Passed
Pull Request — master (#966)
by Konstantin
04:11 queued 01:43
created

ocrd.mets_server.ClientSideOcrdFile.__init__()   A

Complexity

Conditions 1

Size

Total Lines 18
Code Lines 7

Duplication

Lines 0
Ratio 0 %

Importance

Changes 0
Metric Value
eloc 7
dl 0
loc 18
rs 10
c 0
b 0
f 0
cc 1
nop 10

How to fix   Many Parameters   

Many Parameters

Methods with many parameters are not only hard to understand, but their parameters also often become inconsistent when you need more, or different data.

There are several approaches to avoid long parameter lists:

1
"""
2
# METS server functionality
3
"""
4
import re
5
from os import environ, _exit
6
from io import BytesIO
7
from typing import Any, Dict, Optional, Union, List, Tuple
8
9
from fastapi import FastAPI, Request, File, Form, UploadFile
10
from fastapi.responses import JSONResponse
11
from requests import request, Session as requests_session
12
from requests_unixsocket import Session as requests_unixsocket_session
13
from pydantic import BaseModel, Field, ValidationError
14
15
import uvicorn
16
17
from ocrd_models import OcrdMets, OcrdFile, OcrdAgent
18
from ocrd_utils import initLogging, getLogger, deprecated_alias
19
20
#
21
# XXX HACKS TODO
22
#
23
initLogging()
24
25
#
26
# Models
27
#
28
29
class OcrdFileModel(BaseModel):
30
    file_grp : str = Field()
31
    file_id : str = Field()
32
    mimetype : str = Field()
33
    page_id : Union[str, None] = Field()
34
    local_filename : str = Field()
35
36
    @staticmethod
37
    def create(file_grp : str, file_id : str, page_id : Union[str, None], local_filename : str, mimetype : str):
38
        return OcrdFileModel(file_grp=file_grp, file_id=file_id, page_id=page_id, mimetype=mimetype, local_filename=local_filename)
39
40
class OcrdAgentModel(BaseModel):
41
    name : str = Field()
42
    _type : str = Field()
43
    role : str = Field()
44
    otherrole : Optional[str] = Field()
45
    othertype : str = Field()
46
    notes : Optional[List[Tuple[Dict[str, str], Optional[str]]]] = Field()
47
48
    @staticmethod
49
    def create(name : str, _type : str, role : str, otherrole : str, othertype : str, notes : List[Tuple[Dict[str, str], Optional[str]]]):
50
        return OcrdAgentModel(name=name, _type=_type, role=role, otherrole=otherrole, othertype=othertype, notes=notes)
51
52
53
class OcrdFileListModel(BaseModel):
54
    files : List[OcrdFileModel] = Field()
55
56
    @staticmethod
57
    def create(files : List[OcrdFile]):
58
        return OcrdFileListModel(
59
            files=[OcrdFileModel.create(file_grp=f.fileGrp, file_id=f.ID, mimetype=f.mimetype, page_id=f.pageId, local_filename=f.local_filename) for f in files]
60
        )
61
62
class OcrdFileGroupListModel(BaseModel):
63
    file_groups : List[str] = Field()
64
65
    @staticmethod
66
    def create(file_groups : List[str]):
67
        return OcrdFileGroupListModel(file_groups=file_groups)
68
69
class OcrdAgentListModel(BaseModel):
70
    agents : List[OcrdAgentModel] = Field()
71
72
    @staticmethod
73
    def create(agents : List[OcrdAgent]):
74
        return OcrdAgentListModel(
75
            agents=[OcrdAgentModel(name=a.name, _type=a.type, role=a.role, otherrole=a.otherrole, othertype=a.othertype, notes=a.notes) for a in agents]
76
        )
77
78
#
79
# Client
80
#
81
82
class ClientSideOcrdFile:
83
    """
84
    Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdFile`
85
    but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
86
    this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
87
    """
88
89
    def __init__(self, el, mimetype=None, pageId=None, loctype='OTHER', local_filename=None, mets=None, url=None, ID=None, fileGrp=None):
90
        """
91
        Args:
92
            el (): ignored
93
        Keyword Args:
94
            mets (): ignored
95
            mimetype (string): ``@MIMETYPE`` of this ``mets:file``
96
            pageId (string): ``@ID`` of the physical ``mets:structMap`` entry corresponding to this ``mets:file``
97
            loctype (string): ``@LOCTYPE`` of this ``mets:file``
98
            local_filename (): ``@xlink:href`` of this ``mets:file`` - XXX the local file once we have proper mets:FLocat bookkeeping
