ocrd_models.ocrd_file.OcrdFile.__init__() - Code Metrics - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

ocrd_models.ocrd_file.OcrdFile.init() A
last analyzed 2025-12-19 12:30 UTC

↳ Parent: ocrd_models.ocrd_file

Complexity

Conditions

Size

Total Lines	27
Code Lines	14

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	14
dl	0
loc	27
rs	9.2333
c	0
b	0
f	0
cc	5
nop	9

How to fix Many Parameters

"""
API to ``mets:file``
"""
from pathlib import Path
from typing import Any, List, Optional, Union

from ocrd_utils import deprecation_warning

from .ocrd_xml_base import ET  # type: ignore
from .constants import NAMESPACES as NS, TAG_METS_FLOCAT


class OcrdFile():
    """
    Represents a single ``mets:file/mets:FLocat`` (METS file entry).
    """

    def __init__(self, el, mimetype=None, pageId=None, local_filename=None, mets=None, url=None, ID=None, loctype=None):
        """
        Args:
            el (LxmlElement): etree Element of the ``mets:file`` this represents. Create new if not provided
        Keyword Args:
            mets (OcrdMets): Containing :py:class:`ocrd_models.ocrd_mets.OcrdMets`.
            mimetype (string): ``@MIMETYPE`` of this ``mets:file``
            pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
            url (string): original ``@xlink:href`` of this ``mets:file``
            local_filename (string): ``@xlink:href`` pointing to the locally cached version of the file in the workspace
            ID (string): ``@ID`` of this ``mets:file``
            loctype (string): DEPRECATED do not use
        """
        if el is None:
            raise ValueError("Must provide mets:file element this OcrdFile represents")
        if loctype:
            deprecation_warning("'loctype' is not supported in OcrdFile anymore, use 'url' or 'local_filename'")
        self._el = el
        self.mets = mets
        self.ID = ID
        self.mimetype = mimetype
        self.pageId = pageId

        if local_filename:
            self.local_filename = local_filename
        if url:
            self.url = url

    def __str__(self):
        """
        String representation of this ``mets:file``.
        """
        #  props = '\n\t'.join([
        #      ' : '.join([k, getattr(self, k) if getattr(self, k) else '---'])
        #      for k in ['mimetype', 'ID', 'url', 'local_filename']
        #  ])
        #  return 'OcrdFile[' + '\n\t' + props + '\n\t]'
        props = ', '.join([
            '='.join([k, str(getattr(self, k)) if getattr(self, k) else '---'])
            for k in ['ID', 'mimetype', 'url', 'local_filename']
        ])
        try:
            fileGrp = self.fileGrp
        except ValueError:
            fileGrp = '---'
        return '<OcrdFile fileGrp=%s %s]/> ' % (fileGrp, props)

    def __eq__(self, other):
        return (self.ID == other.ID and
                self.url == other.url and
                self.local_filename == other.local_filename  # and
                # EXT_TO_MIME[MIME_TO_EXT[self.mimetype]] == EXT_TO_MIME[MIME_TO_EXT[other.mimetype]] and
                # self.fileGrp == other.fileGrp
                )

    @property
    def basename(self) -> str:
        """
        Get the ``.name`` of the local file
        """
        if not self.local_filename:
            return ''
        return Path(self.local_filename).name

    @property
    def extension(self) -> str:
        if not self.local_filename:
            return ''
        return ''.join(Path(self.local_filename).suffixes)

    @property
    def basename_without_extension(self) -> str:
        """
        Get the ``os.path.basename`` of the local file, if any, with extension removed.
        """
        if not self.local_filename:
            return ''
        return Path(self.local_filename).name[:-len(self.extension)]

    @property
    def ID(self) -> str:
        """
        Get the ``@ID`` of the ``mets:file``.
        """
        return self._el.get('ID')

    @ID.setter
    def ID(self, ID: Optional[str]) -> None:
        """
        Set the ``@ID`` of the ``mets:file`` to :py:attr:`ID`.
        """
        if ID is None:
            return
        if self.mets is None:
            raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
        old_id = self.ID
        self._el.set('ID', ID)
        # also update the references in the physical structmap
        for pageId in self.mets.remove_physical_page_fptr(fileId=old_id):
            self.pageId = pageId

    @property
    def pageId(self) -> str:
        """
        Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
        (physical page manifestation).
        """
        if self.mets is None:
            raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
        return self.mets.get_physical_page_for_file(self)

    @pageId.setter
    def pageId(self, pageId: Optional[str]) -> None:
        """
        Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
        (physical page manifestation) to :py:attr:`pageId`.
        """
        if pageId is None:
            return
        if self.mets is None:
            raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
        self.mets.set_physical_page_for_file(pageId, self)

