临时写了个语法分析的代码，可以顺便输出语法树，以后有时间慢慢解释具体结构吧~
代码：
#include<bits/stdc++.h>
#define xcx(x) printf("ojbk %d\n",(x))
using namespace std;
const int KEY_WORD_NUM = 13 ;
const int OPERATOR_WORD_NUM = 19;
const int DELIMITER_WORD_NUM = 6 ;
const int row_limit = 130;
//词法分析常量
const string _data_type = "data type" ;
const string _statement_keyword = "statement keyword" ;
const string _identifier = "identifier" ;
const string _single_arithmetic_operator = "single arithmetic operator" ;
const string _compare_operator = "compare operator" ;
const string _double_arithmetic_operator = "double arithmetic operator" ;
const string _delimiter = "delimiter" ;
const string _unsigned_number = "unsigned number" ;
const string _const_string = "const string" ;
const string _const_character = "const character";
const string ANNOTATIONL1 = "//" ;
const string ANNOTATIONL2 = "/*" ;
const string key_word [ KEY_WORD_NUM ] = { "break", "if", "else", "for", "while", "do", "int", "double", "char", "read", "write", "bool", "string" } ;
const string operator_word [ OPERATOR_WORD_NUM ]= { "+=", "-=", "*=", "\=", "+", "-", "*", "/", "=", "++", "--", "<", ">", "<=", ">=", "!=", "==", "//", "/*" } ;
const string delimiter_word [ DELIMITER_WORD_NUM ] = { "(", ")", "{", "}", ",", ";"};
//语法分析常量
const string _program = "program" ;
const string _declaration = "declaration" ;
const string _expression = "expression" ;
const string _expression_bool = "expression bool" ;
const string _expression_cal = "expression cal" ;
const string _term = "term" ;
const string _factor = "factor" ;
const string _statement = "statement" ;
const string _statement_if = "if - statement";
const string _statement_while = "while - statement" ;
const string _statement_do_while = "do-while - statement";
const string _statement_for = "for - statement" ;
const string _statement_write = "write - statement" ;
const string _statement_read = "read - statement" ;
const string _statement_compound = "compound statement" ;
const string _statement_expression = "expression statement" ;
const string _constant = "constant" ;
const string _number = "number" ;
struct Node_operator {
string str;
bool is_terminator;
map < char, int > next;
Node_operator(){
str.clear();
is_terminator = false;
next.clear();
}
Node_operator( const string &str ){
this -> next.clear();
this -> str = str;
this -> is_terminator = false;
}
};
struct Word {
string type;
string val;
Word(){
type.clear();
val.clear();
}
Word( const string &str , const string &val ){
this -> type = str;
this -> val = val;
}
void print(){
printf( " %s %s\n", this -> type.c_str(), this -> val.c_str() );
}
bool operator != (const Word &other) const {
return other.type != this->type || other.val != this->val ;
}
bool operator == (const Word &other) const {
return other.type == this->type && other.val == this->val ;
}
};
struct Unit {
vector< Unit > u ;
string val ;
string type;
bool is_error;
Unit(){
is_error = false ;
}
Unit ( const string type ) {
this -> is_error = false ;
this -> type = type;
}
Unit ( const Word &w ) {
this -> is_error = false ;
this -> type = w.type ;
this -> val = w.val ;
u.clear();
}
inline void print() {
printf( " %s ( %s ) " , val.c_str() , type.c_str() ) ;
}
inline void println() {
print();
printf("\n");
}
void push( const Unit &ut ) {
this -> val += ut.val;
this -> u.push_back( ut ) ;
}
void push( const Word &w ) {
this -> val += w.val;
this -> u.push_back( Unit(w) ) ;
}
void clear() {
this -> val.clear();
this -> u.clear();
}
};
struct Node {
Unit x ;
int deep , vl , vr , mid , tl , tr , fa , id;
bool is_level , is_enough_v , is_enough_t ;
Node () {};
Node ( Unit u , int deep , int fa ) {
this -> deep = deep ;
this -> x = u;
