這次實驗被“過來人”們定位很簡單,實驗內容如下:
-----------------------------------------------------------------------------------
對如下工作進行展開描述
(1) 給出語言的詞法規則描述
· 標識符、關鍵字、整常數、字符常數、浮點常數
· 單界符:+,-,×,;,…
· 雙界符:/*,:=,>=,<=,!=,…
· 註釋
(2) 針對這種單詞的狀態轉換圖和程序框圖
(3) 核心數據結構的設計
如符號表、關鍵字等
(4) 錯誤處理
錯誤的位置及類型等
-----------------------------------------------------------------------------------
這次實驗我是用python寫了一個簡單的C語言的詞法分析器,詞法分析器的編寫有很多種方式,比如可以用正則表達式編寫,也可以用LEX工具自動生成,當然,也可以用比較樸素的方式——基於有窮自動機(Finite Automata,FA),也即基於有窮自動機的狀態轉移來編寫程序。
說起這個有窮自動機(Finite Automata,FA),真心感覺是個好東西,分析問題簡單清晰,而且很直觀。記得對有窮自動機有感性認識是在上學期考試分數並不高的《計算機網絡》課上,全龍哥講那個RDT協議的不同版本的時候,用這個自動機來表明遇到不同情況時發送端和接收端要採取的行動。
自動機的形式化定義:
M=(Q,Σ,δ,q0,F)
„ Q→有窮狀態集
„ Σ→有窮輸入字母表
„ δ→從Q×Σ→2Q的映射函數(2Q是Q的冪集)
„ q0∈Q,是唯一的初態
„ F →終態集合,是Q的子集
這裏說說我個人的理解。對有窮自動機的形式化定義的理解比較重要的是上述第三條說明的理解:δ→從Q×Σ→2Q的映射函數。也即這個δ定義了某個具體FA的狀態間轉移關係,或者說定義了某個FA的狀態間轉移的規則。所謂狀態的冪集就是狀態集Q的所有子集構成的集族。則這句話的字面意思是:狀態集和字母表的笛卡爾乘積到狀態集的冪集的映射函數。
比如:M1 = (Q,Σ,δ,q0,F),其中Q = {q,q0,q1,q2...,qn},又((q,a) , {q1,q2,q3})∈δ,也即δ((q,a)) = {q1,q2,q3}。則說明自動機M1有一個狀態q,q在遇到字母a的時候,自動機狀態可能跳轉到q1,q2,q3三個狀態。自動機又分爲有窮自動機和無窮自動機兩種,這裏不再贅述。
有窮自動機可以用狀態圖直觀表示,例子見下文中圖。
至於詞法分析的一些基本知識,簡單敘述一下:
-----------------------------------------------------------------------------------
定義:
詞法分析器的功能輸入源程序,按照構詞規則分解成一系列單詞符號。單詞是語言中具有獨立意義的最小單位,包括關鍵字、標識符、運算符、界符和常量等
(1) 關鍵字 是由程序語言定義的具有固定意義的標識符。例如,Pascal 中的begin,end,if,while都是保留字。這些字通常不用作一般標識符。
(2) 標識符 用來表示各種名字,如變量名,數組名,過程名等等。
(3) 常數 常數的類型一般有整型、實型、布爾型、文字型等。
(4) 運算符 如+、-、*、/等等。
(5) 界符 如逗號、分號、括號、等等。
輸出:
詞法分析器所輸出單詞符號常常表示成如下的二元式:
(單詞種別,單詞符號的屬性值)
單詞種別通常用整數編碼。標識符一般統歸爲一種。常數則宜按類型(整、實、布爾等)分種。關鍵字可將其全體視爲一種。運算符可採用一符一種的方法。界符一般用一符一種的方法。對於每個單詞符號,除了給出了種別編碼之外,還應給出有關單詞符號的屬性信息。單詞符號的屬性是指單詞符號的特性或特徵。
示例:
比如如下的代碼段:
while(i>=j) i--;
經詞法分析器處理後,它將被轉爲如下的單詞符號序列: <while, _>
<(, _>
<id, 指向i的符號表項的指針>
<>=, _>
<id, 指向j的符號表項的指針>
<), _>
<id, 指向i的符號表項的指針>
<--, _>
<;, _>
詞法分析分析器作爲一個獨立子程序:詞法分析是編譯過程中的一個階段,在語法分析前進行。詞法分析作爲一遍,可以簡化設計,改進編譯效率,增加編譯系統的可移植性。也可以和語法分析結合在一起作爲一遍,由語法分析程序調用詞法分析程序來獲得當前單詞供語法分析使用。
-----------------------------------------------------------------------------------
我寫的這個詞法分析器,不是很健全,尤其是錯誤處理機制,像在字符串識別中,'ab'是C語言中不合法的char變量,但是我的詞法分析器不能判斷出錯誤,會死循環;此外,只能識別出有限的關鍵字、有限形式的字符串(相信讀者看懂我的狀態機就知道哪裏有限了),由於時間不夠了,我不想再改了,下面貼出代碼,供大家參考。
對了,貼代碼之前,先說說我的詞法分析器的狀態機的設計。
我對“數字”的詞法分析用了一個狀態機,包括浮點數、整形數,狀態機如下:
對“字符(串)”的識別用了一個狀態機,包括關鍵字、char、以及char *,如下:
當然,對C語言的註釋的識別也用了一個狀態機,必須先把源碼中的註釋cut掉才能進行分析,如下:
我對運算符的識別(包括雙目和單目)沒有采用明顯的狀態機,都是直接分析判斷的,實際從某種意義上來講對它們的分析也是採用了狀態機的原理,只是狀態機結構比較簡單,就沒再顯式用state表示,它們的狀態機實際上如下:
下面上代碼:
Scanner.py,作爲主模塊來執行:
'''
Created on 2012-10-18
@author: liushuai
'''
import string
import Category
import FileAccess
_currentIndex = 0
_Tokens = []
_prog = ""
_categoryNo = -1
_stateNumber = 0
_stateString = 0
_potentialNumber = ""
_potentialString = ""
def readComments(prog):
'''Read the comments of a program'''
state = 0
currentIndex, beginIndex, endIndex = (0, 0, 0)
commentsIndexs = []
for c in prog:
if state == 0:
if c == '/':
beginIndex = currentIndex
state = 1
else:
pass
elif state == 1:
if c == '*':
state = 2
else :
state = 0
elif state == 2:
if c == '*':
state = 3
else:
pass
elif state == 3:
if c == '*':
pass
elif c == '/':
endIndex = currentIndex
commentsIndexs.append([beginIndex, endIndex])
state = 0 #set 0 state
else:
state = 2
currentIndex += 1
return commentsIndexs
def cutComments(prog, commentsIndexs):
'''cut the comments of the program prog'''
num = len(commentsIndexs)
if num == 0:
return prog
else :
comments = []
for i in xrange(num):
comments.append(prog[commentsIndexs[i][0]:commentsIndexs[i][1] + 1])
for item in comments:
