,65608
+ocrd/workspace.py,sha256=cedqK7es2i2nwQCiUiVyWk3j4-nH7bsi6TF7v8siTio,65794
 ocrd/workspace_backup.py,sha256=iab_JjZ_mMP-G8NIUk4PZmfpNlQuGRoqc3NbTSSew1w,3621
 ocrd/workspace_bagger.py,sha256=yU8H3xR5WmQKvgQewac71ie-DUWcfLnMS01D55zsEHQ,11971
 ocrd/cli/__init__.py,sha256=lNR6wMf7JhQ8Jf33tUkowJr0mB3423OMY0_6dkMRLvU,2672
-ocrd/cli/bashlib.py,sha256=XGcO-MmYM3xJBRkSCLEZcGs0hqbw2GR8oyijJPtKnYM,5888
+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=Sqh9Q5-H3u5IPPovp9bpL-QaP01zAHKZDSsi5MhvnQ8,7028
+ocrd/cli/ocrd_tool.py,sha256=qaJgt-LNH0tXkaupMNrEKXasxcgsabHdfLdYESEsomk,7035
 ocrd/cli/process.py,sha256=8KD0i7LT01H9u5CC1vktYMEVpS67da_rp_09_EOECmw,1233
 ocrd/cli/resmgr.py,sha256=bTE-MpF7RRCHhgAbknqZUFHgHScIK6FR3S4h4DEAets,10080
-ocrd/cli/validate.py,sha256=s5GtioCtO0UfRmYi_tdxNkKx0bJIzdAXGMCPA-PTbto,5563
-ocrd/cli/workspace.py,sha256=ETx3qjqUIXJp-Addv8eYmI2T_Js4PJR5UuZ4O6H7mis,40640
+ocrd/cli/validate.py,sha256=nvageDaHCETcE71X5lu7i_4JKpgo9MrvJKinVPLYUTI,5727
+ocrd/cli/workspace.py,sha256=InIQ5rfQWPn4Qsd1s_xA6AC6ndZLCsuyhoAEiqP8bK4,39479
 ocrd/cli/zip.py,sha256=MMJLw3OXWiJVfVtrdJcBkbB8vA1IzSautluazZRuCQ0,5910
 ocrd/decorators/__init__.py,sha256=IJlA1XcdVBO6Hxm9rNDya7QYcqeWcaXXuLtGjfjcen8,7596
 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=4pcBLAFPSpYZLj6r9Yj1GZOQl4r_RWU00pyA4mHwFQk,2621
+ocrd/decorators/ocrd_cli_options.py,sha256=hr2EugwAY_-GJ7F7g77Od9o9eAqhfLBHSpfmCql2OCU,2665
 ocrd/decorators/parameter_option.py,sha256=n8hYw7XVTd3i3tvpK8F1Jx_CqRp6EGF9qJVH95yj92Q,1076
 ocrd/processor/__init__.py,sha256=39ymNwYRdc-b_OJzzKmWCvo2ga3KdsGSYDHE1Hzkn_w,274
-ocrd/processor/base.py,sha256=341APZGx6zCbuxgX-XTkKhPfeQkqblykmC9zSMPH3ss,48843
-ocrd/processor/helpers.py,sha256=Lp9zbHYCLpT3GnPzl-p7UCSFU5Nx99gYEYXwW04v0RI,10157
-ocrd/processor/ocrd_page_result.py,sha256=AazEmnWyPEN47TxXVg0WUQpgFNV_mlIiExwwycUj0nQ,490
+ocrd/processor/base.py,sha256=_TvaxKf_oMaIfuUcaFWas8YimhZ-l1d3RWGELlBJfy8,50307
+ocrd/processor/helpers.py,sha256=vPYUri6ucuhdTNrideywriJ0fCa8UE2QyBXOmS-7RcQ,10232
+ocrd/processor/ocrd_page_result.py,sha256=eDkpyVHcpaBzTHXiGrcNk9PP9Xr-XZru2w_uoX_ZeNA,510
 ocrd/processor/builtin/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-ocrd/processor/builtin/dummy_processor.py,sha256=6ORike_59wb_UUivhA6Iw_Ldg0AaMzX37F7qL9R8S_A,3704
+ocrd/processor/builtin/dummy_processor.py,sha256=iWiw_jJXOqwr7-hFjdkmTCCo1xGr6MLGOshx81PTu-8,3548
 ocrd/processor/builtin/dummy/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 ocrd/processor/builtin/dummy/ocrd-tool.json,sha256=aTA2FZRsRsrkbTctkazFeRu4xsTF6yCdeY07cMzOyt4,677
