Passed
Pull Request — master (#966)
by Konstantin
02:39
created

ocrd.mets_server.OcrdFileGroupListModel.create()   A

Complexity

Conditions 1

Size

Total Lines 3
Code Lines 3

Duplication

Lines 0
Ratio 0 %

Importance

Changes 0
Metric Value
eloc 3
dl 0
loc 3
rs 10
c 0
b 0
f 0
cc 1
nop 1
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
    url : str = Field()
35
36
    @staticmethod
37
    def create(file_grp : str, file_id : str, page_id : Union[str, None], url : str, mimetype : str):
38
        return OcrdFileModel(file_grp=file_grp, file_id=file_id, page_id=page_id, mimetype=mimetype, url=url)
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, url=f.url) 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 (): ignored
99
            url (string): ``@xlink:href`` of this ``mets:file``
100
            ID (string): ``@ID`` of this ``mets:file``
101
        """
102
        self.ID = ID
103
        self.mimetype = mimetype
104
        self.url = url
105
        self.loctype = loctype
106
        self.pageId = pageId
107
        self.fileGrp = fileGrp
108
109
class ClientSideOcrdAgent():
110
    """
111
    Provides the same interface as :py:class:`ocrd_models.ocrd_file.OcrdAgent`
112
    but without attachment to :py:class:`ocrd_models.ocrd_mets.OcrdMets` since
113
    this represents the response of the :py:class:`ocrd.mets_server.OcrdMetsServer`.
114
    """
115
116
    def __init__(self, el, name=None, _type=None, othertype=None, role=None, otherrole=None,
117
                 notes=None):
118
        """
119
        Args:
120
            el (): ignored
121
        Keyword Args:
122
            name (string):
123
            _type (string):
124
            othertype (string):
125
            role (string):
126
            otherrole (string):
127
            notes (dict):
128
        """
129
        self.name = name
130
        self.type = _type
131
        self.othertype = othertype
132
        self.role = role
133
        self.otherrole = otherrole
134
        self.notes = notes
135
136
class ClientSideOcrdMets():
137
    """
138
    Replacement for :py:class:`ocrd_models.ocrd_mets.OcrdMets` with overrides for
139
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.find_files`,
140
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.find_all_files`, and
141
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_agent`,
142
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.agents`,
143
    :py:meth:`ocrd_models.ocrd_mets.OcrdMets.add_file` to query via HTTP a
144
    :py:class:`ocrd.mets_server.OcrdMetsServer`.
145
    """
146
147
    def __init__(self, host, port, socket):
148
        self.log = getLogger('ocrd.mets_client.%s' % ('uds' if socket else 'tcp'))
149
        if socket:
150
            self.url = f'http+unix://{socket.replace("/", "%2F")}'
151
            self.session = requests_unixsocket_session()
152
        else:
153
            self.url = f'http://{host}:{port}'
154
            self.session = requests_session()
155
156
    @deprecated_alias(ID="file_id")
157
    @deprecated_alias(pageId="page_id")
158
    @deprecated_alias(fileGrp="file_grp")
159
    def find_files(self, **kwargs):
160
        if 'pageId' in kwargs:
161
            kwargs['page_id'] = kwargs.pop('pageId')
162
        if 'ID' in kwargs:
163
            kwargs['file_id'] = kwargs.pop('ID')
164
        if 'fileGrp' in kwargs:
165
            kwargs['file_grp'] = kwargs.pop('fileGrp')
166
        r = self.session.request('GET', f'{self.url}/file', params={**kwargs})
167
        for f in r.json()['files']:
168
            yield ClientSideOcrdFile(None, ID=f['file_id'], pageId=f['page_id'], fileGrp=f['file_grp'], url=f['url'], mimetype=f['mimetype'])
169
170
    def find_all_files(self, *args, **kwargs):
171
        return list(self.find_files(*args, **kwargs))
172
173
    def add_agent(self, *args, **kwargs):
174
        return self.session.request('POST', f'{self.url}/agent', json=OcrdAgentModel.create(**kwargs).dict())
175
176
    @property
177
    def agents(self):
178
        # print(self.session.request('GET', f'{self.url}/agent').json())
179
        return [ClientSideOcrdAgent(None, **agent_dict) for agent_dict in self.session.request('GET', f'{self.url}/agent').json()['agents']]
180
181
    @property
182
    def file_groups(self):
183
        return self.session.request('GET', f'{self.url}/file_groups').json()['file_groups']
184
185
    @deprecated_alias(pageId="page_id")
186
    @deprecated_alias(ID="file_id")
187
    def add_file(self, file_grp, content=None, file_id=None, url=None, mimetype=None, page_id=None, **kwargs):
188
        return self.session.request(