99
            ID (string): ``@ID`` of this ``mets:file``
100
        """
101
        self.ID = ID
102
        self.mimetype = mimetype
103
        self.local_filename = local_filename
104
        self.loctype = loctype
105
        self.pageId = pageId
106
        self.fileGrp = fileGrp
107
108
class ClientSideOcrdAgent():
109
    """
110
    Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdAgent`
111
    but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
112
    this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
113
    """
114
115
    def __init__(self, el, name=None, _type=None, othertype=None, role=None, otherrole=None,
116
                 notes=None):
117
        """
118
        Args:
119
            el (): ignored
120
        Keyword Args:
121
            name (string):
122
            _type (string):
123
            othertype (string):
124
            role (string):
125
            otherrole (string):
126
            notes (dict):
127
        """
128
        self.name = name
129
        self.type = _type
130
        self.othertype = othertype
131
        self.role = role
132
        self.otherrole = otherrole
133
        self.notes = notes
134
135
class ClientSideOcrdMets():
136
    """
137
    Replacement for :py:class:`ocrd_models.ocrd_mets.OcrdMets` with overrides for
138
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.find_files`,
139
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.find_all_files`, and
140
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_agent`,
141
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.agents`,
142
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_file` to query via HTTP a
143
    :py:class:`ocrd.mets_server.OcrdMetsServer`.
144
    """
145
146
    def __init__(self, host, port, socket):
147
        self.log = getLogger('ocrd.mets_client.%s' % ('uds' if socket else 'tcp'))
148
        if socket:
149
            self.url = f'http+unix://{socket.replace("/", "%2F")}'
150
            self.session = requests_unixsocket_session()
151
        else:
152
            self.url = f'http://{host}:{port}'
153
            self.session = requests_session()
154
155
    @deprecated_alias(ID="file_id")
156
    @deprecated_alias(pageId="page_id")
157
    @deprecated_alias(fileGrp="file_grp")
158
    def find_files(self, **kwargs):
159
        if 'pageId' in kwargs:
160
            kwargs['page_id'] = kwargs.pop('pageId')
161
        if 'ID' in kwargs:
162
            kwargs['file_id'] = kwargs.pop('ID')
163
        if 'fileGrp' in kwargs:
164
            kwargs['file_grp'] = kwargs.pop('fileGrp')
165
        r = self.session.request('GET', f'{self.url}/file', params={**kwargs})
166
        for f in r.json()['files']:
167
            yield ClientSideOcrdFile(None, ID=f['file_id'], pageId=f['page_id'], fileGrp=f['file_grp'], local_filename=f['local_filename'], mimetype=f['mimetype'])
168
169
    def find_all_files(self, *args, **kwargs):
170
        return list(self.find_files(*args, **kwargs))
171
172
    def add_agent(self, *args, **kwargs):
173
        return self.session.request('POST', f'{self.url}/agent', json=OcrdAgentModel.create(**kwargs).dict())
174
175
    @property
176
    def agents(self):
177
        return [ClientSideOcrdAgent(None, **agent_dict) for agent_dict in self.session.request('GET', f'{self.url}/agent').json()['agents']]
178
179
    @property
180
    def file_groups(self):
181
        return self.session.request('GET', f'{self.url}/file_groups').json()['file_groups']
182
183
    @deprecated_alias(pageId="page_id")
184
    @deprecated_alias(ID="file_id")
185
    def add_file(self, file_grp, content=None, file_id=None, local_filename=None, mimetype=None, page_id=None, **kwargs):
186
        return self.session.request(
187
            'POST',
188
            f'{self.url}/file',
189
            data=OcrdFileModel.create(
190
                file_id=file_id,
191
                file_grp=file_grp,
192
                page_id=page_id,
193
                mimetype=mimetype,
194
                local_filename=local_filename).dict(),
195
        )
196
197
    def save(self):
198
        self.session.request('PUT', self.url)
