ocrd.resolver.Resolver.workspace_from_nothing() - Code Metrics - Inspection of "Use the same logic for handling --mets/--mets-base..." - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#696)

by Konstantin

created 2021-12-20 11:39 UTC

ocrd.resolver.Resolver.workspace_from_nothing() A

↳ Parent: ocrd.resolver

Complexity

Conditions

Size

Total Lines	27
Code Lines	12

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	12
dl	0
loc	27
rs	9.8
c	0
b	0
f	0
cc	4
nop	4

from tempfile import mkdtemp
from pathlib import Path
from warnings import warn

import requests

from ocrd.constants import TMP_PREFIX
from ocrd_utils import (
    getLogger,
    is_local_filename,
    get_local_filename,
    remove_non_path_from_url,
    is_file_in_directory,
    nth_url_segment
)
from ocrd.workspace import Workspace
from ocrd_models import OcrdMets
from ocrd_models.constants import NAMESPACES as NS
from ocrd_models.utils import handle_oai_response

class Resolver():
    """
    Handle uploads, downloads, repository access, and manage temporary directories
    """

    def download_to_directory(self, directory, url, basename=None, if_exists='skip', subdir=None):
        """
        Download a file to a directory.

        Early Shortcut: If `url` is a local file and that file is already in the directory, keep it there.

        If `basename` is not given but subdir is, assume user knows what she's doing and
        use last URL segment as the basename.

        If `basename` is not given and no subdir is given, use the alnum characters in the URL as the basename.

        Args:
            directory (string): Directory to download files to
            basename (string, None): basename part of the filename on disk.
            url (string): URL to download from
            if_exists (string, "skip"): What to do if target file already exists. \
                One of ``skip`` (default), ``overwrite`` or ``raise``
            subdir (string, None): Subdirectory to create within the directory. Think ``mets:fileGrp``.

        Returns:
            Local filename string, *relative* to directory
        """
        log = getLogger('ocrd.resolver.download_to_directory') # pylint: disable=redefined-outer-name
        log.debug("directory=|%s| url=|%s| basename=|%s| if_exists=|%s| subdir=|%s|", directory, url, basename, if_exists, subdir)

        if not url:
            raise Exception("'url' must be a string")
        if not directory:
            raise Exception("'directory' must be a string")  # actually Path would also work

        directory = Path(directory)
        directory.mkdir(parents=True, exist_ok=True)
        directory = str(directory.resolve())

        subdir_path = Path(subdir if subdir else '')
        basename_path = Path(basename if basename else nth_url_segment(url))
        ret = str(Path(subdir_path, basename_path))
        dst_path = Path(directory, ret)

        #  log.info("\n\tdst_path='%s \n\turl=%s", dst_path, url)
        #  print('url=%s', url)
        #  print('directory=%s', directory)
        #  print('subdir_path=%s', subdir_path)
        #  print('basename_path=%s', basename_path)
        #  print('ret=%s', ret)
        #  print('dst_path=%s', dst_path)

        src_path = None
        if is_local_filename(url):
            try:
                # XXX this raises FNFE in Python 3.5 if src_path doesn't exist but not 3.6+
                src_path = Path(get_local_filename(url)).resolve()
            except FileNotFoundError as e:
                log.error("Failed to resolve URL locally: %s --> '%s' which does not exist" % (url, src_path))
                raise e
            if not src_path.exists():
                raise FileNotFoundError("File path passed as 'url' to download_to_directory does not exist: %s" % url)
            if src_path == dst_path:
                log.debug("Stop early, src_path and dst_path are the same: '%s' (url: '%s')" % (src_path, url))
                return ret

        # Respect 'if_exists' arg
        if dst_path.exists():
            if if_exists == 'skip':
                return ret
            if if_exists == 'raise':
                raise FileExistsError("File already exists and if_exists == 'raise': %s" % (dst_path))

        # Create dst_path parent dir
        dst_path.parent.mkdir(parents=True, exist_ok=True)

        # Copy files or download remote assets
        if src_path:
            log.debug("Copying file '%s' to '%s'", src_path, dst_path)
            dst_path.write_bytes(src_path.read_bytes())
        else:
            log.debug("Downloading URL '%s' to '%s'", url, dst_path)
            response = requests.get(url)
            if response.status_code != 200:
                raise Exception("HTTP request failed: %s (HTTP %d)" % (url, response.status_code))
            contents = handle_oai_response(response)
            dst_path.write_bytes(contents)

        return ret

    def workspace_from_url(self, mets_url, dst_dir=None, clobber_mets=False, mets_basename=None, download=False, src_baseurl=None):
        """
        Create a workspace from a METS by URL (i.e. clone if :py:attr:`mets_url` is remote or :py:attr:`dst_dir` is given).