prog = prog.replace(item, "")
return prog
def scan(helper):
'''scan the program, and analysis it'''
global _stateNumber, _stateString, _currentIndex, _Tokens, _prog, _categoryNo, _potentialNumber, _potentialString
currentChar = _prog[_currentIndex]
######################################CHAR STRING######################################
if currentChar == '\'' or currentChar == '\"' or currentChar in string.letters + "_$\\%\@" or (currentChar in string.digits and _stateString != 0):
if _stateString == 0:
if currentChar == '\'':
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 1
_currentIndex += 1
elif currentChar == "\"":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 2
_currentIndex += 1
elif currentChar in string.letters + "$_":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 7
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 1:
if currentChar in string.letters + "#$@%":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 3
_currentIndex += 1
elif currentChar == '\\':
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 9
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 2:
if currentChar in string.letters + "\\% ":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 4
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 3:
if currentChar == '\'':
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 5
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 4:
if currentChar == '\"':
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 6
_currentIndex += 1
elif currentChar in string.letters + "\\% ":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 4
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 7:
if currentChar in string.digits + string.letters + "$_":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 8
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 8:
if currentChar in string.digits + string.letters + "$_":
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 8
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
elif _stateString == 9:
if currentChar in ['b', 'n', 't', '\\', '\'', '\"']:
_potentialString = "%s%s" % (_potentialString, currentChar)
_stateString = 3
_currentIndex += 1
else:
_currentIndex += 1
_stateNumber = 10
else:
_currentIndex += 1
###################################### NUMBERS ######################################
elif currentChar in string.digits + ".":
if _stateNumber == 0:
if currentChar in "123456789":
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 6
_currentIndex += 1
elif currentChar == '0':
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 4
_currentIndex += 1
else:
_stateNumber = 8
_currentIndex += 1
elif _stateNumber == 4:
if currentChar == '.':
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 5
_currentIndex += 1
else:
_stateNumber = 8
_currentIndex += 1
elif _stateNumber == 5:
if currentChar in string.digits:
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 7
_currentIndex += 1
else:
_stateNumber = 8
_currentIndex += 1
elif _stateNumber == 6:
if currentChar in string.digits:
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 6
_currentIndex += 1
elif currentChar == '.':
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 5
_currentIndex += 1
else:
_stateNumber = 8
_currentIndex += 1
elif _stateNumber == 7:
if currentChar in string.digits:
_potentialNumber = "%s%s" % (_potentialNumber, currentChar)
_stateNumber = 7
_currentIndex += 1
else:
_stateNumber = 8
_currentIndex += 1
else:
_currentIndex += 1
