Added ExifInfo (Photos 5 only)
This commit is contained in:
@@ -1,3 +1,3 @@
|
||||
""" version info """
|
||||
|
||||
__version__ = "0.28.16"
|
||||
__version__ = "0.28.17"
|
||||
|
||||
@@ -1,4 +1,3 @@
|
||||
|
||||
"""
|
||||
PhotoInfo class
|
||||
Represents a single photo in the Photos library and provides access to the photo's attributes
|
||||
@@ -6,3 +5,4 @@ PhotosDB.photos() returns a list of PhotoInfo objects
|
||||
"""
|
||||
|
||||
from .photoinfo import PhotoInfo
|
||||
from ._photoinfo_exifinfo import ExifInfo
|
||||
|
||||
94
osxphotos/photoinfo/_photoinfo_exifinfo.py
Normal file
94
osxphotos/photoinfo/_photoinfo_exifinfo.py
Normal file
@@ -0,0 +1,94 @@
|
||||
""" PhotoInfo methods to expose EXIF info from the library """
|
||||
|
||||
import logging
|
||||
from dataclasses import dataclass
|
||||
|
||||
from .._constants import _PHOTOS_4_VERSION
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class ExifInfo:
|
||||
""" EXIF info associated with a photo from the Photos library """
|
||||
|
||||
flash_fired: bool
|
||||
iso: int
|
||||
metering_mode: int
|
||||
sample_rate: int
|
||||
track_format: int
|
||||
white_balance: int
|
||||
aperture: float
|
||||
bit_rate: float
|
||||
duration: float
|
||||
exposure_bias: float
|
||||
focal_length: float
|
||||
fps: float
|
||||
latitude: float
|
||||
longitude: float
|
||||
shutter_speed: float
|
||||
camera_make: str
|
||||
camera_model: str
|
||||
codec: str
|
||||
lens_model: str
|
||||
|
||||
|
||||
@property
|
||||
def exif_info(self):
|
||||
""" Returns an ExifInfo object with the EXIF data for photo
|
||||
Note: the returned EXIF data is the data Photos stores in the database on import;
|
||||
ExifInfo does not provide access to the EXIF info in the actual image file
|
||||
Some or all of the fields may be None
|
||||
Only valid for Photos 5; on earlier database returns None
|
||||
"""
|
||||
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.debug(f"exif_info not implemented for this database version")
|
||||
return None
|
||||
|
||||
try:
|
||||
exif = self._db._db_exifinfo_uuid[self.uuid]
|
||||
exif_info = ExifInfo(
|
||||
iso=exif["ZISO"],
|
||||
flash_fired=True if exif["ZFLASHFIRED"] == 1 else False,
|
||||
metering_mode=exif["ZMETERINGMODE"],
|
||||
sample_rate=exif["ZSAMPLERATE"],
|
||||
track_format=exif["ZTRACKFORMAT"],
|
||||
white_balance=exif["ZWHITEBALANCE"],
|
||||
aperture=exif["ZAPERTURE"],
|
||||
bit_rate=exif["ZBITRATE"],
|
||||
duration=exif["ZDURATION"],
|
||||
exposure_bias=exif["ZEXPOSUREBIAS"],
|
||||
focal_length=exif["ZFOCALLENGTH"],
|
||||
fps=exif["ZFPS"],
|
||||
latitude=exif["ZLATITUDE"],
|
||||
longitude=exif["ZLONGITUDE"],
|
||||
shutter_speed=exif["ZSHUTTERSPEED"],
|
||||
camera_make=exif["ZCAMERAMAKE"],
|
||||
camera_model=exif["ZCAMERAMODEL"],
|
||||
codec=exif["ZCODEC"],
|
||||
lens_model=exif["ZLENSMODEL"],
|
||||
)
|
||||
except KeyError:
|
||||
logging.debug(f"Could not find exif record for uuid {self.uuid}")
|
||||
exif_info = ExifInfo(
|
||||
iso=None,
|
||||
flash_fired=None,
|
||||
metering_mode=None,
|
||||
sample_rate=None,
|
||||
track_format=None,
|
||||
white_balance=None,
|
||||
aperture=None,
|
||||
bit_rate=None,
|
||||
duration=None,
|
||||
exposure_bias=None,
|
||||
focal_length=None,
|
||||
fps=None,
|
||||
latitude=None,
|
||||
longitude=None,
|
||||
shutter_speed=None,
|
||||
camera_make=None,
|
||||
camera_model=None,
|
||||
codec=None,
|
||||
lens_model=None,
|
||||
)
|
||||
|
||||
return exif_info
|
||||
93
osxphotos/photoinfo/_photoinfo_searchinfo.py
Normal file
93
osxphotos/photoinfo/_photoinfo_searchinfo.py
Normal file
