自己动手开发编译器特别篇——用词法分析器解决背诵圣经问题

Imports VBF.Compilers.Scanners
Imports VBF.Compilers.Scanners.RegularExpression
Imports System.IO

Module Program

Sub Main(args As String())
Dim findword = args(0)

Dim bibleLexicon As New Lexicon()
Dim lex = bibleLexicon.DefaultLexer

'定义要寻找单词的词法
Dim TARGET = lex.DefineToken(Literal(findword))
'定义一般单词的词法
Dim WORD = lex.DefineToken((Range("0"c, "9"c) Or
Range("a"c, "z"c) Or
Range("A"c, "Z"c)).Many1)
'定义换行
Dim LF = lex.DefineToken(Symbol(vbLf) Or Literal(vbCrLf))
'定义其他所有符号均忽略
Dim OTHER = lex.DefineToken(Range(ChrW(0), ChrW(255)))

Dim bibleScanner As New PeekableScanner(bibleLexicon.CreateScannerInfo())
bibleScanner.SetSkipTokens(OTHER.Index)

Using sr As New StreamReader("bible.txt")
Dim source As New SourceReader(sr)
bibleScanner.SetSource(source)

Dim scannerWatch As New Stopwatch

Dim lines = 1, columns = 1, totalwords = 0, targetwords = 0
scannerWatch.Start()
Do While bibleScanner.Peek() <> bibleScanner.ScannerInfo.EndOfStreamTokenIndex
Dim x As Lexeme = bibleScanner.Read()

Select Case x.TokenIndex
Case TARGET.Index
Console.WriteLine("第{0}行,第{1}列", lines, columns, x.Value)
columns += 1
targetwords += 1
totalwords += 1
Case WORD.Index
columns += 1
totalwords += 1
Case LF.Index
lines += 1
columns = 1
End Select
Loop
scannerWatch.Stop()

Console.WriteLine("总单词数: " & totalwords)
Console.WriteLine("目标单词出现次数: " & targetwords)
Console.WriteLine("消耗时间: " & scannerWatch.ElapsedMilliseconds)
End Using
End Sub

End Module

第5769行,第29列
第14112行,第8列
第16578行,第14列
第17558行,第8列
第17646行,第25列
第20351行,第34列
第22304行,第23列
第22908行,第31列