ocrd_utils.str.make_xml_id() - Code Metrics - Inspection of ":package: :fire: v3.0.0" - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Push — master ( 71c0c1...5d69e4 )

by Konstantin

created 2025-01-09 10:30 UTC

ocrd_utils.str.make_xml_id() A

↳ Parent: ocrd_utils.str

Complexity

Conditions

Size

Total Lines	11
Code Lines	8

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	8
dl	0
loc	11
rs	10
c	0
b	0
f	0
cc	2
nop	1

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

import re
import json
from typing import List
from .constants import REGEX_FILE_ID, SPARKLINE_CHARS
#from .deprecate import deprecation_warning
from deprecated import deprecated
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',
    'make_xml_id',
    'nth_url_segment',
    'parse_json_file_with_comments',
    'parse_json_string_or_file',
    'parse_json_string_with_comments',
    'remove_non_path_from_url',
    'safe_filename',
]


@deprecated(version='3.0', reason='specify input and output file_grp_cardinality in ocrd-tool.json instead')
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
    return make_xml_id(ret)

def make_xml_id(idstr: str) -> str:
    """
    Turn ``idstr`` into a valid ``xml:id`` literal by replacing ``:`` with ``_``, removing everything non-alphanumeric, ``.`` and ``-`` and prepending `id_` if ``idstr`` starts with a number.
    """
    ret = idstr
    if not REGEX_FILE_ID.fullmatch(ret):
        ret = ret.replace(':', '_')
        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_file_with_comments(val):
    """
    Parse a file of JSON interspersed with #-prefixed full-line comments
    """
    with open(val, 'r', encoding='utf-8') as inputf:
        return parse_json_string_with_comments(inputf.read())