@@ -0,0 +1,93 @@
|
||||
""" Methods and class for PhotoInfo exposing SearchInfo data such as labels
|
||||
Adds the following properties to PhotoInfo (valid only for Photos 5):
|
||||
search_info: returns a SearchInfo object
|
||||
labels: returns list of labels
|
||||
labels_normalized: returns list of normalized labels
|
||||
"""
|
||||
|
||||
from .._constants import _PHOTOS_4_VERSION, SEARCH_CATEGORY_LABEL
|
||||
|
||||
|
||||
@property
|
||||
def search_info(self):
|
||||
""" returns SearchInfo object for photo
|
||||
only valid on Photos 5, on older libraries, returns None
|
||||
"""
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
return None
|
||||
|
||||
# memoize SearchInfo object
|
||||
try:
|
||||
return self._search_info
|
||||
except AttributeError:
|
||||
self._search_info = SearchInfo(self)
|
||||
return self._search_info
|
||||
|
||||
|
||||
@property
|
||||
def labels(self):
|
||||
""" returns list of labels applied to photo by Photos image categorization
|
||||
only valid on Photos 5, on older libraries returns empty list
|
||||
"""
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
return []
|
||||
|
||||
return self.search_info.labels
|
||||
|
||||
|
||||
@property
|
||||
def labels_normalized(self):
|
||||
""" returns normalized list of labels applied to photo by Photos image categorization
|
||||
only valid on Photos 5, on older libraries returns empty list
|
||||
"""
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
return []
|
||||
|
||||
return self.search_info.labels_normalized
|
||||
|
||||
|
||||
class SearchInfo:
|
||||
""" Info about search terms such as machine learning labels that Photos knows about a photo """
|
||||
|
||||
def __init__(self, photo):
|
||||
""" photo: PhotoInfo object """
|
||||
|
||||
if photo._db._db_version <= _PHOTOS_4_VERSION:
|
||||
raise NotImplementedError(
|
||||
f"search info not implemented for this database version"
|
||||
)
|
||||
|
||||
self._photo = photo
|
||||
self.uuid = photo.uuid
|
||||
try:
|
||||
# get search info for this UUID
|
||||
# there might not be any search info data (e.g. if Photo was missing or photoanalysisd not run yet)
|
||||
self._db_searchinfo = photo._db._db_searchinfo_uuid[self.uuid]
|
||||
except KeyError:
|
||||
self._db_searchinfo = None
|
||||
|
||||
@property
|
||||
def labels(self):
|
||||
""" return list of labels associated with Photo """
|
||||
if self._db_searchinfo:
|
||||
labels = [
|
||||
rec["content_string"]
|
||||
for rec in self._db_searchinfo
|
||||
if rec["category"] == SEARCH_CATEGORY_LABEL
|
||||
]
|
||||
else:
|
||||
labels = []
|
||||
return labels
|
||||
|
||||
@property
|
||||
def labels_normalized(self):
|
||||
""" return list of normalized labels associated with Photo """
|
||||
if self._db_searchinfo:
|
||||
labels = [
|
||||
rec["normalized_string"]
|
||||
for rec in self._db_searchinfo
|
||||
if rec["category"] == SEARCH_CATEGORY_LABEL
|
||||
]
|
||||
else:
|
||||
labels = []
|
||||
return labels
|
||||
@@ -19,6 +19,7 @@ from pprint import pformat
|
||||
import yaml
|
||||
from mako.template import Template
|
||||
|
||||
|
||||
from .._constants import (
|
||||
_MAX_IPTC_KEYWORD_LEN,
|
||||
_MOVIE_TYPE,
|
||||
@@ -49,16 +50,22 @@ from ..utils import (
|
||||
get_preferred_uti_extension,
|
||||
)
|
||||
|
||||
# Mixins
|
||||
from .photoinfo_mixin_searchinfo import PhotoInfoMixinSearchInfo, SearchInfo
|
||||
|
||||
|
||||
class PhotoInfo(PhotoInfoMixinSearchInfo):
|
||||
class PhotoInfo:
|
||||
"""
|
||||
Info about a specific photo, contains all the details about the photo
|
||||
including keywords, persons, albums, uuid, path, etc.
