ocrd_models.ocrd_file.OcrdFile.local_filename() - Code Metrics - Inspection of ":memo: changelog" - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 488518...bfe6a0 )

by Konstantin

created 2023-09-11 12:56 UTC

ocrd_models.ocrd_file.OcrdFile.local_filename() A

↳ Parent: ocrd_models.ocrd_file

Complexity

Conditions

Size

Total Lines	8
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	12
dl	0
loc	8
rs	9.8
c	0
b	0
f	0
cc	4
nop	2

"""
API to ``mets:file``
"""
from os.path import splitext, basename
from pathlib import Path

from ocrd_utils import deprecation_warning

from .ocrd_xml_base import ET
from .constants import NAMESPACES as NS, TAG_METS_FLOCAT, TAG_METS_FILE

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``
            local_filename (string): Local filename
            url (string): original ``@xlink:href`` of this ``mets:file``
            local_filename (Path): ``@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 = Path(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
               # 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):
        """
        Get the ``.name`` of the local file
        """
        if not self.local_filename:
            return
        return self.local_filename.name

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

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

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

    @ID.setter
    def ID(self, ID):
        """
        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):
        """
        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):
        """
        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):
        """
        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):
        """
        Get the ``@MIMETYPE`` of the ``mets:file``.
        """
        return self._el.get('MIMETYPE')

    @mimetype.setter
    def mimetype(self, mimetype):
        """
        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):
        """
        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):
        """
        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):
        """
        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:
                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):
        """
        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 Path(el_FLocat.get("{%s}href" % NS["xlink"]))

    @local_filename.setter
    def local_filename(self, fname):
        """
        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
        if el_FLocat is None:
            el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
        el_FLocat.set("{%s}href" % NS["xlink"], str(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, mimetype=None, pageId=None, loctype='OTHER', local_filename=None, mets=None, url=None, ID=None, fileGrp=None):
        """
        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): ignored XXX the remote/original file once we have proper mets:FLocat bookkeeping 
            local_filename (): ``@xlink:href`` of this ``mets:file`` - XXX the local file once we have proper mets:FLocat bookkeeping
            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)