######################################OTEAR OPERATERS######################################
else:
if _stateNumber == 6 or _stateNumber == 4:
helper.outPutToken(_potentialNumber, "INT", Category.IdentifierTable["INT"])
elif _stateNumber == 7:
helper.outPutToken(_potentialNumber, "FLOAT", Category.IdentifierTable["FLOAT"])
elif _stateNumber != 0:
helper.outPutToken("ERROR NUMBER", "None", "None")
_stateNumber = 0
_potentialNumber = ""
if _stateString == 7 or _stateString == 8:
if _potentialString in Category.KeyWordsTable:
helper.outPutToken(_potentialString, _potentialString.upper(), Category.IdentifierTable[_potentialString.upper()])
else:
helper.outPutToken(_potentialString, "IDN" , Category.IdentifierTable["IDN"])
helper.setSymbolTable(_potentialString, "IDN" , Category.IdentifierTable["IDN"])
elif _stateString == 5:
helper.outPutToken(_potentialString, "CHAR", Category.IdentifierTable["CHAR"])
elif _stateString == 6:
helper.outPutToken(_potentialString, "CHAR *", Category.IdentifierTable["CHAR *"])
elif _stateString != 0:
helper.outPutToken("ERROR STRING", "None", "None")
_stateString = 0
_potentialString = ""
if currentChar == " ":
_currentIndex += 1
elif currentChar == '>':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == "=":
helper.outPutToken(">=", ">=", Category.IdentifierTable[">="])
_currentIndex += 1
else :
helper.outPutToken(">", ">", Category.IdentifierTable[">"])
elif currentChar == '<':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == "=":
helper.outPutToken("<=", "<=", Category.IdentifierTable["<="])
_currentIndex += 1
else :
helper.outPutToken("<", "<", Category.IdentifierTable["<"])
elif currentChar == '+':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == '+':
helper.outPutToken("++", "++", Category.IdentifierTable["++"])
_currentIndex += 1
else :
helper.outPutToken("+", "+", Category.IdentifierTable["+"])
elif currentChar == '-':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == '-':
helper.outPutToken("--", "--", Category.IdentifierTable["--"])
else:
helper.outPutToken("-", "-", Category.IdentifierTable["-"])
elif currentChar == '=':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == '=':
helper.outPutToken("==", "==", Category.IdentifierTable["=="])
_currentIndex += 1
else :
helper.outPutToken("=", "=", Category.IdentifierTable["="])
elif currentChar == '!':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == '=':
helper.outPutToken("!=", "!=", Category.IdentifierTable["!="])
_currentIndex += 1
else :
helper.outPutToken("!", "!", Category.IdentifierTable["!"])
elif currentChar == '&':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == '&':
helper.outPutToken("&&", "&&", Category.IdentifierTable["&&"])
_currentIndex += 1
else :
helper.outPutToken("&", "&", Category.IdentifierTable["&"])
elif currentChar == '|':
_currentIndex += 1
currentChar = _prog[_currentIndex]
if currentChar == '|':
helper.outPutToken("||", "||", Category.IdentifierTable["||"])
_currentIndex += 1
else :
helper.outPutToken("|", "|", Category.IdentifierTable["||"])
elif currentChar == '*':
helper.outPutToken("*", "*", Category.IdentifierTable["*"])
_currentIndex += 1
elif currentChar == '/':
helper.outPutToken("/", "/", Category.IdentifierTable["/"])
_currentIndex += 1
elif currentChar == ';':
helper.outPutToken(";", ";", Category.IdentifierTable[";"])
_currentIndex += 1
elif currentChar == ",":
helper.outPutToken(",", ",", Category.IdentifierTable[","])
_currentIndex += 1
elif currentChar == '{':
helper.outPutToken("{", "{", Category.IdentifierTable["{"])