|
||||
"""
|
||||
|
||||
# import additional methods
|
||||
from ._photoinfo_searchinfo import (
|
||||
search_info,
|
||||
labels,
|
||||
labels_normalized,
|
||||
SearchInfo,
|
||||
)
|
||||
from ._photoinfo_exifinfo import exif_info, ExifInfo
|
||||
|
||||
def __init__(self, db=None, uuid=None, info=None):
|
||||
self._uuid = uuid
|
||||
self._info = info
|
||||
|
||||
@@ -1,96 +0,0 @@
|
||||
""" SearchInfo class exposing labels and other search info for Photos 5 databases
|
||||
and
|
||||
PhotoInfoMixinSearchInfo mixin class for PhotoInfo """
|
||||
|
||||
from .._constants import _PHOTOS_4_VERSION, SEARCH_CATEGORY_LABEL
|
||||
|
||||
|
||||
class PhotoInfoMixinSearchInfo:
|
||||
""" Mixin class for PhotoInfo exposing SearchInfo data such as labels
|
||||
Adds the following properties to PhotoInfo (valid only for Photos 5):
|
||||
search_info: returns a SearchInfo object
|
||||
labels: returns list of labels
|
||||
labels_normalized: returns list of normalized labels
|
||||
"""
|
||||
|
||||
@property
|
||||
def search_info(self):
|
||||
""" returns SearchInfo object for photo
|
||||
only valid on Photos 5, on older libraries, returns None
|
||||
"""
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
return None
|
||||
|
||||
# memoize SearchInfo object
|
||||
try:
|
||||
return self._search_info
|
||||
except AttributeError:
|
||||
self._search_info = SearchInfo(self)
|
||||
return self._search_info
|
||||
|
||||
@property
|
||||
def labels(self):
|
||||
""" returns list of labels applied to photo by Photos image categorization
|
||||
only valid on Photos 5, on older libraries returns empty list
|
||||
"""
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
return []
|
||||
|
||||
return self.search_info.labels
|
||||
|
||||
@property
|
||||
def labels_normalized(self):
|
||||
""" returns normalized list of labels applied to photo by Photos image categorization
|
||||
only valid on Photos 5, on older libraries returns empty list
|
||||
"""
|
||||
if self._db._db_version <= _PHOTOS_4_VERSION:
|
||||
return []
|
||||
|
||||
return self.search_info.labels_normalized
|
||||
|
||||
|
||||
class SearchInfo:
|
||||
""" Info about search terms such as machine learning labels that Photos knows about a photo """
|
||||
|
||||
def __init__(self, photo):
|
||||
""" photo: PhotoInfo object """
|
||||
|
||||
if photo._db._db_version <= _PHOTOS_4_VERSION:
|
||||
raise NotImplementedError(
|
||||
f"search info not implemented for this database version"
|
||||
)
|
||||
|
||||
self._photo = photo
|
||||
self.uuid = photo.uuid
|
||||
try:
|
||||
# get search info for this UUID
|
||||
# there might not be any search info data (e.g. if Photo was missing or photoanalysisd not run yet)
|
||||
self._db_searchinfo = photo._db._db_searchinfo_uuid[self.uuid]
|
||||
except KeyError:
|
||||
self._db_searchinfo = None
|
||||
|
||||
@property
|
||||
def labels(self):
|
||||
""" return list of labels associated with Photo """
|
||||
if self._db_searchinfo:
|
||||
labels = [
|
||||
rec["content_string"]
|
||||
for rec in self._db_searchinfo
|
||||
if rec["category"] == SEARCH_CATEGORY_LABEL
|
||||
]
|
||||
else:
|
||||
labels = []
|
||||
return labels
|
||||
|
||||
@property
|
||||
def labels_normalized(self):
|
||||
""" return list of normalized labels associated with Photo """
|
||||
if self._db_searchinfo:
|
||||
labels = [
|
||||
rec["normalized_string"]
|
||||
for rec in self._db_searchinfo
|
||||
if rec["category"] == SEARCH_CATEGORY_LABEL
|
||||
]
|
||||
else:
|
||||
labels = []
|
||||
return labels
|
||||
56
osxphotos/photosdb/_photosdb_process_exif.py
