|
| 1 | +"""Repository data structure.""" |
| 2 | +import json |
| 3 | +from abc import ABC |
| 4 | +from pathlib import Path |
| 5 | +from functools import cached_property |
| 6 | + |
| 7 | +import requests |
| 8 | + |
| 9 | + |
| 10 | +def _check(request: requests.Request, proc: str): |
| 11 | + try: |
| 12 | + request.raise_for_status() |
| 13 | + except requests.HTTPError as err: |
| 14 | + raise requests.HTTPError( |
| 15 | + f"Error while {proc}, info: {request.json()['message']}" |
| 16 | + ) from err |
| 17 | + |
| 18 | + return request |
| 19 | + |
| 20 | +class _SubCommandHandler(ABC): |
| 21 | + def __init__(self, parent): |
| 22 | + self.parent = parent |
| 23 | + |
| 24 | + @property |
| 25 | + def url(self): |
| 26 | + return self.parent.url |
| 27 | + |
| 28 | + @property |
| 29 | + def api_key(self): |
| 30 | + return self.parent.api_key |
| 31 | + |
| 32 | + |
| 33 | +class _File(_SubCommandHandler): |
| 34 | + def __init__(self, parent, name): |
| 35 | + super.__init__(self, parent) |
| 36 | + self.name = name |
| 37 | + |
| 38 | + @property |
| 39 | + def dep_id(self): |
| 40 | + return self.parent.dep_id |
| 41 | + |
| 42 | + @property |
| 43 | + def bucket_url(self): |
| 44 | + return self.parent.bucket_url |
| 45 | + |
| 46 | + def info(self, **params): |
| 47 | + return _check( |
| 48 | + requests.get( |
| 49 | + f"{self.url}/deposit/depositions/{self.dep_id}/files/{self.name}", |
| 50 | + params={**params, "access_token": self.api_key}, |
| 51 | + ), |
| 52 | + f"getting {self.name} file info from deposition {self.dep_id}", |
| 53 | + ) |
| 54 | + |
| 55 | + def update(self, file: Path, **params): |
| 56 | + data = {"name": f"{file.name}"} |
| 57 | + header = {"Content-Type": "application/json"} |
| 58 | + return _check( |
| 59 | + requests.put( |
| 60 | + f"{self.url}/deposit/depositions/{self.dep_id}/files/{self.name}", |
| 61 | + params={**params, "access_token": self.api_key}, |
| 62 | + data=json.dumps(data), |
| 63 | + headers=header, |
| 64 | + ), |
| 65 | + f"updating {self.name} in deposition {self.dep_id}", |
| 66 | + ) |
| 67 | + |
| 68 | + def download(self, dest: Path = Path(), **params): |
| 69 | + info = self.info().json() |
| 70 | + link = info["links"]["download"] |
| 71 | + filename = info["filename"] |
| 72 | + |
| 73 | + request = _check( |
| 74 | + requests.get(link, params={**params, "access_token": self.api_key}), |
| 75 | + f"downloading file {self.name} from deposition {self.dep_id}", |
| 76 | + ) |
| 77 | + |
| 78 | + dest = Path(dest) |
| 79 | + if dest.is_file(): |
| 80 | + raise OSError(f"{dest} is a file which exists. Must be a directory.") |
| 81 | + |
| 82 | + if not dest.isdir(): |
| 83 | + dest.mkdir(parents=True, exist_ok=True) |
| 84 | + |
| 85 | + with (dest / filename).open("wb") as out_file: |
| 86 | + out_file.write(request.content) |
| 87 | + |
| 88 | + return request |
| 89 | + |
| 90 | + def delete(self, **params): |
| 91 | + return _check( |
| 92 | + requests.delete( |
| 93 | + f"{self.url}/deposit/depositions/{self.dep_id}/files/{self.file_id}", |
| 94 | + params={**params, "access_token": self.api_key}, |
| 95 | + ), |
| 96 | + f"deleting file {self.name} from deposition {self.dep_id}", |
| 97 | + ) |
| 98 | + |
| 99 | + |
| 100 | + def upload(self, file: Path, **params): |
| 101 | + file = Path(file) |
| 102 | + |
| 103 | + with file.open("rb") as in_file: |
| 104 | + return _check( |
| 105 | + requests.put( |
| 106 | + f"{self.bucket_url}/{self.name}", |
| 107 | + params={**params, "access_token": self.api_key}, |
| 108 | + data=in_file, |
| 109 | + ), |
| 110 | + f"Uploading file {self.name} to deposition {self.dep_id}" |
| 111 | + ) |
| 112 | + |
| 113 | +class _Files(_SubCommandHandler): |
| 114 | + def __init__(self, parent): |
| 115 | + super.__init__(self, parent) |
| 116 | + |
| 117 | + @property |