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, resolve_preset_file=None):    # 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:
                path = value
                if callable(resolve_preset_file):
                    path = resolve_preset_file(value) or value
                with open(path, '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 : str, end : str) -> List[str]:
    """
    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[:-len(start_num)] != end[:-len(end_num)]:
        raise ValueError(f"Range '{start}..{end}' differ in their non-numeric part: '{start[:-len(start_num)]}' != '{end[:-len(end_num)]}'")
    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
8			from .constants import REGEX_FILE_ID, SPARKLINE_CHARS
9			#from .deprecate import deprecation_warning
10			from deprecated import deprecated
11			from warnings import warn
12			from numpy import array_split
13
14			__all__ = [
15			'assert_file_grp_cardinality',
16			'concat_padded',
17			'get_local_filename',
18			'is_local_filename',
19			'partition_list',
20			'is_string',
21			'make_file_id',
22			'make_xml_id',
23			'nth_url_segment',
24			'parse_json_file_with_comments',
25			'parse_json_string_or_file',
26			'parse_json_string_with_comments',
27			'remove_non_path_from_url',
28			'safe_filename',
29			]
30
31
32			@deprecated(version='3.0', reason='specify input and output file_grp_cardinality in ocrd-tool.json instead')
33			def assert_file_grp_cardinality(grps, n, msg=None):
34			"""
35			Assert that a string of comma-separated fileGrps contains exactly ``n`` entries.
36			"""
37			if isinstance(grps, str):
38			grps = grps.split(',')
39			assert len(grps) == n, \
40			"Expected exactly %d output file group%s%s, but '%s' has %d" % (
41			n,
42			'' if n == 1 else 's',
43			' (%s)' % msg if msg else '',
44			grps,
45			len(grps)
46			)
47
48			def concat_padded(base, *args):
49			"""
50			Concatenate string and zero-padded 4 digit number
51			"""
52			ret = base
53			for n in args:
54			if is_string(n):
55			ret = "%s_%s" % (ret, n)
56			else:
57			ret = "%s_%04i" % (ret, n)
58			return ret
59
60			def remove_non_path_from_url(url):
61			"""
62			Remove everything from URL after path.
63			"""
64			url = url.split('?', 1)[0] # query
65			url = url.split('#', 1)[0] # fragment identifier
66			url = re.sub(r"/+$", "", url) # trailing slashes
67			return url
68
69			def make_file_id(ocrd_file, output_file_grp):
70			"""
71			Derive a new file ID for an output file from an existing input file ``ocrd_file``
72			and the name of the output file's ``fileGrp/@USE``, ``output_file_grp``.
73			If ``ocrd_file``'s ID contains the input file's fileGrp name, then replace it by ``output_file_grp``.
74			Else if ``ocrd_file`` has a ``pageId`` but it is not contained in the ``ocrd_file.ID``, then
75			concatenate ``output_file_grp`` and ``ocrd_file.pageId``.
76			Otherwise concatenate ``output_file_grp`` with the ``ocrd_file.ID``.
77
78			Note: ``make_file_id`` cannot guarantee that the new ID is unique within an actual
79			:py:class:`ocrd_models.ocrd_mets.OcrdMets`.
80			The caller is responsible for ensuring uniqueness of files to be added.
81			Ultimately, ID conflicts will lead to :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_file`
82			raising an exception.
83			This can be avoided if all processors use ``make_file_id`` consistently for ID generation.
84
85			Note: ``make_file_id`` generates page-specific IDs. For IDs representing page segments
86			or ``pc:AlternativeImage`` files, the output of ``make_file_id`` may need to be concatenated
87			with a unique string for that sub-page element, such as `".IMG"` or the segment ID.
88			"""
89			# considerations for this behaviour:
90			# - uniqueness (in spite of different METS and processor conventions)
91			# - predictability (i.e. output name can be anticipated from the input name)
92			# - stability (i.e. output at least as much sorted and consistent as the input)
93			# ... and all this in spite of --page-id selection and --overwrite
94			# (i.e. --overwrite should target the existing ID, and input vs output
95			# IDs should be different, except when overwriting the input fileGrp)
96			ret = ocrd_file.ID.replace(ocrd_file.fileGrp, output_file_grp)
97			if ret == ocrd_file.ID and output_file_grp != ocrd_file.fileGrp:
98			if ocrd_file.pageId and ocrd_file.pageId not in ocrd_file.ID:
99			ret = output_file_grp + '_' + ocrd_file.pageId
100			else:
101			ret = output_file_grp + '_' + ocrd_file.ID
102			return make_xml_id(ret)
103
104			def make_xml_id(idstr: str) -> str:
105			"""
106			Turn ``idstr`` into a valid ``xml:id`` literal by replacing ``:`` with ``_``, removing everything non-alphanumeric, ``.`` and ``-`` and prepending `id_` if ``idstr`` starts with a number.
107			"""
108			ret = idstr
109			if not REGEX_FILE_ID.fullmatch(ret):
110			ret = ret.replace(':', '_')
111			ret = ret.replace('/', '_')
112			ret = re.sub(r'^([^a-zA-Z_])', r'id_\1', ret)
113			ret = re.sub(r'[^\w.-]', r'', ret)
114			return ret
115
116			def nth_url_segment(url, n=-1):
117			"""
118			Return the last /-delimited segment of a URL-like string
119
120			Arguments:
121			url (string):
122			n (integer): index of segment, default: -1
123			"""
124			segments = remove_non_path_from_url(url).split('/')
125			try:
126			return segments[n]
127			except IndexError:
128			return ''
129
130			def get_local_filename(url, start=None):
131			"""
