ocrd_models.ocrd_file.OcrdFile.contentids() - Code Metrics - Inspection of "WorkspaceBagger: Use, in order of preference, f.ba..." - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#1157)

by Konstantin

created 2023-12-14 19:06 UTC

ocrd_models.ocrd_file.OcrdFile.contentids() A

↳ Parent: ocrd_models.ocrd_file

Complexity

Conditions

Size

Total Lines	8
Code Lines	5

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	5
dl	0
loc	8
rs	10
c	0
b	0
f	0
cc	2
nop	1

"""
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
            contentids (string): ``@CONTENTIDS`` of the ``mets:div`` in the ``mets:structMap[@TYPE="PHYSICAL]`` this file manifests
        """
        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 contentids(self):
        """
        Get the ``@CONTENTIDS`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``.
        """
        if self.mets is None:
            raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
        return self.mets.get_contentids_for_file(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, contentids=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``
            fileGrp (string): ``@USE`` of the ``mets:fileGrp`` this file belongs to
        """
        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			contentids (string): ``@CONTENTIDS`` of the ``mets:div`` in the ``mets:structMap[@TYPE="PHYSICAL]`` this file manifests
31			"""
32			if el is None:
33			raise ValueError("Must provide mets:file element this OcrdFile represents")
34			if loctype:
35			deprecation_warning("'loctype' is not supported in OcrdFile anymore, use 'url' or 'local_filename'")
36			self._el = el
37			self.mets = mets
38			self.ID = ID
39			self.mimetype = mimetype
40			self.pageId = pageId
41
42			if local_filename:
43			self.local_filename = Path(local_filename)
44			if url:
45			self.url = url
46
47			def __str__(self):
48			"""
49			String representation of this ``mets:file``.
50			"""
51			# props = '\n\t'.join([
52			# ' : '.join([k, getattr(self, k) if getattr(self, k) else '---'])
53			# for k in ['mimetype', 'ID', 'url', 'local_filename']
54			# ])
55			# return 'OcrdFile[' + '\n\t' + props + '\n\t]'
56			props = ', '.join([
57			'='.join([k, str(getattr(self, k)) if getattr(self, k) else '---'])
58			for k in ['ID', 'mimetype', 'url', 'local_filename']
59			])
60			try:
61			fileGrp = self.fileGrp
62			except ValueError:
63			fileGrp = '---'
64			return '<OcrdFile fileGrp=%s %s]/> ' % (fileGrp, props)
65
66			def __eq__(self, other):
67			return self.ID == other.ID \
68			and self.url == other.url \
69			and self.local_filename == other.local_filename
70			# EXT_TO_MIME[MIME_TO_EXT[self.mimetype]] == EXT_TO_MIME[MIME_TO_EXT[other.mimetype]] and \
71			# self.fileGrp == other.fileGrp
72
73			@property
74			def basename(self):
75			"""
76			Get the ``.name`` of the local file
77			"""
78			if not self.local_filename:
79			return
80			return self.local_filename.name
81
82			@property
83			def extension(self):
84			if not self.local_filename:
85			return
86			return ''.join(self.local_filename.suffixes)
87
88			@property
89			def basename_without_extension(self):
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 self.local_filename.name[:-len(self.extension)]
96
97			@property
98			def ID(self):
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):
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):
121			"""
122			Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file`` (physical page manifestation).
123			"""
124			if self.mets is None:
125			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
126			return self.mets.get_physical_page_for_file(self)
127
128			@pageId.setter
129			def pageId(self, pageId):
130			"""
131			Get the ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file`` (physical page manifestation) to :py:attr:`pageId`.
132			"""
133			if pageId is None:
134			return
135			if self.mets is None:
136			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
137			self.mets.set_physical_page_for_file(pageId, self)
138
139			@property
140			def contentids(self):
141			"""
142			Get the ``@CONTENTIDS`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``.
143			"""
144			if self.mets is None:
145			raise Exception("OcrdFile %s has no member 'mets' pointing to parent OcrdMets" % self)
