Coverage for / home / runner / work / viur-core / viur-core / viur / src / viur / core / modules / file.py: 16%
798 statements
« prev ^ index » next coverage.py v7.14.0, created at 2026-05-21 21:16 +0000
« prev ^ index » next coverage.py v7.14.0, created at 2026-05-21 21:16 +0000
1import base64
2import datetime
3import hashlib
4import hmac
5import html
6import io
7import json
8import logging
9import re
10import string
11import typing as t
12import warnings
13from collections import namedtuple
14from urllib.parse import quote as urlquote, urlencode
15from urllib.request import urlopen
17import PIL
18import PIL.ImageCms
19import google.auth
20import requests
21from PIL import Image
22from google.appengine.api import blobstore, images
23from google.cloud import storage
24from google.oauth2.service_account import Credentials as ServiceAccountCredentials
26from viur.core import conf, current, db, errors, utils, i18n
27from viur.core.bones import BaseBone, BooleanBone, JsonBone, KeyBone, NumericBone, StringBone
29from viur.core.decorators import *
30from viur.core.prototypes.tree import SkelType, Tree, TreeSkel
31from viur.core.skeleton import SkeletonInstance, skeletonByKind
32from viur.core.tasks import CallDeferred, DeleteEntitiesIter, PeriodicTask
34# Globals for connectivity
36VALID_FILENAME_REGEX = re.compile(
37 # || MAY NOT BE THE NAME | MADE OF SPECIAL CHARS | SPECIAL CHARS + `. `|`
38 r"^(?!^(CON|PRN|AUX|NUL|COM[1-9]|LPT[1-9])$)[^\x00-\x1F<>:\"\/\\|?*]*[^\x00-\x1F<>:\"\/\\|?*. ]$",
39 re.IGNORECASE
40)
42_CREDENTIALS, _PROJECT_ID = google.auth.default()
43GOOGLE_STORAGE_CLIENT = storage.Client(_PROJECT_ID, _CREDENTIALS)
45PRIVATE_BUCKET_NAME = f"""{_PROJECT_ID}.appspot.com"""
46PUBLIC_BUCKET_NAME = f"""public-dot-{_PROJECT_ID}"""
47PUBLIC_DLKEY_SUFFIX = "_pub"
49_private_bucket = GOOGLE_STORAGE_CLIENT.lookup_bucket(PRIVATE_BUCKET_NAME)
50_public_bucket = None
52# FilePath is a descriptor for ViUR file components
53FilePath = namedtuple("FilePath", ("dlkey", "is_derived", "filename"))
56def importBlobFromViur2(dlKey, fileName):
57 bucket = conf.main_app.file.get_bucket(dlKey)
59 if not conf.viur2import_blobsource:
60 return False
61 existingImport = db.get(db.Key("viur-viur2-blobimport", dlKey))
62 if existingImport:
63 if existingImport["success"]:
64 return existingImport["dlurl"]
65 return False
66 if conf.viur2import_blobsource["infoURL"]:
67 try:
68 importDataReq = urlopen(conf.viur2import_blobsource["infoURL"] + dlKey)
69 except Exception as e:
70 marker = db.Entity(db.Key("viur-viur2-blobimport", dlKey))
71 marker["success"] = False
72 marker["error"] = "Failed URL-FETCH 1"
73 db.put(marker)
74 return False
75 if importDataReq.status != 200:
76 marker = db.Entity(db.Key("viur-viur2-blobimport", dlKey))
77 marker["success"] = False
78 marker["error"] = "Failed URL-FETCH 2"
79 db.put(marker)
80 return False
81 importData = json.loads(importDataReq.read())
82 oldBlobName = conf.viur2import_blobsource["gsdir"] + "/" + importData["key"]
83 srcBlob = storage.Blob(bucket=bucket,
84 name=conf.viur2import_blobsource["gsdir"] + "/" + importData["key"])
85 else:
86 oldBlobName = conf.viur2import_blobsource["gsdir"] + "/" + dlKey
87 srcBlob = storage.Blob(bucket=bucket, name=conf.viur2import_blobsource["gsdir"] + "/" + dlKey)
88 if not srcBlob.exists():
89 marker = db.Entity(db.Key("viur-viur2-blobimport", dlKey))
90 marker["success"] = False
91 marker["error"] = "Local SRC-Blob missing"
92 marker["oldBlobName"] = oldBlobName
93 db.put(marker)
94 return False
95 bucket.rename_blob(srcBlob, f"{dlKey}/source/{fileName}")
96 marker = db.Entity(db.Key("viur-viur2-blobimport", dlKey))
97 marker["success"] = True
98 marker["old_src_key"] = dlKey
99 marker["old_src_name"] = fileName
100 marker["dlurl"] = conf.main_app.file.create_download_url(dlKey, fileName, False, None)
101 db.put(marker)
102 return marker["dlurl"]
105def thumbnailer(fileSkel, existingFiles, params):
106 file_name = html.unescape(fileSkel["name"])
107 bucket = conf.main_app.file.get_bucket(fileSkel["dlkey"])
109 blob = bucket.get_blob(f"""{fileSkel["dlkey"]}/source/{file_name}""")
110 if not blob:
111 logging.warning(f"""Blob {fileSkel["dlkey"]}/source/{file_name} is missing from cloud storage!""")
112 return
114 source = io.BytesIO()
115 blob.download_to_file(source)
117 result = []
119 for info in params:
120 # Read the image into PIL
121 try:
122 source.seek(0)
123 img = PIL.Image.open(source)
124 except PIL.Image.UnidentifiedImageError: # Can't load this image; so there's no need to try other resolutions
125 break
127 if icc_profile := img.info.get("icc_profile"):
128 # JPEGs might be encoded with a non-standard color-profile; we need to compensate for this if we convert
129 # to WEBp as we'll loose this color-profile information
130 f = io.BytesIO(icc_profile)
131 src_profile = PIL.ImageCms.ImageCmsProfile(f)
132 dst_profile = PIL.ImageCms.createProfile("sRGB")
133 try:
134 img = PIL.ImageCms.profileToProfile(
135 img,
136 inputProfile=src_profile,
137 outputProfile=dst_profile,
138 outputMode="RGBA" if img.has_transparency_data else "RGB")
139 except Exception as e:
140 logging.debug(f"{info=}")
141 logging.exception(e)
142 continue
144 file_extension = info.get("fileExtension", "webp")
145 mimetype = info.get("mimeType", "image/webp")
147 if "width" in info and "height" in info:
148 width = info["width"]
149 height = info["height"]
150 target_filename = f"thumbnail-{width}-{height}.{file_extension}"
152 elif "width" in info:
153 width = info["width"]
154 height = int((float(img.size[1]) * float(width / float(img.size[0]))))
155 target_filename = f"thumbnail-w{width}.{file_extension}"
157 else: # No default fallback - ignore
158 continue
160 # Create resized version of the source
161 target = io.BytesIO()
163 try:
164 img = img.resize((width, height), PIL.Image.LANCZOS)
165 except ValueError as e:
166 # Usually happens to some files, like TIFF-images.