if ( u.u.size() == 0 ) this -> is_level = true;
else this -> is_level = false ;
this -> fa = fa ;
this -> x.type = "[" + u.type + "]";
}
int TypeLength() {
return x.type.length();
}
int ValLength() {
return x.val.length();
}
};
int now_line = 1 , error_type = 0;
char ch = 0;
FILE *fp ;
map < string, int > key_word_id;
map < char, int > operator_word_head;
vector < Node_operator > state_graph;
vector < Word > ans ;
string error_information , fin_name , fout_name ;
Word finish , error_word, wd;
Unit error_unit;
char tree_map[300][row_limit+5];
void read(char &c){
if( fscanf( fp , "%c" , &c ) == EOF ) c = EOF;
if( c == '\n' ) now_line++ ;
}
///词法分析识别函数
inline bool Is_Character ( const char &c ) {
return c>='A'&&c<='Z' || c>='a'&&c<='z' || c=='_';
}
inline bool Is_Number( const char &c ) {
return c>='0'&&c<='9' ;
}
inline bool Is_Operator_head ( const char &c ) {
return operator_word_head.count(c) > 0 ;
}
inline bool Is_Data_type ( const string &s ){
return s=="int" || s=="double" || s=="char" || s=="string" || s=="bool" ;
}
inline bool Is_Blank ( const char &c ) {
return c == ' ' || c == '\t' || c == '\n' ;
}
inline bool Is_Delimiter ( const char &c ) {
return c=='(' || c==')' || c=='{' || c=='}' || c==';' || c==',' ;
}
inline bool Is_Compare_operator ( const string &s ) {
return s=="<" || s==">" || s=="<=" || s==">=" || s=="==" || s=="!=" ;
}
void InitTestScan(){
//读取标识符
key_word_id.clear();
for( int i=0; i<KEY_WORD_NUM; i++ ) {
key_word_id[ key_word[i] ] = i;
}
//读取运算符头字母
operator_word_head.clear();
for( int i=0; i<OPERATOR_WORD_NUM; i++ ) {
operator_word_head[ operator_word[i][0] ] = i;
}
//运算符建图
state_graph.clear();
state_graph.push_back( Node_operator() );
for( int i=0; i<OPERATOR_WORD_NUM; i++ ) {
const string &temp = operator_word[i];
int node_index = 0;
for( int j=0; j<(int)temp.length(); j++ ) {
Node_operator &now_node = state_graph[node_index];
if( now_node.next.count( temp[j] ) == 0 ) {
now_node.next[ temp[j] ] = state_graph.size();
state_graph.push_back( Node_operator( now_node.str +temp[j] ) );
node_index = state_graph.size() -1 ;
}
else{
node_index = now_node.next[ temp[j] ];
}
}
state_graph[node_index].is_terminator = true ;
}
//vector<> 清空
ans.clear();
//当前行起点为1
now_line = 1;
//常量终止Word赋值
finish.type.clear(); finish.val.clear();
//error处理信息
error_type = 0 ;
error_information.clear() ;
error_word = Word( string("error"), string("?") );
error_unit.is_error =true; error_unit.type.clear(); error_unit.val.clear();
}
void PrintError( int op );
int InputTestScan(){
printf( "input the name of the source program...\n" );
cin >> fin_name ;
fp = fopen( fin_name.c_str(), "r" );
if( fp == NULL ) return 1;
//else fclose( fp );
//printf( "input the name of the output file...\n" );
//cin >> fout_name ;
//freopen( fin_name.c_str(), "r", stdin ) ;
return 0;
}
void InitGetWord(){
InitTestScan();
read( ch );
}
///语法分析识别函数
Word GetWord(){ //读取一个标识
Word ret = finish ;
while( ch != EOF ){
while( Is_Blank(ch) ){
read( ch );
if( ch == EOF ) return ret;
}
if( Is_Character(ch) == true ) { //处理标识符
string str;
while( Is_Character(ch) == true || Is_Number(ch) == true ) {
str += ch;
read( ch );
}
if( key_word_id.count(str) > 0 ) {
if( Is_Data_type( str ) == true ) {
ret = Word(_data_type, str);
}
else {
ret = Word(_statement_keyword, str);
}
}
else {
ret = Word(_identifier, str) ;
}
}
else if( Is_Number(ch) == true ) { //处理数字常量
string str;
bool is_decimal = false;
while( Is_Number(ch) == true ) {
str +=ch;
read( ch );
}
if( ch == '.' ) {
is_decimal = true;
str +=ch;
read( ch );
while( Is_Number(ch) == true ) {
str +=ch;
read( ch );
}
if( Is_Character(ch) || ch == '.' ){
error_type = 3;
while( !Is_Blank(ch) && !Is_Delimiter(ch) ) {
