ocrd_models.ocrd_mets.OcrdMets.add_file()

Complexity

Conditions

12

Size

Total Lines	54
Code Lines	31

Duplication

Lines	0
Ratio	0 %

Importance

Changes

0

"""
    def add_file(self, fileGrp : str, mimetype : Optional[str] = None, url : Optional[str] = None, 
                 ID : Optional[str] = None, pageId : Optional[str] = None, force : bool = False, 
                 local_filename : Optional[str] = None, ignore : bool = False, **kwargs) -> OcrdFile:
        """
        Instantiate and add a new :py:class:`ocrd_models.ocrd_file.OcrdFile`.
        Arguments:
            fileGrp (string): ``@USE`` of ``mets:fileGrp`` to add to
        Keyword Args:
            mimetype (string): ``@MIMETYPE`` of the ``mets:file`` to use
            url (string): ``@xlink:href`` (URL or path) of the ``mets:file`` to use
            ID (string): ``@ID`` of the ``mets:file`` to use
            pageId (string): ``@ID`` in the physical ``mets:structMap`` to link to
            force (boolean): Whether to add the file even if a ``mets:file`` with the same ``@ID`` already exists.
            ignore (boolean): Do not look for existing files at all. Shift responsibility for preventing errors from duplicate ID to the user.
            local_filename (string):
        """
        if not ID:
            raise ValueError("Must set ID of the mets:file")
        if not fileGrp:
            raise ValueError("Must set fileGrp of the mets:file")
        if not REGEX_FILE_ID.fullmatch(ID):
            raise ValueError("Invalid syntax for mets:file/@ID %s (not an xs:ID)" % ID)
        if not REGEX_FILE_ID.fullmatch(fileGrp):
            raise ValueError("Invalid syntax for mets:fileGrp/@USE %s (not an xs:ID)" % fileGrp)

        el_fileGrp = self.add_file_group(fileGrp)
        if not ignore:
            mets_file = next(self.find_files(ID=ID, fileGrp=fileGrp), None)
            if mets_file:
                if mets_file.fileGrp == fileGrp and \
                        mets_file.pageId == pageId and \
                        mets_file.mimetype == mimetype:
                    if not force:
                        raise FileExistsError(
                            f"A file with ID=={ID} already exists {mets_file} and neither force nor ignore are set")
                    self.remove_file(ID=ID, fileGrp=fileGrp)
                else:
                    raise FileExistsError(
                        f"A file with ID=={ID} already exists {mets_file} but unrelated - cannot mitigate")

        # To get rid of Python's FutureWarning - checking if v is not None
        kwargs = {k: v for k, v in locals().items()
                  if k in ['url', 'ID', 'mimetype', 'pageId', 'local_filename'] and v is not None}
        # This separation is needed to reuse the same el_mets_file element in the caching if block
        el_mets_file = ET.SubElement(el_fileGrp, TAG_METS_FILE)
        # The caching of the physical page is done in the OcrdFile constructor
        # (which calls us back with set_physical_page_for_file)
        mets_file = OcrdFile(el_mets_file, mets=self, **kwargs)

        if self._cache_flag:
            # Add the file to the file cache
            self._file_cache[fileGrp].update({ID: el_mets_file})

        return mets_file