ocrd_utils.config.OcrdEnvConfig.__init__() - Code Metrics - Inspection of "New processor API" - OCR-D/core - Measure and Improve Code Quality continuously with Scrutinizer

Passed

Pull Request — master (#1240)

unknown

created 2024-08-27 09:11 UTC

ocrd_utils.config.OcrdEnvConfig.init() A

↳ Parent: ocrd_utils.config

Complexity

Conditions

Size

Total Lines	2
Code Lines	2

Duplication

Lines	0
Ratio	0 %

Importance

Changes

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

"""
Most behavior of OCR-D is controlled via command-line flags or keyword args.
Some behavior is global or too cumbersome to handle via explicit code and
better solved by using environment variables.

OcrdEnvConfig is a base class to make this more streamlined, to be subclassed
in the `ocrd` package for the actual values
"""

from os import environ
from pathlib import Path
from tempfile import gettempdir
from textwrap import fill, indent


def _validator_boolean(val):
    return isinstance(val, bool) or str.lower(val) in ('true', 'false', '0', '1')

def _parser_boolean(val):
    return bool(val) if isinstance(val, (int, bool)) else str.lower(val) in ('true', '1')

class OcrdEnvVariable():

    def __init__(self, name, description, parser=str, validator=lambda val: True, default=[False, None]):
        """
        An environment variable for use in OCR-D.

        Args:
            name (str): Name of the environment variable
            description (str): Description of what the variable is used for.

        Keyword Args:
            parser (callable): Function to transform the raw (string) value to whatever is needed.
            validator (callable): Function to validate that the raw (string) value is parseable.
            default (tuple(bool, any)): 2-tuple, first element is a bool whether there is a default
                value defined and second element contains that default value, which can be a callable
                for deferred evaluation
        """
        self.name = name
        self.description = description
        self.parser = parser
        self.validator = validator
        self.has_default = default[0]
        self.default = default[1]

    def __str__(self):
        return f'{self.name}: {self.description}'

    def describe(self, wrap_text=True, indent_text=True):
        desc = self.description
        if self.has_default:
            default = self.default() if callable(self.default) else self.default
            desc += f' (Default: "{default}")'
        ret = ''
        ret  = f'{self.name}\n'
        if wrap_text:
            desc = fill(desc, width=50)
        if indent_text:
            ret = f'  {ret}'
            desc = indent(desc, '    ')
        return ret + desc

class OcrdEnvConfig():

    def __init__(self):
        self._variables = {}

    def add(self, name, *args, **kwargs):
        var = OcrdEnvVariable(name, *args, **kwargs)
        # make visible in ocrd_utils.config docstring (apidoc)
        txt = var.describe(wrap_text=False, indent_text=True)
        globals()['__doc__'] += "\n\n - " + txt + "\n\n"

        self._variables[name] = var
        return self._variables[name]

    def has_default(self, name):
        if not name in self._variables:
            raise ValueError(f"Unregistered env variable {name}")
        return self._variables[name].has_default

    def describe(self, name, *args, **kwargs):
        if not name in self._variables:
            raise ValueError(f"Unregistered env variable {name}")
        return self._variables[name].describe(*args, **kwargs)

    def __getattr__(self, name):
        if not name in self._variables:
            raise ValueError(f"Unregistered env variable {name}")
        var_obj = self._variables[name]
        try:
            raw_value = self.raw_value(name)
        except KeyError as e:
            if var_obj.has_default:
                raw_value = var_obj.default() if callable(var_obj.default) else var_obj.default
            else:
                raise e
        if not var_obj.validator(raw_value):
            raise ValueError(f"'{name}' set to invalid value '{raw_value}'")
        return var_obj.parser(raw_value)

    def is_set(self, name):
        if not name in self._variables:
            raise ValueError(f"Unregistered env variable {name}")
        return name in environ

    def raw_value(self, name):
        if not name in self._variables:
            raise ValueError(f"Unregistered env variable {name}")
        return environ[name]

config = OcrdEnvConfig()

config.add('OCRD_METS_CACHING',
    description='If set to `true`, access to the METS file is cached, speeding in-memory search and modification.',
    validator=_validator_boolean,
    parser=_parser_boolean)

config.add('OCRD_MAX_PROCESSOR_CACHE',
    description="Maximum number of processor instances (for each set of parameters) to be kept in memory (including loaded models) for processing workers or processor servers.",
    parser=int,
    default=(True, 128))

config.add("OCRD_PROFILE",
    description="""\
Whether to enable gathering runtime statistics
on the `ocrd.profile` logger (comma-separated):