 ocrd_modelfactory/__init__.py,sha256=0baYSJXrOCTCguHkE6hBeqpGNVUe3aZUocv64A-DMDk,4094
@@ -42,7 +42,7 @@ ocrd_models/ocrd_exif.py,sha256=5BRLjvB6jg36V68i8jvVnT2SSNnpqLbhLsaMuP51Scw,4583
 ocrd_models/ocrd_file.py,sha256=7lyHezuNnl2FEYV1lV35-QTCrgYAL-3wO2ulFUNq2Ak,9717
 ocrd_models/ocrd_mets.py,sha256=h3y_WI5fVLsbBoUIRNH2ebjuO1-_P6T3BMIULX-ZOIs,42514
 ocrd_models/ocrd_page.py,sha256=sVIvvMeBT8eZnOfW0DTjQUNyu62-llz0v_Ga5Xo-tUM,5393
-ocrd_models/ocrd_page_generateds.py,sha256=XpzRaJAvGc_N58Mb2_Sk42S59pDqMRFiplqce1pUubw,772504
+ocrd_models/ocrd_page_generateds.py,sha256=wfx3vESMAi08rl6-16zNVJe4E3B6APIvL6RCr1roAzg,774092
 ocrd_models/ocrd_xml_base.py,sha256=OW57mXLlwm1nH8CNefvXmwLRws9KL9zSrb-3vH--mX8,1641
 ocrd_models/report.py,sha256=luZxvzAAQyGYOlRNSJQUIUIANG81iGmBW5ag-uXxKCA,2026
 ocrd_models/utils.py,sha256=0_WHf5NEn1WC8MKJc6X_RK8gW-70Z09_mslkKOj7uF8,2369
@@ -95,7 +95,7 @@ ocrd_utils/introspect.py,sha256=gfBlmeEFuRmRUSgdSK0jOxRpYqDRXl2IAE6gv2MZ6as,1977
 ocrd_utils/logging.py,sha256=5_-5T5OWSYicNk8SQyjVqdRj2bVl-gDK1Th-C7oW_HE,8248
 ocrd_utils/ocrd_logging.conf,sha256=kl9x9JS1d8h8F0QZabvrjZtW1iApIaChvkImYafKO5g,3623
 ocrd_utils/os.py,sha256=acRRdDBI8L6BK0Mf773yKEzwdpZSFRBJEKB2crL4EjU,9865
-ocrd_utils/str.py,sha256=JIhsyWphqJuxJAzhRQJUqlZ44AGOeObEPJMFhfWhfhQ,10084
+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
 ocrd_validators/constants.py,sha256=FLP57T3F39weka_XovG40RgVMW1GunnbK04QRQ9tmlE,1802
@@ -118,9 +118,9 @@ 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-3.0.0b3.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
-ocrd-3.0.0b3.dist-info/METADATA,sha256=WZhPkJV0F8A5k-0IVK8HZ5zGWVWwYSa6FuDlpkuh4Xc,10397
-ocrd-3.0.0b3.dist-info/WHEEL,sha256=UvcQYKBHoFqaQd6LKyqHw9fxEolWLQnlzP0h_LgJAfI,91
-ocrd-3.0.0b3.dist-info/entry_points.txt,sha256=tV_gAdO8cbnOjS0GmKfJKbN60xBAV2DQRX6hEjleSjE,94
-ocrd-3.0.0b3.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
-ocrd-3.0.0b3.dist-info/RECORD,,
+ocrd-3.0.0b5.dist-info/LICENSE,sha256=xx0jnfkXJvxRnG63LTGOxlggYnIysveWIZ6H3PNdCrQ,11357
+ocrd-3.0.0b5.dist-info/METADATA,sha256=eIpkAoobj7QocP9VXYASLqE82wN35JlvYuYjSBLk30o,10397
+ocrd-3.0.0b5.dist-info/WHEEL,sha256=5Mi1sN9lKoFv_gxcPtisEVrJZihrm_beibeg5R6xb4I,91
+ocrd-3.0.0b5.dist-info/entry_points.txt,sha256=tV_gAdO8cbnOjS0GmKfJKbN60xBAV2DQRX6hEjleSjE,94
+ocrd-3.0.0b5.dist-info/top_level.txt,sha256=pUgiN42t4KXC5rvpi6V8atza31XP4SCznXpXlVlvomM,75
+ocrd-3.0.0b5.dist-info/RECORD,,