str += ch;
read( ch );
}
if ( ch == '\n' ) now_line-- ;
error_information = str;
return error_word;
}
}
if( is_decimal == true ) {
if( str.back() == '.' ) {
error_information = str ;
error_type = 4;
return error_word;
}
}
ret = Word( _unsigned_number, str ) ;
}
else if ( ch == '\'' ) { //处理单引号
string str;
str +=ch;
read( ch );
while( ch!='\'' && ch!='\n' ) {
if( ch == '\\' ) {
str +=ch;
read( ch );
}
str +=ch;
read( ch );
}
if( ch != '\'' ) {
error_type = 5;
error_information = str;
return error_word;
}
str +=ch;
ret = Word(_const_character, str) ;
read( ch );
if( ret.val.length() > 3 && ret.val[1] !='\\' ) {
error_type = 6;
error_information = str;
return error_word;
}
}
else if ( ch == '"' ) { //处理双引号
string str;
str +=ch;
read( ch );
while( ch!='"' && ch!='\n' ) {
if( ch == '\\' ) {
str +=ch;
read( ch );
}
str +=ch;
read( ch );
}
if( ch != '"' ) {
error_type = 7;
error_information = str;
return error_word;
}
str +=ch;
ret = Word( _const_string, str ) ;
read( ch );
}
else if ( Is_Operator_head(ch) == true ) { //处理运算符
string str ;
int node_index = 0;
while( state_graph[node_index].next.count(ch) > 0 ) {
node_index = state_graph[node_index].next[ch] ;
read( ch );
while( Is_Blank(ch) == true ) read( ch ) ;
}
Node_operator &now_node = state_graph[node_index];
if( now_node.str == ANNOTATIONL1 ) { //处理注释1
while( ch != '\n' ) read( ch );
read( ch );
}
else if( now_node.str == ANNOTATIONL2 ) { //处理注释2
char pre_ch = ch;
read( ch );
while( !(pre_ch == '*' && ch == '/') ) {
pre_ch = ch;
read( ch );
}
read( ch );
}
else if( state_graph[node_index].is_terminator == false ) { //运算符不合法
error_information = state_graph[node_index].str ;
error_type = 8 ;
return error_word;
}
else { //合法运算符
str = now_node.str;
if( str == "++" || str == "--" ) {
ret = Word( _single_arithmetic_operator , str) ;
}
else if( Is_Compare_operator(str) == true ) {
ret = Word( _compare_operator , str) ;
}
else {
ret = Word( _double_arithmetic_operator , str ) ;
}
}
}
else if( Is_Delimiter(ch) == true ) { //处理分界符
string str ;
str = ch;
read( ch ) ;
ret = Word (_delimiter , str );
}
else{ //不合法字符
error_information = ch;
while( Is_Blank(ch) == false ) {
error_information += ch ;
read( ch );
}
error_type = 9;
return error_word;
}
if( error_type != 0 || ( ret.type.length() != 0 || ret.val.length() != 0 ) ) return ret;
}
}
Unit GetProgram() ;
Unit GetDeclaration() ;
Unit GetExpression() ;
Unit GetExpressionBool() ;
Unit GetExpressionCal() ;
Unit GetTerm() ;
Unit GetFactor() ;
Unit GetStatement() ;
Unit GetStatementIf() ;
Unit GetStatementWhile() ;
Unit GetStatementDoWhile() ;
Unit GetStatementFor() ;
Unit GetStatementWrite() ;
Unit GetStatementRead() ;
Unit GetStatementCompound() ;
Unit GetStatementExpression() ;
Unit GetConstant() ;
Unit GetNumber() ;
void PrintTree ( Unit &root ) ; // 想不想看看怎么秀操作
vector < Node > Bfs( Unit &s );
void test(){
fin_name = "in.txt" ;
fp = fopen( fin_name.c_str(), "r" );
InitGetWord();
Unit ans = GetProgram();
PrintError( error_type );
PrintTree ( ans );
/*while( ch != EOF ){
ans.push_back( GetWord() );
ans[ans.size()-1].print();
if( error_type != 0 ) {
printf("error!");
//PrintError( error_type );
error_type = 0;
error_information.clear();
}
}*/
}
int TestScan(){
//int ret = InputTestScan();
//if( ret != 0 ) return ret;
InitGetWord();
while( ch != EOF ){
ans.push_back( GetWord() );
if( error_type != 0 ) {
//PrintError( error_type );
error_type = 0;
error_information.clear();
}
}
return 0;
}
void OutPut(){
freopen( fout_name.c_str(), "w", stdout );
for( int i=0; i<ans.size(); i++ ) {
ans[i].print();
}
/*printf( "----------------------------------------------------------------\n" );*/
}
int main()
{
test(); return 0;
