voussoir/etiquette
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity
etiquette/etiquette/photodb.py
Ethan Dalool e4f686c86f Move thumbnail blobs into the database. 
I've moved the thumbnails around many times over this project
and hopefully it doesn't happen too many more. Once the database has
tens of thousands of items, the thumbnails start to become the biggest
headache on the disk. Backing up, restoring, and sharding files per
directory are slower and more effortful with separate files. In the db
means the db is a larger file, but this is disk space that was already
getting used anyway. Now it's simpler and has atomic transactions.
2023-01-27 17:34:32 -08:00

1266 lines
46 KiB
Python
Raw Blame History

import bcrypt
import hashlib
import json
import os
import random
import tempfile
import types
import typing
from . import constants
from . import decorators
from . import exceptions
from . import helpers
from . import objects
from . import tag_export
from voussoirkit import cacheclass
from voussoirkit import configlayers
from voussoirkit import pathclass
from voussoirkit import progressbars
from voussoirkit import ratelimiter
from voussoirkit import spinal
from voussoirkit import sqlhelpers
from voussoirkit import stringtools
from voussoirkit import timetools
from voussoirkit import vlogging
from voussoirkit import worms
log = vlogging.getLogger(__name__)
RNG = random.SystemRandom()
####################################################################################################
class PDBAlbumMixin:
def __init__(self):
super().__init__()
def get_album(self, id) -> objects.Album:
return self.get_object_by_id(objects.Album, id)
def get_album_count(self) -> int:
return self.select_one_value('SELECT COUNT(*) FROM albums')
def get_albums(self) -> typing.Iterable[objects.Album]:
return self.get_objects(objects.Album)
def get_albums_by_id(self, ids) -> typing.Iterable[objects.Album]:
return self.get_objects_by_id(objects.Album, ids)
def get_albums_by_path(self, directory) -> typing.Iterable[objects.Album]:
'''
Yield Albums with the `associated_directory` of this value,
NOT case-sensitive.
'''
directory = pathclass.Path(directory)
query = 'SELECT albumid FROM album_associated_directories WHERE directory == ?'
bindings = [directory.absolute_path]
album_ids = self.select_column(query, bindings)
return self.get_albums_by_id(album_ids)
def get_albums_by_sql(self, query, bindings=None) -> typing.Iterable[objects.Album]:
return self.get_objects_by_sql(objects.Album, query, bindings)
def get_albums_within_directory(self, directory) -> typing.Iterable[objects.Album]:
# This function is something of a stopgap measure since `search` only
# searches for photos and then yields their containing albums. Thus it
# is not possible for search to find albums that contain no photos.
# I'd like to find a better solution than this separate method.
directory = pathclass.Path(directory)
directory.assert_is_directory()
pattern = directory.absolute_path.rstrip(os.sep)
pattern = f'{pattern}{os.sep}%'
album_ids = self.select_column(
'SELECT DISTINCT albumid FROM album_associated_directories WHERE directory LIKE ?',
[pattern]
)
albums = self.get_albums_by_id(album_ids)
return albums
def get_root_albums(self) -> typing.Iterable[objects.Album]:
'''
Yield Albums that have no parent.
'''
return self.get_root_objects(objects.Album)
@decorators.required_feature('album.new')
@worms.atomic
def new_album(
self,
title=None,
description=None,
*,
associated_directories=None,
author=None,
photos=None,
) -> objects.Album:
'''
Create a new album. Photos can be added now or later.
'''
# These might raise exceptions.
title = objects.Album.normalize_title(title)
description = objects.Album.normalize_description(description)
author_id = self.get_user_id_or_none(author)
# Ok.
album_id = self.generate_id(objects.Album)
log.info('New Album: %s %s.', album_id, title)
data = {
'id': album_id,
'title': title,
'description': description,
'created': timetools.now().timestamp(),
'thumbnail_photo': None,
'author_id': author_id,
}
self.insert(table=objects.Album, pairs=data)
album = self.get_cached_instance(objects.Album, data)
associated_directories = associated_directories or ()
if isinstance(associated_directories, str):
associated_directories = [associated_directories]
album.add_associated_directories(associated_directories)
if photos is not None:
photos = [self.get_photo(photo) for photo in photos]
album.add_photos(photos)
return album
@worms.atomic
def purge_deleted_associated_directories(self, albums=None) -> typing.Iterable[pathclass.Path]:
query = 'SELECT DISTINCT directory FROM album_associated_directories'
directories = self.select_column(query)
directories = (pathclass.Path(d) for d in directories)
directories = [d for d in directories if not d.is_dir]
if not directories:
return
log.info('Purging associated directories %s.', directories)
d_query = sqlhelpers.listify(d.absolute_path for d in directories)
query = f'DELETE FROM album_associated_directories WHERE directory in {d_query}'
if albums is not None:
album_ids = sqlhelpers.listify(a.id for a in albums)
query += f' AND albumid IN {album_ids}'
self.execute(query)
yield from directories
@worms.atomic
def purge_empty_albums(self, albums=None) -> typing.Iterable[objects.Album]:
if albums is None:
to_check = set(self.get_albums())
else:
to_check = set()
for album in albums:
to_check.update(album.walk_children())
while to_check:
album = to_check.pop()
if album.get_children() or album.get_photos():