{ocrd-3.0.0b3.dist-info → ocrd-3.0.0b5.dist-info}/WHEEL

@@ -1,5 +1,5 @@
 Wheel-Version: 1.0
-Generator: setuptools (74.0.0)
+Generator: setuptools (75.0.0)
 Root-Is-Purelib: true
 Tag: py3-none-any
 

ocrd_models/ocrd_page_generateds.py

@@ -2,30 +2,28 @@
 # -*- coding: utf-8 -*-
 
 #
-# Generated Wed Nov  3 12:30:32 2021 by generateDS.py version 2.35.20.
-# Python 3.6.9 (default, Jan 26 2021, 15:33:00)  [GCC 8.4.0]
+# Generated Sat Sep  7 14:17:39 2024 by generateDS.py version 2.35.20.
+# Python 3.8.17+ (heads/3.8-dirty:1663f8ba84, Aug 15 2023, 18:13:01)  [GCC 8.3.0]
 #
 # Command line options:
 #   ('-f', '')
 #   ('--root-element', 'PcGts')
-#   ('-o', 'ocrd_models/ocrd_models/ocrd_page_generateds.py')
+#   ('-o', 'src/ocrd_models/ocrd_page_generateds.py')
 #   ('--silence', '')
 #   ('--export', 'write etree')
 #   ('--disable-generatedssuper-lookup', '')
-#   ('--user-methods', 'ocrd_models/ocrd_page_user_methods.py')
+#   ('--user-methods', 'src/ocrd_page_user_methods.py')
 #
 # Command line arguments:
-#   ocrd_validators/ocrd_validators/page.xsd
+#   src/ocrd_validators/page.xsd
 #
 # Command line:
-#   /home/kba/monorepo/ocrd_all/venv/bin/generateDS -f --root-element="PcGts" -o "ocrd_models/ocrd_models/ocrd_page_generateds.py" --silence --export="write etree" --disable-generatedssuper-lookup --user-methods="ocrd_models/ocrd_page_user_methods.py" ocrd_validators/ocrd_validators/page.xsd
+#   /data/ocr-d/ocrd_all/venv38/bin/generateDS -f --root-element="PcGts" -o "src/ocrd_models/ocrd_page_generateds.py" --silence --export="write etree" --disable-generatedssuper-lookup --user-methods="src/ocrd_page_user_methods.py" src/ocrd_validators/page.xsd
 #
 # Current working directory (os.getcwd()):
 #   core
 #
 