| 118 | + def dep_id(self): |
| 119 | + return self.parent.dep_id |
| 120 | + |
| 121 | + @property |
| 122 | + def bucket_url(self): |
| 123 | + return self.parent.bucket_url |
| 124 | + |
| 125 | + def __getitem__(self, name) -> _File: |
| 126 | + return _File(self, name) |
| 127 | + |
| 128 | + def list(self, **params): |
| 129 | + return _check( |
| 130 | + requests.get( |
| 131 | + f"{self.url}/deposit/depositions/{self.dep_id}/files", |
| 132 | + params={**params, "access_token": self.api_key}, |
| 133 | + ), |
| 134 | + f"listing deposition {self.dep_id} files", |
| 135 | + ) |
| 136 | + |
| 137 | + def sort(self, sorted_ids, **params): |
| 138 | + return _check( |
| 139 | + requests.put( |
| 140 | + f"{self.url}/deposit/depositions/{self.dep_id}/files", |
| 141 | + params={**params, "access_token": self.api_key}, |
| 142 | + data=json.dumps(sorted_ids), |
| 143 | + headers={"Content-Type": "application/json"}, |
| 144 | + ), |
| 145 | + f"sorting files for deposition {self.dep_id}" |
| 146 | + ) |
| 147 | + |
| 148 | + def upload(self, files: dict[str, Path], **params): |
| 149 | + request_list = [] |
| 150 | + for name, file in files.items(): |
| 151 | + file = Path(file) |
| 152 | + |
| 153 | + with file.open("rb") as curr_file: |
| 154 | + request_list.append( |
| 155 | + _check( |
| 156 | + requests.put( |
| 157 | + f"{self.bucket_url}/{name}", |
| 158 | + params={**params, "access_token": self.api_key}, |
| 159 | + data=curr_file, |
| 160 | + ), |
| 161 | + f"Uploading file {self.name} to deposition {self.dep_id}" |
| 162 | + ) |
| 163 | + ) |
| 164 | + |
| 165 | + return request_list |
| 166 | + |
| 167 | + def download(self, dest, **params): |
| 168 | + request = self.list(**params).json() |
| 169 | + files = {file["id"]: file["filename"] for file in request} |
| 170 | + |
| 171 | + for file in files.values(): |
| 172 | + self[file].download(dest, **params) |
| 173 | + |
| 174 | + |
| 175 | +class _Deposition(_SubCommandHandler): |
| 176 | + def __init__(self, parent, dep_id): |
| 177 | + super().__init__(self, parent) |
| 178 | + self.dep_id = dep_id |
| 179 | + |
| 180 | + @property |
| 181 | + def files(self) -> _Files: |
| 182 | + return _Files(self) |
| 183 | + |
| 184 | + @cached_property |
| 185 | + def bucket_url(self): |
| 186 | + return self.get().json()["links"]["bucket"] |
| 187 | + |
| 188 | + def get(self, **params): |
| 189 | + request = _check( |
| 190 | + requests.get( |
| 191 | + f"{self.url}/deposit/depositions/{self.dep_id}", |
| 192 | + params={**params, "access_token": self.api_key}, |
| 193 | + ), |
| 194 | + f"getting deposition {self.dep_id}" |
| 195 | + ) |
| 196 | + self.bucket_url = request.json()["links"]["bucket"] |
| 197 | + return request |
| 198 | + |
| 199 | + |
| 200 | + def create(self, **params): |
| 201 | + return _check( |
| 202 | + requests.post( |
| 203 | + f"{self.url}/deposit/depositions", |
| 204 | + params={**params, "access_token": self.api_key}, |
| 205 | + json={}, |
| 206 | + headers={"Content-Type": "application/json"}, |
| 207 | + ), |
| 208 | + "creating deposition" |
| 209 | + ) |
| 210 | + |
| 211 | + def update(self, data, **params): |
| 212 | + return _check( |
| 213 | + requests.put( |
| 214 | + f"{self.url}/deposit/depositions/{self.dep_id}", |
| 215 | + params={**params, "access_token": self.api_key}, |
| 216 | + data=json.dumps(data), |
| 217 | + headers={"Content-Type": "application/json"}, |
| 218 | + ), |
| 219 | + f"updating deposition {self.dep_id}" |
| 220 | + ) |
| 221 | + |
| 222 | + def delete(self, **params): |
| 223 | + return _check( |
| 224 | + requests.delete( |
| 225 | + f"{self.url}/deposit/depositions/{self.dep_id}", |
| 226 | + params={**params, "access_token": self.api_key}, |
| 227 | + ), |
| 228 | + f"deleting deposition {self.dep_id}" |
| 229 | + ) |
| 230 | + |
