def down_select(syntax_tree, sql, relation2):
if syntax_tree.op == 'SELECT':
condition = syntax_tree.cond
sql = condition.split('&')
relation = []
# 找出关系名
for i in range(len(sql)):
if lookfor(sql[i]) is not None:
relation.append(lookfor(sql[i]))
relation2.append(lookfor2(sql[i]))
syntax_tree = down_select(syntax_tree.lfchild, sql, relation2)
elif syntax_tree.op == 'PROJECTION':
# 跳过投影
syntax_tree.lfchild = down_select(syntax_tree.lfchild, sql, relation2)
elif syntax_tree.op == 'JOIN':
# 有JOIN，将SELECT与关系名下推
first_tree = SyntaxTree()
first_tree.op = 'SELECT'
first_tree.cond = sql[0]
first_tree.lfchild = syntax_tree.lfchild
syntax_tree.lfchild = first_tree
if len(sql) == 1:
return syntax_tree
second_tree = SyntaxTree()
second_tree.op = 'SELECT'
second_tree.cond = sql[1]
second_tree.rchild = syntax_tree.rchild
syntax_tree.rchild = second_tree
return syntax_tree
def parsesql(sql_statement):
sql = sql_statement.split()
tree = SyntaxTree()
index = 0
while True:
if index >= len(sql):
break
elif sql[index] == 'SELECT' or sql[index] == 'PROJECTION':
tree.op = sql[index]
index += 2
condition = ''
# 把[]里的内容全部记录下来
while sql[index] != ']':
condition += sql[index]
condition += ' '
index += 1
index += 1
tree.cond = condition
elif sql[index] == 'JOIN':
# 连接操作创建子树
tree.op = sql[index]
tree.lfchild = SyntaxTree()
tree.lfchild.attr = sql[index - 1]
tree.rchild = SyntaxTree()
tree.rchild.attr = sql[index + 1]
index += 1
elif sql[index] == '(':
# 遇到(再创建一个子树
index += 1
statement = ''
while index < len(sql) and sql[index] != ')':
statement += sql[index]
statement += ' '
index += 1
index += 1
tree.lfchild = parsesql(statement) # 递归构造子树
else:
index += 1
return tree
def __init__(self):
self.map = {}
self.word_tree = WordsTree()
self.syntax_tree = SyntaxTree()
self.mistake_map = {}
def down_proj(syntax_tree, sql, relation2, same):
if syntax_tree.op == 'SELECT':
syntax_tree.lfchild = down_proj(syntax_tree.lfchild, sql, relation2,
same)
elif syntax_tree.op == 'PROJECTION':
# 将投影下推
sql = syntax_tree.cond.split(",")
for i in range(len(sql)):
if lookfor(sql[i]) is not None:
if lookfor2(sql[i]) in relation2:
pass
else:
relation2.append(lookfor2(sql[i]))
# 找相交的元素
same += findthesame(relation2[0], relation2[1])
syntax_tree.lfchild = down_proj(syntax_tree.lfchild, sql, relation2,
same)
elif syntax_tree.op == 'JOIN':
# 有JOIN，将条件和投影下推
if len(sql) > 0 and '=' not in sql[0]:
first_tree = SyntaxTree()
first_tree.op = 'PROJECTION'
first_tree.cond = sql[0] + ', ' + same
first_tree.lfchild = syntax_tree.lfchild
syntax_tree.lfchild = first_tree
if len(sql) > 1:
second_tree = SyntaxTree()
second_tree.op = 'PROJECTION'
second_tree.cond = sql[1] + ', ' + same
second_tree.rchild = syntax_tree.rchild
syntax_tree.rchild = second_tree
elif same != '':
second_tree = SyntaxTree()
second_tree.op = 'PROJECTION'
second_tree.cond = same
second_tree.rchild = syntax_tree.rchild
syntax_tree.rchild = second_tree
return syntax_tree
class Robot:
def __init__(self):
self.map = {}
self.word_tree = WordsTree()
self.syntax_tree = SyntaxTree()
self.mistake_map = {}
def trans_gbk(self, string):
list = []
high= 0
for s in string:
if high != 0:
list.append( (high << 8) + ord(s) )
high = 0
else:
if ord(s) > 0x80:
high = ord(s)
else:
list.append( ord(s) )
if high:
print "trans_gbk ERROR. omit"
return list
# def pre_process(self, words):
def trans_utf8(self, string):
list = []
high = 0
for s in string:
if ord(s) < 0x80:
if high:
list.append( high )
high = 0
list.append( ord(s) )
else:
if ord(s) & 0x40:
if high:
list.append( high )
high = ord(s)
else:
high = (high << 8) + ord(s)
if high:
list.append( high )
return list
def from_file(self, filename):
lines=file(filename, 'r').read().split('\n')
a = None
for l in lines:
l = l.strip()
if len(l) == 0:
continue
if l[0] == '[' and l[-1] == ']':
a=l[1:-1]
else:
if a is None:
print "Error:", l
else:
self.add_question(a, l)
a = None
def walk(self, dir):
import os
for root, dirs, files in os.walk(dir):
for filename in files:
self.from_file( os.path.join(root, filename) )
def add_question(self, quest, answer):
self.map[quest] = answer
def analyse(self):
list = []
for word in self.map:
list += word.split()
list = set(list)
for item in list:
# print "add_item", item
self.word_tree.add( self.trans_gbk(item), True )
for key, value in self.map.items():
# print "begin to insert to syntax tree:", key.split(), '---', value
self.syntax_tree.add( map(tuple, map(self.trans_gbk, key.split() ) ), value)
def ask(self, quest):
words = self.split_word(quest)
# print words
return self.syntax_tree.search( map(tuple, words) )
# print self.syntax_tree.syntax_map
def split_word(self, quest):
sentence = self.trans_gbk(quest)
words = []
pos_start = 0
pos_end = pos_start + 1
length = len( sentence )
# print sentence
while pos_start < length:
word = None
pos_end = pos_start + 1
# print "reset pos-end", pos_start, pos_end
pos, target = self.word_tree.search(sentence[ pos_start : pos_end ])
if pos > 0:
# print "found start:", sentence[ pos_start : pos_end ], pos
if target:
word = sentence[pos_start : pos_end]
# print "found match.", word
for i in xrange( pos_end + 1, len( sentence ) + 1, 1 ):
# print "begin to search:", i, len(sentence), sentence[ pos_start : i ]
_pos, _target = self.word_tree.search( sentence[ pos_start : i ] )
if _pos > pos:
# print "re-found start:", sentence[ pos_start : i ], _pos
pos = _pos
if _target:
word = sentence[pos_start : i]
pos_end = i
# print "re-found match.", word
else:
# print "break"
break
if word:
pos_start = pos_end
# print "reset start", pos_start
words.append( word )
else:
pos_start += 1
else:
# print "restart.", pos_start
pos_start += 1
return words
