统计文件件列表中包含单词的个数的 python 脚本

 

import sys;
import string;

def SearchWord(strFile, strWord):
    print("Search file: ", strFile, " word:", strWord);
    nCount = 0;
    nLine = 0;
    try:
        hFile = open(strFile, "r");
        try:
            while True:
                strLine = hFile.readline();                
                if strLine:
                    #print(strLine);
                    nLineCount = FindStrCount(strLine, strWord);
                    nLine = nLine + 1;
                    print("line ", nLine, " count:", nLineCount);
                    nCount = nCount + nLineCount;
                else:
                    break;
        finally:
            hFile.close();
    except IOError:
        print("open ", strFile , " failed!");

    return nCount;    

def FindStrCount(strLine, strWord):
    nFind = 0;
    nIndex = 0;
    
    while True:        
        nIndex = strLine.find(strWord, nIndex);

        if nIndex < 0:
            break;

        nFind = nFind + 1;
        nIndex = nIndex + 1;

    return nFind;    

if __name__ == "__main__":
    #test for find sub string numbers
    #str = "aa bb cc dd ee ee bb aa aa dd";
    #print(FindStrCount(str, "bb"));
        
    #test for count words
    #print(SearchWord("c:\\c.txt", "=>"));
    #sys.exit();
    
    if 1 == len(sys.argv) or sys.argv[1] in ("-h", "--help"):
        print("usage: wc.py word file1 file2 ... filen");
        sys.exit();

    #count word
    ls = [];
    i = 0;
    for i in range(0, len(sys.argv)):
        if i > 1:
            ls.append(sys.argv[i]);
        elif 1 == i:
            strWord = sys.argv[i];
        else:
            print("work py file:", sys.argv[i]);

    print("search word:", strWord);

    for strFile in ls:
        print("parse file:", strFile);
        print("find ", strWord, " count:", SearchWord(strFile, strWord));