    @property
    def loctypes(self) -> List[str]:
        """
        Get the ``@LOCTYPE``s of the ``mets:file``.
        """
        return [x.get('LOCTYPE') for x in self._el.findall('mets:FLocat', NS)]

    @property
    def mimetype(self) -> str:
        """
        Get the ``@MIMETYPE`` of the ``mets:file``.
        """
        return self._el.get('MIMETYPE')

    @mimetype.setter
    def mimetype(self, mimetype: Optional[str]) -> None:
        """
        Set the ``@MIMETYPE`` of the ``mets:file`` to :py:attr:`mimetype`.
        """
        if mimetype is None:
            return
        self._el.set('MIMETYPE', mimetype)

    @property
    def fileGrp(self) -> str:
        """
        The ``@USE`` of the containing ``mets:fileGrp``
        """
        parent = self._el.getparent()
        if parent is not None:
            return self._el.getparent().get('USE')
        raise ValueError("OcrdFile not related to METS")

    @property
    def url(self) -> str:
        """
        Get the remote/original URL ``@xlink:href`` of this ``mets:file``.
        """
        el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
        if el_FLocat is not None:
            return el_FLocat.get("{%s}href" % NS["xlink"])
        return ''

    @url.setter
    def url(self, url: Optional[str]) -> None:
        """
        Set the remote/original URL ``@xlink:href`` of this ``mets:file`` to :py:attr:`url`.
        """
        el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
        if url is None:
            if el_FLocat is not None:
                self._el.remove(el_FLocat)
            return
        if el_FLocat is None:
            el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
        el_FLocat.set("{%s}href" % NS["xlink"], url)
        el_FLocat.set("LOCTYPE", "URL")

    @property
    def local_filename(self) -> Optional[str]:
        """
        Get the local/cached ``@xlink:href`` of this ``mets:file``.
        """
        el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
        if el_FLocat is not None:
            return el_FLocat.get("{%s}href" % NS["xlink"])
        return None

    @local_filename.setter
    def local_filename(self, fname: Optional[Union[Path, str]]):
        """
        Set the local/cached ``@xlink:href`` of this ``mets:file`` to :py:attr:`local_filename`.
        """
        el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
        if not fname:
            if el_FLocat is not None:
                self._el.remove(el_FLocat)
            return
        else:
            fname = str(fname)
        if el_FLocat is None:
            el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
        el_FLocat.set("{%s}href" % NS["xlink"], fname)
        el_FLocat.set("LOCTYPE", "OTHER")
        el_FLocat.set("OTHERLOCTYPE", "FILE")


class ClientSideOcrdFile:
    """
    Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdFile`
    but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
    this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
    """

    def __init__(
        self,
        el,  # pylint: disable=unused-argument
        mimetype: str = '',
        pageId: str = '',
        loctype: str = 'OTHER',
        local_filename: Optional[str] = None,
        mets: Any = None,  # pylint: disable=unused-argument
        url: str = '',
        ID: str = '',
        fileGrp: str = ''
    ):
        """
        Args:
            el (): ignored
        Keyword Args:
            mets (): ignored
            mimetype (string): ``@MIMETYPE`` of this ``mets:file``
            pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
            loctype (string): ``@LOCTYPE`` of this ``mets:file``
            url (string):  ``@xlink:href`` of this ``mets:file`` (if ``@LOCTYPE==URL``)
            local_filename (): ``@xlink:href`` of this ``mets:file`` (if ``@LOCTYPE==FILE @OTHERLOCTYPE==FILE``)
            ID (string): ``@ID`` of this ``mets:file``
        """
        self.ID = ID
        self.mimetype = mimetype
        self.local_filename = local_filename
        self.url = url
        self.loctype = loctype
        self.pageId = pageId
        self.fileGrp = fileGrp

    def __str__(self):
        props = ', '.join([
            '='.join([k, getattr(self, k) if hasattr(self, k) and getattr(self, k) else '---'])
            for k in ['fileGrp', 'ID', 'mimetype', 'url', 'local_filename']
        ])
        return '<ClientSideOcrdFile %s]/>' % (props)


OcrdFileType = Union[OcrdFile, ClientSideOcrdFile]


