如何检测野牛的转移/减少冲突?

时间:2018-05-01 10:27:40

标签: parsing compilation bison shift-reduce-conflict

我是Bison的新人。 我已经制作了我的.y文件但是在编译中我收到了这个错误:

警告:42次/减少冲突[-Wconflicts-sr]

我已经打开了生成的.output文件,我已经看到了以下内容,但无法理解,无法找出问题所在。

以下是我的.output文件的一部分:

    State 63

   56 expr_decl: KW_NOT expr_decl .  [KW_END, KW_OR, OP_OR, KW_AND, OP_AND, '>', OP_LESSOREQUAL, OP_GREATEROREQUAL, '=', OP_NOTEQUAL, '<', '*', '/', KW_DIV, KW_MOD, ';', ')', '+', '-']
   59          | expr_decl . '*' expr_decl
   60          | expr_decl . '/' expr_decl
   61          | expr_decl . KW_DIV expr_decl
   62          | expr_decl . KW_MOD expr_decl
   63          | expr_decl . '+' expr_decl
   64          | expr_decl . '-' expr_decl
   65          | expr_decl . '=' expr_decl
   66          | expr_decl . '<' expr_decl
   67          | expr_decl . '>' expr_decl
   68          | expr_decl . OP_NOTEQUAL expr_decl
   69          | expr_decl . OP_LESSOREQUAL expr_decl
   70          | expr_decl . OP_GREATEROREQUAL expr_decl
   71          | expr_decl . OP_AND expr_decl
   72          | expr_decl . KW_AND expr_decl
   73          | expr_decl . KW_OR expr_decl
   74          | expr_decl . OP_OR expr_decl

    '+'  shift, and go to state 87
    '-'  shift, and go to state 88

    '+'       [reduce using rule 56 (expr_decl)]
    '-'       [reduce using rule 56 (expr_decl)]
    $default  reduce using rule 56 (expr_decl)

    Conflict between rule 56 and token KW_OR resolved as reduce (KW_OR < KW_NOT).
    Conflict between rule 56 and token OP_OR resolved as reduce (OP_OR < KW_NOT).
    Conflict between rule 56 and token KW_AND resolved as reduce (KW_AND < KW_NOT).
    Conflict between rule 56 and token OP_AND resolved as reduce (OP_AND < KW_NOT).
    Conflict between rule 56 and token '>' resolved as reduce ('>' < KW_NOT).
    Conflict between rule 56 and token OP_LESSOREQUAL resolved as reduce (OP_LESSOREQUAL < KW_NOT).
    Conflict between rule 56 and token OP_GREATEROREQUAL resolved as reduce (OP_GREATEROREQUAL < KW_NOT).
    Conflict between rule 56 and token '=' resolved as reduce ('=' < KW_NOT).
    Conflict between rule 56 and token OP_NOTEQUAL resolved as reduce (OP_NOTEQUAL < KW_NOT).
    Conflict between rule 56 and token '<' resolved as reduce ('<' < KW_NOT).
    Conflict between rule 56 and token '*' resolved as reduce ('*' < KW_NOT).
    Conflict between rule 56 and token '/' resolved as reduce ('/' < KW_NOT).
    Conflict between rule 56 and token KW_DIV resolved as reduce (KW_DIV < KW_NOT).
    Conflict between rule 56 and token KW_MOD resolved as reduce (KW_MOD < KW_NOT).

请帮忙吗?我一直在寻找几个小时但仍然没有...

这是我的声明部分:

%left KW_OR OP_OR
%left KW_AND OP_AND
%left '>' OP_LESSOREQUAL OP_GREATEROREQUAL
%left '=' OP_NOTEQUAL '<'
%left OP_PLUS OP_MINUS
%left '*' '/' KW_DIV KW_MOD
%left cast
%right POSITIVE NEGATIVE
%right KW_NOT '!'

这是我在野牛的规则:

expr_decl : 
        | REAL
        | POSINT
        | '!' expr_decl { $$ = template("!(%s)", $2); }
        | KW_NOT expr_decl { $$ = template("not(%s)", $2); }
        | '+' expr_decl %prec POSITIVE { $$ = template("+(%s)", $2); }
        | '-' expr_decl %prec NEGATIVE { $$ = template("-(%s)", $2); }
        | expr_decl '*' expr_decl { $$ = template("%s * %s", $1, $3); }
        | expr_decl '/' expr_decl { $$ = template("%s / %s", $1, $3); }
        | expr_decl KW_DIV expr_decl { $$ = template("%s div %s", $1, $3); }
        | expr_decl KW_MOD expr_decl { $$ = template("%s mod %s", $1, $3); }
        | expr_decl '+' expr_decl %prec OP_PLUS{ $$ = template("%s + %s", $1, $3); }
        | expr_decl '-' expr_decl %prec OP_MINUS{ $$ = template("%s + %s", $1, $3); }
        | expr_decl '=' expr_decl { $$ = template("%s = %s", $1, $3); }
        | expr_decl '<' expr_decl { $$ = template("%s < %s", $1, $3); }
        | expr_decl '>' expr_decl { $$ = template("%s > %s", $1, $3); }
        | expr_decl OP_NOTEQUAL expr_decl { $$ = template("%s <> %s", $1, $3); }
        | expr_decl OP_LESSOREQUAL expr_decl { $$ = template("%s <= %s", $1, $3); }
        | expr_decl OP_GREATEROREQUAL expr_decl { $$ = template("%s >= %s", $1, $3); }
        | expr_decl OP_AND expr_decl { $$ = template("%s && %s", $1, $3); }
        | expr_decl KW_AND expr_decl { $$ = template("%s and %s", $1, $3); }
            | expr_decl KW_OR expr_decl { $$ = template("%s || %s", $1, $3); }
            | expr_decl OP_OR expr_decl { $$ = template("%s or %s", $1, $3); }
                | '(' expr_decl ')' { $$ = template("(%s)", $2); }
;

1 个答案:

答案 0 :(得分:2)

输出文件的这一部分:

def get_model(input, dropout):
    with tf.name_scope('Conv1'):
        input_4D = tf.reshape(input, [-1, HEIGHT, WIDTH, 1])
        w1 = tf.Variable(tf.truncated_normal([12, 8, 1, 44], stddev=0.01), name='W1')
        b1 = tf.Variable(tf.zeros([44]), name='B1')
        conv1 = tf.nn.conv2d(input_4D, w1, strides=[1, 1, 1, 1], padding='SAME')
        act1 = tf.nn.relu(conv1 + b1)
        drop1 = tf.nn.dropout(act1, dropout)
        max_pool1 = tf.nn.max_pool(drop1, ksize=[1, 2, 2, 1], strides=[1, 2, 2, 1], padding='SAME')
        tf.summary.histogram('weights', w1)
        tf.summary.histogram('biases', b1)
        tf.summary.histogram('activations', act1)
        tf.summary.histogram('dropouts', drop1)

    with tf.name_scope('Conv2'):
        w2 = tf.Variable(tf.truncated_normal([6, 4, 44, 44], stddev=0.01), name='W2')
        b2 = tf.Variable(tf.zeros([44]), name='B2')
        conv2 = tf.nn.conv2d(max_pool1, w2, strides=[1, 1, 1, 1], padding='SAME')
        act2 = tf.nn.relu(conv2 + b2)
        drop2 = tf.nn.dropout(act2, dropout)
        tf.summary.histogram('weights', w2)
        tf.summary.histogram('biases', b2)
        tf.summary.histogram('activations', act2)
        tf.summary.histogram('dropouts', drop2)
        conv_shape = drop2.get_shape()
        count = int(conv_shape[1] * conv_shape[2] * conv_shape[3])
        flat_output = tf.reshape(drop2, [-1, count])

    with tf.name_scope('FC'):
        w3 = tf.Variable(tf.truncated_normal([count, NUM_LABELS], stddev=0.01), name='W3')
        b3 = tf.Variable(tf.zeros([NUM_LABELS]), name='B3')
        fc = tf.add(tf.matmul(flat_output, w3), b3)
        tf.summary.histogram('weights', w3)
        tf.summary.histogram('biases', b3)

    return fc

告诉您在此状态下,在前瞻符号'+' shift, and go to state 87 '-' shift, and go to state 88 '+' [reduce using rule 56 (expr_decl)] '-' [reduce using rule 56 (expr_decl)] '+'上有两个移位/减少冲突。 reduce操作周围的'-' .. [表示该操作已被删除(解决冲突有利于转移)。

消息:

]

告诉您有关语法中优先规则解决的转移/减少冲突(因此不包含在警告中的42个转换/减少冲突中)。

优先规则不能解决Conflict between rule 56 and token KW_OR resolved as reduce (KW_OR < KW_NOT). Conflict between rule 56 and token OP_OR resolved as reduce (OP_OR < KW_NOT). Conflict between rule 56 and token KW_AND resolved as reduce (KW_AND < KW_NOT). : '+'的冲突,因为您没有为这些令牌设置优先级。