int flag = TestScan();
if( flag == 0 ) {
OutPut();
freopen( "CON", "w", stdout );
printf( "\nLexical analysis completed, the results have been saved in the \"%s\" file !\n", fout_name.c_str() );
}
else {
freopen( "CON", "w", stdout );
}
return 0;
}
Unit GetProgram() {
Unit ret = Unit( _program ) , add ;
wd = GetWord();
if ( wd == error_word ) return error_unit;
while( wd != finish ) {
if( wd.type == _data_type ) { // 声明语句
add = GetDeclaration();
}
else { // 语句
add = GetStatement() ;
}
if ( add.is_error == true ) {
return error_unit ;
}
else {
ret.push ( add ) ;
}
}
return ret;
}
Unit GetDeclaration() {
Unit ret = Unit( _declaration ) , add ;
ret.push ( wd ) ;
wd = GetWord();
if ( wd == error_word ) return error_unit ;
if ( wd.type != _identifier ) { // <标识符>
error_type = 10 ;
error_information = ret.val ;
return error_unit ;
}
else {
ret.push ( wd ) ;
wd = GetWord();
if ( wd == error_word ) return error_unit ;
}
if ( wd.val == "=" ) { // = <表达式>
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetExpression();
if( add.is_error == true ){
return error_unit ;
}
else {
ret.push ( add ) ;
}
}
while( wd.val == "," ) { // 循环添加 , <标识符>
ret.push ( wd ) ;
wd = GetWord();
if ( wd == error_word ) return error_unit ;
if ( wd.type != _identifier ) {
error_type = 10 ;
error_information = "," ;
return error_unit ;
}
else {
ret.push ( wd ) ;
wd = GetWord();
if ( wd == error_word ) return error_unit ;
}
if( wd.val == "=" ) {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetExpression();
if( add.is_error == true ){
return error_unit;
}
else {
ret.push ( add ) ;
}
}
}
if( wd.val != ";" ){ // 处理分号
error_type = 11 ;
error_information = ret.val ;
return error_unit;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
return ret;
}
Unit GetExpression() {
Unit ret = Unit ( _expression ) , add ;
if ( wd.type != _identifier ) { // bool表达式
add = GetExpressionBool();
if ( add.is_error == true ) {
return error_unit ;
}
else {
ret.push ( add ) ;
}
}
else{
long fileadd = ftell( fp ) ; // 记录当前位置
char rem_ch = ch ; // 记录当前ch
Word rem_wd = wd; // 记录当前wd
int rem_now_line = now_line ; // 记录当前行数
//printf("rem: %c",ch); wd.print();
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val == "=" || wd.val == "+=" || wd.val == "-=" || wd.val == "\=" || wd.val == "*=" ) { // 赋值表达式
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetExpressionBool();
if ( add.is_error == true ){
return error_unit;
}
else{
ret.push ( add ) ;
}
}
else { //bool 表达式
fseek( fp , fileadd , 0 );// 返回记录文件位置
ch = rem_ch ;
wd = rem_wd;
ret.clear();
now_line = rem_now_line ;
//printf("back : %c",ch); wd.print();
add = GetExpressionBool();
if ( add.is_error == true ) {
return error_unit ;
}
else {
ret.push ( add ) ;
}
}
}
return ret;
}
Unit GetExpressionBool() {
Unit ret = Unit ( _expression_bool ) , add ;
add = GetExpressionCal() ;
if ( add.is_error == true ) {
return error_unit;
}
else {
ret.push ( add ) ;
}
if ( wd.type == _compare_operator ){
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetExpressionCal() ;
if ( add.is_error == true ) {
return error_unit;
}
else {
ret.push ( add ) ;
}
}
return ret;
}
Unit GetExpressionCal() {
Unit ret = Unit ( _expression_cal ) , add ;
add = GetTerm() ;
if ( add.is_error == true ) {
return error_unit;
}
else {
ret.push ( add ) ;
if ( wd.val == "+" || wd.val == "-" ) {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetTerm() ;
if ( add.is_error == true ) {
return error_unit;
}
else {
ret.push ( add ) ;
}
}
}
return ret;
}
Unit GetTerm() {
Unit ret = Unit( _term ) , add ;
add = GetFactor() ;
if ( add.is_error == true ) {
return error_unit ;
}
else {
ret.push ( add ) ;
if ( wd.val == "*" || wd.val == "/" ) {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetTerm() ;