167 logging.debug(f"{info=}")
168 logging.exception(e)
169 break
171 img.save(target, file_extension)
173 # Safe derived target file
174 target_size = target.tell()
175 target.seek(0)
176 target_blob = bucket.blob(f"""{fileSkel["dlkey"]}/derived/{target_filename}""")
177 target_blob.upload_from_file(target, content_type=mimetype)
179 result.append(
180 (target_filename, target_size, mimetype, {"mimetype": mimetype, "width": width, "height": height})
181 )
183 return result
186def cloudfunction_thumbnailer(fileSkel, existingFiles, params):
187 """External Thumbnailer for images.
189 The corresponding cloudfunction can be found here .
190 https://github.com/viur-framework/viur-cloudfunctions/tree/main/thumbnailer
192 You can use it like so:
193 main.py:
195 .. code-block:: python
197 from viur.core.modules.file import cloudfunction_thumbnailer
199 conf.file_thumbnailer_url = "https://xxxxx.cloudfunctions.net/imagerenderer"
200 conf.file_derivations = {"thumbnail": cloudfunction_thumbnailer}
202 conf.derives_pdf = {
203 "thumbnail": [{"width": 1920,"sites":"1,2"}]
204 }
206 skeletons/xxx.py:
207 .. code-block:: python
209 test = FileBone(derive=conf.derives_pdf)
210 """
212 if not conf.file_thumbnailer_url:
213 raise ValueError("conf.file_thumbnailer_url is not set")
215 bucket = conf.main_app.file.get_bucket(fileSkel["dlkey"])
217 def getsignedurl():
218 if conf.instance.is_dev_server:
219 signedUrl = conf.main_app.file.create_download_url(fileSkel["dlkey"], fileSkel["name"])
220 else:
221 path = f"""{fileSkel["dlkey"]}/source/{file_name}"""
222 if not (blob := bucket.get_blob(path)):
223 logging.warning(f"Blob {path} is missing from cloud storage!")
224 return None
225 authRequest = google.auth.transport.requests.Request()
226 expiresAt = datetime.datetime.now() + datetime.timedelta(seconds=60)
227 signing_credentials = google.auth.compute_engine.IDTokenCredentials(authRequest, "")
228 content_disposition = utils.build_content_disposition_header(fileSkel["name"])
229 signedUrl = blob.generate_signed_url(
230 expiresAt,
231 credentials=signing_credentials,
232 response_disposition=content_disposition,
233 version="v4")
234 return signedUrl
236 def make_request():
237 headers = {"Content-Type": "application/json"}
238 data_str = base64.b64encode(json.dumps(dataDict).encode("UTF-8"))
239 sig = conf.main_app.file.hmac_sign(data_str)
240 datadump = json.dumps({"dataStr": data_str.decode('ASCII'), "sign": sig})
241 resp = requests.post(conf.file_thumbnailer_url, data=datadump, headers=headers, allow_redirects=False)
242 if resp.status_code != 200: # Error Handling
243 match resp.status_code:
244 case 302:
245 # The problem is Google resposen 302 to an auth Site when the cloudfunction was not found
246 # https://cloud.google.com/functions/docs/troubleshooting#login
247 logging.error("Cloudfunction not found")
248 case 404:
249 logging.error("Cloudfunction not found")
250 case 403:
251 logging.error("No permission for the Cloudfunction")
252 case _:
253 logging.error(
254 f"cloudfunction_thumbnailer failed with code: {resp.status_code} and data: {resp.content}")
255 return
257 try:
258 response_data = resp.json()
259 except Exception as e:
260 logging.error(f"response could not be converted in json failed with: {e=}")
261 return
262 if "error" in response_data:
263 logging.error(f"cloudfunction_thumbnailer failed with: {response_data.get('error')}")
264 return
266 return response_data
268 file_name = html.unescape(fileSkel["name"])
270 if not (url := getsignedurl()):
271 return
272 dataDict = {
273 "url": url,
274 "name": fileSkel["name"],
275 "params": params,
276 "minetype": fileSkel["mimetype"],
277 "baseUrl": current.request.get().request.host_url.lower(),
278 "targetKey": fileSkel["dlkey"],
279 "nameOnly": True
280 }
281 if not (derivedData := make_request()):
282 return
284 uploadUrls = {}
285 for data in derivedData["values"]:
286 fileName = conf.main_app.file.sanitize_filename(data["name"])
287 blob = bucket.blob(f"""{fileSkel["dlkey"]}/derived/{fileName}""")
288 uploadUrls[fileSkel["dlkey"] + fileName] = blob.create_resumable_upload_session(timeout=60,
289 content_type=data["mimeType"])
291 if not (url := getsignedurl()):
292 return
294 dataDict["url"] = url
295 dataDict["nameOnly"] = False
296 dataDict["uploadUrls"] = uploadUrls
298 if not (derivedData := make_request()):
299 return
300 reslist = []
301 try:
302 for derived in derivedData["values"]:
303 for key, value in derived.items():
304 reslist.append((key, value["size"], value["mimetype"], value["customData"]))
306 except Exception as e:
307 logging.error(f"cloudfunction_thumbnailer failed with: {e=}")
308 return reslist
311class DownloadUrlBone(BaseBone):
312 """
313 This bone is used to inject a freshly signed download url into a FileSkel.
314 """
316 def unserialize(self, skel, name):
317 if "dlkey" in skel.dbEntity and "name" in skel.dbEntity:
318 skel.accessedValues[name] = conf.main_app.file.create_download_url(
319 skel["dlkey"], skel["name"], expires=conf.render_json_download_url_expiration
320 )
321 return True
323 return False
326class FileLeafSkel(TreeSkel):
327 """