        Arguments:
            mets_url (string): Source METS URL or filesystem path
        Keyword Arguments:
            dst_dir (string, None): Target directory for the workspace. \
                By default create a temporary directory under :py:data:`ocrd.constants.TMP_PREFIX`. \
                (The resulting path can be retrieved via :py:attr:`ocrd.Workspace.directory`.)
            clobber_mets (boolean, False): Whether to overwrite existing ``mets.xml``. \
                By default existing ``mets.xml`` will raise an exception.
            download (boolean, False): Whether to also download all the files referenced by the METS
            src_baseurl (string, None): Base URL for resolving relative file locations

        Download (clone) :py:attr:`mets_url` to ``mets.xml`` in :py:attr:`dst_dir`, unless 
        the former is already local and the latter is ``none`` or already identical to its directory name.

        Returns:
            a new :py:class:`~ocrd.workspace.Workspace`
        """
        log = getLogger('ocrd.resolver.workspace_from_url')

        if mets_url is None:
            raise ValueError("Must pass 'mets_url' workspace_from_url")

        # if mets_url is a relative filename, make it absolute
        if is_local_filename(mets_url) and not Path(mets_url).is_absolute():
            mets_url = str(Path(Path.cwd() / mets_url))

        # if mets_basename is not given, use the last URL segment of the mets_url
        if mets_basename is None:
            mets_basename = nth_url_segment(mets_url, -1)

        # If src_baseurl wasn't given, determine from mets_url by removing last url segment
        if not src_baseurl:
            last_segment = nth_url_segment(mets_url)
            src_baseurl = remove_non_path_from_url(remove_non_path_from_url(mets_url)[:-len(last_segment)])

        # resolve dst_dir
        if not dst_dir:
            if is_local_filename(mets_url):
                log.debug("Deriving dst_dir %s from %s", Path(mets_url).parent, mets_url)
                dst_dir = Path(mets_url).parent
            else:
                log.debug("Creating ephemeral workspace '%s' for METS @ <%s>", dst_dir, mets_url)
                dst_dir = mkdtemp(prefix=TMP_PREFIX)
        # XXX Path.resolve is always strict in Python <= 3.5, so create dst_dir unless it exists consistently
        if not Path(dst_dir).exists():
            Path(dst_dir).mkdir(parents=True, exist_ok=False)
        dst_dir = str(Path(dst_dir).resolve())

        log.debug("workspace_from_url\nmets_basename='%s'\nmets_url='%s'\nsrc_baseurl='%s'\ndst_dir='%s'",
            mets_basename, mets_url, src_baseurl, dst_dir)

        self.download_to_directory(dst_dir, mets_url, basename=mets_basename, if_exists='overwrite' if clobber_mets else 'skip')

        workspace = Workspace(self, dst_dir, mets_basename=mets_basename, baseurl=src_baseurl)

        if download:
            for f in workspace.mets.find_files():
                workspace.download_file(f)

        return workspace

    def workspace_from_nothing(self, directory, mets_basename='mets.xml', clobber_mets=False):
        """
        Create an empty workspace.

        Arguments:
            directory (string): Target directory for the workspace. \
                If ``none``, create a temporary directory under :py:data:`ocrd.constants.TMP_PREFIX`. \
                (The resulting path can be retrieved via :py:attr:`ocrd.Workspace.directory`.)
        Keyword Arguments:
            clobber_mets (boolean, False): Whether to overwrite existing ``mets.xml``. \
                By default existing ``mets.xml`` will raise an exception.

        Returns:
            a new :py:class:`~ocrd.workspace.Workspace`
        """
        log = getLogger('ocrd.resolver.workspace_from_nothing')
        if directory is None:
            directory = mkdtemp(prefix=TMP_PREFIX)
        Path(directory).mkdir(parents=True, exist_ok=True)
        mets_path = Path(directory, mets_basename)
        if mets_path.exists() and not clobber_mets:
            raise FileExistsError("METS '%s' already exists in '%s' and clobber_mets not set." % (mets_basename, directory))
        mets = OcrdMets.empty_mets()
        log.info("Writing METS to %s", mets_path)
        mets_path.write_bytes(mets.to_xml(xmllint=True))

        return Workspace(self, directory, mets, mets_basename=mets_basename)

    def resolve_mets_arguments(self, directory, mets_url, mets_basename):
        """
        Resolve the ``--mets``, ``--mets-basename`` and `--directory`` argument
        into a coherent set of arguments according to https://github.com/OCR-D/core/issues/517
        """
        log = getLogger('ocrd.resolver.resolve_mets_arguments')
        mets_is_remote = mets_url and (mets_url.startswith('http://') or mets_url.startswith('https://'))