- `CPU`: yields CPU and wall-time,
- `RSS`: also yields peak memory (resident set size)
- `PSS`: also yields peak memory (proportional set size)

""",
  validator=lambda val : all(t in ('', 'CPU', 'RSS', 'PSS') for t in val.split(',')),
  default=(True, ''))

config.add("OCRD_PROFILE_FILE",
    description="If set, then the CPU profile is written to this file for later peruse with a analysis tools like snakeviz")

config.add("OCRD_DOWNLOAD_RETRIES",
    description="Number of times to retry failed attempts for downloads of resources or workspace files.",
    validator=int,
    parser=int)

def _ocrd_download_timeout_parser(val):
    timeout = val.split(',')
    if len(timeout) > 1:
        timeout = tuple(float(x) for x in timeout)
    else:
        timeout = float(timeout[0])
    return timeout

config.add("OCRD_DOWNLOAD_TIMEOUT",
    description="Timeout in seconds for connecting or reading (comma-separated) when downloading.",
    parser=_ocrd_download_timeout_parser)

config.add("OCRD_DOWNLOAD_INPUT",
    description="Whether to download files not present locally during processing",
    default=(True, True),
    validator=_validator_boolean,
    parser=_parser_boolean)

config.add("OCRD_MISSING_INPUT",
    description="""\
How to deal with missing input files (for some fileGrp/pageId) during processing:

 - `SKIP`: ignore and proceed with next page's input
 - `ABORT`: throw :py:class:`.MissingInputFile`

""",
    default=(True, 'SKIP'),
    validator=lambda val: val in ['SKIP', 'ABORT'],
    parser=str)

config.add("OCRD_MISSING_OUTPUT",
    description="""\
How to deal with missing output files (for some fileGrp/pageId) during processing:

 - `SKIP`: ignore and proceed processing next page
 - `COPY`: fall back to copying input PAGE to output fileGrp for page
 - `ABORT`: re-throw whatever caused processing to fail

""",
    default=(True, 'SKIP'),
    validator=lambda val: val in ['SKIP', 'COPY', 'ABORT'],
    parser=str)

config.add("OCRD_MAX_MISSING_OUTPUTS",
    description="Maximal rate of skipped/fallback pages among all processed pages before aborting (decimal fraction, ignored if negative).",
    default=(True, 0.1),
    parser=float)

config.add("OCRD_EXISTING_OUTPUT",
    description="""\
How to deal with already existing output files (for some fileGrp/pageId) during processing:

 - `SKIP`: ignore and proceed processing next page
 - `OVERWRITE`: force writing result to output fileGrp for page
 - `ABORT`: re-throw :py:class:`FileExistsError`

""",
    default=(True, 'SKIP'),
    validator=lambda val: val in ['SKIP', 'OVERWRITE', 'ABORT'],
    parser=str)

config.add("OCRD_NETWORK_SERVER_ADDR_PROCESSING",
        description="Default address of Processing Server to connect to (for `ocrd network client processing`).",
        default=(True, ''))

config.add("OCRD_NETWORK_CLIENT_POLLING_SLEEP",
           description="How many seconds to sleep before trying again.",
           parser=int,
           default=(True, 30))

config.add("OCRD_NETWORK_CLIENT_POLLING_TIMEOUT",
           description="Timeout for a blocking ocrd network client (in seconds).",
           parser=int,
           default=(True, 3600))

config.add("OCRD_NETWORK_SERVER_ADDR_WORKFLOW",
        description="Default address of Workflow Server to connect to (for `ocrd network client workflow`).",
        default=(True, ''))

config.add("OCRD_NETWORK_SERVER_ADDR_WORKSPACE",
        description="Default address of Workspace Server to connect to (for `ocrd network client workspace`).",
        default=(True, ''))

config.add("OCRD_NETWORK_RABBITMQ_CLIENT_CONNECT_ATTEMPTS",
    description="Number of attempts for a RabbitMQ client to connect before failing.",
    parser=int,
    default=(True, 3))