| 231 | + def publish(self, **params): |
| 232 | + return _check( |
| 233 | + requests.post( |
| 234 | + f"{self.url}/deposit/depositions/{self.dep_id}/actions/publish", |
| 235 | + params={**params, "access_token": self.api_key}, |
| 236 | + ), |
| 237 | + f"publishing deposition {self.dep_id}", |
| 238 | + ) |
| 239 | + |
| 240 | + def edit(self, **params): |
| 241 | + return _check( |
| 242 | + requests.post( |
| 243 | + f"{self.url}/deposit/depositions/{self.dep_id}/actions/edit", |
| 244 | + params={**params, "access_token": self.api_key}, |
| 245 | + ), |
| 246 | + f"editing deposition {self.dep_id}", |
| 247 | + ) |
| 248 | + |
| 249 | + def discard(self, **params): |
| 250 | + return _check( |
| 251 | + requests.post( |
| 252 | + f"{self.url}/deposit/depositions/{self.dep_id}/actions/discard", |
| 253 | + params={**params, "access_token": self.api_key}, |
| 254 | + ), |
| 255 | + f"discarding deposition {self.dep_id}", |
| 256 | + ) |
| 257 | + |
| 258 | + def new_version(self, **params): |
| 259 | + return _check( |
| 260 | + requests.post( |
| 261 | + f"{self.url}/deposit/depositions/{self.dep_id}/actions/newversion", |
| 262 | + params={**params, "access_token": self.api_key}, |
| 263 | + ), |
| 264 | + f"setting new version for deposition {self.dep_id}", |
| 265 | + ) |
| 266 | + |
| 267 | +class _Repository(_SubCommandHandler): |
| 268 | + def __getitem__(self, dep_id): |
| 269 | + return _Deposition(self, dep_id) |
| 270 | + |
| 271 | + def list(self, **params): |
| 272 | + return _check( |
| 273 | + requests.get( |
| 274 | + f"{self.url}/deposit/depositions", |
| 275 | + params={**params, "access_token": self.api_key}, |
| 276 | + ), |
| 277 | + "listing depositions" |
| 278 | + ) |
| 279 | + |
| 280 | + |
| 281 | +class _Records(_SubCommandHandler): |
| 282 | + def get(self, rec_id, **params): |
| 283 | + return _check( |
| 284 | + requests.get( |
| 285 | + f"{self.url}/records/{rec_id}", |
| 286 | + params={**params, "access_token": self.api_key}, |
| 287 | + ), |
| 288 | + f"getting record {rec_id}", |
| 289 | + ) |
| 290 | + |
| 291 | + def list(self, **params): |
| 292 | + return _check( |
| 293 | + requests.get( |
| 294 | + f"{self.url}/records", params={**params, "access_token": self.api_key} |
| 295 | + ), |
| 296 | + "listing records", |
| 297 | + ) |
| 298 | + |
| 299 | +class _Licenses(_SubCommandHandler): |
| 300 | + def get(self, lic_id, **params): |
| 301 | + return _check( |
| 302 | + requests.get( |
| 303 | + f"{self.url}/licenses/{lic_id}", |
| 304 | + params={**params, "access_token": self.api_key}, |
| 305 | + ), |
| 306 | + f"getting license {lic_id}", |
| 307 | + ) |
| 308 | + |
| 309 | + def list(self, **params): |
| 310 | + return _check( |
| 311 | + requests.get( |
| 312 | + f"{self.url}/licenses/", params={**params, "access_token": self.api_key} |
| 313 | + ), |
| 314 | + "listing licenses", |
| 315 | + ) |
| 316 | + |
| 317 | + |
| 318 | +class InvenioRepository: |
| 319 | + """Handler for Invenio-like repositories. |
| 320 | +
|
| 321 | + Handles pushing info to e.g. Zenodo |
| 322 | +
|
| 323 | + Parameters |
| 324 | + ---------- |
| 325 | + url |
| 326 | + Repository URL. |
| 327 | + api_key : str |
| 328 | + API key with appropriate permissions. |
| 329 | +
|
| 330 | + Examples |
| 331 | + -------- |
| 332 | + .. code-block:: |
| 333 | + my_repo = InvenioRepository("abc123", "companyname.website") |
| 334 | + my_repo.depositions["my_repo"].files["file"].upload(my_file) |
| 335 | + my_repo.records.get() |
| 336 | + my_repo.liceses.list() |
| 337 | + """ |
| 338 | + def __init__(self, url: str, api_key: str): |
| 339 | + self.url = url |
| 340 | + self.api_key = api_key |
| 341 | + |
| 342 | + self.depositions = _Repository(self) |
| 343 | + self.records = _Records(self) |
| 344 | + self.licenses = _Licenses(self) |
0 commit comments