continue
# This may have been the last child of an otherwise empty parent.
to_check.update(album.get_parents())
album.delete()
yield album
####################################################################################################
class PDBBookmarkMixin:
def __init__(self):
super().__init__()
def get_bookmark(self, id) -> objects.Bookmark:
return self.get_object_by_id(objects.Bookmark, id)
def get_bookmark_count(self) -> int:
return self.select_one_value('SELECT COUNT(*) FROM bookmarks')
def get_bookmarks(self) -> typing.Iterable[objects.Bookmark]:
return self.get_objects(objects.Bookmark)
def get_bookmarks_by_id(self, ids) -> typing.Iterable[objects.Bookmark]:
return self.get_objects_by_id(objects.Bookmark, ids)
def get_bookmarks_by_sql(self, query, bindings=None) -> typing.Iterable[objects.Bookmark]:
return self.get_objects_by_sql(objects.Bookmark, query, bindings)
@decorators.required_feature('bookmark.new')
@worms.atomic
def new_bookmark(self, url, title=None, *, author=None) -> objects.Bookmark:
# These might raise exceptions.
title = objects.Bookmark.normalize_title(title)
url = objects.Bookmark.normalize_url(url)
author_id = self.get_user_id_or_none(author)
# Ok.
bookmark_id = self.generate_id(objects.Bookmark)
log.info('New Bookmark: %s %s %s.', bookmark_id, title, url)
data = {
'id': bookmark_id,
'title': title,
'url': url,
'created': timetools.now().timestamp(),
'author_id': author_id,
}
self.insert(table=objects.Bookmark, pairs=data)
bookmark = self.get_cached_instance(objects.Bookmark, data)
return bookmark
####################################################################################################
class PDBGroupableMixin:
def __init__(self):
super().__init__()
def get_root_objects(self, object_class):
'''
For Groupable types, yield objects which have no parent.
'''
object_table = object_class.table
group_table = object_class.group_table
query = f'''
SELECT * FROM {object_table}
WHERE NOT EXISTS (
SELECT 1 FROM {group_table}
WHERE memberid == {object_table}.id
)
'''
rows = self.select(query)
for row in rows:
instance = self.get_cached_instance(object_class, row)
yield instance
####################################################################################################
class PDBPhotoMixin:
def __init__(self):
super().__init__()
def assert_no_such_photo_by_path(self, filepath) -> None:
try:
existing = self.get_photo_by_path(filepath)
except exceptions.NoSuchPhoto:
return
else:
raise exceptions.PhotoExists(existing)
def get_photo(self, id) -> objects.Photo:
return self.get_object_by_id(objects.Photo, id)
def get_photo_by_path(self, filepath) -> objects.Photo:
filepath = pathclass.Path(filepath)
query = 'SELECT * FROM photos WHERE filepath == ?'
bindings = [filepath.absolute_path]
photo_row = self.select_one(query, bindings)
if photo_row is None:
raise exceptions.NoSuchPhoto(filepath)
photo = self.get_cached_instance(objects.Photo, photo_row)
return photo
def get_photo_count(self) -> int:
return self.select_one_value('SELECT COUNT(*) FROM photos')
def get_photos(self) -> typing.Iterable[objects.Photo]:
return self.get_objects(objects.Photo)
def get_photos_by_id(self, ids) -> typing.Iterable[objects.Photo]:
return self.get_objects_by_id(objects.Photo, ids)
def get_photos_by_recent(self, count=None) -> typing.Iterable[objects.Photo]:
'''
Yield photo objects in order of creation time.
'''
if count is not None and count <= 0:
return
query = 'SELECT * FROM photos ORDER BY created DESC'
photo_rows = self.select(query)
for photo_row in photo_rows:
photo = self.get_cached_instance(objects.Photo, photo_row)
yield photo
if count is None:
continue
count -= 1
if count <= 0:
break
def get_photos_by_hash(self, sha256) -> typing.Iterable[objects.Photo]:
if not isinstance(sha256, str) or len(sha256) != 64:
raise TypeError(f'sha256 shoulbe the 64-character hexdigest string.')
query = 'SELECT * FROM photos WHERE sha256 == ?'