Normal file
56
osxphotos/photosdb/_photosdb_process_exif.py
Normal file
@@ -0,0 +1,56 @@
|
||||
""" PhotosDB method for processing exif info
|
||||
Do not import this module directly """
|
||||
|
||||
import logging
|
||||
|
||||
from .._constants import _PHOTOS_4_VERSION
|
||||
from ..utils import _db_is_locked, _debug, _open_sql_file
|
||||
|
||||
|
||||
def _process_exifinfo(self):
|
||||
""" load the exif data from the database
|
||||
this is a PhotosDB method that should be imported in
|
||||
the PhotosDB class definition in photosdb.py
|
||||
"""
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
_process_exifinfo_4(self)
|
||||
else:
|
||||
_process_exifinfo_5(self)
|
||||
|
||||
|
||||
# The following methods do not get imported into PhotosDB
|
||||
# but will get called by _process_exifinfo
|
||||
|
||||
|
||||
def _process_exifinfo_4(photosdb):
|
||||
""" process exif info for Photos <= 4
|
||||
photosdb: PhotosDB instance """
|
||||
photosdb._db_exifinfo_uuid = {}
|
||||
raise NotImplementedError(f"search info not implemented for this database version")
|
||||
|
||||
|
||||
def _process_exifinfo_5(photosdb):
|
||||
""" process exif info for Photos >= 5
|
||||
photosdb: PhotosDB instance """
|
||||
|
||||
db = photosdb._tmp_db
|
||||
|
||||
(conn, cursor) = _open_sql_file(db)
|
||||
|
||||
result = conn.execute(
|
||||
"""
|
||||
SELECT ZGENERICASSET.ZUUID, ZEXTENDEDATTRIBUTES.*
|
||||
FROM ZGENERICASSET
|
||||
JOIN ZEXTENDEDATTRIBUTES
|
||||
ON ZEXTENDEDATTRIBUTES.ZASSET = ZGENERICASSET.Z_PK
|
||||
"""
|
||||
)
|
||||
|
||||
photosdb._db_exifinfo_uuid = {}
|
||||
cols = [c[0] for c in result.description]
|
||||
for row in result.fetchall():
|
||||
record = dict(zip(cols, row))
|
||||
uuid = record["ZUUID"]
|
||||
if uuid in photosdb._db_exifinfo_uuid:
|
||||
logging.warning(f"duplicate exifinfo record found for uuid {uuid}")
|
||||
photosdb._db_exifinfo_uuid[uuid] = record
|
||||
197
osxphotos/photosdb/_photosdb_process_searchinfo.py
Normal file
197
osxphotos/photosdb/_photosdb_process_searchinfo.py
Normal file
@@ -0,0 +1,197 @@
|
||||
""" Methods for PhotosDB to add Photos 5 search info such as machine learning labels
|
||||
Kudos to Simon Willison who figured out how to extract this data from psi.sql
|
||||
ref: https://github.com/dogsheep/photos-to-sqlite/issues/16
|
||||
"""
|
||||
|
||||
import logging
|
||||
import pathlib
|
||||
import uuid as uuidlib
|
||||
from pprint import pformat
|
||||
|
||||
from .._constants import _PHOTOS_4_VERSION, SEARCH_CATEGORY_LABEL
|
||||
from ..utils import _db_is_locked, _debug, _open_sql_file
|
||||
|
||||
"""
|
||||
This module should be imported in the class defintion of PhotosDB in photosdb.py
|
||||
Do not import this module directly
|
||||
This module adds the following method to PhotosDB:
|
||||
_process_searchinfo: process search terms from psi.sqlite
|
||||
|
||||
The following properties are added to PhotosDB
|
||||
labels: list of all labels in the library
|
||||
labels_normalized: list of all labels normalized in the library
|
||||
labels_as_dict: dict of {label: count of photos} in reverse sorted order (most photos first)
|
||||
labels_normalized_as_dict: dict of {normalized label: count of photos} in reverse sorted order (most photos first)
|
||||
|
||||
The following data structures are added to PhotosDB
|
||||
self._db_searchinfo_categories
|
||||
self._db_searchinfo_uuid
|
||||
self._db_searchinfo_labels
|
||||
self._db_searchinfo_labels_normalized
|
||||
|
||||
These methods only work on Photos 5 databases. Will print warning on earlier library versions.
