Moved to single file storage (appending chunks to existing file)

Added file finalize actions
Added error handling for files
Moved everything to pathlib
Simplified models
Squashed all migrations to single operation

.gitignore

@@ -94,4 +94,5 @@ venv.bak/
 
 lockbox/media
 lockbox/staticfiles
-TODO.txt
+TODO.txt
+FILES

lockbox/common/constants.py

@@ -7,18 +7,15 @@ class UPLOAD_STATUS_TYPES:
     UPLOADING = "uploading"
     COMPLETED = "completed"
     ABANDONED = "abandoned"
-    PROCESSING = "processing"
+    ERROR = "error"
+
+class UPLOAD_ERROR_CODES:
+    FILE_MISSING = "file_missing"
+    CHUNK_MISMATCH = "chunk_mismatch"
 
 # Config
 
 CONFIG_KEYS = {
-    "EXPIRATION_DELTA_MINUTES": {
-        "description": "Date created + this delta at which file expires",
-        "verbose_name": "File expiration delta (minutes)",
-        "native_type": int,
-        "sensitive": False,
-        "default": 120,
-    },
     "ABANDONED_DELTA_MINUTES": {
         "description": "Date created + this delta at which a file is marked as abandoned",
         "verbose_name": "Uncompleted file abandoned max age",

lockbox/lockbox/settings.py

@@ -120,6 +120,7 @@ STORAGES = {
 # Storage
 MEDIA_ROOT = Path("/home/kitty/src/lockbox/FILES")
 MEDIA_URL = "files/"
+INCOMPLETE_EXT = ".incomplete"
 
 validate_paths(MEDIA_ROOT)
 

lockbox/lockbox/setup.py

@@ -1,11 +1,10 @@
-import os
+from pathlib import Path
 
 # TODO: LOG MEEEEE
 # TODO: Figure out file owner in system, permissions, GUID
 # Whats the default path if not provided? // docker volume
 def validate_paths(media_path):
-    if not os.path.isdir(media_path):
         try:
-            os.makedirs(media_path)
+            Path(media_path).mkdir(exist_ok=True)
         except Exception as e:
             raise e

lockbox/storage/migrations/0001_initial.py

@@ -1,4 +1,4 @@
-# Generated by Django 4.2.15 on 2024-09-16 11:24
+# Generated by Django 4.2.15 on 2024-09-17 06:52
 
