ocrd_utils.str.get_local_filename() - Code Metrics - Inspection of ":memo: changelog" - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( ae6efc...6ec3c6 )

by Konstantin

created 2024-01-30 17:33 UTC

ocrd_utils.str.get_local_filename() B

↳ Parent: ocrd_utils.str

Complexity

Conditions

Size

Total Lines	22
Code Lines	14

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	14
dl	0
loc	22
rs	7.3333
c	0
b	0
f	0
cc	8
nop	2

"""
Utility functions for strings, paths and URL.
"""

import re
import json
from typing import List, Union
from .constants import REGEX_FILE_ID, SPARKLINE_CHARS
from .deprecate import deprecation_warning
from warnings import warn
from numpy import array_split

__all__ = [
    'assert_file_grp_cardinality',
    'concat_padded',
    'get_local_filename',
    'is_local_filename',
    'partition_list',
    'is_string',
    'make_file_id',
    'nth_url_segment',
    'parse_json_string_or_file',
    'parse_json_string_with_comments',
    'remove_non_path_from_url',
    'safe_filename',
]


def assert_file_grp_cardinality(grps, n, msg=None):
    """
    Assert that a string of comma-separated fileGrps contains exactly ``n`` entries.
    """
    if isinstance(grps, str):
        grps = grps.split(',')
    assert len(grps) == n, \
            "Expected exactly %d output file group%s%s, but '%s' has %d" % (
                n,
                '' if n == 1 else 's',
                ' (%s)' % msg if msg else '',
                grps,
                len(grps)
            )

def concat_padded(base, *args):
    """
    Concatenate string and zero-padded 4 digit number
    """
    ret = base
    for n in args:
        if is_string(n):
            ret = "%s_%s" % (ret, n)
        else:
            ret = "%s_%04i"  % (ret, n)
    return ret

def remove_non_path_from_url(url):
    """
    Remove everything from URL after path.
    """
    url = url.split('?', 1)[0]    # query
    url = url.split('#', 1)[0]    # fragment identifier
    url = re.sub(r"/+$", "", url) # trailing slashes
    return url

def make_file_id(ocrd_file, output_file_grp):
    """
    Derive a new file ID for an output file from an existing input file ``ocrd_file``
    and the name of the output file's ``fileGrp/@USE``, ``output_file_grp``.
    If ``ocrd_file``'s ID contains the input file's fileGrp name, then replace it by ``output_file_grp``.
    Else if ``ocrd_file`` has a ``pageId`` but it is not contained in the ``ocrd_file.ID``, then
        concatenate ``output_file_grp`` and ``ocrd_file.pageId``.
    Otherwise concatenate ``output_file_grp`` with the ``ocrd_file.ID``.

    Note: ``make_file_id`` cannot guarantee that the new ID is unique within an actual
    :py:class:`ocrd_models.ocrd_mets.OcrdMets`.
    The caller is responsible for ensuring uniqueness of files to be added.
    Ultimately, ID conflicts will lead to :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_file`
    raising an exception.
    This can be avoided if all processors use ``make_file_id`` consistently for ID generation.

    Note: ``make_file_id`` generates page-specific IDs. For IDs representing page segments
    or ``pc:AlternativeImage`` files, the output of ``make_file_id`` may need to be concatenated
    with a unique string for that sub-page element, such as `".IMG"` or the segment ID.
    """
    # considerations for this behaviour:
    # - uniqueness (in spite of different METS and processor conventions)
    # - predictability (i.e. output name can be anticipated from the input name)
    # - stability (i.e. output at least as much sorted and consistent as the input)
    # ... and all this in spite of --page-id selection and --overwrite
    # (i.e. --overwrite should target the existing ID, and input vs output
    #  IDs should be different, except when overwriting the input fileGrp)
    ret = ocrd_file.ID.replace(ocrd_file.fileGrp, output_file_grp)
    if ret == ocrd_file.ID and output_file_grp != ocrd_file.fileGrp:
        if ocrd_file.pageId and ocrd_file.pageId not in ocrd_file.ID:
            ret = output_file_grp + '_' + ocrd_file.pageId
        else:
            ret = output_file_grp + '_' + ocrd_file.ID
    if not REGEX_FILE_ID.fullmatch(ret):
        ret = ret.replace(':', '_')
        ret = re.sub(r'^([^a-zA-Z_])', r'id_\1', ret)
        ret = re.sub(r'[^\w.-]', r'', ret)
    return ret

def nth_url_segment(url, n=-1):
    """
    Return the last /-delimited segment of a URL-like string

    Arguments:
        url (string):
        n (integer): index of segment, default: -1
    """
    segments = remove_non_path_from_url(url).split('/')
    try:
        return segments[n]
    except IndexError:
        return ''

def get_local_filename(url, start=None):
    """
    Return local filename, optionally relative to ``start``