        # XXX we might want to be more strict like this but it might break # legacy code
        # Allow --mets and --directory together iff --mets is a remote URL
        # if directory and mets_url and not mets_is_remote:
        #     raise ValueError("Use either --mets or --directory, not both")

        # If --mets is a URL, a directory must be explicitly provided (not strictly necessary, but retained for legacy behavior)
        if not directory and mets_is_remote:
            raise ValueError("--mets is an http(s) URL but no --directory was given")

        # Determine --mets-basename
        if not mets_basename and mets_url:
            mets_basename = Path(mets_url).name
        elif not mets_basename and not mets_url:
            mets_basename = 'mets.xml'
        elif mets_basename and mets_url:
            raise ValueError("Use either --mets or --mets-basename, not both")
        else:
            warn("--mets-basename is deprecated. Use --mets/--directory instead", DeprecationWarning)

        # Determine --directory and --mets-url
        if not directory and not mets_url:
            directory = Path.cwd()
            mets_url = Path(directory, mets_basename)
        elif directory and not mets_url:
            directory = Path(directory).resolve()
            mets_url = directory / mets_basename
        elif not directory and mets_url:
            mets_url = Path(mets_url).resolve()
            directory = mets_url.parent
        else: # == directory and mets_url:
            directory = Path(directory).resolve()
            if not mets_is_remote:
                # --mets is just a basename and --directory is set, so treat --mets as --mets-basename
                if Path(mets_url).parent == Path('.'):
                    mets_url = directory / mets_url
                else:
                    mets_url = Path(mets_url).resolve()
                    if not is_file_in_directory(directory, mets_url):
                        raise ValueError("--mets '%s' has a directory part inconsistent with --directory '%s'" % (mets_url, directory))

        return str(Path(directory).resolve()), str(mets_url), str(mets_basename)