|
||||
"""
|
||||
|
||||
|
||||
def _process_searchinfo(self):
|
||||
""" load machine learning/search term label info from a Photos library
|
||||
db_connection: a connection to the SQLite database file containing the
|
||||
search terms. In Photos 5, this is called psi.sqlite
|
||||
Note: Only works on Photos version == 5.0 """
|
||||
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
raise NotImplementedError(
|
||||
f"search info not implemented for this database version"
|
||||
)
|
||||
|
||||
search_db_path = pathlib.Path(self._dbfile).parent / "search" / "psi.sqlite"
|
||||
if not search_db_path.exists():
|
||||
raise FileNotFoundError(f"could not find search db: {search_db_path}")
|
||||
|
||||
if _db_is_locked(search_db_path):
|
||||
search_db = self._copy_db_file(search_db_path)
|
||||
else:
|
||||
search_db = search_db_path
|
||||
|
||||
(conn, c) = _open_sql_file(search_db)
|
||||
|
||||
result = conn.execute(
|
||||
"""
|
||||
select
|
||||
ga.rowid,
|
||||
assets.uuid_0,
|
||||
assets.uuid_1,
|
||||
groups.rowid as groupid,
|
||||
groups.category,
|
||||
groups.owning_groupid,
|
||||
groups.content_string,
|
||||
groups.normalized_string,
|
||||
groups.lookup_identifier
|
||||
from
|
||||
ga
|
||||
join groups on groups.rowid = ga.groupid
|
||||
join assets on ga.assetid = assets.rowid
|
||||
order by
|
||||
ga.rowid
|
||||
"""
|
||||
)
|
||||
|
||||
# _db_searchinfo_uuid is dict in form {uuid : [list of associated search info records]
|
||||
_db_searchinfo_uuid = {}
|
||||
|
||||
# _db_searchinfo_categories is dict in form {search info category id: list normalized strings for the category
|
||||
# right now, this is mostly for debugging to easily see which search terms are in the library
|
||||
_db_searchinfo_categories = {}
|
||||
|
||||
# _db_searchinfo_labels is dict in form {normalized label: [list of photo uuids]}
|
||||
# this serves as a reverse index from label to photos containing the label
|
||||
# _db_searchinfo_labels_normalized is the same but with normalized (lower case) version of the label
|
||||
_db_searchinfo_labels = {}
|
||||
_db_searchinfo_labels_normalized = {}
|
||||
|
||||
cols = [c[0] for c in result.description]
|
||||
for row in result.fetchall():
|
||||
record = dict(zip(cols, row))
|
||||
uuid = ints_to_uuid(record["uuid_0"], record["uuid_1"])
|
||||
# strings have null character appended, so strip it
|
||||
for key in record:
|
||||
if isinstance(record[key], str):
|
||||
record[key] = record[key].replace("\x00", "")
|
||||
try:
|
||||
_db_searchinfo_uuid[uuid].append(record)
|
||||
except KeyError:
|
||||
_db_searchinfo_uuid[uuid] = [record]
|
||||
|
||||
category = record["category"]
|
||||
try:
|
||||
_db_searchinfo_categories[record["category"]].append(
|
||||
record["normalized_string"]
|
||||
)
|
||||
except KeyError:
|
||||
_db_searchinfo_categories[record["category"]] = [
|
||||
record["normalized_string"]
|
||||
]
|
||||
|
||||
if record["category"] == SEARCH_CATEGORY_LABEL:
|
||||
label = record["content_string"]
|
||||
label_norm = record["normalized_string"]
|
||||
try:
|
||||
_db_searchinfo_labels[label].append(uuid)
|
||||
_db_searchinfo_labels_normalized[label_norm].append(uuid)
|
||||
except KeyError:
|
||||
_db_searchinfo_labels[label] = [uuid]
|
||||
_db_searchinfo_labels_normalized[label_norm] = [uuid]
|
||||
|
||||
self._db_searchinfo_categories = _db_searchinfo_categories
|
||||
self._db_searchinfo_uuid = _db_searchinfo_uuid
|
||||
self._db_searchinfo_labels = _db_searchinfo_labels
|
||||
self._db_searchinfo_labels_normalized = _db_searchinfo_labels_normalized
|
||||
|
||||
if _debug():
|
||||
logging.debug(
|
||||
"_db_searchinfo_categories: \n" + pformat(self._db_searchinfo_categories)
|
||||
)
|
||||