    Arguments:
        url (string): filename or URL
        start (string): Base path to remove from filename. Raise an exception if not a prefix of url
    """
    if url.startswith('https://') or url.startswith('http:'):
        raise ValueError("Can't determine local filename of http(s) URL")
    if url.startswith('file://'):
        url = url[len('file://'):]
    # Goobi/Kitodo produces those, they are always absolute
    if url.startswith('file:/'):
        url = url[len('file:'):]
    if start:
        if not url.startswith(start):
            raise ValueError("Cannot remove prefix %s from url %s" % (start, url))
        if not start.endswith('/'):
            start += '/'
        url = url[len(start):]
    return url

def is_local_filename(url):
    """
    Whether a url is a local filename.
    """
    # deprecation_warning("Deprecated so we spot inconsistent URL/file handling")
    return url.startswith('file://') or not('://' in url)

def is_string(val):
    """
    Return whether a value is a ``str``.
    """
    return isinstance(val, str)


def parse_json_string_with_comments(val):
    """
    Parse a string of JSON interspersed with #-prefixed full-line comments
    """
    jsonstr = re.sub(r'^\s*#.*$', '', val, flags=re.MULTILINE)
    return json.loads(jsonstr)

def parse_json_string_or_file(*values):    # pylint: disable=unused-argument
    """
    Parse a string as either the path to a JSON object or a literal JSON object.

    Empty strings are equivalent to '{}'
    """
    ret = {}
    for value in values:
        err = None
        value_parsed = None
        if re.fullmatch(r"\s*", value):
            continue
        try:
            try:
                with open(value, 'r') as f:
                    value_parsed = parse_json_string_with_comments(f.read())
            except (FileNotFoundError, OSError):
                value_parsed = parse_json_string_with_comments(value.strip())
            if not isinstance(value_parsed, dict):
                err = ValueError("Not a valid JSON object: '%s' (parsed as '%s')" % (value, value_parsed))
        except json.decoder.JSONDecodeError as e:
            err = ValueError("Error parsing '%s': %s" % (value, e))
        if err:
            raise err       # pylint: disable=raising-bad-type
        ret = {**ret, **value_parsed}
    return ret

def safe_filename(url):
    """
    Sanitize input to be safely used as the basename of a local file.
    """
    ret = re.sub(r'[^\w]+', '_', url)
    ret = re.sub(r'^\.*', '', ret)
    ret = re.sub(r'\.\.*', '.', ret)
    #  print('safe filename: %s -> %s' % (url, ret))
    return ret

def generate_range(start, end):
    """
    Generate a list of strings by incrementing the number part of ``start`` until including ``end``.
    """
    ret = []
    try:
        start_num, end_num = re.findall(r'\d+', start)[-1], re.findall(r'\d+', end)[-1]
    except IndexError:
        raise ValueError("Range '%s..%s': could not find numeric part" % (start, end))
    if start_num == end_num:
        warn("Range '%s..%s': evaluates to the same number")
    for i in range(int(start_num), int(end_num) + 1):
        ret.append(start.replace(start_num, str(i).zfill(len(start_num))))
    return ret


def partition_list(lst, chunks, chunk_index=None):
    """
    Partition a list into roughly equally-sized chunks

    Args:
        lst (list): list to partition
        chunks (int): number of chunks to generate (not per chunk!)

    Keyword Args:
        chunk_index (None|int): If provided, return only a list consisting of this chunk

    Returns:
        list(list())
    """
    if not lst:
        return []
    # Catch potential empty ranges returned by numpy.array_split
    #  which are problematic in the ocr-d scope
    if chunks > len(lst):
        raise ValueError("Amount of chunks bigger than list size")
    ret = [x.tolist() for x in array_split(lst, chunks)]
    if chunk_index is not None:
        return [ret[chunk_index]]
    return ret

def sparkline(values : List[int]) -> str:
    """
    Render a list of points with block characters
    """
    if any(x is None or not isinstance(x, (int, float)) or x < 0 for x in values):
        # return an empty string on non-positive-int values, better not to
        # output a sparkline than to cancel execution due to problematic input
        return ''
    max_value = max(values)
    max_mapping = len(SPARKLINE_CHARS) - 1
    # normalize to 0..1 and convert to index in SPARKLINE_CHARS
    mapped = [int(x / max_value * max_mapping) for x in values]
    return ''.join(SPARKLINE_CHARS[x] for x in mapped)