if ( add.is_error == true ) {
return error_unit;
}
else {
ret.push ( add ) ;
}
}
}
return ret;
}
Unit GetFactor() {
Unit ret = Unit ( _factor ) , add ;
if ( wd.val == "(" ) { //表达式
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetExpression() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push( add ) ;
}
if ( wd.val != ")" ) {
error_type = 13 ;
error_information = ret.val;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
}
else if ( wd.type == _identifier ) { //标识符
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.type == _single_arithmetic_operator ) { // 自运算
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
}
else if ( wd.val == "-" || wd.type == _unsigned_number ) { //数字常量
if( wd.type == _unsigned_number ) {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.type == _unsigned_number ) {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
else {
error_type = 13 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
}
}
else if ( wd.type == _const_character || wd.type == _const_string ) { //字符或字符串常量
ret.push ( wd ) ;
wd = GetWord();
if ( wd == error_word ) return error_unit ;
}
else { //不合法factor
error_type = 13 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
return ret;
}
Unit GetStatement() {
Unit ret = Unit( _statement ) , add ;
if ( wd.type == _statement_keyword ) {
if( wd.val == "if" ) { // if-statement
add = GetStatementIf() ;
}
else if( wd.val == "while" ) { // while-statement
add = GetStatementWhile() ;
}
else if( wd.val == "do" ) { // do_while-statement
add = GetStatementDoWhile() ;
}
else if( wd.val == "for" ) { // for-statement
add = GetStatementFor() ;
}
else if( wd.val == "write" ) { // write-statement
add = GetStatementWrite() ;
}
else if( wd.val == "read" ) { // read-statement
add = GetStatementRead() ;
}
else { // break 语句
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val != ";" ) {
error_type = 11 ;
error_information = ret.val ;
return error_unit ;
}
else {
ret.push ( wd ) ;
wd = GetWord();
if ( wd == error_word ) return error_unit ;
}
return ret;
}
}
else {
if( wd.val == "{" ) { //复合语句
add = GetStatementCompound() ;
}
else { //表达式语句
add = GetStatementExpression() ;
}
}
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
return ret;
}
Unit GetStatementIf() {
Unit ret = Unit( _statement_if ) , add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val != "(" ) {
error_type = 15 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetExpression() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
if ( wd.val != ")" ) {
error_type = 13 ;
error_information = add.val ;
ret.is_error = true;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetStatement();
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
if ( wd.val == "else" ) { //else 语句
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetStatement() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
}
return ret;
}
Unit GetStatementWhile() {
Unit ret = Unit( _statement_while ) , add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val != "(" ) {
error_type = 13 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetExpression() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
if ( wd.val != ")" ) {
error_type = 15 ;
error_information = add.val ;
ret.is_error = true ;
return ret ;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetStatement() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret ;
}
else {
ret.push ( add ) ;
}
return ret;
}
Unit GetStatementDoWhile() {
Unit ret = Unit( _statement_do_while ) , add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
add = GetStatement() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret ;