bindings = [sha256]
yield from self.get_photos_by_sql(query, bindings)
def get_photos_by_sql(self, query, bindings=None) -> typing.Iterable[objects.Photo]:
return self.get_objects_by_sql(objects.Photo, query, bindings)
@decorators.required_feature('photo.new')
@worms.atomic
def new_photo(
self,
filepath,
*,
author=None,
do_metadata=True,
do_thumbnail=True,
hash_kwargs=None,
known_hash=None,
searchhidden=False,
tags=None,
trusted_file=False,
) -> objects.Photo:
'''
Given a filepath, determine its attributes and create a new Photo object
in the database. Tags may be applied now or later.
hash_kwargs:
Additional kwargs passed into spinal.hash_file. Notably, you may
wish to set bytes_per_second to keep system load low.
known_hash:
If the sha256 of the file is already known, you may provide it here
so it does not need to be recalculated. This is primarily intended
for digest_directory since it will look for hash matches first
before creating new photos and thus can provide the known hash.
Returns the Photo object.
'''
# These might raise exceptions
filepath = pathclass.Path(filepath)
if not filepath.is_file:
raise FileNotFoundError(filepath.absolute_path)
self.assert_no_such_photo_by_path(filepath=filepath)
author_id = self.get_user_id_or_none(author)
if known_hash is None:
pass
elif not isinstance(known_hash, str) or len(known_hash) != 64:
raise TypeError(f'known_hash should be the 64-character sha256 hexdigest string.')
# Ok.
photo_id = self.generate_id(objects.Photo)
log.info('New Photo: %s %s.', photo_id, filepath.absolute_path)
data = {
'id': photo_id,
'filepath': filepath.absolute_path,
'override_filename': None,
'created': timetools.now().timestamp(),
'tagged_at': None,
'author_id': author_id,
'searchhidden': searchhidden,
# These will be filled in during the metadata stage.
'mtime': None,
'sha256': known_hash,
'bytes': None,
'width': None,
'height': None,
'duration': None,
}
self.insert(table=objects.Photo, pairs=data)
photo = self.get_photo(photo_id)
if do_metadata:
hash_kwargs = hash_kwargs or {}
photo.reload_metadata(hash_kwargs=hash_kwargs, trusted_file=trusted_file)
if do_thumbnail:
photo.generate_thumbnail(trusted_file=trusted_file)
tags = tags or []
tags = [self.get_tag(name=tag) for tag in tags]
for tag in tags:
photo.add_tag(tag)
return photo
@worms.atomic
def purge_deleted_files(self, photos=None) -> typing.Iterable[objects.Photo]:
'''
Delete Photos whose corresponding file on disk is missing.
photos:
An iterable of Photo objects to check.
If not provided, all photos are checked.
If you only want to delete photos that have not been tagged, consider
P.purge_deleted_files(P.search(has_tags=False, is_searchhidden=None)).
'''
if photos is None:
photos = self.get_photos_by_recent()
for photo in photos:
if photo.real_path.exists:
continue
photo.delete()
yield photo
def search(self, **kwargs):
return objects.Search(photodb=self, kwargs=kwargs)
####################################################################################################
class PDBTagMixin:
def __init__(self):
super().__init__()
def assert_no_such_tag(self, name) -> None:
try:
existing_tag = self.get_tag_by_name(name)
except exceptions.NoSuchTag:
return
else:
raise exceptions.TagExists(existing_tag)
def _get_all_tag_names(self):
query = 'SELECT name FROM tags'
names = set(self.select_column(query))
return names
def get_all_tag_names(self) -> set[str]:
'''
Return a set containing the names of all tags as strings.
Useful for when you don't want the overhead of actual Tag objects.
'''
return self.get_cached_tag_export(self._get_all_tag_names)
def _get_all_synonyms(self):
query = 'SELECT name, mastername FROM tag_synonyms'
syn_rows = self.select(query)
synonyms = {syn: tag for (syn, tag) in syn_rows}
return synonyms
def get_all_synonyms(self) -> dict:
'''
Return a dict mapping {synonym: mastertag} as strings.
'''
return self.get_cached_tag_export(self._get_all_synonyms)
def get_cached_tag_export(self, function, **kwargs):
if isinstance(function, str):
function = getattr(tag_export, function)
if 'tags' in kwargs:
kwargs['tags'] = tuple(kwargs['tags'])
key = (function.__name__,) + helpers.dict_to_tuple(kwargs)
try:
exp = self.caches['tag_exports'][key]
return exp
except KeyError:
exp = function(**kwargs)
if isinstance(exp, types.GeneratorType):
exp = tuple(exp)
self.caches['tag_exports'][key] = exp
return exp
def get_root_tags(self) -> typing.Iterable[objects.Tag]:
'''
Yield Tags that have no parent.
'''
return self.get_root_objects(objects.Tag)
def get_tag(self, name=None, id=None) -> objects.Tag:
'''
Redirect to get_tag_by_id or get_tag_by_name.