1			"""
2			Utility functions for strings, paths and URL.
3			"""
4
5			import re
6			import json
7			from typing import List, Union
8			from .constants import REGEX_FILE_ID, SPARKLINE_CHARS
9			from .deprecate import deprecation_warning
10			from warnings import warn
11			from numpy import array_split
12
13			__all__ = [
14			'assert_file_grp_cardinality',
15			'concat_padded',
16			'get_local_filename',
17			'is_local_filename',
18			'partition_list',
19			'is_string',
20			'make_file_id',
21			'nth_url_segment',
22			'parse_json_string_or_file',
23			'parse_json_string_with_comments',
24			'remove_non_path_from_url',
25			'safe_filename',
26			]
27
28
29			def assert_file_grp_cardinality(grps, n, msg=None):
30			"""
31			Assert that a string of comma-separated fileGrps contains exactly ``n`` entries.
32			"""
33			if isinstance(grps, str):
34			grps = grps.split(',')
35			assert len(grps) == n, \
36			"Expected exactly %d output file group%s%s, but '%s' has %d" % (
37			n,
38			'' if n == 1 else 's',
39			' (%s)' % msg if msg else '',
40			grps,
41			len(grps)
42			)
43
44			def concat_padded(base, *args):
45			"""
46			Concatenate string and zero-padded 4 digit number
47			"""
48			ret = base
49			for n in args:
50			if is_string(n):
51			ret = "%s_%s" % (ret, n)
52			else:
53			ret = "%s_%04i" % (ret, n)
54			return ret
55
56			def remove_non_path_from_url(url):
57			"""
58			Remove everything from URL after path.
59			"""
60			url = url.split('?', 1)[0] # query
61			url = url.split('#', 1)[0] # fragment identifier
62			url = re.sub(r"/+$", "", url) # trailing slashes
63			return url
64
65			def make_file_id(ocrd_file, output_file_grp):
66			"""
67			Derive a new file ID for an output file from an existing input file ``ocrd_file``
68			and the name of the output file's ``fileGrp/@USE``, ``output_file_grp``.
69			If ``ocrd_file``'s ID contains the input file's fileGrp name, then replace it by ``output_file_grp``.
70			Else if ``ocrd_file`` has a ``pageId`` but it is not contained in the ``ocrd_file.ID``, then
71			concatenate ``output_file_grp`` and ``ocrd_file.pageId``.
72			Otherwise concatenate ``output_file_grp`` with the ``ocrd_file.ID``.
73
74			Note: ``make_file_id`` cannot guarantee that the new ID is unique within an actual
75			:py:class:`ocrd_models.ocrd_mets.OcrdMets`.
76			The caller is responsible for ensuring uniqueness of files to be added.
77			Ultimately, ID conflicts will lead to :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_file`
78			raising an exception.
79			This can be avoided if all processors use ``make_file_id`` consistently for ID generation.
80
81			Note: ``make_file_id`` generates page-specific IDs. For IDs representing page segments
82			or ``pc:AlternativeImage`` files, the output of ``make_file_id`` may need to be concatenated
83			with a unique string for that sub-page element, such as `".IMG"` or the segment ID.
84			"""
85			# considerations for this behaviour:
86			# - uniqueness (in spite of different METS and processor conventions)
87			# - predictability (i.e. output name can be anticipated from the input name)
88			# - stability (i.e. output at least as much sorted and consistent as the input)
89			# ... and all this in spite of --page-id selection and --overwrite
90			# (i.e. --overwrite should target the existing ID, and input vs output
91			# IDs should be different, except when overwriting the input fileGrp)
92			ret = ocrd_file.ID.replace(ocrd_file.fileGrp, output_file_grp)
93			if ret == ocrd_file.ID and output_file_grp != ocrd_file.fileGrp:
94			if ocrd_file.pageId and ocrd_file.pageId not in ocrd_file.ID:
95			ret = output_file_grp + '_' + ocrd_file.pageId
96			else:
97			ret = output_file_grp + '_' + ocrd_file.ID
98			if not REGEX_FILE_ID.fullmatch(ret):
99			ret = ret.replace(':', '_')
100			ret = re.sub(r'^([^a-zA-Z_])', r'id_\1', ret)
101			ret = re.sub(r'[^\w.-]', r'', ret)
102			return ret
103
104			def nth_url_segment(url, n=-1):
105			"""
106			Return the last /-delimited segment of a URL-like string
107
108			Arguments:
109			url (string):
110			n (integer): index of segment, default: -1
111			"""
112			segments = remove_non_path_from_url(url).split('/')
113			try:
114			return segments[n]
115			except IndexError:
116			return ''
117
118			def get_local_filename(url, start=None):
119			"""
120			Return local filename, optionally relative to ``start``
121
122			Arguments:
123			url (string): filename or URL
124			start (string): Base path to remove from filename. Raise an exception if not a prefix of url
125			"""
126			if url.startswith('https://') or url.startswith('http:'):