328 Default file leaf skeleton.
329 """
330 kindName = "file"
332 name = StringBone(
333 descr="Filename",
334 caseSensitive=False,
335 searchable=True,
336 vfunc=lambda val: None if File.is_valid_filename(val) else "Invalid filename provided",
337 )
339 alt = StringBone(
340 descr=i18n.translate(
341 "viur.core.image.alt",
342 defaultText="Alternative description",
343 ),
344 searchable=True,
345 languages=conf.i18n.available_languages,
346 )
348 size = NumericBone(
349 descr="Filesize in Bytes",
350 readOnly=True,
351 searchable=True,
352 )
354 dlkey = StringBone(
355 descr="Download-Key",
356 readOnly=True,
357 )
359 mimetype = StringBone(
360 descr="MIME-Type",
361 readOnly=True,
362 )
364 weak = BooleanBone(
365 descr="Weak reference",
366 readOnly=True,
367 visible=False,
368 )
370 pending = BooleanBone(
371 descr="Pending upload",
372 readOnly=True,
373 visible=False,
374 defaultValue=False,
375 )
377 width = NumericBone(
378 descr="Width",
379 readOnly=True,
380 searchable=True,
381 )
383 height = NumericBone(
384 descr="Height",
385 readOnly=True,
386 searchable=True,
387 )
389 downloadUrl = DownloadUrlBone(
390 descr="Download-URL",
391 readOnly=True,
392 visible=False,
393 )
395 derived = JsonBone(
396 descr="Derived Files",
397 readOnly=True,
398 visible=False,
399 )
401 pendingparententry = KeyBone(
402 descr="Pending key Reference",
403 readOnly=True,
404 visible=False,
405 )
407 crc32c_checksum = StringBone(
408 descr="CRC32C checksum",
409 readOnly=True,
410 )
412 md5_checksum = StringBone(
413 descr="MD5 checksum",
414 readOnly=True,
415 )
417 public = BooleanBone(
418 descr="Public File",
419 readOnly=True,
420 defaultValue=False,
421 )
423 serving_url = StringBone(
424 descr="Serving-URL",
425 readOnly=True,
426 params={
427 "tooltip": "The 'serving_url' is only available in public file repositories.",
428 }
429 )
431 @classmethod
432 def _inject_serving_url(cls, skel: SkeletonInstance) -> None:
433 """Inject the serving url for public image files into a FileSkel"""
434 if (
435 skel["public"]
436 and skel["mimetype"]
437 and skel["mimetype"].startswith("image/")
438 and not skel["serving_url"]
439 ):
440 bucket = File.get_bucket(skel["dlkey"])
441 filename = f"/gs/{bucket.name}/{skel['dlkey']}/source/{utils.string.unescape(skel['name'])}"
443 # Trying this on local development server will raise a
444 # `google.appengine.runtime.apiproxy_errors.RPCFailedError`
445 if conf.instance.is_dev_server:
446 logging.warning(f"Can't inject serving_url for {filename!r} on local development server")
447 return
449 try:
450 skel["serving_url"] = images.get_serving_url(None, secure_url=True, filename=filename)
452 except Exception as e:
453 logging.warning(f"Failed to create serving_url for {filename!r} with exception {e!r}")
454 logging.exception(e)
456 def preProcessBlobLocks(self, locks):
457 """
458 Ensure that our dlkey is locked even if we don't have a filebone here
459 """
460 if not self["weak"] and self["dlkey"]:
461 locks.add(self["dlkey"])
462 return locks
464 @classmethod
465 def refresh(cls, skel):
466 super().refresh(skel)
467 if conf.viur2import_blobsource:
468 importData = importBlobFromViur2(skel["dlkey"], skel["name"])
469 if importData:
470 if not skel["downloadUrl"]:
471 skel["downloadUrl"] = importData
472 skel["pendingparententry"] = None
474 cls._inject_serving_url(skel)
476 @classmethod
477 def write(cls, skel, **kwargs):
478 cls._inject_serving_url(skel)
479 return super().write(skel, **kwargs)
482class FileNodeSkel(TreeSkel):
483 """
484 Default file node skeleton.
485 """
486 kindName = "file_rootNode" # FIXME: VIUR4, don't use "_rootNode" kindname
488 name = StringBone(
489 descr="Name",
490 required=True,
491 searchable=True
492 )
494 rootNode = BooleanBone(
495 descr="Is RootNode",
496 defaultValue=False,
497 readOnly=True,
498 visible=False,
499 )
501 public = BooleanBone(
502 descr="Is public?",
503 defaultValue=False,
504 readOnly=True,
505 visible=False,
506 )
508 viurCurrentSeoKeys = None
511class File(Tree):
512 PENDING_POSTFIX = " (pending)"
513 DOWNLOAD_URL_PREFIX = "/file/download/"
514 INTERNAL_SERVING_URL_PREFIX = "/file/serve/"
515 MAX_FILENAME_LEN = 256
516 IMAGE_META_MAX_SIZE: t.Final[int] = 10 * 1024 ** 2
517 """Maximum size of image files that should be analysed in :meth:`set_image_meta`.
518 Default: 10 MiB"""
520 leafSkelCls = FileLeafSkel
521 nodeSkelCls = FileNodeSkel
523 handler = "tree.simple.file"
524 adminInfo = {
525 "icon": "folder-fill",
526 "handler": handler, # fixme: Use static handler; Remove with VIUR4!
527 }
529 roles = {
530 "*": "view",
531 "editor": ("add", "edit"),
532 "admin": "*",
533 }
535 default_order = "name"
537 # Helper functions currently resist here
539 @staticmethod
540 def get_bucket(dlkey: str) -> google.cloud.storage.bucket.Bucket:
541 """
542 Retrieves a Google Cloud Storage bucket for the given dlkey.
543 """
544 global _public_bucket
545 if dlkey and dlkey.endswith(PUBLIC_DLKEY_SUFFIX):
546 if _public_bucket or (_public_bucket := GOOGLE_STORAGE_CLIENT.lookup_bucket(PUBLIC_BUCKET_NAME)):
547 return _public_bucket
549 raise ValueError(
550 f"""The bucket '{PUBLIC_BUCKET_NAME}' does not exist! Please create it with ACL access."""
551 )
553 return _private_bucket
555 @classmethod
556 def is_valid_filename(cls, filename: str) -> bool:
557 """
558 Verifies a valid filename.
560 The filename should be valid on Linux, Mac OS and Windows.
561 It should not be longer than MAX_FILENAME_LEN chars.
563 Rule set: https://stackoverflow.com/a/31976060/3749896
564 Regex test: https://regex101.com/r/iBYpoC/1
565 """
566 if not filename.strip():
567 return False
569 if len(filename) > cls.MAX_FILENAME_LEN:
570 return False
572 return bool(re.match(VALID_FILENAME_REGEX, filename))
574 @staticmethod
575 def hmac_sign(data: t.Any) -> str:
576 assert conf.file_hmac_key is not None, "No hmac-key set!"
577 if not isinstance(data, bytes): 577 ↛ 578line 577 didn't jump to line 578 because the condition on line 577 was never true
578 data = str(data).encode("UTF-8")
579 return hmac.new(conf.file_hmac_key, msg=data, digestmod=hashlib.sha3_384).hexdigest()
581 @classmethod
582 def hmac_verify(cls, data: t.Any, signature: str) -> bool:
583 try:
584 return hmac.compare_digest(cls.hmac_sign(data.encode("ASCII")), signature)
585 except (TypeError, UnicodeEncodeError):
586 return False
588 @classmethod
589 def create_internal_serving_url(
590 cls,
591 serving_url: str,
592 size: int = 0,
593 filename: str = "",
594 options: str = "",
595 download: bool = False
596 ) -> str:
597 """
598 Helper function to generate an internal serving url (endpoint: /file/serve) from a Google serving url.
600 This is needed to hide requests to Google as they are internally be routed, and can be the result of a
601 legal requirement like GDPR.
603 :param serving_url: Is the original serving URL as generated from FileLeafSkel._inject_serving_url()
604 :param size: Optional size setting
605 :param filename: Optonal filename setting
606 :param options: Additional options parameter-pass through to /file/serve
607 :param download: Download parameter-pass through to /file/serve
608 """
610 # Split a serving URL into its components, used by serve function.
611 res = re.match(
612 r"^https:\/\/(.*?)\.googleusercontent\.com\/(.*?)$",
613 serving_url
614 )
616 if not res:
617 raise ValueError(f"Invalid {serving_url=!r} provided")
619 # Create internal serving URL
620 serving_url = cls.INTERNAL_SERVING_URL_PREFIX + "/".join(res.groups())
622 # Append additional parameters
623 if params := {
624 k: v for k, v in {
625 "download": download,
626 "filename": filename,
627 "options": options,
628 "size": size,
629 }.items() if v
630 }:
631 serving_url += f"?{urlencode(params)}"
633 return serving_url
635 @classmethod
636 def create_download_url(
637 cls,
638 dlkey: str,
639 filename: str,
640 derived: bool = False,
641 expires: t.Optional[datetime.timedelta | int] = datetime.timedelta(hours=1),
642 download_filename: t.Optional[str] = None
643 ) -> str:
644 """
645 Utility function that creates a signed download-url for the given folder/filename combination
647 :param folder: The GCS-Folder (= the download-key) for that file
648 :param filename: The name of the file. Either the original filename or the name of a derived file.