132			Return local filename, optionally relative to ``start``
133
134			Arguments:
135			url (string): filename or URL
136			start (string): Base path to remove from filename. Raise an exception if not a prefix of url
137			"""
138			if url.startswith('https://') or url.startswith('http:'):
139			raise ValueError("Can't determine local filename of http(s) URL")
140			if url.startswith('file://'):
141			url = url[len('file://'):]
142			# Goobi/Kitodo produces those, they are always absolute
143			if url.startswith('file:/'):
144			url = url[len('file:'):]
145			if start:
146			if not url.startswith(start):
147			raise ValueError("Cannot remove prefix %s from url %s" % (start, url))
148			if not start.endswith('/'):
149			start += '/'
150			url = url[len(start):]
151			return url
152
153			def is_local_filename(url):
154			"""
155			Whether a url is a local filename.
156			"""
157			# deprecation_warning("Deprecated so we spot inconsistent URL/file handling")
158			return url.startswith('file://') or not('://' in url)
159
160			def is_string(val):
161			"""
162			Return whether a value is a ``str``.
163			"""
164			return isinstance(val, str)
165
166
167			def parse_json_file_with_comments(val):
168			"""
169			Parse a file of JSON interspersed with #-prefixed full-line comments
170			"""
171			with open(val, 'r', encoding='utf-8') as inputf:
172			return parse_json_string_with_comments(inputf.read())
173
174			def parse_json_string_with_comments(val):
175			"""
176			Parse a string of JSON interspersed with #-prefixed full-line comments
177			"""
178			jsonstr = re.sub(r'^\s#.$', '', val, flags=re.MULTILINE)
179			return json.loads(jsonstr)
180
181			def parse_json_string_or_file(*values, resolve_preset_file=None): # pylint: disable=unused-argument
182			"""
183			Parse a string as either the path to a JSON object or a literal JSON object.
184
185			Empty strings are equivalent to '{}'
186			"""
187			ret = {}
188			for value in values:
189			err = None
190			value_parsed = None
191			if re.fullmatch(r"\s*", value):
192			continue
193			try:
194			try:
195			path = value
196			if callable(resolve_preset_file):
197			path = resolve_preset_file(value) or value
198			with open(path, 'r') as f:
199			value_parsed = parse_json_string_with_comments(f.read())
200			except (FileNotFoundError, OSError):
201			value_parsed = parse_json_string_with_comments(value.strip())
202			if not isinstance(value_parsed, dict):
203			err = ValueError("Not a valid JSON object: '%s' (parsed as '%s')" % (value, value_parsed))
204			except json.decoder.JSONDecodeError as e:
205			err = ValueError("Error parsing '%s': %s" % (value, e))
206			if err:
207			raise err # pylint: disable=raising-bad-type
208			ret = {ret, value_parsed}
209			return ret
210
211			def safe_filename(url):
212			"""
213			Sanitize input to be safely used as the basename of a local file.
214			"""
215			ret = re.sub(r'[^\w]+', '_', url)
216			ret = re.sub(r'^\.*', '', ret)
217			ret = re.sub(r'\.\.*', '.', ret)
218			# print('safe filename: %s -> %s' % (url, ret))
219			return ret
220
221			def generate_range(start : str, end : str) -> List[str]:
222			"""
223			Generate a list of strings by incrementing the number part of ``start`` until including ``end``.
224			"""
225			ret = []
226			try:
227			start_num, end_num = re.findall(r'\d+', start)[-1], re.findall(r'\d+', end)[-1]
228			except IndexError:
229			raise ValueError("Range '%s..%s': could not find numeric part" % (start, end))
230			if start[:-len(start_num)] != end[:-len(end_num)]:
231			raise ValueError(f"Range '{start}..{end}' differ in their non-numeric part: '{start[:-len(start_num)]}' != '{end[:-len(end_num)]}'")
232			if start_num == end_num:
233			warn("Range '%s..%s': evaluates to the same number")
234			for i in range(int(start_num), int(end_num) + 1):
235			ret.append(start.replace(start_num, str(i).zfill(len(start_num))))
236			return ret
237
238
239			def partition_list(lst, chunks, chunk_index=None):
240			"""
241			Partition a list into roughly equally-sized chunks
242
243			Args:
244			lst (list): list to partition
245			chunks (int): number of chunks to generate (not per chunk!)
246
247			Keyword Args:
248			chunk_index (None\|int): If provided, return only a list consisting of this chunk
249
250			Returns:
251			list(list())
252			"""
253			if not lst:
254			return []
255			# Catch potential empty ranges returned by numpy.array_split
256			# which are problematic in the ocr-d scope
257			if chunks > len(lst):
258			raise ValueError("Amount of chunks bigger than list size")
259			ret = [x.tolist() for x in array_split(lst, chunks)]
260			if chunk_index is not None:
261			return [ret[chunk_index]]
262			return ret
263
264			def sparkline(values : List[int]) -> str:
265			"""
266			Render a list of points with block characters
267			"""
268			if any(x is None or not isinstance(x, (int, float)) or x < 0 for x in values):
269			# return an empty string on non-positive-int values, better not to
270			# output a sparkline than to cancel execution due to problematic input
271			return ''
272			max_value = max(values)
273			max_mapping = len(SPARKLINE_CHARS) - 1
274			# normalize to 0..1 and convert to index in SPARKLINE_CHARS
275			mapped = [int(x / max_value * max_mapping) for x in values]
276			return ''.join(SPARKLINE_CHARS[x] for x in mapped)
277

OCR-D / core

Push — master ( 71c0c1...5d69e4 )

ocrd_utils.str.make_xml_id() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like