1			"""
2			API to ``mets:file``
3			"""
4			from pathlib import Path
5			from typing import Any, List, Optional, Union
6
7			from ocrd_utils import deprecation_warning
8
9			from .ocrd_xml_base import ET # type: ignore
10			from .constants import NAMESPACES as NS, TAG_METS_FLOCAT
11
12
13			class OcrdFile():
14			"""
15			Represents a single ``mets:file/mets:FLocat`` (METS file entry).
16			"""
17
18			def __init__(self, el, mimetype=None, pageId=None, local_filename=None, mets=None, url=None, ID=None, loctype=None):
19			"""
20			Args:
21			el (LxmlElement): etree Element of the ``mets:file`` this represents. Create new if not provided
22			Keyword Args:
23			mets (OcrdMets): Containing :py:class:`ocrd_models.ocrd_mets.OcrdMets`.
24			mimetype (string): ``@MIMETYPE`` of this ``mets:file``
25			pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
26			url (string): original ``@xlink:href`` of this ``mets:file``
27			local_filename (string): ``@xlink:href`` pointing to the locally cached version of the file in the workspace
28			ID (string): ``@ID`` of this ``mets:file``
29			loctype (string): DEPRECATED do not use
30			"""
31			if el is None:
32			raise ValueError("Must provide mets:file element this OcrdFile represents")
33			if loctype:
34			deprecation_warning("'loctype' is not supported in OcrdFile anymore, use 'url' or 'local_filename'")
35			self._el = el
36			self.mets = mets
37			self.ID = ID
38			self.mimetype = mimetype
39			self.pageId = pageId
40
41			if local_filename:
42			self.local_filename = local_filename
43			if url:
44			self.url = url
45
46			def __str__(self):
47			"""
48			String representation of this ``mets:file``.
49			"""
50			# props = '\n\t'.join([
51			# ' : '.join([k, getattr(self, k) if getattr(self, k) else '---'])
52			# for k in ['mimetype', 'ID', 'url', 'local_filename']
53			# ])
54			# return 'OcrdFile[' + '\n\t' + props + '\n\t]'
55			props = ', '.join([
56			'='.join([k, str(getattr(self, k)) if getattr(self, k) else '---'])
57			for k in ['ID', 'mimetype', 'url', 'local_filename']
58			])
59			try:
60			fileGrp = self.fileGrp
61			except ValueError:
62			fileGrp = '---'
63			return '<OcrdFile fileGrp=%s %s]/> ' % (fileGrp, props)
64
65			def __eq__(self, other):
66			return (self.ID == other.ID and
67			self.url == other.url and
68			self.local_filename == other.local_filename # and
69			# EXT_TO_MIME[MIME_TO_EXT[self.mimetype]] == EXT_TO_MIME[MIME_TO_EXT[other.mimetype]] and
70			# self.fileGrp == other.fileGrp
71			)
72
73			@property
74			def basename(self) -> str:
75			"""
76			Get the ``.name`` of the local file
77			"""
78			if not self.local_filename:
79			return ''
80			return Path(self.local_filename).name
81
82			@property
83			def extension(self) -> str:
84			if not self.local_filename:
85			return ''
86			return ''.join(Path(self.local_filename).suffixes)
87
88			@property
89			def basename_without_extension(self) -> str:
90			"""
91			Get the ``os.path.basename`` of the local file, if any, with extension removed.
92			"""
93			if not self.local_filename:
94			return ''
95			return Path(self.local_filename).name[:-len(self.extension)]
96
97			@property
98			def ID(self) -> str:
99			"""
100			Get the ``@ID`` of the ``mets:file``.
101			"""
102			return self._el.get('ID')
103
104			@ID.setter
105			def ID(self, ID: Optional[str]) -> None:
106			"""
107			Set the ``@ID`` of the ``mets:file`` to :py:attr:`ID`.
108			"""
109			if ID is None:
110			return
111			if self.mets is None:
112			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
113			old_id = self.ID
114			self._el.set('ID', ID)
115			# also update the references in the physical structmap
116			for pageId in self.mets.remove_physical_page_fptr(fileId=old_id):
117			self.pageId = pageId
118
119			@property
120			def pageId(self) -> str:
121			"""
122			Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
123			(physical page manifestation).
124			"""
125			if self.mets is None:
126			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
127			return self.mets.get_physical_page_for_file(self)
128
129			@pageId.setter
130			def pageId(self, pageId: Optional[str]) -> None:
131			"""
132			Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
133			(physical page manifestation) to :py:attr:`pageId`.
134			"""
135			if pageId is None:
136			return
137			if self.mets is None:
138			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
139			self.mets.set_physical_page_for_file(pageId, self)
140
141			@property
142			def loctypes(self) -> List[str]:
143			"""