649 :param derived: True, if it points to a derived file, False if it points to the original uploaded file
650 :param expires:
651 None if the file is supposed to be public (which causes it to be cached on the google ede caches),
652 otherwise a datetime.timedelta of how long that link should be valid
653 :param download_filename: If set, browser is enforced to download this blob with the given alternate
654 filename
655 :return: The signed download-url relative to the current domain (eg /download/...)
656 """
657 if isinstance(expires, int): 657 ↛ 658line 657 didn't jump to line 658 because the condition on line 657 was never true
658 expires = datetime.timedelta(minutes=expires)
660 filename = html.unescape(filename)
661 filepath = f"""{dlkey}/{"derived" if derived else "source"}/{filename}"""
663 if download_filename: 663 ↛ 664line 663 didn't jump to line 664 because the condition on line 663 was never true
664 if not cls.is_valid_filename(download_filename):
665 raise errors.UnprocessableEntity(f"Invalid download_filename {download_filename!r} provided")
667 download_filename = urlquote(download_filename)
669 expires = (datetime.datetime.now() + expires).strftime("%Y%m%d%H%M") if expires else 0
671 data = base64.urlsafe_b64encode(f"""{filepath}\0{expires}\0{download_filename or ""}""".encode("UTF-8"))
672 sig = cls.hmac_sign(data)
674 return f"""{cls.DOWNLOAD_URL_PREFIX}{data.decode("ASCII")}?sig={sig}"""
676 @classmethod
677 def parse_download_url(cls, url) -> t.Optional[FilePath]:
678 """
679 Parses a file download URL in the format `/file/download/xxxx?sig=yyyy` into its FilePath.
681 If the URL cannot be parsed, the function returns None.
683 :param url: The file download URL to be parsed.
684 :return: A FilePath on success, None otherwise.
685 """
686 if not url.startswith(cls.DOWNLOAD_URL_PREFIX) or "?" not in url: 686 ↛ 687line 686 didn't jump to line 687 because the condition on line 686 was never true
687 return None
689 data, sig = url.removeprefix(cls.DOWNLOAD_URL_PREFIX).split("?", 1) # Strip "/file/download/" and split on "?"
690 sig = sig.removeprefix("sig=")
692 if not cls.hmac_verify(data, sig): 692 ↛ 694line 692 didn't jump to line 694 because the condition on line 692 was never true
693 # Invalid signature
694 return None
696 # Split the blobKey into the individual fields it should contain
697 data = base64.urlsafe_b64decode(data).decode("UTF-8")
699 match data.count("\0"):
700 case 2: 700 ↛ 702line 700 didn't jump to line 702 because the pattern on line 700 always matched
701 dlpath, valid_until, _ = data.split("\0")
702 case 1:
703 # It's the old format, without an downloadFileName
704 dlpath, valid_until = data.split("\0")
705 case _:
706 # Invalid path
707 return None
709 if valid_until != "0" and datetime.strptime(valid_until, "%Y%m%d%H%M") < datetime.now(): 709 ↛ 711line 709 didn't jump to line 711 because the condition on line 709 was never true
710 # Signature expired
711 return None
713 if dlpath.count("/") != 2: 713 ↛ 715line 713 didn't jump to line 715 because the condition on line 713 was never true
714 # Invalid path
715 return None
717 dlkey, derived, filename = dlpath.split("/")
718 return FilePath(dlkey, derived != "source", filename)
720 @classmethod
721 def create_src_set(
722 cls,
723 file: t.Union["SkeletonInstance", dict, str],
724 expires: t.Optional[datetime.timedelta | int] = datetime.timedelta(hours=1),
725 width: t.Optional[int] = None,
726 height: t.Optional[int] = None,
727 language: t.Optional[str] = None,
728 ) -> str:
729 """
730 Generates a string suitable for use as the srcset tag in html. This functionality provides the browser
731 with a list of images in different sizes and allows it to choose the smallest file that will fill it's
732 viewport without upscaling.
734 :param file: The file skeleton (or if multiple=True a single value from it) to generate the srcset.
735 :param expires:
736 None if the file is supposed to be public (which causes it to be cached on the google edecaches),
737 otherwise it's lifetime in seconds
738 :param width:
739 A list of widths that should be included in the srcset.
740 If a given width is not available, it will be skipped.
741 :param height: A list of heights that should be included in the srcset. If a given height is not available,
742 it will be skipped.
743 :param language: Language overwrite if file has multiple languages, and we want to explicitly specify one
744 :return: The srctag generated or an empty string if a invalid file object was supplied
745 """
746 if not width and not height:
747 logging.error("Neither width or height supplied")
748 return ""
750 if isinstance(file, str):
751 file = db.Query("file").filter("dlkey =", file).order(("creationdate", db.SortOrder.Ascending)).getEntry()
753 if not file:
754 return ""
756 if isinstance(file, i18n.LanguageWrapper):
757 language = language or current.language.get()
758 if not language or not (file := cls.get(language)):
759 return ""
761 if "dlkey" not in file and "dest" in file:
762 file = file["dest"]
764 from viur.core.skeleton import SkeletonInstance # avoid circular imports
766 if not (
767 isinstance(file, (SkeletonInstance, dict))
768 and "dlkey" in file
769 and "derived" in file
770 ):
771 logging.error("Invalid file supplied")
772 return ""
774 if not isinstance(file["derived"], dict):
775 logging.error("No derives available")
776 return ""
778 src_set = []
779 for filename, derivate in file["derived"]["files"].items():
780 customData = derivate.get("customData", {})
782 if width and customData.get("width") in width:
783 src_set.append(
784 f"""{cls.create_download_url(file["dlkey"], filename, True, expires)} {customData["width"]}w"""
785 )
787 if height and customData.get("height") in height:
788 src_set.append(
789 f"""{cls.create_download_url(file["dlkey"], filename, True, expires)} {customData["height"]}h"""
790 )
792 return ", ".join(src_set)
794 def write(
795 self,
796 filename: str,
797 content: t.Any,
798 mimetype: str = "text/plain",
799 *,
800 width: int = None,
801 height: int = None,
802 public: bool = False,
803 rootnode: t.Optional[db.Key] = None,
804 folder: t.Iterable[str] | str = (),
805 ) -> db.Key:
806 """
807 Write a file from any bytes-like object into the file module.
809 If *folder* and *rootnode* are both set, the file is added to the repository in that folder.
810 If only *folder* is set, the file is added to the default repository in that folder.
811 If only *rootnode* is set, the file is added to that repository in the root folder.
813 If both are not set, the file is added without a path or repository as a weak file.
814 It will not be visible in admin in this case.
816 :param filename: Filename to be written.
817 :param content: The file content to be written, as bytes-like object.
818 :param mimetype: The file's mimetype.
819 :param width: Optional width information for the file.
820 :param height: Optional height information for the file.
821 :param public: True if the file should be publicly accessible.
822 :param rootnode: Optional root-node of the repository to add the file to
823 :param folder: Optional folder the file should be written into.