logging.debug("_db_searchinfo_uuid: \n" + pformat(self._db_searchinfo_uuid))
|
||||
logging.debug("_db_searchinfo_labels: \n" + pformat(self._db_searchinfo_labels))
|
||||
logging.debug(
|
||||
"_db_searchinfo_labels_normalized: \n"
|
||||
+ pformat(self._db_searchinfo_labels_normalized)
|
||||
)
|
||||
|
||||
|
||||
@property
|
||||
def labels(self):
|
||||
""" return list of all search info labels found in the library """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return []
|
||||
|
||||
return list(self._db_searchinfo_labels.keys())
|
||||
|
||||
|
||||
@property
|
||||
def labels_normalized(self):
|
||||
""" return list of all normalized search info labels found in the library """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return []
|
||||
|
||||
return list(self._db_searchinfo_labels_normalized.keys())
|
||||
|
||||
|
||||
@property
|
||||
def labels_as_dict(self):
|
||||
""" return labels as dict of label: count in reverse sorted order (descending) """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return dict()
|
||||
|
||||
labels = {k: len(v) for k, v in self._db_searchinfo_labels.items()}
|
||||
labels = dict(sorted(labels.items(), key=lambda kv: kv[1], reverse=True))
|
||||
return labels
|
||||
|
||||
|
||||
@property
|
||||
def labels_normalized_as_dict(self):
|
||||
""" return normalized labels as dict of label: count in reverse sorted order (descending) """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return dict()
|
||||
labels = {k: len(v) for k, v in self._db_searchinfo_labels_normalized.items()}
|
||||
labels = dict(sorted(labels.items(), key=lambda kv: kv[1], reverse=True))
|
||||
return labels
|
||||
|
||||
|
||||
# The following method is not imported into PhotosDB
|
||||
|
||||
|
||||
def ints_to_uuid(uuid_0, uuid_1):
|
||||
""" convert two signed ints into a UUID strings
|
||||
uuid_0, uuid_1: the two int components of an RFC 4122 UUID """
|
||||
|
||||
# assumes uuid imported as uuidlib (to avoid namespace conflict with other uses of uuid)
|
||||
|
||||
bytes_ = uuid_0.to_bytes(8, "little", signed=True) + uuid_1.to_bytes(
|
||||
8, "little", signed=True
|
||||
)
|
||||
return str(uuidlib.UUID(bytes=bytes_)).upper()
|
||||
@@ -44,20 +44,27 @@ from ..utils import (
|
||||
get_last_library_path,
|
||||
)
|
||||
|
||||
# mixins
|
||||
from .photosdb_mixin_searchinfo import PhotosDBMixinSearchInfo
|
||||
|
||||
# TODO: Add test for imageTimeZoneOffsetSeconds = None
|
||||
# TODO: Fix command line so multiple --keyword, etc. are AND (instead of OR as they are in .photos())
|
||||
# Or fix the help text to match behavior
|
||||
# TODO: Add test for __str__
|
||||
# TODO: Add special albums and magic albums
|
||||
# TODO: fix "if X not in y" dictionary checks to use try/except EAFP style
|
||||
|
||||
|
||||
class PhotosDB(PhotosDBMixinSearchInfo):
|
||||
class PhotosDB:
|
||||
""" Processes a Photos.app library database to extract information about photos """
|
||||
|
||||
# import additional methods
|
||||
from ._photosdb_process_exif import _process_exifinfo
|
||||
from ._photosdb_process_searchinfo import (
|
||||
_process_searchinfo,
|
||||
labels,
|
||||
labels_normalized,
|
||||
labels_as_dict,
|
||||
labels_normalized_as_dict,
|
||||
)
|
||||
|
||||
def __init__(self, *dbfile_, dbfile=None):
|
||||
""" create a new PhotosDB object
|
||||
path to photos library or database may be specified EITHER as first argument or as named argument dbfile=path
|
||||
@@ -1848,6 +1855,9 @@ class PhotosDB(PhotosDBMixinSearchInfo):
|
||||
# process search info
|
||||
self._process_searchinfo()
|
||||
|
||||
# process exif info
|
||||
self._process_exifinfo()
|
||||
|
||||
# done processing, dump debug data if requested
|
||||
if _debug():
|
||||
logging.debug("Faces (_dbfaces_uuid):")
