web_parser.py

web parser的範例程式

               #if PAL_count != 0:
                   #PAL_count = PAL_count - 1
                outPutFile.write("Freq.Tp:")
                outPutFile.write(PAL_dic["Freq.Tp"])
                outPutFile.write("  ")
                outPutFile.write("Provider Name Channel Name:")
                outPutFile.write(PAL_dic["Provider Name Channel Name"])
                outPutFile.write("  ") 
                outPutFile.write("Video Encryption:")
                outPutFile.write(PAL_dic["Video Encryption"])
                outPutFile.write("  ") 
                outPutFile.write("Audio:")
                outPutFile.write(PAL_dic["Audio"])
                outPutFile.write("\n")  
            elif List.count("PAL") == 0 and rowspan_count != rowspan_ori and PAL_count != 0:
                PAL_dic["Provider Name Channel Name"] = List[2]
                PAL_dic["Audio"] = List[5] 
               # print PAL_dic
               # raw_input('List.count("PAL") == 0 and rowspan_count != rowspan_ori and PAL_count != 0')
                if rowspan_count != 0:
                    rowspan_count = rowspan_count - 1
                #if PAL_count != 0:
                #    PAL_count = PAL_count - 1
                outPutFile.write("Freq.Tp:")
                outPutFile.write(PAL_dic["Freq.Tp"])
                outPutFile.write("  ")
                outPutFile.write("Provider Name Channel Name:")
                outPutFile.write(PAL_dic["Provider Name Channel Name"])
                outPutFile.write("  ") 
                outPutFile.write("Video Encryption:")
                outPutFile.write(PAL_dic["Video Encryption"])
                outPutFile.write("  ") 
                outPutFile.write("Audio:")
                outPutFile.write(PAL_dic["Audio"])
                outPutFile.write("\n")  
             #print List[5]
             #raw_input('List[5]')
             #if List.count("DVB") > 0 and rowspan_count == rowspan_ori:
            if len(List) >= 6:    
                if List[5].count("DVB") > 0 and rowspan_count == rowspan_ori: 
                    PAL_count = 0
                    DVB_dic_1["Freq.Tp"] = List[1]
                    DVB_dic_1["Provider Name Channel Name"] = List[3]
                    DVB_dic_1["SR-FEC"] = List[6]
                    DVB_dic_1["NID-TID"] = List[7]  
                    if List[5].count("DVB-S2  -  8PSK") > 0:
                        print List
                        DVB_dic_1["Video Encryption"] = "DVB-S2-8PSK"
                    elif List[5].count("DVB-S2") > 0:
                        print List
                        DVB_dic_1["Video Encryption"] = "DVB-S2"
                   # print DVB_dic_1
                   # raw_input('List[5].count("DVB") > 0 and rowspan_count == rowspan_ori')
                    if rowspan_count != 0:
                        rowspan_count = rowspan_count - 1
                    outPutFile.write("Freq.Tp:")
                    outPutFile.write(DVB_dic_1["Freq.Tp"])
                    outPutFile.write("  ")
                    outPutFile.write("Provider Name Channel Name:")
                    outPutFile.write(DVB_dic_1["Provider Name Channel Name"])
                    outPutFile.write("  ") 
                    outPutFile.write("Video Encryption:")
                    outPutFile.write(DVB_dic_1["Video Encryption"])
                    outPutFile.write("  ") 
                    outPutFile.write("SR-FEC:")
                    outPutFile.write(DVB_dic_1["SR-FEC"])
                    outPutFile.write("  ")  
                    outPutFile.write("NID-TID:")
                    outPutFile.write(DVB_dic_1["NID-TID"])
                    outPutFile.write("\n")  
                    
                    freq = DVB_dic_1["Freq.Tp"][:7]
                    fileDatabase.write(freq) 
                    fileDatabase.write(" ")
                    fileDatabase.write(DVB_dic_1["Video Encryption"])
                    fileDatabase.write(" ")
                    sr_fec = string.replace(DVB_dic_1["SR-FEC"],"-"," ")
                    if len(sr_fec) > 9:
                        if sr_fec[9] != " ":
                            sr_fec = sr_fec[0:10]
                        else:
                            sr_fec = sr_fec[0:9]
                    else:
                        sr_fec = sr_fec[0:9]       
                    fileDatabase.write(sr_fec)   
                    fileDatabase.write("\n")
                    
                    
                    resultFile.write(freq) 
                    resultFile.write(" ")
                    resultFile.write(DVB_dic_1["Video Encryption"])
                    resultFile.write(" ")
                    resultFile.write(sr_fec)   
                    resultFile.write("\n")