-# type: ignore
-
 from itertools import zip_longest
 import os
 import sys
@@ -223,7 +221,7 @@ class GeneratedsSuper(object):
             try:
                 int(value)
             except (TypeError, ValueError):
-                raise_parse_error(node, 'Requires sequence of integer values')
+                raise_parse_error(node, 'Requires sequence of integer valuess')
         return values
     def gds_format_float(self, input_data, input_name=''):
         return ('%.15f' % input_data).rstrip('0')
@@ -1230,9 +1228,10 @@ class PcGtsType(GeneratedsSuper):
         return hash(self.id)
     @property
     def id(self):
+        from ocrd_utils import make_xml_id
         if hasattr(self, 'pcGtsId'):
             return self.pcGtsId or ''
-        return self.imageFilename
+        return make_xml_id(self.imageFilename)
     def get_AllAlternativeImagePaths(self, page=True, region=True, line=True, word=True, glyph=True):
         """
         Get all the ``pc:AlternativeImage/@filename`` paths referenced in the PAGE-XML document.
@@ -3116,9 +3115,10 @@ class PageType(GeneratedsSuper):
         return hash(self.id)
     @property
     def id(self):
+        from ocrd_utils import make_xml_id
         if hasattr(self, 'pcGtsId'):
             return self.pcGtsId or ''
-        return self.imageFilename
+        return make_xml_id(self.imageFilename)
     # pylint: disable=line-too-long,invalid-name,protected-access,missing-module-docstring
     def _region_class(self, x): # pylint: disable=unused-argument
         return x.__class__.__name__.replace('RegionType', '')
@@ -3314,6 +3314,39 @@ class PageType(GeneratedsSuper):
                 ret += lines if lo in ['top-to-bottom', 'left-to-right'] else list(reversed(lines))
         return ret
     
+    def get_ReadingOrderGroups(self) -> dict:
+        """
+        Aggregate recursive ReadingOrder into a dictionary, mapping each regionRef
+        (i.e. segment `@id`) to its referring group object (i.e one of
+    
+        \b
+        - :py:class:`.RegionRefType`
+        - :py:class:`.RegionRefIndexedType`
+        - :py:class:`.OrderedGroupType`
+        - :py:class:`.OrderedGroupIndexedType`
+        - :py:class:`.UnoderedGroupType`
+        - :py:class:`.UnoderedGroupIndexedType`
+        """
+        def get_groupdict(group):
+            regionrefs = list()
+            if isinstance(group, (OrderedGroupType, OrderedGroupIndexedType)):
+                regionrefs = (group.get_RegionRefIndexed() +
+                              group.get_OrderedGroupIndexed() +
+                              group.get_UnorderedGroupIndexed())
+            if isinstance(group, (UnorderedGroupType, UnorderedGroupIndexedType)):
+                regionrefs = (group.get_RegionRef() +
+                              group.get_OrderedGroup() +
+                              group.get_UnorderedGroup())
+            refdict = {}
+            for elem in regionrefs:
+                refdict[elem.get_regionRef()] = elem
+                if not isinstance(elem, (RegionRefType, RegionRefIndexedType)):
+                    refdict = {**refdict, **get_groupdict(elem)}
+            return refdict
+        ro = self.get_ReadingOrder()
+        if ro is None:
+            return {}
+        return get_groupdict(ro.get_OrderedGroup() or ro.get_UnorderedGroup())
     def set_orientation(self, orientation):
         """
         Set deskewing angle to given `orientation` number.

ocrd_utils/str.py

@@ -108,10 +108,11 @@ def make_xml_id(idstr: str) -> str:
     ret = idstr
     if not REGEX_FILE_ID.fullmatch(ret):
         ret = ret.replace(':', '_')
+        ret = ret.replace('/', '_')
         ret = re.sub(r'^([^a-zA-Z_])', r'id_\1', ret)
         ret = re.sub(r'[^\w.-]', r'', ret)
     return ret
-    
+
 def nth_url_segment(url, n=-1):
     """
     Return the last /-delimited segment of a URL-like string