'''
if not helpers.is_xor(id, name):
raise exceptions.NotExclusive(['id', 'name'])
if id is not None:
return self.get_tag_by_id(id)
else:
return self.get_tag_by_name(name)
def get_tag_by_id(self, id) -> objects.Tag:
return self.get_object_by_id(objects.Tag, id)
def get_tag_by_name(self, tagname) -> objects.Tag:
if isinstance(tagname, objects.Tag):
if tagname.photodb == self:
return tagname
tagname = tagname.name
try:
# TODO: this logic is flawed because tags that were created in
# the past may have had different normalization.
# At the same time, I don't want to just pass the input directly
# into the query, we should still do SOME assumed normalization
# like whitespace strip.
tagname = self.normalize_tagname(tagname)
except (exceptions.TagTooShort, exceptions.TagTooLong):
raise exceptions.NoSuchTag(tagname)
while True:
# Return if it's a toplevel...
tag_row = self.select_one('SELECT * FROM tags WHERE name == ?', [tagname])
if tag_row is not None:
break
# ...or resolve the synonym and try again.
query = 'SELECT mastername FROM tag_synonyms WHERE name == ?'
bindings = [tagname]
mastername = self.select_one_value(query, bindings)
if mastername is None:
# was not a master tag or synonym
raise exceptions.NoSuchTag(tagname)
tagname = mastername
tag = self.get_cached_instance(objects.Tag, tag_row)
return tag
def get_tag_count(self) -> int:
return self.select_one_value('SELECT COUNT(*) FROM tags')
def get_tags(self) -> typing.Iterable[objects.Tag]:
'''
Yield all Tags in the database.
'''
return self.get_objects(objects.Tag)
def get_tags_by_id(self, ids) -> typing.Iterable[objects.Tag]:
return self.get_objects_by_id(objects.Tag, ids)
def get_tags_by_sql(self, query, bindings=None) -> typing.Iterable[objects.Tag]:
return self.get_objects_by_sql(objects.Tag, query, bindings)
@decorators.required_feature('tag.new')
@worms.atomic
def new_tag(self, tagname, description=None, *, author=None) -> objects.Tag:
'''
Register a new tag and return the Tag object.
'''
# These might raise exceptions.
tagname = self.normalize_tagname(tagname)
self.assert_no_such_tag(name=tagname)
description = objects.Tag.normalize_description(description)
author_id = self.get_user_id_or_none(author)
# Ok.
tag_id = self.generate_id(objects.Tag)
log.info('New Tag: %s %s.', tag_id, tagname)
self.caches['tag_exports'].clear()
data = {
'id': tag_id,
'name': tagname,
'description': description,
'created': timetools.now().timestamp(),
'author_id': author_id,
}
self.insert(table=objects.Tag, pairs=data)
tag = self.get_cached_instance(objects.Tag, data)
return tag
def normalize_tagname(self, tagname) -> str:
tagname = objects.Tag.normalize_name(
tagname,
# valid_chars=self.config['tag']['valid_chars'],
min_length=self.config['tag']['min_length'],
max_length=self.config['tag']['max_length'],
)
return tagname
####################################################################################################
class PDBUserMixin:
def __init__(self):
super().__init__()
def assert_no_such_user(self, username) -> None:
try:
existing_user = self.get_user(username=username)
except exceptions.NoSuchUser:
return
else:
raise exceptions.UserExists(existing_user)
def assert_valid_password(self, password) -> None:
if not isinstance(password, bytes):
raise TypeError(f'Password must be {bytes}, not {type(password)}.')
if len(password) < self.config['user']['min_password_length']:
raise exceptions.PasswordTooShort(min_length=self.config['user']['min_password_length'])
def assert_valid_username(self, username) -> None:
if not isinstance(username, str):
raise TypeError(f'Username must be {str}, not {type(username)}.')
if len(username) < self.config['user']['min_username_length']:
raise exceptions.UsernameTooShort(
username=username,
min_length=self.config['user']['min_username_length']
)
if len(username) > self.config['user']['max_username_length']:
raise exceptions.UsernameTooLong(
username=username,
max_length=self.config['user']['max_username_length']
)
badchars = [c for c in username if c not in self.config['user']['valid_chars']]
if badchars:
raise exceptions.InvalidUsernameChars(username=username, badchars=badchars)
def get_user(self, username=None, id=None) -> objects.User:
'''
Redirect to get_user_by_id or get_user_by_username.