                     
                   #TotalList.append(DVB_dic_1) 
                elif List.count("DVB") == 0 and List.count("PAL") == 0 and len(List) == 7:    
                #elif List.count("DVB") == 0 and rowspan_count != rowspan_ori:
                #elif rowspan_count != rowspan_ori:    
                    DVB_dic_2["Provider Name Channel Name"] = List[2]
                    DVB_dic_2["SID-VPID"] = List[5]
                    DVB_dic_2["Audio"] = List[6] 
                    if List[4].count("MPEG-4/HD") > 0:
                        DVB_dic_2["Video Encryption"] = "DVB-S2"
                   # print DVB_dic_2
                   # raw_input('List.count("DVB") == 0 and len(List) == 7 ')
                    if rowspan_count != 0:
                        rowspan_count = rowspan_count - 1
                    outPutFile.write("Freq.Tp:")
                    outPutFile.write(DVB_dic_2["Freq.Tp"])
                    outPutFile.write("  ")
                    outPutFile.write("Provider Name Channel Name:")
                    outPutFile.write(DVB_dic_2["Provider Name Channel Name"])
                    outPutFile.write("  ") 
                    outPutFile.write("Video Encryption:")
                    outPutFile.write(DVB_dic_2["Video Encryption"])
                    outPutFile.write("  ") 
                    outPutFile.write("SID-VPID:")
                    outPutFile.write(DVB_dic_2["SID-VPID"])
                    outPutFile.write("  ")  
                    outPutFile.write("Audio:")
                    outPutFile.write(DVB_dic_2["Audio"])
                    outPutFile.write("\n")   
                   #TotalList.append(DVB_dic_2)
                elif List.count("DVB") == 0 and List.count("PAL") == 0 and len(List) == 6 and PAL_count == 0:
                    DVB_dic_2["Provider Name Channel Name"] = List[1]
                    DVB_dic_2["SID-VPID"] = List[4]
                    DVB_dic_2["Audio"] = List[5] 
                    if List[3].count("MPEG-4/HD") > 0:
                        DVB_dic_2["Video Encryption"] = "DVB-S2"
                   # print DVB_dic_2
                   # raw_input('List.count("DVB") == 0 and List.count("PAL") == 0 and len(List) == 6 and PAL_count == 0')
                    if rowspan_count != 0:
                        rowspan_count = rowspan_count - 1
                    outPutFile.write("Freq.Tp:")
                    outPutFile.write(DVB_dic_2["Freq.Tp"])
                    outPutFile.write("  ")
                    outPutFile.write("Provider Name Channel Name:")
                    outPutFile.write(DVB_dic_2["Provider Name Channel Name"])
                    outPutFile.write("  ") 
                    outPutFile.write("Video Encryption:")
                    outPutFile.write(DVB_dic_2["Video Encryption"])
                    outPutFile.write("  ") 
                    outPutFile.write("SID-VPID:")
                    outPutFile.write(DVB_dic_2["SID-VPID"])
                    outPutFile.write("  ")  
                    outPutFile.write("Audio:")
                    outPutFile.write(DVB_dic_2["Audio"])
                    outPutFile.write("\n")     
                   #TotalList.append(DVB_dic_2)
          #  print "After this stage: rawspan_count= "
          #  print rowspan_count
          #  print "\nrowspan_ori="
          #  print rowspan_ori
            if rowspan_count == 0:
                rowspan_ori = 0
                rowspan_flag  = 1  
                DVB_dic_1["Video Encryption"] = "DVB"
                DVB_dic_2["Video Encryption"] = "DVB"
          #  print TotalList    
         
            List = []      
    
        try:

                if hasattr(Parser, 'stop') and Parser.stop:
                    break
                Parser.feed(line)
            
        except HTMLParser.HTMLParseError, data:
            print "# Parser error : " + data.msg

        Parser.close()
        combine = string.join(item)
       #print combine
        List.append(combine)
       #print List    
       #raw_input("Hit Enter to write report: ")
    
    file.close()
    outPutFile.close()
    fileDatabase.close()  
resultFile.close()
import shutil
outName = "web_contentX.txt"
for index in range(1,fileNum+1):  
    fileOutPutName = string.replace(outputName,"X",str(index)) 
    fileOutName = string.replace(outName,"X",str(index))
    databaseName = string.replace(database,"X",str(index))
    shutil.move(fileOutPutName,".\\other output\\")
    shutil.move(fileOutName,".\\other output\\")
    shutil.move(databaseName,".\\other output\\")