199
200
201
#
202
# Server
203
#
204
205
class OcrdMetsServer():
206
207
    def __init__(self, workspace, host, port, socket):
208
        self.workspace = workspace
209
        if socket and host:
210
            raise ValueError("Expecting either socket or host/port")
211
        if not socket and not(host and port):
212
            raise ValueError("Expecting both host and port")
213
        self.host = host
214
        self.port = port
215
        self.socket = socket
216
        self.log = getLogger('ocrd.workspace_client')
217
218
    def shutdown(self):
219
        _exit(0)
220
221
    def startup(self):
222
223
        # XXX HACK 
224
        # circumventing dependency injection like this is bad and
225
        # needs to be refactored once it's all runnign
226
        workspace = self.workspace
227
228
        app = FastAPI(
229
            title="OCR-D METS Server",
230
            description="Providing simultaneous write-access to mets.xml for OCR-D",
231
        )
232
233
        @app.exception_handler(ValidationError)
234
        async def exception_handler_validation_error(request: Request, exc: ValidationError):
235
            return JSONResponse(status_code=400, content=exc.errors())
236
237
        @app.exception_handler(FileExistsError)
238
        async def exception_handler_file_exists(request: Request, exc: FileExistsError):
239
            return JSONResponse(status_code=400, content=str(exc))
240
241
        @app.exception_handler(re.error)
242
        async def exception_handler_invalid_regex(request: Request, exc: re.error):
243
            return JSONResponse(status_code=400, content=f'invalid regex: {exc}')
244
245
        @app.get("/file", response_model=OcrdFileListModel)
246
        async def find_files(
247
            file_grp : Union[str, None] = None,
248
            file_id : Union[str, None] = None,
249
            page_id : Union[str, None] = None,
250
            mimetype : Union[str, None] = None,
251
        ):
252
            """
253
            Find files in the mets
254
            """
255
            found = workspace.mets.find_all_files(fileGrp=file_grp, ID=file_id, pageId=page_id, mimetype=mimetype)
256
            return OcrdFileListModel.create(found)
257
258
        @app.put('/')
259
        def save():
260
            return workspace.save_mets()
261
262
        @app.post('/file', response_model=OcrdFileModel)
263
        async def add_file(
264
            file_grp : str = Form(),
265
            file_id : str = Form(),
266
            page_id : Union[str, None] = Form(),
267
            mimetype : str = Form(),
268
            local_filename : str = Form(),
269
        ):
270
            """
271
            Add a file
272
            """
273
            # Validate
274
            file_resource = OcrdFileModel.create(file_grp=file_grp, file_id=file_id, page_id=page_id, mimetype=mimetype, local_filename=local_filename)
275
            # Add to workspace
276
            kwargs = file_resource.dict()
277
            kwargs['page_id'] = page_id
278
            workspace.add_file(**kwargs)
279
            return file_resource
280
281
        @app.get('/file_groups', response_model=OcrdFileGroupListModel)
282
        async def file_groups():
283
            return {'file_groups': workspace.mets.file_groups}
284
285
        @app.post('/agent', response_model=OcrdAgentModel)
286
        async def add_agent(agent : OcrdAgentModel):
287
            kwargs = agent.dict()
288
            workspace.mets.add_agent(**kwargs)
289
            return agent
290
291
        @app.get('/agent', response_model=OcrdAgentListModel)
292
        async def agents():
293
            return OcrdAgentListModel.create(workspace.mets.agents)
294
295
        @app.delete('/')
296
        async def stop():
297
            """
298
            Stop the server
299
            """
300
            getLogger('ocrd_models.ocrd_mets').info('Shutting down')
301
            workspace.save_mets()
302
            # XXX HACK os._exit to not trigger SystemExit caught by uvicorn with sys.exit
303
            _exit(0)
304
305
306
        uvicorn.run(app, host=self.host, port=self.port, uds=self.socket)
307
308
309