'''
if not helpers.is_xor(id, username):
raise exceptions.NotExclusive(['id', 'username'])
if id:
return self.get_user_by_id(id)
else:
return self.get_user_by_username(username)
def get_user_by_id(self, id) -> objects.User:
return self.get_object_by_id(objects.User, id)
def get_user_by_username(self, username) -> objects.User:
user_row = self.select_one('SELECT * FROM users WHERE username == ?', [username])
if user_row is None:
raise exceptions.NoSuchUser(username)
return self.get_cached_instance(objects.User, user_row)
def get_user_count(self) -> int:
return self.select_one_value('SELECT COUNT(*) FROM users')
def get_user_id_or_none(self, user_obj_or_id) -> typing.Optional[str]:
'''
For methods that create photos, albums, etc., we sometimes associate
them with an author but sometimes not. The callers of those methods
might be trying to submit a User object, or a user's ID, or maybe they
left it None.
This method converts those inputs into a User's ID if possible, or else
returns None, hiding validation that those methods would otherwise have
to duplicate.
Exceptions like NoSuchUser can still be raised if the input appears to
be workable but fails.
'''
if user_obj_or_id is None:
return None
elif isinstance(user_obj_or_id, objects.User):
if user_obj_or_id.photodb != self:
raise ValueError('That user does not belong to this photodb.')
author_id = user_obj_or_id.id
elif isinstance(user_obj_or_id, str):
# Confirm that this string is a valid ID and not junk.
author_id = self.get_user(id=user_obj_or_id).id
else:
raise TypeError(f'Unworkable type {type(user_obj_or_id)}.')
return author_id
def get_users(self) -> typing.Iterable[objects.User]:
return self.get_objects(objects.User)
def get_users_by_id(self, ids) -> typing.Iterable[objects.User]:
return self.get_objects_by_id(objects.User, ids)
def get_users_by_sql(self, query, bindings=None) -> typing.Iterable[objects.User]:
return self.get_objects_by_sql(objects.User, query, bindings)
@decorators.required_feature('user.new')
@worms.atomic
def new_user(self, username, password, *, display_name=None) -> objects.User:
# These might raise exceptions.
self.assert_valid_username(username)
self.assert_no_such_user(username=username)
if not isinstance(password, bytes):
password = password.encode('utf-8')
self.assert_valid_password(password)
display_name = objects.User.normalize_display_name(
display_name,
max_length=self.config['user']['max_display_name_length'],
)
# Ok.
user_id = self.generate_id(objects.User)
log.info('New User: %s %s.', user_id, username)
hashed_password = bcrypt.hashpw(password, bcrypt.gensalt())
data = {
'id': user_id,
'username': username,
'password': hashed_password,
'display_name': display_name,
'created': timetools.now().timestamp(),
}
self.insert(table=objects.User, pairs=data)
return self.get_cached_instance(objects.User, data)
####################################################################################################
class PDBUtilMixin:
def __init__(self):
super().__init__()
@worms.atomic
def digest_directory(
self,
directory,
*,
exclude_directories=None,
exclude_filenames=None,
glob_directories=None,
glob_filenames=None,
hash_kwargs=None,
make_albums=True,
natural_sort=True,
new_photo_kwargs=None,
new_photo_ratelimit=None,
recurse=True,
yield_albums=True,
yield_new_photos=True,
yield_old_photos=True,
):
'''
Walk the directory and create Photos for every file.
If a Photo object already exists for a file, it will be added to the
correct album.
exclude_directories:
A list of basenames or absolute paths of directories to ignore.
This list works in addition to, not instead of, the
digest_exclude_dirs config value.
exclude_filenames:
A list of basenames or absolute paths of filenames to ignore.
This list works in addition to, not instead of, the
digest_exclude_files config value.
hash_kwargs:
Additional kwargs passed into spinal.hash_file. Notably, you may
wish to set bytes_per_second to keep system load low.
make_albums:
If True, every directory that is digested will be turned into an
Album, and the directory path will be added to the Album's
associated_directories list. Child directories will become child
albums.
If there already exists an Album associated with the directory,
the newly digested photos will be added to that album.
Because album/directory relationships are not unique, there might
be multiple albums associated with a directory, in which case they
will all get the photos.
natural_sort:
If True, the list of files will be natural sorted before digest.
This way, the `created` timestamps on every Photo correspond to the
same order that the files are listed when natural sorted. This is
essentially an aesthetic preference, that when you are viewing the
photos sorted by timestamp they are also natural sorted.
See stringtools.natural_sorter.
new_photo_kwargs:
A dict of kwargs to pass into every call of new_photo.
new_photo_ratelimit:
A ratelimiter.Ratelimiter object, or an int/float number of seconds
to wait between every photo digest.
It is worth noting that timestamp resolution / accuracy varies by
system. If you digest photos very quickly, you might have many with
the exact same created timestamp. This doesn't cause any technical
problems, but it is another somewhat aesthetic choice. If you start
with with a reference photo and then query
`SELECT FROM photos WHERE created > reference`, you could miss
several photos with the exact same timestamp, unless you use >= and
then ignore the reference photo.
recurse:
If True, walk the whole directory tree. If False, only digest the
photos from the given directory and not its subdirectories.
yield_albums:
If True, yield Albums as they are processed, new or not.
yield_photos:
If True, yield Photos as they are processed, new or not.
