|
| 1 | +import hashlib |
| 2 | +from copy import deepcopy |
| 3 | +from random import choice |
| 4 | +from string import ascii_letters |
| 5 | +from typing import BinaryIO |
| 6 | + |
| 7 | +import requests |
| 8 | + |
| 9 | +from csfunctions.service.base import BaseService |
| 10 | +from csfunctions.service.file_upload_schemas import ( |
| 11 | + CompleteFileUploadRequest, |
| 12 | + CreateNewFileRequest, |
| 13 | + CreateNewFileResponse, |
| 14 | + GeneratePresignedUrlRequest, |
| 15 | + PresignedWriteUrls, |
| 16 | +) |
| 17 | + |
| 18 | + |
| 19 | +def _generate_lock_id(): |
| 20 | + return "".join(choice(ascii_letters) for i in range(12)) # nosec |
| 21 | + |
| 22 | + |
| 23 | +class FileUploadService(BaseService): |
| 24 | + def _create_new_file(self, filename: str, parent_object_id: str, persno: str, check_access: bool = True) -> str: |
| 25 | + """Creates a new (empty) file attached to the parent object and returns the cdb_object_id.""" |
| 26 | + response_json = self.request( |
| 27 | + endpoint="/file_upload/create", |
| 28 | + method="POST", |
| 29 | + json=CreateNewFileRequest( |
| 30 | + parent_object_id=parent_object_id, filename=filename, persno=persno, check_access=check_access |
| 31 | + ).model_dump(), |
| 32 | + ) |
| 33 | + data = CreateNewFileResponse.model_validate(response_json) |
| 34 | + return data.file_object_id |
| 35 | + |
| 36 | + def _get_presigned_write_urls( |
| 37 | + self, file_object_id: str, filesize: int, lock_id: str, persno: str, check_access: bool = True |
| 38 | + ) -> PresignedWriteUrls: |
| 39 | + response_json = self.request( |
| 40 | + endpoint=f"/file_upload/{file_object_id}/generate_presigned_url", |
| 41 | + method="POST", |
| 42 | + json=GeneratePresignedUrlRequest( |
| 43 | + check_access=check_access, persno=persno, filesize=filesize, lock_id=lock_id |
| 44 | + ).model_dump(), |
| 45 | + ) |
| 46 | + |
| 47 | + return PresignedWriteUrls.model_validate(response_json) |
| 48 | + |
| 49 | + def _upload_file_content( |
| 50 | + self, presigned_urls: PresignedWriteUrls, stream: BinaryIO |
| 51 | + ) -> tuple[PresignedWriteUrls, str]: |
| 52 | + """Upload file stream in chunks using presigned URLs and return updated context + sha256 hash.""" |
| 53 | + etags: list[str] = [] |
| 54 | + sha256 = hashlib.sha256() |
| 55 | + for url in presigned_urls.urls: |
| 56 | + data: bytes = stream.read(presigned_urls.chunksize) |
| 57 | + sha256.update(data) |
| 58 | + resp = requests.put(url, data=data, headers=presigned_urls.headers, timeout=20) |
| 59 | + # 20 second timeout to stay below 30s max execution time of the Function |
| 60 | + # otherwise we won't get a proper error message in the logs |
| 61 | + resp.raise_for_status() |
| 62 | + etag = resp.headers.get("ETag") |
| 63 | + if etag: |
| 64 | + etags.append(etag) |
| 65 | + updated = deepcopy(presigned_urls) |
| 66 | + if etags: |
| 67 | + updated.etags = etags |
| 68 | + return updated, sha256.hexdigest() |
| 69 | + |
| 70 | + @staticmethod |
| 71 | + def _get_stream_size(stream: BinaryIO) -> int: |
| 72 | + if not stream.seekable(): |
| 73 | + raise ValueError("Stream is not seekable; size cannot be determined.") |
| 74 | + current_pos = stream.tell() |
| 75 | + stream.seek(0, 2) |
| 76 | + size = stream.tell() |
| 77 | + stream.seek(current_pos) |
| 78 | + return size |
| 79 | + |
| 80 | + def _complete_upload( |
| 81 | + self, |
| 82 | + file_object_id: str, |
| 83 | + filesize: int, |
| 84 | + lock_id: str, |
| 85 | + presigned_urls: PresignedWriteUrls, |
| 86 | + persno: str, |
| 87 | + check_access: bool = True, |
| 88 | + sha256: str | None = None, |
| 89 | + delete_derived_files: bool = True, |
| 90 | + ) -> None: |
| 91 | + self.request( |
| 92 | + endpoint=f"/file_upload/{file_object_id}/complete", |
| 93 | + method="POST", |
| 94 | + json=CompleteFileUploadRequest( |
| 95 | + filesize=filesize, |
| 96 | + check_access=check_access, |
| 97 | + persno=persno, |
| 98 | + presigned_write_urls=presigned_urls, |
| 99 | + lock_id=lock_id, |
| 100 | + sha256=sha256, |
| 101 | + delete_derived_files=delete_derived_files, |
| 102 | + ).model_dump(), |
| 103 | + ) |
| 104 | + |
| 105 | + def upload_file_content( |
| 106 | + self, |
| 107 | + file_object_id: str, |
| 108 | + stream: BinaryIO, |
| 109 | + persno: str | None = None, |
| 110 | + check_access: bool = True, |
| 111 | + filesize: int | None = None, |
| 112 | + delete_derived_files: bool = True, |
| 113 | + ) -> None: |
| 114 | + persno = persno or self.metadata.app_user |
| 115 | + if filesize is None: |
| 116 | + filesize = self._get_stream_size(stream) |
| 117 | + lock_id = _generate_lock_id() |
| 118 | + presigned = self._get_presigned_write_urls( |
| 119 | + file_object_id=file_object_id, |
| 120 | + filesize=filesize, |
| 121 | + lock_id=lock_id, |
| 122 | + persno=persno, |
| 123 | + check_access=check_access, |
| 124 | + ) |
| 125 | + presigned_with_etags, sha256 = self._upload_file_content(presigned_urls=presigned, stream=stream) |
| 126 | + self._complete_upload( |
| 127 | + file_object_id=file_object_id, |
| 128 | + filesize=filesize, |
| 129 | + lock_id=lock_id, |
| 130 | + presigned_urls=presigned_with_etags, |
| 131 | + persno=persno, |
| 132 | + check_access=check_access, |
| 133 | + sha256=sha256, |
| 134 | + delete_derived_files=delete_derived_files, |
| 135 | + ) |
| 136 | + |
| 137 | + def upload_new_file( |
| 138 | + self, |
| 139 | + parent_object_id: str, |
| 140 | + filename: str, |
| 141 | + stream: BinaryIO, |
| 142 | + persno: str | None = None, |
| 143 | + check_access: bool = True, |
| 144 | + filesize: int | None = None, |
| 145 | + ) -> str: |
| 146 | + persno = persno or self.metadata.app_user |
| 147 | + file_object_id = self._create_new_file( |
| 148 | + filename=filename, |
| 149 | + parent_object_id=parent_object_id, |
| 150 | + persno=persno, |
| 151 | + check_access=check_access, |
| 152 | + ) |
| 153 | + self.upload_file_content( |
| 154 | + file_object_id=file_object_id, |
| 155 | + stream=stream, |
| 156 | + persno=persno, |
| 157 | + check_access=check_access, |
| 158 | + filesize=filesize, |
| 159 | + delete_derived_files=False, |
| 160 | + ) |
| 161 | + return file_object_id |
0 commit comments