 import common.utils
 from django.conf import settings
@@ -23,15 +23,16 @@ class Migration(migrations.Migration):
                 ('lid', models.UUIDField(default=uuid.uuid4, primary_key=True, serialize=False, verbose_name='lockbox ID')),
                 ('date_created', models.DateTimeField(blank=True, help_text='date at which this object was created', verbose_name='date created')),
                 ('date_updated', models.DateTimeField(blank=True, help_text='date at which this object was last updated', verbose_name='date updated')),
-                ('filename', models.CharField(help_text='Name of the file', max_length=255, verbose_name='name')),
-                ('extension', models.CharField(blank=True, help_text='reported filesystem extension (not mime type)', max_length=128, null=True, verbose_name='extension')),
-                ('file', models.FileField(blank=True, help_text='actual file', null=True, upload_to='', verbose_name='file')),
-                ('status', models.CharField(choices=[('uploading', 'uploading'), ('completed', 'completed'), ('processing', 'processing'), ('abandoned', 'abandoned')], default='uploading', help_text='upload status for file', max_length=10, verbose_name='status')),
-                ('date_completed', models.DateTimeField(blank=True, help_text="datetime at which this file's upload was completed", null=True, verbose_name='completed on')),
+                ('mime_type', models.CharField(blank=True, help_text='reported mime-type', max_length=128, null=True, verbose_name='mime-type')),
+                ('file', models.FileField(blank=True, help_text='actual file', null=True, upload_to=storage.models.upload_to_fielpath, verbose_name='file')),
+                ('status', models.CharField(choices=[('uploading', 'uploading'), ('completed', 'completed'), ('abandoned', 'abandoned'), ('error', 'error')], default='uploading', help_text='upload status for file', max_length=10, verbose_name='status')),
+                ('datetime_completed', models.DateTimeField(blank=True, help_text="datetime at which this file's upload was completed", null=True, verbose_name='completed on')),
                 ('expires', models.BooleanField(default=False, help_text="will be scrubbed on 'date_expires'", verbose_name='expires')),
                 ('delete_on_expiration', models.BooleanField(default=False, help_text='will be deleted if expired and expires is true', verbose_name='delete on expiration')),
-                ('size_on_disk', models.PositiveBigIntegerField(blank=True, help_text='total size on disk for this file', null=True, verbose_name='size on disk (bytes)')),
+                ('size', models.PositiveBigIntegerField(blank=True, help_text='total size on disk for this file', null=True, verbose_name='size (bytes)')),
+                ('expected_size', models.PositiveBigIntegerField(blank=True, help_text='expected file size', null=True, verbose_name='expected size (bytes)')),
                 ('max_size_chunk_bytes', models.PositiveBigIntegerField(default=common.utils.get_max_size_chunk_bytes, help_text='max size of each individual chunk for this file', verbose_name='maximum size of chunks (bytes)')),
+                ('last_end_bytes', models.BigIntegerField(blank=True, help_text='last uploaded bytes position', null=True, verbose_name='last end bytes')),
                 ('owner', models.ForeignKey(blank=True, help_text='Who owns this file', null=True, on_delete=django.db.models.deletion.SET_NULL, related_name='files_owned', to=settings.AUTH_USER_MODEL, verbose_name='owner')),
             ],
             options={
@@ -39,23 +40,4 @@ class Migration(migrations.Migration):
                 'verbose_name_plural': 'files',
             },
         ),
-        migrations.CreateModel(
-            name='FileChunk',
-            fields=[
-                ('lid', models.UUIDField(default=uuid.uuid4, primary_key=True, serialize=False, verbose_name='lockbox ID')),
-                ('date_created', models.DateTimeField(blank=True, help_text='date at which this object was created', verbose_name='date created')),
-                ('date_updated', models.DateTimeField(blank=True, help_text='date at which this object was last updated', verbose_name='date updated')),
-                ('chunk', models.FileField(help_text='chunk file', upload_to=storage.models.get_upload_path_chunk, verbose_name='chunk file')),
-                ('chunk_id', models.BigIntegerField(help_text='chunk id', verbose_name='chunk id')),
-                ('size', models.BigIntegerField(help_text='chunk size', verbose_name='size')),
-                ('start_bytes', models.BigIntegerField(help_text='part of file start', verbose_name='start bytes')),
-                ('end_bytes', models.BigIntegerField(help_text='part of file end', verbose_name='end bytes')),
-                ('file', models.ForeignKey(on_delete=django.db.models.deletion.CASCADE, related_name='chunks', to='storage.file')),
-            ],
-            options={
-                'verbose_name': 'file chunk',
-                'verbose_name_plural': 'file chunks',
-                'unique_together': {('file', 'chunk_id')},
-            },
-        ),
     ]

lockbox/storage/models.py

@@ -1,47 +1,33 @@
 from datetime import timedelta
 from pathlib import Path
-
-from common.constants import UPLOAD_STATUS_TYPES
+from common.constants import UPLOAD_STATUS_TYPES, UPLOAD_ERROR_CODES
 from common.models import LockboxBase
 from common.utils import get_config, get_max_size_chunk_bytes
-from django.conf import settings
 from django.core.files.uploadedfile import UploadedFile
 from django.db import models, transaction
 from django.utils import timezone
 from django.utils.translation import gettext_lazy as _
+from django.conf import settings
+from hashlib import md5
 
 
-def get_upload_path_chunk(instance, filename):
-    # TODO: How do we reconcile storage?
-    # TODO: Do we autodetect existing files task?
-    # TODO: Figure out absolute storage :(, custom storage and custom filefield? why is this not a def behaviour?
+class UploadError(Exception):
+    
+    def __init__(self, *args, **kwargs):
+        self.code = kwargs.pop("code")
+        super().__init__(*args, **kwargs)
 
