Working with manual sync

This commit is contained in:
Tris Forster 2026-05-24 13:51:23 +10:00
parent ef981e06e8
commit cec64ecd2f
4 changed files with 41 additions and 12 deletions

View File

@ -1,4 +1,5 @@
from .base import * # noqa
from os import environ
DEBUG = True
@ -8,3 +9,20 @@ SECRET_KEY = "DO NOT USE IN PRODUCTION"
INSTALLED_APPS.append("debug_toolbar") # noqa
MIDDLEWARE.insert(1, "debug_toolbar.middleware.DebugToolbarMiddleware") # noqa
INTERNAL_IPS = ["127.0.0.1"]
LOGGING = {
"version": 1,
"disable_existing_loggers": False,
"handlers": {
"console": {
"class": "logging.StreamHandler",
"level": environ.get("DEBUG_LEVEL", "WARNING"),
},
},
"loggers": {
"polyphonic": {
"handlers": ["console"],
"level": environ.get("DEBUG_LEVEL", "WARNING"),
}
},
}

View File

@ -1,4 +1,5 @@
from polyphonic.library.models import Collection, Work, WorkMeta, Document
from byostorage.models import UserStorage
import logging
@ -6,12 +7,11 @@ logger = logging.getLogger(__name__)
def sync_work(work: Work):
logger.info("Syncing '%s'", work.name)
folder_id = work.meta_info.get(name="folderid").value
storage = work.collection.storage.instance()
prefix = work.collection.storage.name
_, files = storage.listdir(folder_id)
logger.info("Syncing '%s' from %r", work.name, folder_id)
storage = UserStorage.objects.get(name="gdrive").instance()
existing = set(
[
@ -21,6 +21,9 @@ def sync_work(work: Work):
)
logger.debug("%d existing documents", len(existing))
_, files = storage.listdir(folder_id)
logger.debug("Remote files: %r", files)
for file in files:
if file.id in existing:
logger.debug("%30s: Skipping existing (%s)", file.name, file.id)
@ -32,7 +35,7 @@ def sync_work(work: Work):
continue
logger.info("%40s: Adding", file.name)
doc = work.docs.create(upload=f"{prefix}:{file}", doctype=Document.DOCTYPE_PDF)
doc = work.docs.create(upload=f"gdrive:{file}", doctype=Document.DOCTYPE_PDF)
doc.auto_tag()
for uri in existing:
@ -45,8 +48,10 @@ def sync_collection(collection: Collection, sync_existing: bool = False):
if not collection.storage.storage.endswith("GDriveLinkStorage"):
raise RuntimeError("Not a gdrive storage")
if not collection.prefix:
raise KeyError("Prefix must store folder id")
try:
folder_id = collection.settings["folder_id"]
except KeyError:
raise KeyError("Missing 'folder_id' in settings")
existing = dict(
WorkMeta.objects.filter(
@ -55,9 +60,12 @@ def sync_collection(collection: Collection, sync_existing: bool = False):
)
storage = collection.storage.instance()
folders, _ = storage.listdir(collection.prefix)
folders, _ = storage.listdir(folder_id)
for folder in folders:
if folder[0] == "_":
continue
if folder.id in existing:
if sync_existing:
logger.info("%40s: Syncing (%s)", folder.name, folder.id[:12])

View File

@ -8,7 +8,7 @@ import logging
logger = logging.getLogger(__name__)
SHARED_FOLDER = re.compile(r"https://drive.google.com/drive/folders/(\w+)")
SHARED_FOLDER = re.compile(r"https://drive.google.com/drive[u0-9\/]+folders/([\w\-]+)")
SHARED_FILE = re.compile(r"https://drive.google.com/file/d/([\w\-]+)")
FILES_API = "https://www.googleapis.com/drive/v3/files"
@ -50,12 +50,11 @@ class GDriveLinkStorage(Storage):
return data
def listdir(self, path) -> tuple[list[str], list[str]]:
# used to test for valid connection parameters - should do something to validate API key here
logger.debug("listdir: %s", path)
if path == "":
return [], []
logger.debug("LISTDIR: %s", path)
folder_id = self.parse_id(path)
url = f"{FILES_API}?q='{folder_id}'+in+parents&key={self.api_key}"
data = self.get_json(url)

View File

@ -6,6 +6,8 @@ from polyphonic.library.models import Work, Document
from polyphonic.library import forms
from byostorage.models import UserStorage
from . import sync_work
class WorkGDriveView(CollectionMixin, SingleObjectMixin, FormView):
model = Work
@ -37,11 +39,13 @@ class WorkGDriveView(CollectionMixin, SingleObjectMixin, FormView):
self.object.meta_info.update_or_create(
name="folderid", defaults={"value": folderid}
)
sync_work(self.object)
return redirect("work_detail", self.collection.pk, self.kwargs["pk"])
link = storage.import_link(link)
if link is None:
form.add_error("link", str(e))
form.add_error("link", "Not a valid link")
return self.form_invalid(form)
work = self.collection.works.get(pk=self.kwargs["pk"])