config.add(name="OCRD_NETWORK_SOCKETS_ROOT_DIR",
           description="The root directory where all mets server related socket files are created",
           parser=lambda val: Path(val),
           default=(True, Path(gettempdir(), "ocrd_network_sockets")))
config.OCRD_NETWORK_SOCKETS_ROOT_DIR.mkdir(parents=True, exist_ok=True)

config.add(name="OCRD_NETWORK_LOGS_ROOT_DIR",
           description="The root directory where all ocrd_network related file logs are stored",
           parser=lambda val: Path(val),
           default=(True, Path(gettempdir(), "ocrd_network_logs")))
config.OCRD_NETWORK_LOGS_ROOT_DIR.mkdir(parents=True, exist_ok=True)

config.add("HOME",
    description="Directory to look for `ocrd_logging.conf`, fallback for unset XDG variables.",
    # description="HOME directory, cf. https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html",
    validator=lambda val: Path(val).is_dir(),
    parser=lambda val: Path(val),
    default=(True, lambda: Path.home()))

config.add("XDG_DATA_HOME",
    description="Directory to look for `./ocrd-resources/*` (i.e. `ocrd resmgr` data location)",
    parser=lambda val: Path(val),
    default=(True, lambda: Path(config.HOME, '.local/share')))

config.add("XDG_CONFIG_HOME",
    description="Directory to look for `./ocrd/resources.yml` (i.e. `ocrd resmgr` user database)",
    parser=lambda val: Path(val),
    default=(True, lambda: Path(config.HOME, '.config')))

config.add("OCRD_LOGGING_DEBUG",
    description="Print information about the logging setup to STDERR",
    default=(True, False),
    validator=_validator_boolean,
    parser=_parser_boolean)