825 :return: Returns the key of the file object written. This can be associated e.g. with a FileBone.
826 """
827 # logging.info(f"{filename=} {mimetype=} {width=} {height=} {public=}")
828 if not self.is_valid_filename(filename):
829 raise ValueError(f"{filename=} is invalid")
831 # Folder mode?
832 if folder:
833 # Validate correct folder naming
834 if isinstance(folder, str):
835 folder = folder, # make it a tuple
837 for foldername in folder:
838 if not self.is_valid_filename(foldername):
839 raise ValueError(f"{foldername=} is invalid")
841 # When in folder-mode, a rootnode must exist!
842 if rootnode is None:
843 rootnode = self.ensureOwnModuleRootNode()
845 parentrepokey = rootnode.key
846 parentfolderkey = rootnode.key
848 for foldername in folder:
849 query = self.addSkel("node").all()
850 query.filter("parentrepo", parentrepokey)
851 query.filter("parententry", parentfolderkey)
852 query.filter("name", foldername)
854 if folder_skel := query.getSkel():
855 # Skip existing folder
856 parentfolderkey = folder_skel["key"]
857 else:
858 # Create new folder
859 folder_skel = self.addSkel("node")
861 folder_skel["name"] = foldername
862 folder_skel["parentrepo"] = parentrepokey
863 folder_skel["parententry"] = parentfolderkey
864 folder_skel.write()
866 parentfolderkey = folder_skel["key"]
868 else:
869 parentrepokey = None
870 parentfolderkey = None
872 # Write the file
873 dl_key = utils.string.random()
875 if public:
876 dl_key += PUBLIC_DLKEY_SUFFIX # mark file as public
878 bucket = self.get_bucket(dl_key)
880 blob = bucket.blob(f"{dl_key}/source/{filename}")
881 blob.upload_from_file(io.BytesIO(content), content_type=mimetype)
883 fileskel = self.addSkel("leaf")
885 fileskel["parentrepo"] = parentrepokey
886 fileskel["parententry"] = parentfolderkey
887 fileskel["name"] = filename
888 fileskel["size"] = blob.size
889 fileskel["mimetype"] = mimetype
890 fileskel["dlkey"] = dl_key
891 fileskel["weak"] = bool(parentrepokey)
892 fileskel["public"] = public
893 fileskel["width"] = width
894 fileskel["height"] = height
895 fileskel["crc32c_checksum"] = base64.b64decode(blob.crc32c).hex()
896 fileskel["md5_checksum"] = base64.b64decode(blob.md5_hash).hex()
897 fileskel["pending"] = False
899 return fileskel.write()["key"]
901 def read(
902 self,
903 key: db.Key | int | str | None = None,
904 path: str | None = None,
905 ) -> tuple[io.BytesIO, str]:
906 """
907 Read a file from the Cloud Storage.
909 If a key and a path are provided, the key is preferred.
910 This means that the entry in the db is searched first and if this is not found, the path is used.
912 :param key: Key of the LeafSkel that contains the "dlkey" and the "name".
913 :param path: The path of the file in the Cloud Storage Bucket.
915 :return: Returns the file as a io.BytesIO buffer and the content-type
916 """
917 if not key and not path:
918 raise ValueError("Please provide a key or a path")
920 if key:
921 skel = self.viewSkel("leaf")
922 if not skel.read(db.key_helper(key, skel.kindName)):
923 if not path:
924 raise ValueError("This skeleton is not in the database!")
925 else:
926 path = f"""{skel["dlkey"]}/source/{skel["name"]}"""
928 bucket = self.get_bucket(skel["dlkey"])
929 else:
930 bucket = self.get_bucket(path.split("/", 1)[0]) # path's first part is dlkey plus eventual postfix
932 blob = bucket.blob(path)
933 return io.BytesIO(blob.download_as_bytes()), blob.content_type
935 @CallDeferred
936 def deleteRecursive(self, parentKey):
937 files = db.Query(self.leafSkelCls().kindName).filter("parentdir =", parentKey).iter()
938 for fileEntry in files:
939 self.mark_for_deletion(fileEntry["dlkey"])
940 skel = self.leafSkelCls()
942 if skel.read(str(fileEntry.key())):
943 skel.delete()
944 dirs = db.Query(self.nodeSkelCls().kindName).filter("parentdir", parentKey).iter()
945 for d in dirs:
946 self.deleteRecursive(d.key)
947 skel = self.nodeSkelCls()
948 if skel.read(d.key):
949 skel.delete()
951 @exposed
952 @skey
953 def getUploadURL(
954 self,
955 fileName: str,
956 mimeType: str,
957 size: t.Optional[int] = None,
958 node: t.Optional[str | db.Key] = None,
959 authData: t.Optional[str] = None,
960 authSig: t.Optional[str] = None,
961 public: bool = False,
962 ):
963 filename = fileName.strip() # VIUR4 FIXME: just for compatiblity of the parameter names
965 if not self.is_valid_filename(filename):
966 raise errors.UnprocessableEntity(f"Invalid filename {filename!r} provided")
968 # Validate the mimetype from the client seems legit
969 mimetype = mimeType.strip().lower()
970 if not (
971 mimetype
972 and mimetype.count("/") == 1
973 and all(ch in string.ascii_letters + string.digits + "/-.+" for ch in mimetype)
974 ):
975 raise errors.UnprocessableEntity(f"Invalid mime-type {mimetype!r} provided")
977 # Validate authentication data
978 if authData and authSig:
979 # First, validate the signature, otherwise we don't need to proceed further
980 if not self.hmac_verify(authData, authSig):
981 raise errors.Unauthorized()
983 authData = json.loads(base64.b64decode(authData.encode("ASCII")).decode("UTF-8"))
985 if datetime.datetime.strptime(authData["validUntil"], "%Y%m%d%H%M") < datetime.datetime.now():
986 raise errors.Gone("The upload URL has expired")
988 if authData["validMimeTypes"]:
989 for validMimeType in authData["validMimeTypes"]:
990 if (
991 validMimeType == mimetype
992 or (validMimeType.endswith("*") and mimetype.startswith(validMimeType[:-1]))
993 ):
994 break
995 else:
996 raise errors.UnprocessableEntity(f"Invalid mime-type {mimetype} provided")
998 node = authData["node"]
999 maxSize = authData["maxSize"]
1001 else:
1002 rootNode = None
1003 if node and not (rootNode := self.getRootNode(node)):
1004 raise errors.NotFound(f"No valid root node found for {node=}")
1006 if not self.canAdd("leaf", rootNode):
1007 raise errors.Forbidden()
1009 if rootNode and public != bool(rootNode.get("public")):
1010 raise errors.Forbidden("Cannot upload a public file into private repository or vice versa")
1012 maxSize = None # The user has some file/add permissions, don't restrict fileSize
1014 if maxSize:
1015 if size > maxSize:
1016 raise errors.UnprocessableEntity(f"Size {size} exceeds maximum size {maxSize}")
1017 else:
1018 size = None
1020 # Create upload-URL and download key
1021 dlkey = utils.string.random() # let's roll a random key
1023 if public:
1024 dlkey += PUBLIC_DLKEY_SUFFIX # mark file as public
1026 blob = self.get_bucket(dlkey).blob(f"{dlkey}/source/{filename}")
1027 upload_url = blob.create_resumable_upload_session(content_type=mimeType, size=size, timeout=60)
1029 # Create a corresponding file-lock object early, otherwise we would have to ensure that the file-lock object
1030 # the user creates matches the file he had uploaded
1031 file_skel = self.addSkel("leaf")
1033 file_skel["name"] = filename + self.PENDING_POSTFIX
1034 file_skel["size"] = 0
1035 file_skel["mimetype"] = "application/octetstream"
1036 file_skel["dlkey"] = dlkey
1037 file_skel["parentdir"] = None
1038 file_skel["pendingparententry"] = db.key_helper(node, self.addSkel("node").kindName) if node else None
1039 file_skel["pending"] = True
1040 file_skel["weak"] = True
1041 file_skel["public"] = public
1042 file_skel["width"] = 0
1043 file_skel["height"] = 0
1045 file_skel.write()
1046 key = str(file_skel["key"])
1048 # Mark that entry dirty as we might never receive an add
1049 self.mark_for_deletion(dlkey)
1051 # In this case, we'd have to store the key in the users session so he can call add() later on
1052 if authData and authSig:
1053 session = current.session.get()
1055 if "pendingFileUploadKeys" not in session:
1056 session["pendingFileUploadKeys"] = []
1058 session["pendingFileUploadKeys"].append(key)
1060 # Clamp to the latest 50 pending uploads
1061 session["pendingFileUploadKeys"] = session["pendingFileUploadKeys"][-50:]
1062 session.markChanged()
1064 return self.render.view({
1065 "uploadKey": key,
1066 "uploadUrl": upload_url,
1067 })
1069 @exposed
1070 def download(self, blobKey: str, fileName: str = "", download: bool = False, sig: str = "", *args, **kwargs):
1071 """