146			return self.mets.get_contentids_for_file(self)
147
148			@property
149			def loctypes(self):
150			"""
151			Get the ``@LOCTYPE``s of the ``mets:file``.
152			"""
153			return [x.get('LOCTYPE') for x in self._el.findall('mets:FLocat', NS)]
154
155			@property
156			def mimetype(self):
157			"""
158			Get the ``@MIMETYPE`` of the ``mets:file``.
159			"""
160			return self._el.get('MIMETYPE')
161
162			@mimetype.setter
163			def mimetype(self, mimetype):
164			"""
165			Set the ``@MIMETYPE`` of the ``mets:file`` to :py:attr:`mimetype`.
166			"""
167			if mimetype is None:
168			return
169			self._el.set('MIMETYPE', mimetype)
170
171			@property
172			def fileGrp(self):
173			"""
174			The ``@USE`` of the containing ``mets:fileGrp``
175			"""
176			parent = self._el.getparent()
177			if parent is not None:
178			return self._el.getparent().get('USE')
179			raise ValueError("OcrdFile not related to METS")
180
181			@property
182			def url(self):
183			"""
184			Get the remote/original URL ``@xlink:href`` of this ``mets:file``.
185			"""
186			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
187			if el_FLocat is not None:
188			return el_FLocat.get("{%s}href" % NS["xlink"])
189			return ''
190
191			@url.setter
192			def url(self, url):
193			"""
194			Set the remote/original URL ``@xlink:href`` of this ``mets:file`` to :py:attr:`url`.
195			"""
196			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="URL"]', NS)
197			if url is None:
198			if el_FLocat:
199			self._el.remove(el_FLocat)
200			return
201			if el_FLocat is None:
202			el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
203			el_FLocat.set("{%s}href" % NS["xlink"], url)
204			el_FLocat.set("LOCTYPE", "URL")
205
206			@property
207			def local_filename(self):
208			"""
209			Get the local/cached ``@xlink:href`` of this ``mets:file``.
210			"""
211			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
212			if el_FLocat is not None:
213			return Path(el_FLocat.get("{%s}href" % NS["xlink"]))
214
215			@local_filename.setter
216			def local_filename(self, fname):
217			"""
218			Set the local/cached ``@xlink:href`` of this ``mets:file`` to :py:attr:`local_filename`.
219			"""
220			el_FLocat = self._el.find('mets:FLocat[@LOCTYPE="OTHER"][@OTHERLOCTYPE="FILE"]', NS)
221			if not fname:
222			if el_FLocat is not None:
223			self._el.remove(el_FLocat)
224			return
225			if el_FLocat is None:
226			el_FLocat = ET.SubElement(self._el, TAG_METS_FLOCAT)
227			el_FLocat.set("{%s}href" % NS["xlink"], str(fname))
228			el_FLocat.set("LOCTYPE", "OTHER")
229			el_FLocat.set("OTHERLOCTYPE", "FILE")
230
231
232			class ClientSideOcrdFile:
233			"""
234			Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdFile`
235			but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
236			this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
237			"""
238
239			def __init__(self, el, mimetype=None, pageId=None, loctype='OTHER', local_filename=None, mets=None, url=None, ID=None, fileGrp=None, contentids=None):
240			"""
241			Args:
242			el (): ignored
243			Keyword Args:
244			mets (): ignored
245			mimetype (string): ``@MIMETYPE`` of this ``mets:file``
246			pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
247			loctype (string): ``@LOCTYPE`` of this ``mets:file``
248			url (string): ignored XXX the remote/original file once we have proper mets:FLocat bookkeeping
249			local_filename (): ``@xlink:href`` of this ``mets:file`` - XXX the local file once we have proper mets:FLocat bookkeeping
250			ID (string): ``@ID`` of this ``mets:file``
251			fileGrp (string): ``@USE`` of the ``mets:fileGrp`` this file belongs to
252			"""
253			self.ID = ID
254			self.mimetype = mimetype
255			self.local_filename = local_filename
256			self.url = url
257			self.loctype = loctype
258			self.pageId = pageId
259			self.fileGrp = fileGrp
260
261			def __str__(self):
262			props = ', '.join([
263			'='.join([k, getattr(self, k) if hasattr(self, k) and getattr(self, k) else '---'])
264			for k in ['fileGrp', 'ID', 'mimetype', 'url', 'local_filename']
265			])
266			return '<ClientSideOcrdFile %s]/>' % (props)
267

OCR-D / core

Pull Request — master (#1157)

ocrd_models.ocrd_file.OcrdFile.contentids() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like