ext_open/ext_open/searcher.py

import sqlite3
import os
import re
import update as _update
from config import config
import time
# https://stackoverflow.com/a/5365533
osp = os.path
DB_PATH = osp.join(osp.dirname(__file__),'ext.db')
ext_database = sqlite3.connect(DB_PATH)
def init_db():
    cur = ext_database.cursor()
    cur.execute('CREATE TABLE IF NOT EXISTS updates ( `ext` TEXT, `update_time` INTEGER, `file_count` INTEGER);')

def regexp(expr, item,case):
    if case:
        return (1 if re.search(expr,item) is not None else 0)
    else:
        return (1 if re.search(expr,item,flags=re.I) is not None else 0)
def _in(expr,item,case):
    if case:
        return (1 if expr in item else 0)
    else:
        return (1 if expr.lower() in item else 0)

ext_database.create_function("REGEXP", 3, regexp)
ext_database.create_function("INSIDE", 3, _in)

class Searcher:

    def __init__(self,ext,case_sensitive=True):
        self.cur = ext_database.cursor()
        self.ext = ext
        self.case = case_sensitive
        self.index = '{}_index'.format(ext)
        self.__init_table__()

    def __init_table__(self):
        self.cur.execute('CREATE TABLE IF NOT EXISTS {} (`name` TEXT,`fullpath` TEXT, `update_time` INTEGER,UNIQUE(name,fullpath));'.format(self.ext))
        self.cur.execute('CREATE INDEX IF NOT EXISTS {} ON {} ( `name`, `fullpath` )'.format(self.index,self.ext))
        self.cur.execute(
            'INSERT INTO updates (ext,update_time,file_count) SELECT :ext, 0, 0 WHERE NOT EXISTS(SELECT 1 FROM updates WHERE ext = :ext AND update_time = 0 AND file_count = 0);',
            {'ext':self.ext}
            )
        self.commit()
    def update(self,update_mtimes):
        self.cur.execute('SELECT MAX(update_time) FROM updates WHERE ext=?;',[self.ext])
        last_update = self.cur.fetchone()[0]
        # input(str(last_update))
        _time = int(time.time())
        files = _update.update(
            self.ext,
            config.getlist(self.ext,'paths'),
            last_update,
            update_mtimes
            )
        self.cur.executemany('INSERT OR IGNORE INTO {} VALUES (?,?,?)'.format(self.ext),(row + [_time] for row in files))
        self.cur.execute('INSERT INTO updates VALUES (?, ?, ?)',[self.ext,_time,len(files)])
        self.commit()

    def commit(self):
        ext_database.commit()

    def close(self):
        ext_database.close()

    def search(self,query,regex=False):
        if regex:
            self.cur.execute('SELECT name,fullpath FROM {} WHERE REGEXP(?,name,?)=1;'.format(self.ext),[query,self.case])
        else:
            self.cur.execute('SELECT name,fullpath FROM {} WHERE INSIDE(?,name,?)=1;'.format(self.ext),[query,self.case])
        return self.cur.fetchall()

    def clean(self,backhistory):
        self.cur.execute('SELECT update_time FROM updates WHERE ext=? ORDER BY update_time DESC LIMIT ?;',[self.ext,backhistory+1])
        res = self.cur.fetchall()
        threshold = res[-1][0]
        self.cur.execute('DELETE FROM {} WHERE update_time < ?;'.format(self.ext),[threshold])
        self.commit()

    def clear(self):
        self.cur.execute('DELETE FROM {};'.format(self.ext))
        self.commit()

if __name__ == "__main__":
    init_db()
    s = Searcher('py')
    # s.update(True)