def set_checksums(apps, schema_editor):
document_model = apps.get_model("documents", "Document")
if not document_model.objects.all().exists():
return
print(
colourise(
"\n\n"
" This is a one-time only migration to generate checksums for all\n"
" of your existing documents. If you have a lot of documents\n"
" though, this may take a while, so a coffee break may be in\n"
" order."
"\n",
opts=("bold", )))
sums = {}
for d in document_model.objects.all():
document = Document(d)
print(" {} {} {}".format(
colourise("*", fg="green"),
colourise("Generating a checksum for", fg="white"),
colourise(document.file_name, fg="cyan")))
with document.source_file as encrypted:
checksum = hashlib.md5(GnuPG.decrypted(encrypted)).hexdigest()
if checksum in sums:
error = "\n{line}{p1}\n\n{doc1}\n{doc2}\n\n{p2}\n\n{code}\n\n{p3}{line}".format(
p1=colourise(
"It appears that you have two identical documents in your collection and \nPaperless no longer supports this (see issue #97). The documents in question\nare:",
fg="yellow"),
p2=colourise(
"To fix this problem, you'll have to remove one of them from the database, a task\nmost easily done by running the following command in the same\ndirectory as manage.py:",
fg="yellow"),
p3=colourise(
"When that's finished, re-run the migrate, and provided that there aren't any\nother duplicates, you should be good to go.",
fg="yellow"),
doc1=colourise(" * {} (id: {})".format(
sums[checksum][1], sums[checksum][0]),
fg="red"),
doc2=colourise(" * {} (id: {})".format(
document.file_name, document.pk),
fg="red"),
code=colourise(
" $ echo 'DELETE FROM documents_document WHERE id = {pk};' | ./manage.py dbshell"
.format(pk=document.pk),
fg="green"),
line=colourise("\n{}\n".format("=" * 80),
fg="white",
opts=("bold", )))
raise RuntimeError(error)
sums[checksum] = (document.pk, document.file_name)
document_model.objects.filter(pk=document.pk).update(checksum=checksum)
def move_documents_and_create_thumbnails(apps, schema_editor):
documents = os.listdir(os.path.join(settings.MEDIA_ROOT, "documents"))
if set(documents) == {"originals", "thumbnails"}:
return
print(colourise(
"\n\n"
" This is a one-time only migration to generate thumbnails for all of your\n"
" documents so that future UIs will have something to work with. If you have\n"
" a lot of documents though, this may take a while, so a coffee break may be\n"
" in order."
"\n", opts=("bold",)
))
for f in sorted(documents):
if not f.endswith("gpg"):
continue
print(" {} {} {}".format(
colourise("*", fg="green"),
colourise("Generating a thumbnail for", fg="white"),
colourise(f, fg="cyan")
))
thumb_temp = tempfile.mkdtemp(
prefix="paperless", dir=settings.SCRATCH_DIR)
orig_temp = tempfile.mkdtemp(
prefix="paperless", dir=settings.SCRATCH_DIR)
orig_source = os.path.join(settings.MEDIA_ROOT, "documents", f)
orig_target = os.path.join(orig_temp, f.replace(".gpg", ""))
with open(orig_source, "rb") as encrypted:
with open(orig_target, "wb") as unencrypted:
unencrypted.write(GnuPG.decrypted(encrypted))
subprocess.Popen((
settings.CONVERT_BINARY,
"-scale", "500x5000",
"-alpha", "remove",
orig_target,
os.path.join(thumb_temp, "convert-%04d.png")
)).wait()
thumb_source = os.path.join(thumb_temp, "convert-0000.png")
thumb_target = os.path.join(
settings.MEDIA_ROOT,
"documents",
"thumbnails",
re.sub(r"(\d+)\.\w+(\.gpg)", "\\1.png\\2", f)
)
with open(thumb_source, "rb") as unencrypted:
with open(thumb_target, "wb") as encrypted:
encrypted.write(GnuPG.encrypted(unencrypted))
shutil.rmtree(thumb_temp)
shutil.rmtree(orig_temp)
shutil.move(
os.path.join(settings.MEDIA_ROOT, "documents", f),
os.path.join(settings.MEDIA_ROOT, "documents", "originals", f),
)
def set_checksums(apps, schema_editor):
document_model = apps.get_model("documents", "Document")
if not document_model.objects.all().exists():
return
print(colourise(
"\n\n"
" This is a one-time only migration to generate checksums for all\n"
" of your existing documents. If you have a lot of documents\n"
" though, this may take a while, so a coffee break may be in\n"
" order."
"\n", opts=("bold",)
))
sums = {}
for d in document_model.objects.all():
document = Document(d)
print(" {} {} {}".format(
colourise("*", fg="green"),
colourise("Generating a checksum for", fg="white"),
colourise(document.file_name, fg="cyan")
))
with document.source_file as encrypted:
checksum = hashlib.md5(GnuPG.decrypted(encrypted)).hexdigest()
if checksum in sums:
error = "\n{line}{p1}\n\n{doc1}\n{doc2}\n\n{p2}\n\n{code}\n\n{p3}{line}".format(
p1=colourise("It appears that you have two identical documents in your collection and \nPaperless no longer supports this (see issue #97). The documents in question\nare:", fg="yellow"),
p2=colourise("To fix this problem, you'll have to remove one of them from the database, a task\nmost easily done by running the following command in the same\ndirectory as manage.py:", fg="yellow"),
p3=colourise("When that's finished, re-run the migrate, and provided that there aren't any\nother duplicates, you should be good to go.", fg="yellow"),
doc1=colourise(" * {} (id: {})".format(sums[checksum][1], sums[checksum][0]), fg="red"),
doc2=colourise(" * {} (id: {})".format(document.file_name, document.pk), fg="red"),
code=colourise(" $ echo 'DELETE FROM documents_document WHERE id = {pk};' | ./manage.py dbshell".format(pk=document.pk), fg="green"),
line=colourise("\n{}\n".format("=" * 80), fg="white", opts=("bold",))
)
raise RuntimeError(error)
sums[checksum] = (document.pk, document.file_name)
document_model.objects.filter(pk=document.pk).update(checksum=checksum)