_currentIndex += 1
elif currentChar == '}':
helper.outPutToken("}", "}", Category.IdentifierTable["}"])
_currentIndex += 1
elif currentChar == '[':
helper.outPutToken("[", "[", Category.IdentifierTable["["])
_currentIndex += 1
elif currentChar == ']':
helper.outPutToken("]", "]", Category.IdentifierTable["]"])
_currentIndex += 1
elif currentChar == '(':
helper.outPutToken("(", "(", Category.IdentifierTable["("])
_currentIndex += 1
elif currentChar == ')':
helper.outPutToken(")", ")", Category.IdentifierTable[")"])
_currentIndex += 1
if __name__ == '__main__':
helper = FileAccess.FileHelper("H://test.c", "H://token.txt", "H://symbol_table.txt")
prog = helper.readProg()
print prog
comments = readComments(prog)
_prog = cutComments(prog, comments)
print _prog
while _currentIndex < len(_prog):
scan(helper)
helper.closeFiles()
Category.py,這個模塊裏面定義了一些C語言中的關鍵字、運算符等等,是種別碼錶:
'''
Created on 2012-10-18
@author: liushuai
'''
IdentifierTable = {"INT":1,"FLOAT":2,"CHAR":3,"IDN":4,"WHILE":5,"FOR":6,"DO":7,"BREAK":31,"CONTINUE":32,"CHAR *":33,"IF":37,
"*":8,"/":9,"+":10,"-":11,">":12,"<":13,"=":14,
"++":15,"--":16,"==":17,"!=":18,">=":19,"<=":20,
"&&":28,"||":29,"!":30,"&":35,"|":36,
";":21,",":34,
"{":22,"}":23,"[":24,"]":25,"(":26,")":27}
KeyWordsTable = ("int", "float", "char", "while", "for", "do","break","continue","char *","if")
'''
Created on 2012-10-23
@author: liushuai
'''
import Category
class FileHelper(object):
def __init__(self,progPath,tokenPath,symbolTablePath):
self.progPath = progPath
self.tokenPath = tokenPath
self.symbolTablePath = symbolTablePath
self.tokenFp = open(self.tokenPath,"w")
self.symbolTableFp = open(self.symbolTablePath,"w")
self.symbolTable = {}.fromkeys(Category.KeyWordsTable) #initialize symbol table
print self.symbolTable
def readProg(self):
'''read the program into the RAM'''
fp = open(self.progPath, "r+")
prog = ""
for eachLine in fp.readlines():
#print eachLine
prog = "%s%s" % (prog, eachLine.strip())
fp.close()
return prog
def outPutToken(self,tokenSelf,tokenInner,tokenNo):
'''output token into a file'''
self.tokenFp.write("(" + tokenInner + "," + tokenSelf + ")" + "\n")
print "(" + tokenInner + "," + tokenSelf + ")"
def setSymbolTable(self,tokenSelf,tokenInner,tokenNo):
'''output symbol into symbol table'''
if not self.symbolTable.has_key(tokenSelf):
self.symbolTable[tokenSelf] = None
def writeSymbolToFile(self):
for k in self.symbolTable:
self.symbolTableFp.write(k + "\n")
def closeFiles(self):
'''close token Files'''
self.writeSymbolToFile()
if self.tokenFp != None:
self.tokenFp.close()
if self.symbolTableFp != None:
self.symbolTableFp.close()
下面是我的測試C語言程序:
int main ()
{
char str[10000];
int num[30]={0};
char std[28]={"abcdefghijklmnopqrstuvwxyz"};
int i,j,temp;
float test=-0.34;
float test1=23.45;
char tom;
for (i=0;i<1000;i++) /*字符的讀入*/
{
str[i]=getchar();
if (str[i]=='\n')
break;
}
for (i=0;i<10000;i++) /*字符的統計數量*/
{
if (str[i]=='\n')
break;
j=str[i] - 97; /*-97 or sth - 97?*/
num[j]++;
}
for (i=0;i<27;i++) /*字符的按出現頻率排序*/
{
for (j=i+1;j<26;j++)
{
if (num[j]>num[i])
{
temp=num[j];
num[j]=num[i];
num[i]=temp;
tom=std[j];
std[j]=std[i];
std[i]=tom;
}
}
}
for (i=0;i<27;i++) /*字符的按字母表順序排序*/
{
for (j=i+1;j<26;j++)
{
if (num[i]==num[j])
{
if (std[j]<std[i])
{
tom=std[j];
std[j]=std[i];
std[i]=tom;
}
}
}
}
for (i=0;i<29;i++)
{
if (num[i]==0)
break;
}
}
最後腳本輸出如下:
Token.txt:
Symbol_table.txt:
這次實驗就算做完了。
雖然完成的很水,但是正確情況下的輸出還是令人滿意的,詞法分析完事後,等着句法分析調用它的輸出結果吧。對了,至於符號表在整個編譯階段的作用,龍書第一版2.4節和6.7節有介紹,請讀者查閱。
寫的很水很水,如有不足,歡迎指正。