'''
def _normalize_directory(directory):
directory = pathclass.Path(directory)
directory.assert_is_directory()
directory.correct_case()
return directory
def _normalize_exclude_directories(exclude_directories):
exclude_directories = exclude_directories or []
exclude_directories.extend(self.config['digest_exclude_dirs'])
return exclude_directories
def _normalize_exclude_filenames(exclude_filenames):
exclude_filenames = exclude_filenames or []
exclude_filenames.extend(self.config['digest_exclude_files'])
return exclude_filenames
def _normalize_new_photo_kwargs(new_photo_kwargs):
if new_photo_kwargs is None:
new_photo_kwargs = {}
else:
new_photo_kwargs = new_photo_kwargs.copy()
new_photo_kwargs.pop('commit', None)
new_photo_kwargs.pop('filepath', None)
new_photo_kwargs.setdefault('hash_kwargs', hash_kwargs)
return new_photo_kwargs
def _normalize_new_photo_ratelimit(new_photo_ratelimit):
if new_photo_ratelimit is None:
return new_photo_ratelimit
elif isinstance(new_photo_ratelimit, ratelimiter.Ratelimiter):
return new_photo_ratelimit
elif isinstance(new_photo_ratelimit, (int, float)):
new_photo_ratelimit = ratelimiter.Ratelimiter(allowance=1, period=new_photo_ratelimit)
return new_photo_ratelimit
raise TypeError(new_photo_ratelimit)
def check_renamed(filepath):
'''
We'll do our best to determine if this file is actually a rename of
a file that's already in the database.
'''
same_meta = self.get_photos_by_sql(
'SELECT * FROM photos WHERE mtime != 0 AND mtime == ? AND bytes == ?',
[filepath.stat.st_mtime, filepath.stat.st_size]
)
same_meta = [photo for photo in same_meta if not photo.real_path.is_file]
if len(same_meta) == 1:
photo = same_meta[0]
log.debug('Found mtime+bytesize match %s.', photo)
return photo
log.loud('Hashing file %s to check for rename.', filepath)
if filepath.size > 100 * (2 ** 20):
progressbar = progressbars.bar1_bytestring()
else:
progressbar = None
sha256 = spinal.hash_file(
filepath,
hash_class=hashlib.sha256, **hash_kwargs,
progressbar=progressbar,
).hexdigest()
same_hash = self.get_photos_by_hash(sha256)
same_hash = [photo for photo in same_hash if not photo.real_path.is_file]
# fwiw, I'm not checking byte size since it's a hash match.
if len(same_hash) > 1:
same_hash = [photo for photo in same_hash if photo.mtime == filepath.stat.st_mtime]
if len(same_hash) == 1:
return same_hash[0]
# Although we did not find a match, we can still benefit from our
# hash work by passing this as the known_hash to new_photo.
return {'sha256': sha256}
def create_or_fetch_photo(filepath):
'''
Given a filepath, find the corresponding Photo object if it exists,
otherwise create it and then return it.
'''
try:
return (self.get_photo_by_path(filepath), False)
except exceptions.NoSuchPhoto:
pass
result = check_renamed(filepath)
if isinstance(result, objects.Photo):
result.relocate(filepath.absolute_path)
return (result, False)
elif isinstance(result, dict) and 'sha256' in result:
sha256 = result['sha256']
else:
sha256 = None
photo = self.new_photo(filepath, known_hash=sha256, **new_photo_kwargs)
if new_photo_ratelimit is not None:
new_photo_ratelimit.limit()
return (photo, True)
def create_or_fetch_current_albums(albums_by_path, current_directory):
current_albums = albums_by_path.get(current_directory.absolute_path, None)
if current_albums is not None:
return current_albums
current_albums = list(self.get_albums_by_path(current_directory.absolute_path))
if not current_albums:
current_albums = [self.new_album(
associated_directories=current_directory.absolute_path,
title=current_directory.basename,
)]
albums_by_path[current_directory.absolute_path] = current_albums
return current_albums
def orphan_join_parent_albums(albums_by_path, current_albums, current_directory):
'''
If the current album is an orphan, let's check if there exists an
album for the parent directory. If so, add the current album to it.