|
||||
|
||||
@@ -1,192 +0,0 @@
|
||||
""" Mixin class for PhotosDB to add Photos 5 search info such as machine learning labels
|
||||
Kudos to Simon Willison who figured out how to extract this data from psi.sql
|
||||
ref: https://github.com/dogsheep/photos-to-sqlite/issues/16
|
||||
"""
|
||||
|
||||
import logging
|
||||
import pathlib
|
||||
import uuid as uuidlib
|
||||
from pprint import pformat
|
||||
|
||||
from .._constants import _PHOTOS_4_VERSION, SEARCH_CATEGORY_LABEL
|
||||
from ..utils import _db_is_locked, _debug, _open_sql_file
|
||||
|
||||
|
||||
class PhotosDBMixinSearchInfo:
|
||||
""" Mixin class to extend PhotosDB to process search info terms
|
||||
This mixin adds the following method to PhotosDB:
|
||||
_process_searchinfo: process search terms from psi.sqlite
|
||||
|
||||
The following properties are added to PhotosDB
|
||||
labels: list of all labels in the library
|
||||
labels_normalized: list of all labels normalized in the library
|
||||
labels_as_dict: dict of {label: count of photos} in reverse sorted order (most photos first)
|
||||
labels_normalized_as_dict: dict of {normalized label: count of photos} in reverse sorted order (most photos first)
|
||||
|
||||
The following data structures are added to PhotosDB
|
||||
self._db_searchinfo_categories
|
||||
self._db_searchinfo_uuid
|
||||
self._db_searchinfo_labels
|
||||
self._db_searchinfo_labels_normalized
|
||||
|
||||
These methods only work on Photos 5 databases. Will print warning on earlier library versions.
|
||||
"""
|
||||
|
||||
def _process_searchinfo(self):
|
||||
""" load machine learning/search term label info from a Photos library
|
||||
db_connection: a connection to the SQLite database file containing the
|
||||
search terms. In Photos 5, this is called psi.sqlite
|
||||
Note: Only works on Photos version == 5.0 """
|
||||
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
raise NotImplementedError(
|
||||
f"search info not implemented for this database version"
|
||||
)
|
||||
|
||||
search_db_path = pathlib.Path(self._dbfile).parent / "search" / "psi.sqlite"
|
||||
if not search_db_path.exists():
|
||||
raise FileNotFoundError(f"could not find search db: {search_db_path}")
|
||||
|
||||
if _db_is_locked(search_db_path):
|
||||
search_db = self._copy_db_file(search_db_path)
|
||||
else:
|
||||
search_db = search_db_path
|
||||
|
||||
(conn, c) = _open_sql_file(search_db)
|
||||
|
||||
result = conn.execute(
|
||||
"""
|
||||
select
|
||||
ga.rowid,
|
||||
assets.uuid_0,
|
||||
assets.uuid_1,
|
||||
groups.rowid as groupid,
|
||||
groups.category,
|
||||
groups.owning_groupid,
|
||||
groups.content_string,
|
||||
groups.normalized_string,
|
||||
groups.lookup_identifier
|
||||
from
|
||||
ga
|
||||
join groups on groups.rowid = ga.groupid
|
||||
join assets on ga.assetid = assets.rowid
|
||||
order by
|
||||
ga.rowid
|
||||
"""
|
||||
)
|
||||
|
||||
# _db_searchinfo_uuid is dict in form {uuid : [list of associated search info records]
|
||||
_db_searchinfo_uuid = {}
|
||||
|
||||
# _db_searchinfo_categories is dict in form {search info category id: list normalized strings for the category
|
||||
# right now, this is mostly for debugging to easily see which search terms are in the library
|
||||
_db_searchinfo_categories = {}
|
||||
|
||||
# _db_searchinfo_labels is dict in form {normalized label: [list of photo uuids]}
|
||||
# this serves as a reverse index from label to photos containing the label
|
||||
# _db_searchinfo_labels_normalized is the same but with normalized (lower case) version of the label
|
||||
_db_searchinfo_labels = {}
|
||||
_db_searchinfo_labels_normalized = {}
|
||||
|
||||
cols = [c[0] for c in result.description]
|
||||
for row in result.fetchall():
|