-    filename = f"{instance.chunk_id}.chunk"
-    chunk_dir = settings.MEDIA_ROOT / str(instance.file.lid)
-
-    if not Path.exists(chunk_dir):
-        Path.mkdir(chunk_dir)
-
-    target_path = Path(chunk_dir) / Path(filename)
-    print(target_path)
-    return target_path
+def upload_to_fielpath(instance, filename):
+    return Path(str(instance.lid)).joinpath(f"{filename}{settings.INCOMPLETE_EXT}")
 
 
 class File(LockboxBase):
-    filename = models.CharField(
-        max_length=255,
-        null=False,
-        blank=False,
-        verbose_name = _("name"),
-        help_text=_("Name of the file"),
-    )
-
-    extension = models.CharField(
+    mime_type = models.CharField(
         max_length=128,
         blank=True,
         null=True,
-        verbose_name=_("extension"),
-        help_text=_("reported filesystem extension (not mime type)"),
+        verbose_name=_("mime-type"),
+        help_text=_("reported mime-type"),
     )
 
     file = models.FileField(
@@ -49,14 +35,14 @@ class File(LockboxBase):
         blank=True,
         verbose_name=_("file"),
         help_text=_("actual file"),
+        upload_to=upload_to_fielpath
     )
 
-    # TODO: Make this an FSM
     UPLOAD_CHOICES = (
         (UPLOAD_STATUS_TYPES.UPLOADING, _(UPLOAD_STATUS_TYPES.UPLOADING)),
         (UPLOAD_STATUS_TYPES.COMPLETED, _(UPLOAD_STATUS_TYPES.COMPLETED)),
-        (UPLOAD_STATUS_TYPES.PROCESSING, _(UPLOAD_STATUS_TYPES.PROCESSING)),
         (UPLOAD_STATUS_TYPES.ABANDONED, _(UPLOAD_STATUS_TYPES.ABANDONED)),
+        (UPLOAD_STATUS_TYPES.ERROR, _(UPLOAD_STATUS_TYPES.ERROR)),
     )
 
     status = models.CharField(
@@ -69,7 +55,7 @@ class File(LockboxBase):
         help_text=_("upload status for file"),
     )
 
-    date_completed = models.DateTimeField(
+    datetime_completed = models.DateTimeField(
         null=True,
         blank=True,
         verbose_name=_("completed on"),
@@ -102,13 +88,20 @@ class File(LockboxBase):
         help_text=_("will be deleted if expired and expires is true"),
     )
 
-    size_on_disk = models.PositiveBigIntegerField(
+    size = models.PositiveBigIntegerField(
         null=True,
         blank=True,
-        verbose_name=_("size on disk (bytes)"),
+        verbose_name=_("size (bytes)"),
         help_text=_("total size on disk for this file"),
     )
 
+    expected_size = models.PositiveBigIntegerField(
+        null=True,
+        blank=True,
+        verbose_name=_("expected size (bytes)"),
+        help_text=_("expected file size"),
+    )
+
     max_size_chunk_bytes = models.PositiveBigIntegerField(
         null=False,
         blank=False,
@@ -117,157 +110,122 @@ class File(LockboxBase):
         help_text=_("max size of each individual chunk for this file"),
     )
 
+    last_end_bytes = models.BigIntegerField(
+        null=True,
+        blank=True,
+        verbose_name=("last end bytes"),
+        help_text=_("last uploaded bytes position"),
+    )
+
     readonly_fields = [
-        "extension",
+        "mime_type",
         "status",
-        "date_completed",
-        "size_on_disk",
+        "datetime_completed",
+        "size",
         "file",
         "max_size_chunk_bytes",
+        "last_end_bytes",
         *LockboxBase.readonly_fields,
     ]
 
     def __str__(self):
-        return self.filename
+        name = "NO NAME"
+
+        if self.file:
+            name = self.file.name
+        return f"{name} ({self.lid})"
+
 
     class Meta:
         verbose_name = _("file")
         verbose_name_plural = _("files")
 