'''
orphans = [album for album in current_albums if not album.has_any_parent()]
if not orphans:
return
parents = albums_by_path.get(current_directory.parent.absolute_path, None)
if not parents:
return
for parent in parents:
parent.add_children(orphans)
directory = _normalize_directory(directory)
exclude_directories = _normalize_exclude_directories(exclude_directories)
exclude_filenames = _normalize_exclude_filenames(exclude_filenames)
hash_kwargs = hash_kwargs or {}
new_photo_kwargs = _normalize_new_photo_kwargs(new_photo_kwargs)
new_photo_ratelimit = _normalize_new_photo_ratelimit(new_photo_ratelimit)
albums_by_path = {}
log.info('Digesting directory "%s".', directory.absolute_path)
walk_generator = spinal.walk(
directory,
exclude_directories=exclude_directories,
exclude_filenames=exclude_filenames,
glob_directories=glob_directories,
glob_filenames=glob_filenames,
recurse=recurse,
yield_style='nested',
)
for (current_directory, subdirectories, files) in walk_generator:
if natural_sort:
files = sorted(files, key=lambda f: stringtools.natural_sorter(f.basename))
photos = [create_or_fetch_photo(file) for file in files]
# Note, this means that empty folders will not get an Album.
# At this time this behavior is intentional. Furthermore, due to
# the glob/exclude rules, we don't want albums being created if
# they don't contain any files of interest, even if they do contain
# other files.
if not photos:
continue
for (photo, is_new) in photos:
if (is_new and yield_new_photos) or (not is_new and yield_old_photos):
yield photo
if not make_albums:
continue
current_albums = create_or_fetch_current_albums(albums_by_path, current_directory)
orphan_join_parent_albums(albums_by_path, current_albums, current_directory)
for album in current_albums:
album.add_photos(photo for (photo, is_new) in photos)
if yield_albums:
yield from current_albums
@worms.atomic
def easybake(self, ebstring, author=None):
'''
Easily create tags, groups, and synonyms with a string like
"group1.group2.tag+synonym"
"family.parents.dad+father"
etc
'''
output_notes = []
def create_or_get(name):
try:
item = self.get_tag(name=name)
note = ('existing_tag', item.name)
except exceptions.NoSuchTag:
item = self.new_tag(name, author=author)
note = ('new_tag', item.name)
output_notes.append(note)
return item
(tagname, synonym, rename_to) = helpers.split_easybake_string(ebstring)
if rename_to:
tag = self.get_tag(name=tagname)
old_name = tag.name
tag.rename(rename_to)
note = ('rename_tag', f'{old_name}={tag.name}')
output_notes.append(note)
else:
tag_parts = tagname.split('.')
tags = [create_or_get(t) for t in tag_parts]
for (higher, lower) in zip(tags, tags[1:]):
try:
higher.add_child(lower)
note = ('join_group', f'{higher.name}.{lower.name}')
output_notes.append(note)
except exceptions.GroupExists:
pass
tag = tags[-1]
if synonym:
synonym = tag.add_synonym(synonym)
note = ('add_synonym', f'{tag.name}+{synonym}')
output_notes.append(note)
return output_notes
####################################################################################################
class PhotoDB(
PDBAlbumMixin,
PDBBookmarkMixin,
PDBGroupableMixin,
PDBPhotoMixin,
PDBTagMixin,
PDBUserMixin,
PDBUtilMixin,
worms.DatabaseWithCaching,
):
def __init__(
self,
data_directory=None,
*,
create=False,
ephemeral=False,
skip_version_check=False,
):
'''
data_directory:
This directory will contain the sql file, config file,
generated thumbnails, etc. The directory is the database for all
intents and purposes.
create:
If True, the data_directory will be created if it does not exist.
If False, we expect that data_directory and the sql file exist.
ephemeral:
Use an in-memory sql database, and a temporary directory for
everything else, so that the whole PhotoDB disappears after closing.
Requires that data_directory=None.
skip_version_check:
Skip the version check so that you don't get DatabaseOutOfDate.
Beware of modifying any data in this state.
'''
super().__init__()
# Used by decorators.required_feature.
self._photodb = self
ephemeral = bool(ephemeral)
if data_directory is not None and ephemeral:
raise exceptions.NotExclusive(['data_directory', 'ephemeral'])
self.ephemeral = ephemeral
# DATA DIR PREP
if data_directory is not None:
pass
elif self.ephemeral:
# In addition to the data_dir as a pathclass object, keep the
# TempDir object so we can use the cleanup method later.
self.ephemeral_directory = tempfile.TemporaryDirectory(prefix='etiquette_ephem_')
data_directory = self.ephemeral_directory.name
else:
data_directory = pathclass.cwd().with_child(constants.DEFAULT_DATADIR)
if isinstance(data_directory, str):
data_directory = helpers.remove_path_badchars(data_directory, allowed=':/\\')
self.data_directory = pathclass.Path(data_directory)
if self.data_directory.exists and not self.data_directory.is_dir:
raise exceptions.BadDataDirectory(self.data_directory.absolute_path)
# DATABASE / WORMS
self._init_sql(create=create, skip_version_check=skip_version_check)
# THUMBNAIL DIRECTORY
self.thumbnail_directory = self.data_directory.with_child(constants.DEFAULT_THUMBDIR)
self.thumbnail_directory.makedirs(exist_ok=True)
# CONFIG
self.config_filepath = self.data_directory.with_child(constants.DEFAULT_CONFIGNAME)
self.load_config()
# WORMS
self._init_column_index()
self._init_caches()
def _check_version(self):
'''
Compare database's user_version against constants.DATABASE_VERSION,
raising exceptions.DatabaseOutOfDate if not correct.