}
else {
ret.push ( add ) ;
}
if ( wd.val != "while" ) {
error_type = 18 ;
error_information = ret.val ;
ret.is_error = true ;
return ret ;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
if( wd.val != "(" ) { // 左括号
error_type = 15;
error_information = ret.val;
return error_unit ;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit;
}
add = GetExpressionBool() ;
if ( add.is_error == true ) {
ret.is_error = true ;
return ret ;
}
else {
ret.push ( add ) ;
}
if( wd.val != ")" ) { // 右括号
error_type = 13;
error_information = ret.val;
return error_unit ;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit;
}
if ( wd.val != ";" ) {
error_type = 11 ;
error_information = add.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
return ret;
}
Unit GetStatementFor() {
Unit ret = Unit( _statement_for ), add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val != "(" ) {
error_type = 15 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetExpression() ; // 表达式一
if ( add.is_error == true ) {
ret.is_error = true ;
return ret ;
}
else {
ret.push ( add ) ;
}
if ( wd.val != ";" ) { // 分号一
error_type = 11;
error_information = add.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetExpression() ; // 表达式二
if ( add.is_error == true ) {
ret.is_error = true ;
return ret ;
}
else {
ret.push ( add ) ;
}
if ( wd.val != ";" ) { //分号二
error_type = 11;
error_information = add.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetExpression() ; // 表达式三
if ( add.is_error == true ) {
ret.is_error = true ;
return ret ;
}
else {
ret.push ( add ) ;
}
if ( wd.val != ")" ) { // 右括号
error_type = 13 ;
error_information = add.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetStatement() ; // 循环语句
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
return ret ;
}
Unit GetStatementWrite() {
Unit ret = Unit( _statement_write ) , add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val != "(" ) {
error_type = 15 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
add = GetExpressionCal() ;//算术表达式
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
if ( wd.val != ")" ) {
error_type = 13 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
if ( wd.val != ";" ) {
error_type = 11 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
return ret;
}
Unit GetStatementRead() {
Unit ret = Unit( _statement_read ) , add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
if ( wd.val != "(" ) {
error_type = 15 ;
error_information = ret.val ;
ret.is_error = true ;
return ret ;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
if ( wd.type != _identifier ) {
error_type = 17 ;
error_information = wd.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
if ( wd.val != ")" ) {
error_type = 13 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
if ( wd.val != ";" ) {
error_type = 11 ;
error_information = ret.val ;
ret.is_error = true ;
return ret;
}
else {
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
}
return ret;
}
Unit GetStatementCompound() {
Unit ret = Unit( _statement_compound ), add ;
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
while( wd.val != "}" ) {
add = GetStatement();
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
}
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
return ret ;
}
Unit GetStatementExpression() {
Unit ret = Unit( _statement_expression ) , add ;
while( wd.val != ";" ) {
add = GetExpression();
if ( add.is_error == true ) {
ret.is_error = true ;
return ret;
}
else {
ret.push ( add ) ;
}
}
ret.push ( wd ) ;
wd = GetWord() ;
if ( wd == error_word ) return error_unit ;
return ret;
}
void PrintError( int op ){
///文件操作相关
if( op == 1 ) { //输入文件打开失败