||||
record = dict(zip(cols, row))
|
||||
uuid = ints_to_uuid(record["uuid_0"], record["uuid_1"])
|
||||
# strings have null character appended, so strip it
|
||||
for key in record:
|
||||
if isinstance(record[key], str):
|
||||
record[key] = record[key].replace("\x00", "")
|
||||
try:
|
||||
_db_searchinfo_uuid[uuid].append(record)
|
||||
except KeyError:
|
||||
_db_searchinfo_uuid[uuid] = [record]
|
||||
|
||||
category = record["category"]
|
||||
try:
|
||||
_db_searchinfo_categories[record["category"]].append(
|
||||
record["normalized_string"]
|
||||
)
|
||||
except KeyError:
|
||||
_db_searchinfo_categories[record["category"]] = [
|
||||
record["normalized_string"]
|
||||
]
|
||||
|
||||
if record["category"] == SEARCH_CATEGORY_LABEL:
|
||||
label = record["content_string"]
|
||||
label_norm = record["normalized_string"]
|
||||
try:
|
||||
_db_searchinfo_labels[label].append(uuid)
|
||||
_db_searchinfo_labels_normalized[label_norm].append(uuid)
|
||||
except KeyError:
|
||||
_db_searchinfo_labels[label] = [uuid]
|
||||
_db_searchinfo_labels_normalized[label_norm] = [uuid]
|
||||
|
||||
self._db_searchinfo_categories = _db_searchinfo_categories
|
||||
self._db_searchinfo_uuid = _db_searchinfo_uuid
|
||||
self._db_searchinfo_labels = _db_searchinfo_labels
|
||||
self._db_searchinfo_labels_normalized = _db_searchinfo_labels_normalized
|
||||
|
||||
if _debug():
|
||||
logging.debug(
|
||||
"_db_searchinfo_categories: \n"
|
||||
+ pformat(self._db_searchinfo_categories)
|
||||
)
|
||||
logging.debug("_db_searchinfo_uuid: \n" + pformat(self._db_searchinfo_uuid))
|
||||
logging.debug(
|
||||
"_db_searchinfo_labels: \n" + pformat(self._db_searchinfo_labels)
|
||||
)
|
||||
logging.debug(
|
||||
"_db_searchinfo_labels_normalized: \n"
|
||||
+ pformat(self._db_searchinfo_labels_normalized)
|
||||
)
|
||||
|
||||
@property
|
||||
def labels(self):
|
||||
""" return list of all search info labels found in the library """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return []
|
||||
|
||||
return list(self._db_searchinfo_labels.keys())
|
||||
|
||||
@property
|
||||
def labels_normalized(self):
|
||||
""" return list of all normalized search info labels found in the library """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return []
|
||||
|
||||
return list(self._db_searchinfo_labels_normalized.keys())
|
||||
|
||||
@property
|
||||
def labels_as_dict(self):
|
||||
""" return labels as dict of label: count in reverse sorted order (descending) """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return dict()
|
||||
|
||||
labels = {k: len(v) for k, v in self._db_searchinfo_labels.items()}
|
||||
labels = dict(sorted(labels.items(), key=lambda kv: kv[1], reverse=True))
|
||||
return labels
|
||||
|
||||
@property
|
||||
def labels_normalized_as_dict(self):
|
||||
""" return normalized labels as dict of label: count in reverse sorted order (descending) """
|
||||
if self._db_version <= _PHOTOS_4_VERSION:
|
||||
logging.warning(f"SearchInfo not implemented for this library version")
|
||||
return dict()
|
||||
labels = {k: len(v) for k, v in self._db_searchinfo_labels_normalized.items()}
|
||||
labels = dict(sorted(labels.items(), key=lambda kv: kv[1], reverse=True))
|
||||
return labels
|
||||
|
||||
|
||||
def ints_to_uuid(uuid_0, uuid_1):
|
||||
""" convert two signed ints into a UUID strings
|
||||
uuid_0, uuid_1: the two int components of an RFC 4122 UUID """
|
||||
|
||||
# assumes uuid imported as uuidlib (to avoid namespace conflict with other uses of uuid)
|
||||
|
||||
bytes_ = uuid_0.to_bytes(8, "little", signed=True) + uuid_1.to_bytes(
|
||||
8, "little", signed=True
|
||||
)
|
||||
return str(uuidlib.UUID(bytes=bytes_)).upper()
|
||||
Reference in New Issue
Block a user