'''
existing = self.pragma_read('user_version')
if existing != constants.DATABASE_VERSION:
raise exceptions.DatabaseOutOfDate(
existing=existing,
new=constants.DATABASE_VERSION,
filepath=self.data_directory,
)
def _first_time_setup(self):
log.info('Running first-time database setup.')
with self.transaction:
self._load_pragmas()
self.pragma_write('user_version', constants.DATABASE_VERSION)
self.executescript(constants.DB_INIT)
def _init_caches(self):
self.caches = {
objects.Album: cacheclass.Cache(maxlen=self.config['cache_size']['album']),
objects.Bookmark: cacheclass.Cache(maxlen=self.config['cache_size']['bookmark']),
objects.Photo: cacheclass.Cache(maxlen=self.config['cache_size']['photo']),
objects.Tag: cacheclass.Cache(maxlen=self.config['cache_size']['tag']),
objects.User: cacheclass.Cache(maxlen=self.config['cache_size']['user']),
'tag_exports': cacheclass.Cache(maxlen=100),
}
def _init_column_index(self):
self.COLUMNS = constants.SQL_COLUMNS
self.COLUMN_INDEX = constants.SQL_INDEX
def _init_sql(self, create, skip_version_check):
if self.ephemeral:
existing_database = False
self.sql_write = self._make_sqlite_write_connection(':memory:')
self.sql_read = self._make_sqlite_read_connection(':memory:')
self._first_time_setup()
return
self.database_filepath = self.data_directory.with_child(constants.DEFAULT_DBNAME)
existing_database = self.database_filepath.exists
if not existing_database and not create:
msg = f'"{self.database_filepath.absolute_path}" does not exist and create is off.'
raise FileNotFoundError(msg)
self.data_directory.makedirs(exist_ok=True)
self.sql_write = self._make_sqlite_write_connection(self.database_filepath)
self.sql_read = self._make_sqlite_read_connection(self.database_filepath)
if existing_database:
if not skip_version_check:
self._check_version()
with self.transaction:
self._load_pragmas()
else:
self._first_time_setup()
def _load_pragmas(self):
log.debug('Reloading pragmas.')
self.pragma_write('cache_size', 10000)
self.pragma_write('foreign_keys', 'on')
# Will add -> PhotoDB when forward references are supported
@classmethod
def closest_photodb(cls, path='.', *args, **kwargs):
'''
Starting from the given path and climbing upwards towards the filesystem
root, look for an existing Etiquette data directory and return the
PhotoDB object. If none exists, raise exceptions.NoClosestPhotoDB.
'''
path = pathclass.Path(path)
starting = path
while True:
possible = path.with_child(constants.DEFAULT_DATADIR)
if possible.is_dir:
break
parent = path.parent
if path == parent:
raise exceptions.NoClosestPhotoDB(starting.absolute_path)
path = parent
path = possible
log.debug('Found closest PhotoDB at "%s".', path.absolute_path)
try:
photodb = cls(
data_directory=path,
create=False,
*args,
**kwargs,
)
except FileNotFoundError:
raise exceptions.NoClosestPhotoDB(starting.absolute_path)
return photodb
def __del__(self):
self.close()
def __repr__(self):
if self.ephemeral:
return 'PhotoDB(ephemeral=True)'
else:
return f'PhotoDB(data_directory={self.data_directory})'
def close(self) -> None:
log.debug('Closing PhotoDB.')
super().close()
if getattr(self, 'ephemeral', False):
self.ephemeral_directory.cleanup()
def generate_id(self, thing_class) -> int:
'''
Create a new ID number that is unique to the given table.
'''
if not issubclass(thing_class, objects.ObjectBase):
raise TypeError(thing_class)
table = thing_class.table
length = self.config['id_bits']
for retry in range(10):
id = RNG.getrandbits(length)
if not self.exists(f'SELECT 1 FROM {table} WHERE id == ?', [id]):
return id
raise exceptions.GenerateIDFailed(table=table)
def load_config(self) -> None:
log.debug('Loading config file.')
(config, needs_rewrite) = configlayers.load_file(
filepath=self.config_filepath,
default_config=constants.DEFAULT_CONFIGURATION,
)
self.config = config
if needs_rewrite:
self.save_config()
def save_config(self) -> None:
log.debug('Saving config file.')
with self.config_filepath.open('w', encoding='utf-8') as handle:
handle.write(json.dumps(self.config, indent=4, sort_keys=True))
Reference in a new issue View git blame Copy permalink