ocrd_utils.str.make_file_id() - Code Metrics - Inspection of "ocrd_utils: separate make_xml_id, PcGtsType.id: ap..." - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#1271)

unknown

created 2024-08-16 09:00 UTC

ocrd_utils.str.make_file_id() A

↳ Parent: ocrd_utils.str

Complexity

Conditions

Size

Total Lines	34
Code Lines	7

Duplication

Lines	0
Ratio	0 %

Importance

Changes

Metric	Value
eloc	7
dl	0
loc	34
rs	9.3333
c	0
b	0
f	0
cc	5
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',
    'make_xml_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
    return make_xml_id(ret)

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

OCR-D / core

Pull Request — master (#1271)

ocrd_utils.str.make_file_id() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like