     @property
-    def checksum(self):
-        return 0
-
-    @property
-    def date_expires(self):
-        return self.date_created + timedelta(minutes=get_config("EXPIRATION_DELTA_MINUTES"))
+    def md5(self):
+        if self.exists:
+            return md5(open(self.file, "rb").read()).hexdigest()
+        return None
 
     @property
     def abandoned(self):
-        return self.date_created + timedelta(minutes=get_config("ABANDONED_DELTA_MINUTES"))
+        return self.date_created <= timezone.now() + timedelta(minutes=get_config("ABANDONED_DELTA_MINUTES"))
+
+    @classmethod
+    def abandoned_condition():
+        return models.Q(date_created__lte=timezone.now() + timedelta(minutes=get_config("ABANDONED_DELTA_MINUTES")))
 
     @property
     def expired(self):
         return self.date_expires <= timezone.now()
-    
-    @property
-    def last_chunk_id(self):
-        last_chunk_id = self.chunks.order_by("-chunk_id").values("chunk_id").first()
-        if last_chunk_id:
-            return last_chunk_id.get("chunk_id")
-        return - 1
-    
-    def create_chunk(self, chunk_file, chunk_data):
-        chunk = FileChunk(
-            file=self,
-            chunk=chunk_file,
-            chunk_id=self.last_chunk_id,
-            **chunk_data
-        )
 
-        chunk.save()
-        return chunk
-    
+    @classmethod
+    def expired_conditon():
+        return models.Q(date_expires__lte=timezone.now())
+
+    @property
+    def exists(self):
+        if not self.file:
+            return False
+        return Path(self.file.path).is_file()
+
+    def append_chunk(self, chunk_file, chunk_data):
+        # Override in case recently abandoned
+        # Will persist if it does not error out.
+        self.status = UPLOAD_STATUS_TYPES.UPLOADING
+
+        # Do not rely on DB file state, check for actual file.
+        if not self.exists:
+            # Oh oh, we are uploading a n + 1 chunk but theres no file
+            if chunk_data["start_bytes"] != 0:
+                self.status = UPLOAD_STATUS_TYPES.ERROR
+                self.save()
+                raise UploadError("File for uploaded chunk no longer exists", code=UPLOAD_ERROR_CODES.FILE_MISSING)
+
+        if self.last_end_bytes and self.last_end_bytes + 1 != chunk_data["start_bytes"]:
+            # Client screwed up, this is not where we left
+            raise UploadError("Mismatch in expected chunk", code=UPLOAD_ERROR_CODES.CHUNK_MISMATCH)
+
+        self.last_end_bytes = chunk_data["end_bytes"]
+        if self.expected_size == self.last_end_bytes:
+            # File is one shot chunk.
+            if chunk_data["start_bytes"] == 0:
+                self.file = chunk_file
+            self.save()
+            self.finalize()
+            return
+
+        # This is an n + 1 chunk.
+        print("Appending bytes yo")
+        chunk_file.seek(0)
+        self.save()
+
+    def finalize(self):
+        self.refresh_from_db()
+        self.status = UPLOAD_STATUS_TYPES.COMPLETED
+        self.datetime_completed = timezone.now()
+        final_name = self.file.name.replace(settings.INCOMPLETE_EXT, "")
+        final_path = settings.MEDIA_ROOT / final_name
+        with transaction.atomic():
+            Path(self.file.path).rename(final_path)
+            self.file.name = final_name
+            self.save()
 
     def save(self, *args, **kwargs):
-        self.max_size_chunk_bytes = get_max_size_chunk_bytes()
+        if not self.max_size_chunk_bytes:
+            self.max_size_chunk_bytes = get_max_size_chunk_bytes()
         return super().save(*args, **kwargs)
         
 
     def delete(self, *args, **kwargs):
 