1072 Download a file.
1073 :param blobKey: The unique blob key of the file.
1074 :param fileName: Optional filename to provide in the header.
1075 :param download: Set header to attachment retrival, set explictly to "1" if download is wanted.
1076 """
1077 if filename := fileName.strip():
1078 if not self.is_valid_filename(filename):
1079 raise errors.UnprocessableEntity(f"The provided filename {filename!r} is invalid!")
1081 try:
1082 values = base64.urlsafe_b64decode(blobKey).decode("UTF-8").split("\0")
1083 except ValueError:
1084 raise errors.BadRequest(f"Invalid encoding of blob key {blobKey!r}!")
1085 try:
1086 dlPath, validUntil, *download_filename = values
1087 # Maybe it's the old format, without a download_filename
1088 download_filename = download_filename[0] if download_filename else ""
1089 except ValueError:
1090 logging.error(f"Encoding of {blobKey=!r} OK. {values=} invalid.")
1091 raise errors.BadRequest(f"The blob key {blobKey!r} has an invalid amount of encoded values!")
1093 bucket = self.get_bucket(dlPath.split("/", 1)[0])
1095 if not sig:
1096 # Check if the current user has the right to download *any* blob present in this application.
1097 # blobKey is then the path inside cloudstore - not a base64 encoded tuple
1098 if not (usr := current.user.get()):
1099 raise errors.Unauthorized()
1100 if "root" not in usr["access"] and "file-view" not in usr["access"]:
1101 raise errors.Forbidden()
1102 validUntil = "-1" # Prevent this from being cached down below
1103 blob = bucket.get_blob(blobKey)
1105 else:
1106 # We got an request including a signature (probably a guest or a user without file-view access)
1107 # First, validate the signature, otherwise we don't need to proceed any further
1108 if not self.hmac_verify(blobKey, sig):
1109 raise errors.Forbidden()
1111 if validUntil != "0" and datetime.datetime.strptime(validUntil, "%Y%m%d%H%M") < datetime.datetime.now():
1112 blob = None
1113 else:
1114 blob = bucket.get_blob(dlPath)
1116 if not blob:
1117 raise errors.Gone("The requested blob has expired.")
1119 if not filename:
1120 filename = download_filename or urlquote(blob.name.rsplit("/", 1)[-1])
1122 content_disposition = utils.build_content_disposition_header(filename, attachment=download)
1124 if isinstance(_CREDENTIALS, ServiceAccountCredentials):
1125 expiresAt = datetime.datetime.now() + datetime.timedelta(seconds=60)
1126 signedUrl = blob.generate_signed_url(expiresAt, response_disposition=content_disposition, version="v4")
1127 raise errors.Redirect(signedUrl)
1129 elif conf.instance.is_dev_server: # No Service-Account to sign with - Serve everything directly
1130 response = current.request.get().response
1131 response.headers["Content-Type"] = blob.content_type
1132 if content_disposition:
1133 response.headers["Content-Disposition"] = content_disposition
1134 return blob.download_as_bytes()
1136 if validUntil == "0" or blobKey.endswith(PUBLIC_DLKEY_SUFFIX): # Its an indefinitely valid URL
1137 if blob.size < 5 * 1024 * 1024: # Less than 5 MB - Serve directly and push it into the ede caches
1138 response = current.request.get().response
1139 response.headers["Content-Type"] = blob.content_type
1140 response.headers["Cache-Control"] = "public, max-age=604800" # 7 Days
1141 if content_disposition:
1142 response.headers["Content-Disposition"] = content_disposition
1143 return blob.download_as_bytes()
1145 # Default fallback - create a signed URL and redirect
1146 authRequest = google.auth.transport.requests.Request()
1147 expiresAt = datetime.datetime.now() + datetime.timedelta(seconds=60)
1148 signing_credentials = google.auth.compute_engine.IDTokenCredentials(authRequest, "")
1149 signedUrl = blob.generate_signed_url(
1150 expiresAt,
1151 credentials=signing_credentials,
1152 response_disposition=content_disposition,
1153 version="v4")
1155 raise errors.Redirect(signedUrl)
1157 SERVE_VALID_OPTIONS = {
1158 "c",
1159 "p",
1160 "fv",
1161 "fh",
1162 "r90",
1163 "r180",
1164 "r270",
1165 "nu",
1166 }
1167 """
1168 Valid modification option shorts for the serve-function.
1169 This is passed-through to the Google UserContent API, and hast to be supported there.
1170 """
1172 SERVE_VALID_FORMATS = {
1173 "jpg": "rj",
1174 "jpeg": "rj",
1175 "png": "rp",
1176 "webp": "rw",
1177 }
1178 """
1179 Valid file-formats to the serve-function.
1180 This is passed-through to the Google UserContent API, and hast to be supported there.
1181 """
1183 @exposed
1184 def serve(
1185 self,
1186 host: str,
1187 key: str,
1188 size: t.Optional[int] = None,
1189 filename: t.Optional[str] = None,
1190 options: str = "",
1191 download: bool = False,
1192 ):
1193 """
1194 Requests an image using the serving url to bypass direct Google requests.
1196 :param host: the google host prefix i.e. lh3
1197 :param key: the serving url key
1198 :param size: the target image size
1199 :param filename: a random string with an extention, valid extentions are (defined in File.SERVE_VALID_FORMATS).
1200 :param options: - seperated options (defined in File.SERVE_VALID_OPTIONS).