printf( "return = %d ERROR: fail to open the source program !\n" , op );
}
if( op == 2 ) { //输出文件不存在， 理论不可能发生
printf( "return = %d ERROR: fail to open the output file !\n" , op );
}
///词法分析相关
if( op == 3 ) { //数字常量不合法
printf( "return = %d ERROR: %s is not a number in line %d !\n" , op , error_information.c_str() , now_line );
}
if( op == 4 ) { //小数点后缺少数字
printf( "return = %d ERROR: Missing Numbers behind the decimal point in line %d !\n" , op , now_line );
}
if( op == 5 ) { //单引号不匹配
printf( "return = %d ERROR: missing terminating \' behind %s character in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 6 ) { // 字符常量超长
printf( "return = %d ERROR: Character constants %s are excessively long in line %d !\n", error_information.c_str() , now_line );
}
if( op == 7 ) { //双引号不匹配
printf( "return = %d ERROR: missing terminating \" behind %s character in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 8 ) { //运算符不合法
printf( "return = %d ERROR: the operator %s is not valid in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 9 ) { //非合法字符
printf( "return = %d ERROR: %s was not legal character in line %d !\n", op , error_information.c_str() , now_line );
}
///语法分析相关
if( op == 10 ) { //缺少标识符
printf( "return = %d ERROR: miss declaration after %s in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 11 ) { //缺少分号
printf( "return = %d ERROR: miss ';' after %s in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 12 ) { //不是合法表达式
printf( "return = %d ERROR: %s is not a expression in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 13 ) { //缺少右括号
printf( "return = %d ERROR: miss ')' after %s in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 14 ) { //不是因子
printf( "return = %d ERROR: %s is not a factor in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 15 ) { //缺少左括号
printf( "return = %d ERROR: miss '(' before %s in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 16 ) { //write括号中信息不合法
printf( "return = %d ERROR: %s is not a declaration in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 17 ) { //read括号中信息不合法
printf( "return = %d ERROR: %s is not a expression in line %d !\n", op , error_information.c_str() , now_line );
}
if( op == 18 ) { //缺少while
printf( "return = %d ERROR: miss while after 'do' in line %d !\n", op , now_line );
}
}
vector < Node > Bfs( Unit &s ) {
queue < Node > q ;
vector < Node > ret ;
q.push( Node( s , 0 , -1 ) ) ;
while( !q.empty() ) {
Node now = q.front() ;
ret.push_back( now ) ;
q.pop();
for ( int i=0; i<now.x.u.size(); i++ ) {
q.push ( Node( now.x.u[i] , now.deep+1 , ret.size()-1 ) ) ;
}
}
/*for ( int i=0; i<ret.size(); i++ ) { printf("%d ",ret[i].deep); ret[i].x.println(); }*/
return ret;
}
void draw( int row_num = 100 ) {
for ( int i=0; i<row_num; i++ ) {
for ( int j=0; j<row_limit; j++ ) {
if ( tree_map[i][j] != 0 )
printf( "%c" , tree_map[i][j] ) ;
else {
printf(" ");
}
}
printf("\n");
}
}
void PrintTree ( Unit &root ) {
///build tree sequence
vector < Node > temp = Bfs ( root );
vector < Node > e_vector ;
vector< vector < Node > > a ;
for ( int i=0; i<temp.size(); i++ ) {
temp[i].id = i;
if ( temp[i].deep >= a.size() ) {
a.push_back( e_vector );
}
a[ temp[i].deep ].push_back( temp[i] ) ;
}
for ( int i=0; i<a.size(); i++ ){
int len = row_limit / a[i].size();
int limit = len - 2;
for( int j=0; j<a[i].size(); j++ ) {
Node &now = a[i][j] ;
now.mid = j*len + len/2 ;
if ( now.TypeLength() > limit ) {
now.tl = j*len +1;
now.tr = now.tl + limit -1 ;
now.is_enough_t = false ;
}
else {
now.tl = now.mid - now.TypeLength()/2 ;
now.tr = now.tl + now.TypeLength() - 1 ;
now.is_enough_t = true ;
}
if ( now.ValLength() > limit ) {
now.vl = j*len +1;
now.vr = now.vl + limit -1 ;
now.is_enough_v = false ;