189
            'POST',
190
            f'{self.url}/file',
191
            data=OcrdFileModel.create(
192
                file_id=file_id,
193
                file_grp=file_grp,
194
                page_id=page_id,
195
                mimetype=mimetype,
196
                url=url).dict(),
197
            files={'data': content}
198
        )
199
200
    def save(self):
201
        self.session.request('PUT', self.url)
202
203
204
#
205
# Server
206
#
207
208
class OcrdMetsServer():
209
210
    def __init__(self, workspace, host, port, socket):
211
        self.workspace = workspace
212
        if socket and host:
213
            raise ValueError("Expecting either socket or host/port")
214
        if not socket and not(host and port):
215
            raise ValueError("Expecting both host and port")
216
        self.host = host
217
        self.port = port
218
        self.socket = socket
219
        self.log = getLogger('ocrd.workspace_client')
220
221
    def shutdown(self):
222
        _exit(0)
223
224
    def startup(self):
225
226
        # XXX HACK 
227
        # circumventing dependency injection like this is bad and
228
        # needs to be refactored once it's all runnign
229
        workspace = self.workspace
230
231
        app = FastAPI(
232
            title="OCR-D METS Server",
233
            description="Providing simultaneous write-access to mets.xml for OCR-D",
234
        )
235
236
        @app.exception_handler(ValidationError)
237
        async def exception_handler_validation_error(request: Request, exc: ValidationError):
238
            return JSONResponse(status_code=400, content=exc.errors())
239
240
        @app.exception_handler(FileExistsError)
241
        async def exception_handler_file_exists(request: Request, exc: FileExistsError):
242
            return JSONResponse(status_code=400, content=str(exc))
243
244
        @app.exception_handler(re.error)
245
        async def exception_handler_invalid_regex(request: Request, exc: re.error):
246
            return JSONResponse(status_code=400, content=f'invalid regex: {exc}')
247
248
        @app.get("/file", response_model=OcrdFileListModel)
249
        async def find_files(
250
            file_grp : Union[str, None] = None,
251
            file_id : Union[str, None] = None,
252
            page_id : Union[str, None] = None,
253
            mimetype : Union[str, None] = None,
254
        ):
255
            """
256
            Find files in the mets
257
            """
258
            found = workspace.mets.find_all_files(fileGrp=file_grp, ID=file_id, pageId=page_id, mimetype=mimetype)
259
            return OcrdFileListModel.create(found)
260
261
        @app.put('/')
262
        def save():
263
            return workspace.save_mets()
264
265
        @app.post('/file', response_model=OcrdFileModel)
266
        async def add_file(
267
            file_grp : str = Form(),
268
            file_id : str = Form(),
269
            page_id : Union[str, None] = Form(),
270
            mimetype : str = Form(),
271
            url : str = Form(),
272
        ):
273
            """
274
            Add a file
275
            """
276
            # Validate
277
            file_resource = OcrdFileModel.create(file_grp=file_grp, file_id=file_id, page_id=page_id, mimetype=mimetype, url=url)
278
            # Add to workspace
279
            kwargs = file_resource.dict()
280
            kwargs['page_id'] = page_id
281
            kwargs['local_filename'] = kwargs.pop('url')
282
            workspace.add_file(**kwargs)
283
            return file_resource
284
285
        @app.get('/file_groups', response_model=OcrdFileGroupListModel)
286
        async def file_groups():
287
            return {'file_groups': workspace.mets.file_groups}
288
289
        @app.post('/agent', response_model=OcrdAgentModel)
290
        async def add_agent(agent : OcrdAgentModel):
291
            kwargs = agent.dict()
292
            workspace.mets.add_agent(**kwargs)
293
            return agent
294
295
        @app.get('/agent', response_model=OcrdAgentListModel)
296
        async def agents():
297
            return OcrdAgentListModel.create(workspace.mets.agents)
298
299
        @app.delete('/')
300
        async def stop():
301
            """
302
            Stop the server
303
            """
304
            getLogger('ocrd_models.ocrd_mets').info('Shutting down')
305
            workspace.save_mets()
306
            # XXX HACK os._exit to not trigger SystemExit caught by uvicorn with sys.exit
307
            _exit(0)
308
309
310
        uvicorn.run(app, host=self.host, port=self.port, uds=self.socket)
311
312
313