1			from tempfile import mkdtemp
2			from pathlib import Path
3			from warnings import warn
4
5			import requests
6
7			from ocrd.constants import TMP_PREFIX
8			from ocrd_utils import (
9			getLogger,
10			is_local_filename,
11			get_local_filename,
12			remove_non_path_from_url,
13			is_file_in_directory,
14			nth_url_segment
15			)
16			from ocrd.workspace import Workspace
17			from ocrd_models import OcrdMets
18			from ocrd_models.constants import NAMESPACES as NS
19			from ocrd_models.utils import handle_oai_response
20
21			class Resolver():
22			"""
23			Handle uploads, downloads, repository access, and manage temporary directories
24			"""
25
26			def download_to_directory(self, directory, url, basename=None, if_exists='skip', subdir=None):
27			"""
28			Download a file to a directory.
29
30			Early Shortcut: If `url` is a local file and that file is already in the directory, keep it there.
31
32			If `basename` is not given but subdir is, assume user knows what she's doing and
33			use last URL segment as the basename.
34
35			If `basename` is not given and no subdir is given, use the alnum characters in the URL as the basename.
36
37			Args:
38			directory (string): Directory to download files to
39			basename (string, None): basename part of the filename on disk.
40			url (string): URL to download from
41			if_exists (string, "skip"): What to do if target file already exists. \
42			One of ``skip`` (default), ``overwrite`` or ``raise``
43			subdir (string, None): Subdirectory to create within the directory. Think ``mets:fileGrp``.
44
45			Returns:
46			Local filename string, relative to directory
47			"""
48			log = getLogger('ocrd.resolver.download_to_directory') # pylint: disable=redefined-outer-name
49			log.debug("directory=\|%s\| url=\|%s\| basename=\|%s\| if_exists=\|%s\| subdir=\|%s\|", directory, url, basename, if_exists, subdir)
50
51			if not url:
52			raise Exception("'url' must be a string")
53			if not directory:
54			raise Exception("'directory' must be a string") # actually Path would also work
55
56			directory = Path(directory)
57			directory.mkdir(parents=True, exist_ok=True)
58			directory = str(directory.resolve())
59
60			subdir_path = Path(subdir if subdir else '')
61			basename_path = Path(basename if basename else nth_url_segment(url))
62			ret = str(Path(subdir_path, basename_path))
63			dst_path = Path(directory, ret)
64
65			# log.info("\n\tdst_path='%s \n\turl=%s", dst_path, url)
66			# print('url=%s', url)
67			# print('directory=%s', directory)
68			# print('subdir_path=%s', subdir_path)
69			# print('basename_path=%s', basename_path)
70			# print('ret=%s', ret)
71			# print('dst_path=%s', dst_path)
72
73			src_path = None
74			if is_local_filename(url):
75			try:
76			# XXX this raises FNFE in Python 3.5 if src_path doesn't exist but not 3.6+
77			src_path = Path(get_local_filename(url)).resolve()
78			except FileNotFoundError as e:
79			log.error("Failed to resolve URL locally: %s --> '%s' which does not exist" % (url, src_path))
80			raise e
81			if not src_path.exists():
82			raise FileNotFoundError("File path passed as 'url' to download_to_directory does not exist: %s" % url)
83			if src_path == dst_path:
84			log.debug("Stop early, src_path and dst_path are the same: '%s' (url: '%s')" % (src_path, url))
85			return ret
86
87			# Respect 'if_exists' arg
88			if dst_path.exists():
89			if if_exists == 'skip':
90			return ret
91			if if_exists == 'raise':
92			raise FileExistsError("File already exists and if_exists == 'raise': %s" % (dst_path))
93
94			# Create dst_path parent dir
95			dst_path.parent.mkdir(parents=True, exist_ok=True)
96
97			# Copy files or download remote assets
98			if src_path:
99			log.debug("Copying file '%s' to '%s'", src_path, dst_path)
100			dst_path.write_bytes(src_path.read_bytes())
101			else:
102			log.debug("Downloading URL '%s' to '%s'", url, dst_path)
103			response = requests.get(url)
104			if response.status_code != 200:
105			raise Exception("HTTP request failed: %s (HTTP %d)" % (url, response.status_code))
106			contents = handle_oai_response(response)
107			dst_path.write_bytes(contents)
108
109			return ret
110
111			def workspace_from_url(self, mets_url, dst_dir=None, clobber_mets=False, mets_basename=None, download=False, src_baseurl=None):
112			"""
113			Create a workspace from a METS by URL (i.e. clone if :py:attr:`mets_url` is remote or :py:attr:`dst_dir` is given).
114
115			Arguments:
116			mets_url (string): Source METS URL or filesystem path
117			Keyword Arguments:
118			dst_dir (string, None): Target directory for the workspace. \
119			By default create a temporary directory under :py:data:`ocrd.constants.TMP_PREFIX`. \
120			(The resulting path can be retrieved via :py:attr:`ocrd.Workspace.directory`.)
121			clobber_mets (boolean, False): Whether to overwrite existing ``mets.xml``. \
122			By default existing ``mets.xml`` will raise an exception.
123			download (boolean, False): Whether to also download all the files referenced by the METS
124			src_baseurl (string, None): Base URL for resolving relative file locations
125
126			Download (clone) :py:attr:`mets_url` to ``mets.xml`` in :py:attr:`dst_dir`, unless
127			the former is already local and the latter is ``none`` or already identical to its directory name.
128
129			Returns:
130			a new :py:class:`~ocrd.workspace.Workspace`
131			"""
132			log = getLogger('ocrd.resolver.workspace_from_url')
133
134			if mets_url is None:
135			raise ValueError("Must pass 'mets_url' workspace_from_url")
136