1			"""
2			API to ``mets:file``
3			"""
4			from os.path import splitext, basename
5			from pathlib import Path
6
7			from ocrd_utils import deprecation_warning
8
9			from .ocrd_xml_base import ET
10			from .constants import NAMESPACES as NS, TAG_METS_FLOCAT, TAG_METS_FILE
11
12			class OcrdFile():
13			"""
14			Represents a single ``mets:file/mets:FLocat`` (METS file entry).
15			"""
16
17			def __init__(self, el, mimetype=None, pageId=None, local_filename=None, mets=None, url=None, ID=None, loctype=None):
18			"""
19			Args:
20			el (LxmlElement): etree Element of the ``mets:file`` this represents. Create new if not provided
21			Keyword Args:
22			mets (OcrdMets): Containing :py:class:`ocrd_models.ocrd_mets.OcrdMets`.
23			mimetype (string): ``@MIMETYPE`` of this ``mets:file``
24			pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
25			local_filename (string): Local filename
26			url (string): original ``@xlink:href`` of this ``mets:file``
27			local_filename (Path): ``@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 = Path(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 \
67			and self.url == other.url \
68			and self.local_filename == other.local_filename
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			@property
73			def basename(self):
74			"""
75			Get the ``.name`` of the local file
76			"""
77			if not self.local_filename:
78			return
79			return self.local_filename.name
80
81			@property
82			def extension(self):
83			if not self.local_filename:
84			return
85			return ''.join(self.local_filename.suffixes)
86
87			@property
88			def basename_without_extension(self):
89			"""
90			Get the ``os.path.basename`` of the local file, if any, with extension removed.
91			"""
92			if not self.local_filename:
93			return
94			return self.local_filename.name[:-len(self.extension)]
95
96			@property
97			def ID(self):
98			"""
99			Get the ``@ID`` of the ``mets:file``.
100			"""
101			return self._el.get('ID')
102
103			@ID.setter
104			def ID(self, ID):
105			"""
106			Set the ``@ID`` of the ``mets:file`` to :py:attr:`ID`.
107			"""
108			if ID is None:
109			return
110			if self.mets is None:
111			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
112			old_id = self.ID
113			self._el.set('ID', ID)
114			# also update the references in the physical structmap
115			for pageId in self.mets.remove_physical_page_fptr(fileId=old_id):
116			self.pageId = pageId
117
118			@property
119			def pageId(self):
120			"""
121			Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file`` (physical page manifestation).
122			"""
123			if self.mets is None:
124			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
125			return self.mets.get_physical_page_for_file(self)
126
127			@pageId.setter
128			def pageId(self, pageId):
129			"""
130			Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file`` (physical page manifestation) to :py:attr:`pageId`.
131			"""
132			if pageId is None:
133			return
134			if self.mets is None:
135			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
136			self.mets.set_physical_page_for_file(pageId, self)
137
138			@property
139			def loctypes(self):
140			"""
141			Get the ``@LOCTYPE``s of the ``mets:file``.
142			"""
143			return [x.get('LOCTYPE') for x in self._el.findall('mets:FLocat', NS)]
144
145			@property
146			def mimetype(self):
147			"""
148			Get the ``@MIMETYPE`` of the ``mets:file``.
149			"""
150			return self._el.get('MIMETYPE')
151
152			@mimetype.setter
153			def mimetype(self, mimetype):
154			"""
155			Set the ``@MIMETYPE`` of the ``mets:file`` to :py:attr:`mimetype`.
156			"""
157			if mimetype is None:
158			return
159			self._el.set('MIMETYPE', mimetype)
160
161			@property
162			def fileGrp(self):
163			"""
164			The ``@USE`` of the containing ``mets:fileGrp``
165			"""
166			parent = self._el.getparent()
167			if parent is not None:
168			return self._el.getparent().get('USE')
169			raise ValueError("OcrdFile not related to METS")
170
171			@property
172			def url(self):
173			"""
174			Get the remote/original URL ``@xlink:href`` of this ``mets:file``.
175			"""
176			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
177			if el_FLocat is not None:
178			return el_FLocat.get("{%s}href" % NS["xlink"])
179			return ''
180
181			@url.setter
182			def url(self, url):
183			"""
184			Set the remote/original URL ``@xlink:href`` of this ``mets:file`` to :py:attr:`url`.
185			"""
186			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
187			if url is None:
188			if el_FLocat:
189			self._el.remove(el_FLocat)
190			return
191			if el_FLocat is None:
192			el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
193			el_FLocat.set("{%s}href" % NS["xlink"], url)
194			el_FLocat.set("LOCTYPE", "URL")
195
196			@property
197			def local_filename(self):
198			"""
199			Get the local/cached ``@xlink:href`` of this ``mets:file``.
200			"""
201			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
202			if el_FLocat is not None:
203			return Path(el_FLocat.get("{%s}href" % NS["xlink"]))
204
205			@local_filename.setter
206			def local_filename(self, fname):
207			"""
208			Set the local/cached ``@xlink:href`` of this ``mets:file`` to :py:attr:`local_filename`.
209			"""
210			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
211			if not fname:
212			if el_FLocat is not None:
213			self._el.remove(el_FLocat)
214			return
215			if el_FLocat is None:
216			el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
217			el_FLocat.set("{%s}href" % NS["xlink"], str(fname))
218			el_FLocat.set("LOCTYPE", "OTHER")
219			el_FLocat.set("OTHERLOCTYPE", "FILE")
220
221
222			class ClientSideOcrdFile:
223			"""
224			Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdFile`
225			but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
226			this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
227			"""
228
229			def __init__(self, el, mimetype=None, pageId=None, loctype='OTHER', local_filename=None, mets=None, url=None, ID=None, fileGrp=None):
230			"""
231			Args:
232			el (): ignored
233			Keyword Args:
234			mets (): ignored
235			mimetype (string): ``@MIMETYPE`` of this ``mets:file``
236			pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
237			loctype (string): ``@LOCTYPE`` of this ``mets:file``
238			url (string): ignored XXX the remote/original file once we have proper mets:FLocat bookkeeping
239			local_filename (): ``@xlink:href`` of this ``mets:file`` - XXX the local file once we have proper mets:FLocat bookkeeping
240			ID (string): ``@ID`` of this ``mets:file``
241			"""
242			self.ID = ID
243			self.mimetype = mimetype
244			self.local_filename = local_filename
245			self.url = url
246			self.loctype = loctype
247			self.pageId = pageId
248			self.fileGrp = fileGrp
249
250			def __str__(self):
251			props = ', '.join([
252			'='.join([k, getattr(self, k) if hasattr(self, k) and getattr(self, k) else '---'])
253			for k in ['fileGrp', 'ID', 'mimetype', 'url', 'local_filename']
254			])
255			return '<ClientSideOcrdFile %s]/>' % (props)
256

OCR-D / core

Push — master ( 488518...bfe6a0 )

ocrd_models.ocrd_file.OcrdFile.local_filename() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like