-        if self.file:
-            storage, path = self.file.storage, self.file.path
-
-        if self.file:
-            # TODO: Figure out if file exists and try to delete it if error, report error.
-            storage.delete(path)
-
         with transaction.atomic():
-            self.chunks.all().delete()
+            if self.file:
+                if Path(self.file.path).is_file():
+                    self.file.storage.delete(self.file.path)
+                    self.file.storage.delete(Path(self.file.path).parent)
             result = super().delete(*args, **kwargs)
         return result
 
-    def handler_bytes(self):
-        # TODO: This is a naive approach, we almost never want to do this.
-        self.file.close()
-        self.file.open(mode="rb")
-        return UploadedFile(file=self.file, name=self.filename, size=self.offset)
-
-class FileChunk(LockboxBase):
-    file = models.ForeignKey(
-        "storage.File",
-        null=False,
-        blank=False,
-        on_delete=models.CASCADE,
-        related_name="chunks",
-    )
-
-    chunk = models.FileField(
-        upload_to=get_upload_path_chunk,
-        null=False,
-        blank=False,
-        verbose_name=_("chunk file"),
-        help_text=_("chunk file"),
-    )
-
-    chunk_id = models.BigIntegerField(
-        null=False,
-        blank=False,
-        verbose_name=_("chunk id"),
-        help_text=_("chunk id"),
-    )
-
-    size = models.BigIntegerField(
-        null=False,
-        blank=False,
-        verbose_name=("size"),
-        help_text=_("chunk size"),
-    )
-
-    start_bytes = models.BigIntegerField(
-        null=False,
-        blank=False,
-        verbose_name=("start bytes"),
-        help_text=_("part of file start"),
-    )
-
-    end_bytes = models.BigIntegerField(
-        null=False,
-        blank=False,
-        verbose_name=("end bytes"),
-        help_text=_("part of file end"),
-    )
-
-    readonly_fields = [
-        "file",
-        "chunk_id",
-        "start",
-        "end",
-        "size",
-        *LockboxBase.readonly_fields,
-    ]
-
-    def __str__(self):
-        return f"{self.file.filename}.{self.chunk_id}.chunk"
-
-    class Meta:
-        verbose_name = _("file chunk")
-        verbose_name_plural = _("file chunks")
-        unique_together = ("file", "chunk_id")
-
-    def save(self, *args, **kwargs):
-        self.chunk_id = self.file.last_chunk_id + 1
-        return super().save(*args, **kwargs)
-    
-    def delete(self, *args, **kwargs):
-        if self.file:
-            storage, path = self.file.storage, self.file.path
-
-        if self.file:
-            # TODO: Figure out if file exists and try to delete it if error, report error.
-            storage.delete(path)
-        return super().delete(*args, **kwargs)
 
 # class FileShare(LockboxBase):
 #     file = models.ForeignKey(

lockbox/storage/serializers.py

@@ -1,6 +1,6 @@
 from rest_framework import serializers
 
-from storage.models import File, FileChunk
+from storage.models import File
 
 
 class FileSerializer(serializers.ModelSerializer):
@@ -9,21 +9,3 @@ class FileSerializer(serializers.ModelSerializer):
         model = File
         fields = "__all__"
         read_only_fields = File.readonly_fields
-
-
-class FileChunkSerializer(serializers.ModelSerializer):
-
-    class Meta:
-        model = FileChunk
-        fields = "__all__"
-        read_only_fields = FileChunk.readonly_fields
-
-    def validate(self, data):
-        data = super().validate(data)
-        file = File.objects.get(lid=data["file"])
-
-        if data["size"] > file.max_size_chunk_bytes:
-            detail = f"'size' param is larger than max chunk size for file:\
-                {data["size"]} > {file.max_size_chunk_bytes}"
-            raise serializers.ValidationError(detail)
-        return data

lockbox/storage/urls.py

@@ -7,11 +7,7 @@ from storage import views_api, views_client
 router = SimpleRouter()
 router.register(r'files', views_api.FileModelViewSet)
 
