1
0
mirror of https://github.com/arsenetar/dupeguru.git synced 2024-11-10 17:59:02 +00:00
dupeguru/core_pe/cache.py
2012-03-15 14:28:40 -04:00

137 lines
4.6 KiB
Python

# Created By: Virgil Dupras
# Created On: 2006/09/14
# Copyright 2012 Hardcoded Software (http://www.hardcoded.net)
#
# This software is licensed under the "BSD" License as described in the "LICENSE" file,
# which should be included with this package. The terms are also available at
# http://www.hardcoded.net/licenses/bsd_license
import os
import logging
import sqlite3 as sqlite
from ._cache import string_to_colors
def colors_to_string(colors):
"""Transform the 3 sized tuples 'colors' into a hex string.
[(0,100,255)] --> 0064ff
[(1,2,3),(4,5,6)] --> 010203040506
"""
return ''.join(['%02x%02x%02x' % (r,g,b) for r,g,b in colors])
# This function is an important bottleneck of dupeGuru PE. It has been converted to C.
# def string_to_colors(s):
# """Transform the string 's' in a list of 3 sized tuples.
# """
# result = []
# for i in xrange(0, len(s), 6):
# number = int(s[i:i+6], 16)
# result.append((number >> 16, (number >> 8) & 0xff, number & 0xff))
# return result
class Cache(object):
"""A class to cache picture blocks.
"""
def __init__(self, db=':memory:'):
self.dbname = db
self.con = None
self._create_con()
def __contains__(self, key):
sql = "select count(*) from pictures where path = ?"
result = self.con.execute(sql, [key]).fetchall()
return result[0][0] > 0
def __delitem__(self, key):
if key not in self:
raise KeyError(key)
sql = "delete from pictures where path = ?"
self.con.execute(sql, [key])
# Optimized
def __getitem__(self, key):
if isinstance(key, int):
sql = "select blocks from pictures where rowid = ?"
else:
sql = "select blocks from pictures where path = ?"
result = self.con.execute(sql, [key]).fetchone()
if result:
result = string_to_colors(result[0])
return result
else:
raise KeyError(key)
def __iter__(self):
sql = "select path from pictures"
result = self.con.execute(sql)
return (row[0] for row in result)
def __len__(self):
sql = "select count(*) from pictures"
result = self.con.execute(sql).fetchall()
return result[0][0]
def __setitem__(self, key, value):
value = colors_to_string(value)
if key in self:
sql = "update pictures set blocks = ? where path = ?"
else:
sql = "insert into pictures(blocks,path) values(?,?)"
try:
self.con.execute(sql, [value, key])
except sqlite.OperationalError:
logging.warning('Picture cache could not set value for key %r', key)
except sqlite.DatabaseError as e:
logging.warning('DatabaseError while setting value for key %r: %s', key, str(e))
def _create_con(self, second_try=False):
def create_tables():
sql = "create table pictures(path TEXT, blocks TEXT)"
self.con.execute(sql);
sql = "create index idx_path on pictures (path)"
self.con.execute(sql)
self.con = sqlite.connect(self.dbname, isolation_level=None)
try:
self.con.execute("select * from pictures where 1=2")
except sqlite.OperationalError: # new db
create_tables()
except sqlite.DatabaseError as e: # corrupted db
if second_try:
raise # Something really strange is happening
logging.warning('Could not create picture cache because of an error: %s', str(e))
self.con.close()
os.remove(self.dbname)
self._create_con(second_try=True)
def clear(self):
self.close()
if self.dbname != ':memory:':
os.remove(self.dbname)
self._create_con()
def close(self):
if self.con is not None:
self.con.close()
self.con = None
def filter(self, func):
to_delete = [key for key in self if not func(key)]
for key in to_delete:
del self[key]
def get_id(self, path):
sql = "select rowid from pictures where path = ?"
result = self.con.execute(sql, [path]).fetchone()
if result:
return result[0]
else:
raise ValueError(path)
def get_multiple(self, rowids):
sql = "select rowid, blocks from pictures where rowid in (%s)" % ','.join(map(str, rowids))
cur = self.con.execute(sql)
return ((rowid, string_to_colors(blocks)) for rowid, blocks in cur)