1			"""
2			Most behavior of OCR-D is controlled via command-line flags or keyword args.
3			Some behavior is global or too cumbersome to handle via explicit code and
4			better solved by using environment variables.
5
6			OcrdEnvConfig is a base class to make this more streamlined, to be subclassed
7			in the `ocrd` package for the actual values
8			"""
9
10			from os import environ
11			from pathlib import Path
12			from tempfile import gettempdir
13			from textwrap import fill, indent
14
15
16			def _validator_boolean(val):
17			return isinstance(val, bool) or str.lower(val) in ('true', 'false', '0', '1')
18
19			def _parser_boolean(val):
20			return bool(val) if isinstance(val, (int, bool)) else str.lower(val) in ('true', '1')
21
22			class OcrdEnvVariable():
23
24			def __init__(self, name, description, parser=str, validator=lambda val: True, default=[False, None]):
25			"""
26			An environment variable for use in OCR-D.
27
28			Args:
29			name (str): Name of the environment variable
30			description (str): Description of what the variable is used for.
31
32			Keyword Args:
33			parser (callable): Function to transform the raw (string) value to whatever is needed.
34			validator (callable): Function to validate that the raw (string) value is parseable.
35			default (tuple(bool, any)): 2-tuple, first element is a bool whether there is a default
36			value defined and second element contains that default value, which can be a callable
37			for deferred evaluation
38			"""
39			self.name = name
40			self.description = description
41			self.parser = parser
42			self.validator = validator
43			self.has_default = default[0]
44			self.default = default[1]
45
46			def __str__(self):
47			return f'{self.name}: {self.description}'
48
49			def describe(self, wrap_text=True, indent_text=True):
50			desc = self.description
51			if self.has_default:
52			default = self.default() if callable(self.default) else self.default
53			desc += f' (Default: "{default}")'
54			ret = ''
55			ret = f'{self.name}\n'
56			if wrap_text:
57			desc = fill(desc, width=50)
58			if indent_text:
59			ret = f' {ret}'
60			desc = indent(desc, ' ')
61			return ret + desc
62
63			class OcrdEnvConfig():
64
65			def __init__(self):
66			self._variables = {}
67
68			def add(self, name, args, *kwargs):
69			var = OcrdEnvVariable(name, args, *kwargs)
70			# make visible in ocrd_utils.config docstring (apidoc)
71			txt = var.describe(wrap_text=False, indent_text=True)
72			globals()['__doc__'] += "\n\n - " + txt + "\n\n"
			0 ignored issues – show Comprehensibility Best Practice introduced 2024-08-21 20:54 UTC by Report Bug Copy Issue Report The variable `globals` does not seem to be defined. Loading history...
73			self._variables[name] = var
74			return self._variables[name]
75
76			def has_default(self, name):
77			if not name in self._variables:
78			raise ValueError(f"Unregistered env variable {name}")
79			return self._variables[name].has_default
80
81			def describe(self, name, args, *kwargs):
82			if not name in self._variables:
83			raise ValueError(f"Unregistered env variable {name}")
84			return self._variables[name].describe(args, *kwargs)
85
86			def __getattr__(self, name):
87			if not name in self._variables:
88			raise ValueError(f"Unregistered env variable {name}")
89			var_obj = self._variables[name]
90			try:
91			raw_value = self.raw_value(name)
92			except KeyError as e:
93			if var_obj.has_default:
94			raw_value = var_obj.default() if callable(var_obj.default) else var_obj.default
95			else:
96			raise e
97			if not var_obj.validator(raw_value):
98			raise ValueError(f"'{name}' set to invalid value '{raw_value}'")
99			return var_obj.parser(raw_value)
100
101			def is_set(self, name):
102			if not name in self._variables:
103			raise ValueError(f"Unregistered env variable {name}")
104			return name in environ
105
106			def raw_value(self, name):
107			if not name in self._variables:
108			raise ValueError(f"Unregistered env variable {name}")
109			return environ[name]
110
111			config = OcrdEnvConfig()
112
113			config.add('OCRD_METS_CACHING',
114			description='If set to `true`, access to the METS file is cached, speeding in-memory search and modification.',
115			validator=_validator_boolean,
116			parser=_parser_boolean)
117
118			config.add('OCRD_MAX_PROCESSOR_CACHE',
119			description="Maximum number of processor instances (for each set of parameters) to be kept in memory (including loaded models) for processing workers or processor servers.",
120			parser=int,
121			default=(True, 128))
122
123			config.add("OCRD_PROFILE",
124			description="""\
125			Whether to enable gathering runtime statistics
126			on the `ocrd.profile` logger (comma-separated):
127
128			- `CPU`: yields CPU and wall-time,
129			- `RSS`: also yields peak memory (resident set size)
130			- `PSS`: also yields peak memory (proportional set size)
131
132			""",
133			validator=lambda val : all(t in ('', 'CPU', 'RSS', 'PSS') for t in val.split(',')),
134			default=(True, ''))
135
136			config.add("OCRD_PROFILE_FILE",
137			description="If set, then the CPU profile is written to this file for later peruse with a analysis tools like snakeviz")
138
139			config.add("OCRD_DOWNLOAD_RETRIES",
140			description="Number of times to retry failed attempts for downloads of resources or workspace files.",