137			# if mets_url is a relative filename, make it absolute
138			if is_local_filename(mets_url) and not Path(mets_url).is_absolute():
139			mets_url = str(Path(Path.cwd() / mets_url))
140
141			# if mets_basename is not given, use the last URL segment of the mets_url
142			if mets_basename is None:
143			mets_basename = nth_url_segment(mets_url, -1)
144
145			# If src_baseurl wasn't given, determine from mets_url by removing last url segment
146			if not src_baseurl:
147			last_segment = nth_url_segment(mets_url)
148			src_baseurl = remove_non_path_from_url(remove_non_path_from_url(mets_url)[:-len(last_segment)])
149
150			# resolve dst_dir
151			if not dst_dir:
152			if is_local_filename(mets_url):
153			log.debug("Deriving dst_dir %s from %s", Path(mets_url).parent, mets_url)
154			dst_dir = Path(mets_url).parent
155			else:
156			log.debug("Creating ephemeral workspace '%s' for METS @ <%s>", dst_dir, mets_url)
157			dst_dir = mkdtemp(prefix=TMP_PREFIX)
158			# XXX Path.resolve is always strict in Python <= 3.5, so create dst_dir unless it exists consistently
159			if not Path(dst_dir).exists():
160			Path(dst_dir).mkdir(parents=True, exist_ok=False)
161			dst_dir = str(Path(dst_dir).resolve())
162
163			log.debug("workspace_from_url\nmets_basename='%s'\nmets_url='%s'\nsrc_baseurl='%s'\ndst_dir='%s'",
164			mets_basename, mets_url, src_baseurl, dst_dir)
165
166			self.download_to_directory(dst_dir, mets_url, basename=mets_basename, if_exists='overwrite' if clobber_mets else 'skip')
167
168			workspace = Workspace(self, dst_dir, mets_basename=mets_basename, baseurl=src_baseurl)
169
170			if download:
171			for f in workspace.mets.find_files():
172			workspace.download_file(f)
173
174			return workspace
175
176			def workspace_from_nothing(self, directory, mets_basename='mets.xml', clobber_mets=False):
177			"""
178			Create an empty workspace.
179
180			Arguments:
181			directory (string): Target directory for the workspace. \
182			If ``none``, create a temporary directory under :py:data:`ocrd.constants.TMP_PREFIX`. \
183			(The resulting path can be retrieved via :py:attr:`ocrd.Workspace.directory`.)
184			Keyword Arguments:
185			clobber_mets (boolean, False): Whether to overwrite existing ``mets.xml``. \
186			By default existing ``mets.xml`` will raise an exception.
187
188			Returns:
189			a new :py:class:`~ocrd.workspace.Workspace`
190			"""
191			log = getLogger('ocrd.resolver.workspace_from_nothing')
192			if directory is None:
193			directory = mkdtemp(prefix=TMP_PREFIX)
194			Path(directory).mkdir(parents=True, exist_ok=True)
195			mets_path = Path(directory, mets_basename)
196			if mets_path.exists() and not clobber_mets:
197			raise FileExistsError("METS '%s' already exists in '%s' and clobber_mets not set." % (mets_basename, directory))
198			mets = OcrdMets.empty_mets()
199			log.info("Writing METS to %s", mets_path)
200			mets_path.write_bytes(mets.to_xml(xmllint=True))
201
202			return Workspace(self, directory, mets, mets_basename=mets_basename)
203
204			def resolve_mets_arguments(self, directory, mets_url, mets_basename):
205			"""
206			Resolve the ``--mets``, ``--mets-basename`` and `--directory`` argument
207			into a coherent set of arguments according to https://github.com/OCR-D/core/issues/517
208			"""
209			log = getLogger('ocrd.resolver.resolve_mets_arguments')
210			mets_is_remote = mets_url and (mets_url.startswith('http://') or mets_url.startswith('https://'))
211
212			# XXX we might want to be more strict like this but it might break # legacy code
213			# Allow --mets and --directory together iff --mets is a remote URL
214			# if directory and mets_url and not mets_is_remote:
215			# raise ValueError("Use either --mets or --directory, not both")
216
217			# If --mets is a URL, a directory must be explicitly provided (not strictly necessary, but retained for legacy behavior)
218			if not directory and mets_is_remote:
219			raise ValueError("--mets is an http(s) URL but no --directory was given")
220
221			# Determine --mets-basename
222			if not mets_basename and mets_url:
223			mets_basename = Path(mets_url).name
224			elif not mets_basename and not mets_url:
225			mets_basename = 'mets.xml'
226			elif mets_basename and mets_url:
227			raise ValueError("Use either --mets or --mets-basename, not both")
228			else:
229			warn("--mets-basename is deprecated. Use --mets/--directory instead", DeprecationWarning)
230
231			# Determine --directory and --mets-url
232			if not directory and not mets_url:
233			directory = Path.cwd()
234			mets_url = Path(directory, mets_basename)
235			elif directory and not mets_url:
236			directory = Path(directory).resolve()
237			mets_url = directory / mets_basename
238			elif not directory and mets_url:
239			mets_url = Path(mets_url).resolve()
240			directory = mets_url.parent
241			else: # == directory and mets_url:
242			directory = Path(directory).resolve()
243			if not mets_is_remote:
244			# --mets is just a basename and --directory is set, so treat --mets as --mets-basename
245			if Path(mets_url).parent == Path('.'):
246			mets_url = directory / mets_url
247			else:
248			mets_url = Path(mets_url).resolve()
249			if not is_file_in_directory(directory, mets_url):
250			raise ValueError("--mets '%s' has a directory part inconsistent with --directory '%s'" % (mets_url, directory))
251
252			return str(Path(directory).resolve()), str(mets_url), str(mets_basename)
253
254
255

OCR-D / core

Pull Request — master (#696)

ocrd.resolver.Resolver.workspace_from_nothing() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like