1201 c - crop
1202 p - face crop
1203 fv - vertrical flip
1204 fh - horizontal flip
1205 rXXX - rotate 90, 180, 270
1206 nu - no upscale
1207 :param download: Serves the content as download (Content-Disposition) or not.
1209 :return: Returns the requested content on success, raises a proper HTTP exception otherwise.
1210 """
1212 if any(c not in conf.search_valid_chars for c in host):
1213 raise errors.BadRequest("key contains invalid characters")
1215 # extract format from filename
1216 file_fmt = "webp"
1218 if filename:
1219 fmt = filename.rsplit(".", 1)[-1].lower()
1220 if fmt in self.SERVE_VALID_FORMATS:
1221 file_fmt = fmt
1222 else:
1223 raise errors.UnprocessableEntity(f"Unsupported filetype {fmt}")
1225 url = f"https://{host}.googleusercontent.com/{key}"
1227 if options and not all(param in self.SERVE_VALID_OPTIONS for param in options.split("-")):
1228 raise errors.BadRequest("Invalid options provided")
1230 options += f"-{self.SERVE_VALID_FORMATS[file_fmt]}"
1232 if size:
1233 options = f"s{size}-" + options
1235 url += "=" + options
1237 response = current.request.get().response
1238 response.headers["Content-Type"] = f"image/{file_fmt}"
1239 response.headers["Cache-Control"] = "public, max-age=604800" # 7 Days
1240 response.headers["Content-Disposition"] = utils.build_content_disposition_header(filename, attachment=download)
1242 answ = requests.get(url, timeout=20)
1243 if not answ.ok:
1244 logging.error(f"{answ.status_code} {answ.text}")
1245 raise errors.BadRequest("Unable to fetch a file with these parameters")
1247 return answ.content
1249 @exposed
1250 @force_ssl
1251 @force_post
1252 @skey(allow_empty=True)
1253 def add(self, skelType: SkelType, node: db.Key | int | str | None = None, *args, **kwargs):
1254 # We can't add files directly (they need to be uploaded
1255 if skelType == "leaf": # We need to handle leafs separately here
1256 targetKey = kwargs.get("key")
1257 skel = self.addSkel("leaf")
1259 if not skel.read(targetKey):
1260 raise errors.NotFound()
1262 if not skel["pending"]:
1263 raise errors.PreconditionFailed()
1265 skel["pending"] = False
1266 skel["parententry"] = skel["pendingparententry"]
1268 if skel["parententry"]:
1269 rootNode = self.getRootNode(skel["parententry"])
1270 else:
1271 rootNode = None
1273 if not self.canAdd("leaf", rootNode):
1274 # Check for a marker in this session (created if using a signed upload URL)
1275 session = current.session.get()
1276 if targetKey not in (session.get("pendingFileUploadKeys") or []):
1277 raise errors.Forbidden()
1278 session["pendingFileUploadKeys"].remove(targetKey)
1279 session.markChanged()
1281 # Now read the blob from the dlkey folder
1282 bucket = self.get_bucket(skel["dlkey"])
1284 blobs = list(bucket.list_blobs(prefix=f"""{skel["dlkey"]}/"""))
1285 if len(blobs) != 1:
1286 logging.error("Invalid number of blobs in folder")
1287 logging.error(targetKey)
1288 raise errors.PreconditionFailed()
1290 # only one item is allowed here!
1291 blob = blobs[0]
1293 # update the corresponding file skeleton
1294 skel["name"] = skel["name"].removesuffix(self.PENDING_POSTFIX)
1295 skel["mimetype"] = utils.string.escape(blob.content_type)
1296 skel["size"] = blob.size
1297 skel["parentrepo"] = rootNode["key"] if rootNode else None
1298 skel["weak"] = rootNode is None
1299 skel["crc32c_checksum"] = base64.b64decode(blob.crc32c).hex()
1300 skel["md5_checksum"] = base64.b64decode(blob.md5_hash).hex()
1301 self.onAdd("leaf", skel)
1302 skel.write()
1303 self.onAdded("leaf", skel)
1305 # Add updated download-URL as the auto-generated isn't valid yet
1306 skel["downloadUrl"] = self.create_download_url(skel["dlkey"], skel["name"])
1308 return self.render.addSuccess(skel)
1310 return super().add(skelType, node, *args, **kwargs)
1312 @exposed
1313 def get_download_url(
1314 self,
1315 key: t.Optional[db.Key] = None,
1316 dlkey: t.Optional[str] = None,
1317 filename: t.Optional[str] = None,
1318 derived: bool = False,
1319 ):
1320 """
1321 Request a download url for a given file
1322 :param key: The key of the file
1323 :param dlkey: The download key of the file
1324 :param filename: The filename to be given. If no filename is provided
1325 downloadUrls for all derived files are returned in case of `derived=True`.
1326 :param derived: True, if a derived file download URL is being requested.
1327 """
1328 skel = self.viewSkel("leaf")
1329 if dlkey is not None:
1330 skel = skel.all().filter("dlkey", dlkey).getSkel()
1331 elif key is None and dlkey is None:
1332 raise errors.BadRequest("No key or dlkey provided")
1334 if not (skel and skel.read(key)):
1335 raise errors.NotFound()
1337 if not self.canView("leaf", skel):
1338 raise errors.Unauthorized()
1340 dlkey = skel["dlkey"]
1342 if derived and filename is None:
1343 res = {}
1344 for filename in skel["derived"]["files"]:
1345 res[filename] = self.create_download_url(dlkey, filename, derived)
1346 else:
1347 if derived:
1348 # Check if Filename exist in the Derives. We sign nothing that not exist.
1349 if filename not in skel["derived"]["files"]:
1350 raise errors.NotFound("File not in derives")
1351 else:
1352 if filename is None:
1353 filename = skel["name"]
1354 elif filename != skel["name"]:
1355 raise errors.NotFound("Filename not match")
1357 res = self.create_download_url(dlkey, filename, derived)
1359 return self.render.view(res)
1361 def onEdit(self, skelType: SkelType, skel: SkeletonInstance):
1362 super().onEdit(skelType, skel)
1364 if skelType == "leaf":
1365 old_skel = self.editSkel(skelType)
1366 old_skel.setEntity(skel.dbEntity)
1368 if old_skel["name"] == skel["name"]: # name not changed we can return
1369 return
1371 # Move Blob to new name
1372 # https://cloud.google.com/storage/docs/copying-renaming-moving-objects
1373 old_path = f"""{skel["dlkey"]}/source/{html.unescape(old_skel["name"])}"""
1374 new_path = f"""{skel["dlkey"]}/source/{html.unescape(skel["name"])}"""
1376 bucket = self.get_bucket(skel["dlkey"])
1378 if not (old_blob := bucket.get_blob(old_path)):
1379 raise errors.Gone()
1381 bucket.copy_blob(old_blob, bucket, new_path, if_generation_match=0)
1382 bucket.delete_blob(old_path)
1384 def onAdded(self, skelType: SkelType, skel: SkeletonInstance) -> None:
1385 if skelType == "leaf" and skel["mimetype"].startswith("image/"):
1386 if skel["size"] > self.IMAGE_META_MAX_SIZE:
1387 logging.warning(f"File size {skel['size']} exceeds limit {self.IMAGE_META_MAX_SIZE=}")
1388 return
1389 self.set_image_meta(skel["key"])
1391 super().onAdded(skelType, skel)
1393 @CallDeferred
1394 def set_image_meta(self, key: db.Key) -> None:
1395 """Write image metadata (height and width) to FileSkel"""
1396 skel = self.editSkel("leaf", key)
1397 if not skel.read(key):
1398 logging.error(f"File {key} does not exist")
1399 return
1400 if skel["width"] and skel["height"]:
1401 logging.info(f'File {skel["key"]} has already {skel["width"]=} and {skel["height"]=}')
1402 return
1403 file_name = html.unescape(skel["name"])
1404 blob = self.get_bucket(skel["dlkey"]).get_blob(f"""{skel["dlkey"]}/source/{file_name}""")
1405 if not blob:
1406 logging.error(f'Blob {skel["dlkey"]}/source/{file_name} is missing in Cloud Storage!')