-chunk_router = NestedSimpleRouter(router, r'files', lookup="file")
-chunk_router.register(r'chunks', views_api.FileChunkViewSet, basename="file-chunks")
-
 urlpatterns = [
     path("api/", include(router.urls)),
-    path("api/", include(chunk_router.urls)),
     path("upload/", views_client.FileUploadView.as_view(), name="client-fileupload"),
 ]

lockbox/storage/views_api.py

@@ -8,12 +8,14 @@ from rest_framework import status
 from rest_framework.decorators import action
 from rest_framework.response import Response
 from rest_framework.viewsets import ModelViewSet
-from rest_framework.exceptions import NotFound, ValidationError
+from rest_framework.exceptions import NotFound
+from rest_framework.exceptions import ValidationError as UserValidationError
 from rest_framework.parsers import FileUploadParser
 
-# from user.models import LockboxUser
-from storage.models import File, FileChunk
-from storage.serializers import FileChunkSerializer, FileSerializer
+from django.core.exceptions import ValidationError
+
+from storage.models import File, UploadError
+from storage.serializers import FileSerializer
 
 
 class FileModelViewSet(ModelViewSet):
@@ -21,48 +23,38 @@ class FileModelViewSet(ModelViewSet):
     queryset = File.objects.all()
     serializer_class = FileSerializer
 
-    @action(detail=True, methods=["GET"])
-    def last_chunk_position(self, request, pk=None):
-        file = self.get_object()
-        last_chunk_id = file.last_chunk_id
-        last_postion = 0
-        if  last_chunk_id != -1:
-            last_chunk = self.chunks.order_by("-chunk_id").values("end_bytes").first()
-            if last_chunk:
-                last_postion = last_chunk_id.get("end_bytes")
-        return Response({"last_chunk_position": last_postion}, status=status.HTTP_200_OK)
+    @action(detail=True, methods=["PUT"])
+    def append_chunk(self, request, filename="DUMMY", format=None, pk=None):
+        try:
+            file = File.objects.filter(lid=pk).first()
+        except ValidationError:
+            raise UserValidationError(f"UUID {pk} is not a valid UUID")
 
-
-class FileChunkViewSet(ModelViewSet):
-    model = FileChunk
-    queryset = FileChunk.objects.all()
-    serializer_class = FileChunkSerializer
-    parser_classes = (FileUploadParser,)
-
-    def create(self, request, filename="DUMMY", format=None, file_pk=None):
-        file = File.objects.filter(lid=str(file_pk)).first()
         if not file:
-            raise NotFound(f"File with ID {file_pk} was not found")
+            raise NotFound(f"File with ID {pk} was not found")
 
         chunk_data = self.get_content_range(request)
         if not chunk_data:
-            raise ValidationError(
+            raise UserValidationError(
                 f"Missing content range headers"
             )
 
-        chunk_file = request.FILES["file"]
+        chunk_file = request.FILES["Content"]
         if chunk_file.size > file.max_size_chunk_bytes:
-            raise ValidationError(
+            raise UserValidationError(
                 f"Chunk size is greater than files max chunk size: {chunk_file.size} > {file.max_size_chunk_bytes}")
         
         range_size = chunk_data["end_bytes"] - chunk_data["start_bytes"]
         if chunk_file.size != range_size:
-            raise ValidationError(
+            raise UserValidationError(
                 f"Actual chunk size mismatches content-range header: {chunk_file.size} != {range_size}"
             )
 
         chunk_data["size"] = chunk_file.size
-        file.create_chunk(chunk_file=chunk_file, chunk_data=chunk_data)
+        try:
+            file.append_chunk(chunk_file, chunk_data)
+        except UploadError as e:
+            return Response({"code": e.code}, status=status.HTTP_400_BAD_REQUEST)
         return Response(status=status.HTTP_201_CREATED)
 
     def get_content_range(self, request):
@@ -77,4 +69,4 @@ class FileChunkViewSet(ModelViewSet):
         return {
             "start_bytes": int(match.group('start')),
             "end_bytes": int(match.group('end')),
-        }
+        }