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

Passed

Push — master ( 0ea71f...5b7c9f )

by Konstantin

created 2023-12-05 13:58 UTC

ocrd_utils.str.batched() A

↳ Parent: ocrd_utils.str

Complexity

Conditions

Size

Total Lines	8
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	8
dl	0
loc	8
rs	10
c	0
b	0
f	0
cc	3
nop	2

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

import re
import json
from .constants import REGEX_FILE_ID
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


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

OCR-D / core

Push — master ( 0ea71f...5b7c9f )

ocrd_utils.str.batched() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like