141			validator=int,
142			parser=int)
143
144			def _ocrd_download_timeout_parser(val):
145			timeout = val.split(',')
146			if len(timeout) > 1:
147			timeout = tuple(float(x) for x in timeout)
148			else:
149			timeout = float(timeout[0])
150			return timeout
151
152			config.add("OCRD_DOWNLOAD_TIMEOUT",
153			description="Timeout in seconds for connecting or reading (comma-separated) when downloading.",
154			parser=_ocrd_download_timeout_parser)
155
156			config.add("OCRD_DOWNLOAD_INPUT",
157			description="Whether to download files not present locally during processing",
158			default=(True, True),
159			validator=_validator_boolean,
160			parser=_parser_boolean)
161
162			config.add("OCRD_MISSING_INPUT",
163			description="""\
164			How to deal with missing input files (for some fileGrp/pageId) during processing:
165
166			- `SKIP`: ignore and proceed with next page's input
167			- `ABORT`: throw :py:class:`.MissingInputFile`
168
169			""",
170			default=(True, 'SKIP'),
171			validator=lambda val: val in ['SKIP', 'ABORT'],
172			parser=str)
173
174			config.add("OCRD_MISSING_OUTPUT",
175			description="""\
176			How to deal with missing output files (for some fileGrp/pageId) during processing:
177
178			- `SKIP`: ignore and proceed processing next page
179			- `COPY`: fall back to copying input PAGE to output fileGrp for page
180			- `ABORT`: re-throw whatever caused processing to fail
181
182			""",
183			default=(True, 'SKIP'),
184			validator=lambda val: val in ['SKIP', 'COPY', 'ABORT'],
185			parser=str)
186
187			config.add("OCRD_MAX_MISSING_OUTPUTS",
188			description="Maximal rate of skipped/fallback pages among all processed pages before aborting (decimal fraction, ignored if negative).",
189			default=(True, 0.1),
190			parser=float)
191
192			config.add("OCRD_EXISTING_OUTPUT",
193			description="""\
194			How to deal with already existing output files (for some fileGrp/pageId) during processing:
195
196			- `SKIP`: ignore and proceed processing next page
197			- `OVERWRITE`: force writing result to output fileGrp for page
198			- `ABORT`: re-throw :py:class:`FileExistsError`
199
200			""",
201			default=(True, 'SKIP'),
202			validator=lambda val: val in ['SKIP', 'OVERWRITE', 'ABORT'],
203			parser=str)
204
205			config.add("OCRD_NETWORK_SERVER_ADDR_PROCESSING",
206			description="Default address of Processing Server to connect to (for `ocrd network client processing`).",
207			default=(True, ''))
208
209			config.add("OCRD_NETWORK_CLIENT_POLLING_SLEEP",
210			description="How many seconds to sleep before trying again.",
211			parser=int,
212			default=(True, 30))
213
214			config.add("OCRD_NETWORK_CLIENT_POLLING_TIMEOUT",
215			description="Timeout for a blocking ocrd network client (in seconds).",
216			parser=int,
217			default=(True, 3600))
218
219			config.add("OCRD_NETWORK_SERVER_ADDR_WORKFLOW",
220			description="Default address of Workflow Server to connect to (for `ocrd network client workflow`).",
221			default=(True, ''))
222
223			config.add("OCRD_NETWORK_SERVER_ADDR_WORKSPACE",
224			description="Default address of Workspace Server to connect to (for `ocrd network client workspace`).",
225			default=(True, ''))
226
227			config.add("OCRD_NETWORK_RABBITMQ_CLIENT_CONNECT_ATTEMPTS",
228			description="Number of attempts for a RabbitMQ client to connect before failing.",
229			parser=int,
230			default=(True, 3))
231
232			config.add(name="OCRD_NETWORK_SOCKETS_ROOT_DIR",
233			description="The root directory where all mets server related socket files are created",
234			parser=lambda val: Path(val),
235			default=(True, Path(gettempdir(), "ocrd_network_sockets")))
236			config.OCRD_NETWORK_SOCKETS_ROOT_DIR.mkdir(parents=True, exist_ok=True)
237
238			config.add(name="OCRD_NETWORK_LOGS_ROOT_DIR",
239			description="The root directory where all ocrd_network related file logs are stored",
240			parser=lambda val: Path(val),
241			default=(True, Path(gettempdir(), "ocrd_network_logs")))
242			config.OCRD_NETWORK_LOGS_ROOT_DIR.mkdir(parents=True, exist_ok=True)
243
244			config.add("HOME",
245			description="Directory to look for `ocrd_logging.conf`, fallback for unset XDG variables.",
246			# description="HOME directory, cf. https://specifications.freedesktop.org/basedir-spec/basedir-spec-latest.html",
247			validator=lambda val: Path(val).is_dir(),
248			parser=lambda val: Path(val),
249			default=(True, lambda: Path.home()))
250
251			config.add("XDG_DATA_HOME",
252			description="Directory to look for `./ocrd-resources/*` (i.e. `ocrd resmgr` data location)",
253			parser=lambda val: Path(val),
254			default=(True, lambda: Path(config.HOME, '.local/share')))
255
256			config.add("XDG_CONFIG_HOME",
257			description="Directory to look for `./ocrd/resources.yml` (i.e. `ocrd resmgr` user database)",
258			parser=lambda val: Path(val),
259			default=(True, lambda: Path(config.HOME, '.config')))
260
261			config.add("OCRD_LOGGING_DEBUG",
262			description="Print information about the logging setup to STDERR",
263			default=(True, False),
264			validator=_validator_boolean,
265			parser=_parser_boolean)
266

OCR-D / core

Pull Request — master (#1240)

ocrd_utils.config.OcrdEnvConfig.__init__() A

Complexity

Size

Duplication

Importance

Duplication Side-by-Side

Filter issues like

ocrd_utils.config.OcrdEnvConfig.init() A