127			raise ValueError("Can't determine local filename of http(s) URL")
128			if url.startswith('file://'):
129			url = url[len('file://'):]
130			# Goobi/Kitodo produces those, they are always absolute
131			if url.startswith('file:/'):
132			url = url[len('file:'):]
133			if start:
134			if not url.startswith(start):
135			raise ValueError("Cannot remove prefix %s from url %s" % (start, url))
136			if not start.endswith('/'):
137			start += '/'
138			url = url[len(start):]
139			return url
140
141			def is_local_filename(url):
142			"""
143			Whether a url is a local filename.
144			"""
145			# deprecation_warning("Deprecated so we spot inconsistent URL/file handling")
146			return url.startswith('file://') or not('://' in url)
147
148			def is_string(val):
149			"""
150			Return whether a value is a ``str``.
151			"""
152			return isinstance(val, str)
153
154
155			def parse_json_string_with_comments(val):
156			"""
157			Parse a string of JSON interspersed with #-prefixed full-line comments
158			"""
159			jsonstr = re.sub(r'^\s#.$', '', val, flags=re.MULTILINE)
160			return json.loads(jsonstr)
161
162			def parse_json_string_or_file(*values): # pylint: disable=unused-argument
163			"""
164			Parse a string as either the path to a JSON object or a literal JSON object.
165
166			Empty strings are equivalent to '{}'
167			"""
168			ret = {}
169			for value in values:
170			err = None
171			value_parsed = None
172			if re.fullmatch(r"\s*", value):
173			continue
174			try:
175			try:
176			with open(value, 'r') as f:
177			value_parsed = parse_json_string_with_comments(f.read())
178			except (FileNotFoundError, OSError):
179			value_parsed = parse_json_string_with_comments(value.strip())
180			if not isinstance(value_parsed, dict):
181			err = ValueError("Not a valid JSON object: '%s' (parsed as '%s')" % (value, value_parsed))
182			except json.decoder.JSONDecodeError as e:
183			err = ValueError("Error parsing '%s': %s" % (value, e))
184			if err:
185			raise err # pylint: disable=raising-bad-type
186			ret = {ret, value_parsed}
187			return ret
188
189			def safe_filename(url):
190			"""
191			Sanitize input to be safely used as the basename of a local file.
192			"""
193			ret = re.sub(r'[^\w]+', '_', url)
194			ret = re.sub(r'^\.*', '', ret)
195			ret = re.sub(r'\.\.*', '.', ret)
196			# print('safe filename: %s -> %s' % (url, ret))
197			return ret
198
199			def generate_range(start, end):
200			"""
201			Generate a list of strings by incrementing the number part of ``start`` until including ``end``.
202			"""
203			ret = []
204			try:
205			start_num, end_num = re.findall(r'\d+', start)[-1], re.findall(r'\d+', end)[-1]
206			except IndexError:
207			raise ValueError("Range '%s..%s': could not find numeric part" % (start, end))
208			if start_num == end_num:
209			warn("Range '%s..%s': evaluates to the same number")
210			for i in range(int(start_num), int(end_num) + 1):
211			ret.append(start.replace(start_num, str(i).zfill(len(start_num))))
212			return ret
213
214
215			def partition_list(lst, chunks, chunk_index=None):
216			"""
217			Partition a list into roughly equally-sized chunks
218
219			Args:
220			lst (list): list to partition
221			chunks (int): number of chunks to generate (not per chunk!)
222
223			Keyword Args:
224			chunk_index (None\|int): If provided, return only a list consisting of this chunk
225
226			Returns:
227			list(list())
228			"""
229			if not lst:
230			return []
231			# Catch potential empty ranges returned by numpy.array_split
232			# which are problematic in the ocr-d scope
233			if chunks > len(lst):
234			raise ValueError("Amount of chunks bigger than list size")
235			ret = [x.tolist() for x in array_split(lst, chunks)]
236			if chunk_index is not None:
237			return [ret[chunk_index]]
238			return ret
239
240			def sparkline(values : List[int]) -> str:
241			"""
242			Render a list of points with block characters
243			"""
244			if any(x is None or not isinstance(x, (int, float)) or x < 0 for x in values):
245			# return an empty string on non-positive-int values, better not to
246			# output a sparkline than to cancel execution due to problematic input
247			return ''
248			max_value = max(values)
249			max_mapping = len(SPARKLINE_CHARS) - 1
250			# normalize to 0..1 and convert to index in SPARKLINE_CHARS
251			mapped = [int(x / max_value * max_mapping) for x in values]
252			return ''.join(SPARKLINE_CHARS[x] for x in mapped)
253
254

OCR-D / core

Push — master ( ae6efc...6ec3c6 )

ocrd_utils.str.get_local_filename() B

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like