144			Get the ``@LOCTYPE``s of the ``mets:file``.
145			"""
146			return [x.get('LOCTYPE') for x in self._el.findall('mets:FLocat', NS)]
147
148			@property
149			def mimetype(self) -> str:
150			"""
151			Get the ``@MIMETYPE`` of the ``mets:file``.
152			"""
153			return self._el.get('MIMETYPE')
154
155			@mimetype.setter
156			def mimetype(self, mimetype: Optional[str]) -> None:
157			"""
158			Set the ``@MIMETYPE`` of the ``mets:file`` to :py:attr:`mimetype`.
159			"""
160			if mimetype is None:
161			return
162			self._el.set('MIMETYPE', mimetype)
163
164			@property
165			def fileGrp(self) -> str:
166			"""
167			The ``@USE`` of the containing ``mets:fileGrp``
168			"""
169			parent = self._el.getparent()
170			if parent is not None:
171			return self._el.getparent().get('USE')
172			raise ValueError("OcrdFile not related to METS")
173
174			@property
175			def url(self) -> str:
176			"""
177			Get the remote/original URL ``@xlink:href`` of this ``mets:file``.
178			"""
179			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
180			if el_FLocat is not None:
181			return el_FLocat.get("{%s}href" % NS["xlink"])
182			return ''
183
184			@url.setter
185			def url(self, url: Optional[str]) -> None:
186			"""
187			Set the remote/original URL ``@xlink:href`` of this ``mets:file`` to :py:attr:`url`.
188			"""
189			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
190			if url is None:
191			if el_FLocat is not None:
192			self._el.remove(el_FLocat)
193			return
194			if el_FLocat is None:
195			el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
196			el_FLocat.set("{%s}href" % NS["xlink"], url)
197			el_FLocat.set("LOCTYPE", "URL")
198
199			@property
200			def local_filename(self) -> Optional[str]:
201			"""
202			Get the local/cached ``@xlink:href`` of this ``mets:file``.
203			"""
204			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
205			if el_FLocat is not None:
206			return el_FLocat.get("{%s}href" % NS["xlink"])
207			return None
208
209			@local_filename.setter
210			def local_filename(self, fname: Optional[Union[Path, str]]):
211			"""
212			Set the local/cached ``@xlink:href`` of this ``mets:file`` to :py:attr:`local_filename`.
213			"""
214			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
215			if not fname:
216			if el_FLocat is not None:
217			self._el.remove(el_FLocat)
218			return
219			else:
220			fname = str(fname)
221			if el_FLocat is None:
222			el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
223			el_FLocat.set("{%s}href" % NS["xlink"], fname)
224			el_FLocat.set("LOCTYPE", "OTHER")
225			el_FLocat.set("OTHERLOCTYPE", "FILE")
226
227
228			class ClientSideOcrdFile:
229			"""
230			Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdFile`
231			but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
232			this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
233			"""
234
235			def __init__(
236			self,
237			el, # pylint: disable=unused-argument
238			mimetype: str = '',
239			pageId: str = '',
240			loctype: str = 'OTHER',
241			local_filename: Optional[str] = None,
242			mets: Any = None, # pylint: disable=unused-argument
243			url: str = '',
244			ID: str = '',
245			fileGrp: str = ''
246			):
247			"""
248			Args:
249			el (): ignored
250			Keyword Args:
251			mets (): ignored
252			mimetype (string): ``@MIMETYPE`` of this ``mets:file``
253			pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
254			loctype (string): ``@LOCTYPE`` of this ``mets:file``
255			url (string): ``@xlink:href`` of this ``mets:file`` (if ``@LOCTYPE==URL``)
256			local_filename (): ``@xlink:href`` of this ``mets:file`` (if ``@LOCTYPE==FILE @OTHERLOCTYPE==FILE``)
257			ID (string): ``@ID`` of this ``mets:file``
258			"""
259			self.ID = ID
260			self.mimetype = mimetype
261			self.local_filename = local_filename
262			self.url = url
263			self.loctype = loctype
264			self.pageId = pageId
265			self.fileGrp = fileGrp
266
267			def __str__(self):
268			props = ', '.join([
269			'='.join([k, getattr(self, k) if hasattr(self, k) and getattr(self, k) else '---'])
270			for k in ['fileGrp', 'ID', 'mimetype', 'url', 'local_filename']
271			])
272			return '<ClientSideOcrdFile %s]/>' % (props)
273
274
275			OcrdFileType = Union[OcrdFile, ClientSideOcrdFile]
276

OCR-D / core

ocrd_models.ocrd_file.OcrdFile.__init__() A last analyzed 2025-12-19 12:30 UTC

Complexity

Size

Duplication

Importance

How to fix Many Parameters

Many Parameters

Duplication Side-by-Side

Filter issues like

ocrd_models.ocrd_file.OcrdFile.init() A
last analyzed 2025-12-19 12:30 UTC