}
else {
now.vl = now.mid - now.ValLength()/2 ;
now.vr = now.vl + now.ValLength() - 1 ;
now.is_enough_v = true ;
}
}
}
for ( int i=0; i<a.size(); i++ ) {
for ( int j=0; j<a[i].size(); j++ ) {
temp[ a[i][j].id ] = a[i][j];
//printf("%d %d %d \n",i,a[i][j].id ,temp[a[i][j].id].mid);
}
}
///paint tree map
memset ( tree_map, 0, sizeof(tree_map) ) ;
int line_pos = 0 , pre_l = row_limit , pre_r = -1 ;
for( int i=0; i<a.size(); i++ ) {
if ( i!=0 ) {
for ( int j=0; j<a[i].size(); j++ ) {// 划横线
if ( j==0 || a[i][j].fa!=a[i][j-1].fa ) {
pre_l = min( a[i][j].mid , temp[a[i][j].fa].mid ) ;
}
if ( j==a[i].size()-1 || a[i][j].fa!=a[i][j+1].fa ) {
pre_r = max( a[i][j].mid , temp[a[i][j].fa].mid ) ;
//printf("%d %d \n",pre_l,pre_r);
for ( int k=pre_l; k<=pre_r; k++ ) {
tree_map [ line_pos ] [ k ] = '-' ;
}
}
}
line_pos++ ;
for ( int j=0; j<a[i].size(); j++ ) { // 划竖线
tree_map [ line_pos ] [ a[i][j].mid ] = '|';
}
line_pos++ ;
}
for ( int j=0; j<a[i].size(); j++ ) {// 输出 Val
Node &now = a[i][j] ;
if ( now.is_enough_v == false ) {
for ( int k=now.vl, p=0; k<=now.vr; k++, p++ ) {
tree_map[line_pos][k] = now.x.val[p] ;
}
for ( int k=now.vr+1; k<=now.vr+3; k++ ) {
tree_map[line_pos][k] = '.' ;
}
}
else {
for ( int k=now.vl, p=0 ; p<now.x.val.length(); k++, p++ ) {
tree_map[line_pos][k] = now.x.val[p] ;
}
}
}
line_pos++ ;
for ( int j=0; j<a[i].size(); j++ ) {// 输出 Type
Node &now = a[i][j] ;
if ( now.is_enough_t == false ) {
for ( int k=now.tl, p=0; k<=now.tr; k++, p++ ) {
tree_map[line_pos][k] = now.x.type[p] ;
}
for ( int k=now.tr+1; k<=now.tr+3; k++ ) {
tree_map[line_pos][k] = '.' ;
}
}
else {
for ( int k=now.tl, p=0 ; p<now.x.type.length(); k++, p++ ) {
tree_map[line_pos][k] = now.x.type[p] ;
}
}
}
line_pos++ ;
for ( int j=0; j<a[i].size(); j++ ) { // 划竖线（后）
if ( a[i][j].is_level == false )
tree_map [ line_pos ] [ a[i][j].mid ] = '|';
}
line_pos++ ;
}
draw ( line_pos ) ;
}
测试数据
//合法输入：
//www*= a++;
k += c++ -2*x*-3.4;
/*
write(-3.2*(i-3)*(j-4));
int a , b , c = 12 , d = "sadas" ;
*/
/*
while(t--){
if(3==5-2) write(-3.0*x*y+w*2);
}
*/
/*k = c++ ;
write(-3*i*(j-4));
9.0.0.2.
double c = -6.2;
for( i=-2 ; i<=w ; i++ ) {
for( j = 3.8; j<i ;j ++) {
read(d);
write(-3*i*(j-4));
}
}
c++;
b = a++;
int a , b , c = 12 , d = "sadas" ;
int a;
b = a++;
if( c == d ) {
c = w;
}
while( w-- != k*2.2 );
while ( T-- < = 2 ) {
t=t+1;
}
read(w);
write(432);
write("sds");
write(4/sd+sd*4);
do{
i++;
if(k!=2) break;
}while(23+3==s);
*/
运行截图：
k+=c++-2*x*-3.4;
[program]
|
-
|
k+=c++-2*x*-3.4;
[statement]
|
-
|
k+=c++-2*x*-3.4;
[expression statement]
|
------------------------------------------------------------------
| |
k+=c++-2*x*-3.4 ;
[expression] [delimiter]
|
---------------------------------------------------------------------------------------
| | |
k += c++-2*x*-3.4
[identifier] [double arithmetic operator] [expression bool]
|
-------------------------------------------
|
c++-2*x*-3.4
[expression cal]
|
---------------------------------------------------------------------------------------
| | |
c++ - 2*x*-3.4
[term] [double arithmetic operator] [term]
| |
------ -----------------------------------------------------------------
| | | |
c++ 2 * x*-3.4
[factor] [factor] [double arithmetic operator] [term]
| | |
---------------------- ----- -------------------------------------------
| | | | | |
c ++ 2 x * -3.4
[identifier] [single arithmetic ...[unsigned number] [factor] [double arithmetic ... [term]
| |
------------------------------------------ -------------------
| |
x -3.4
[identifier] [factor]
|
------------------------------------------------------------------
| |
- 3.4
[double arithmetic operator] [unsigned number]
京公网安备 11010502036488号