1407 return
1409 file_obj = io.BytesIO()
1410 blob.download_to_file(file_obj)
1411 file_obj.seek(0)
1412 try:
1413 img = Image.open(file_obj)
1414 except Image.UnidentifiedImageError as e: # Can't load this image
1415 logging.exception(f'Cannot open {skel["key"]} | {skel["name"]} to set image meta data: {e}')
1416 return
1418 skel.patch(
1419 values={
1420 "width": img.width,
1421 "height": img.height,
1422 },
1423 )
1425 def mark_for_deletion(self, dlkey: str) -> None:
1426 """
1427 Adds a marker to the datastore that the file specified as *dlkey* can be deleted.
1429 Once the mark has been set, the data store is checked four times (default: every 4 hours)
1430 if the file is in use somewhere. If it is still in use, the mark goes away, otherwise
1431 the mark and the file are removed from the datastore. These delayed checks are necessary
1432 due to database inconsistency.
1434 :param dlkey: Unique download-key of the file that shall be marked for deletion.
1435 """
1436 fileObj = db.Query("viur-deleted-files").filter("dlkey", dlkey).getEntry()
1438 if fileObj: # Its allready marked
1439 return
1441 fileObj = db.Entity(db.Key("viur-deleted-files"))
1442 fileObj["itercount"] = 0
1443 fileObj["dlkey"] = str(dlkey)
1445 db.put(fileObj)
1448@PeriodicTask(interval=datetime.timedelta(hours=4))
1449def startCheckForUnreferencedBlobs():
1450 """
1451 Start searching for blob locks that have been recently freed
1452 """
1453 doCheckForUnreferencedBlobs()
1456@CallDeferred
1457def doCheckForUnreferencedBlobs(cursor=None):
1458 def getOldBlobKeysTxn(dbKey):
1459 obj = db.get(dbKey)
1460 res = obj["old_blob_references"] or []
1461 if obj["is_stale"]:
1462 db.delete(dbKey)
1463 else:
1464 obj["has_old_blob_references"] = False
1465 obj["old_blob_references"] = []
1466 db.put(obj)
1467 return res
1469 query = db.Query("viur-blob-locks").filter("has_old_blob_references", True).setCursor(cursor)
1470 for lockObj in query.run(100):
1471 oldBlobKeys = db.run_in_transaction(getOldBlobKeysTxn, lockObj.key)
1472 for blobKey in oldBlobKeys:
1473 if db.Query("viur-blob-locks").filter("active_blob_references =", blobKey).getEntry():
1474 # This blob is referenced elsewhere
1475 logging.info(f"Stale blob is still referenced, {blobKey}")
1476 continue
1477 # Add a marker and schedule it for deletion
1478 fileObj = db.Query("viur-deleted-files").filter("dlkey", blobKey).getEntry()
1479 if fileObj: # Its already marked
1480 logging.info(f"Stale blob already marked for deletion, {blobKey}")
1481 return
1482 fileObj = db.Entity(db.Key("viur-deleted-files"))
1483 fileObj["itercount"] = 0
1484 fileObj["dlkey"] = str(blobKey)
1485 logging.info(f"Stale blob marked dirty, {blobKey}")
1486 db.put(fileObj)
1487 newCursor = query.getCursor()
1488 if newCursor:
1489 doCheckForUnreferencedBlobs(newCursor)
1492@PeriodicTask(interval=datetime.timedelta(hours=4))
1493def startCleanupDeletedFiles():
1494 """
1495 Increase deletion counter on each blob currently not referenced and delete
1496 it if that counter reaches maxIterCount
1497 """
1498 doCleanupDeletedFiles()
1501@CallDeferred
1502def doCleanupDeletedFiles(cursor=None):
1503 maxIterCount = 2 # How often a file will be checked for deletion
1504 query = db.Query("viur-deleted-files")
1505 if cursor:
1506 query.setCursor(cursor)
1507 for file in query.run(100):
1508 if "dlkey" not in file:
1509 db.delete(file.key)
1510 elif db.Query("viur-blob-locks").filter("active_blob_references =", file["dlkey"]).getEntry():
1511 logging.info(f"""is referenced, {file["dlkey"]}""")
1512 db.delete(file.key)
1513 else:
1514 if file["itercount"] > maxIterCount:
1515 logging.info(f"""Finally deleting, {file["dlkey"]}""")
1516 bucket = conf.main_app.file.get_bucket(file["dlkey"])
1517 blobs = bucket.list_blobs(prefix=f"""{file["dlkey"]}/""")
1518 for blob in blobs:
1519 blob.delete()
1520 db.delete(file.key)
1521 # There should be exactly 1 or 0 of these
1522 for f in skeletonByKind("file")().all().filter("dlkey =", file["dlkey"]).fetch(99):
1523 f.delete()
1525 if f["serving_url"]:
1526 bucket = conf.main_app.file.get_bucket(f["dlkey"])
1527 blob_key = blobstore.create_gs_key(
1528 f"/gs/{bucket.name}/{f['dlkey']}/source/{f['name']}"
1529 )
1530 images.delete_serving_url(blob_key) # delete serving url
1531 else:
1532 logging.debug(f"""Increasing count, {file["dlkey"]}""")
1533 file["itercount"] += 1
1534 db.put(file)
1535 newCursor = query.getCursor()
1536 if newCursor:
1537 doCleanupDeletedFiles(newCursor)
1540@PeriodicTask(interval=datetime.timedelta(hours=4))
1541def start_delete_pending_files():
1542 """
1543 Start deletion of pending FileSkels that are older than 7 days.
1544 """
1545 DeleteEntitiesIter.startIterOnQuery(
1546 FileLeafSkel().all()
1547 .filter("pending =", True)
1548 .filter("creationdate <", utils.utcNow() - datetime.timedelta(days=7))
1549 )
1552# DEPRECATED ATTRIBUTES HANDLING
1554def __getattr__(attr: str) -> object:
1555 if entry := { 1555 ↛ 1559line 1555 didn't jump to line 1559 because the condition on line 1555 was never true
1556 # stuff prior viur-core < 3.7
1557 "GOOGLE_STORAGE_BUCKET": ("conf.main_app.file.get_bucket()", _private_bucket),
1558 }.get(attr):
1559 msg = f"{attr} was replaced by {entry[0]}"
1560 warnings.warn(msg, DeprecationWarning, stacklevel=2)
1561 logging.warning(msg, stacklevel=2)
1562 return entry[1]
1564 return super(__import__(__name__).__class__).__getattribute__(attr)