解析FTP LIST命令返回的目录列表

有个程序要分析FTP服务器返回的目录列表，本来以为比较简单，也在网上查了几个帖子，可都是一知半解的。于是下载了Filezilla的源代码，她的源文件
directorylistingparser.h
directorylistingparser.cpp
就是解析目录列表的，有同样需求的不妨看一看，还是挺费事的，不同平台要都要特殊处理。我把头文件贴出来：

#ifndef __DIRECTORYLISTINGPARSER_H__
#define __DIRECTORYLISTINGPARSER_H__

/* This class is responsible for parsing the directory listings returned by
 * the server.
 * Unfortunatly, RFC959 did not specify the format of directory listings, so
 * each server uses its own format. In addition to that, in most cases the 
 * listings were not designed to be machine-parsable, they were meant to be
 * human readable by users of that particular server.
 * By far the most common format is the one returned by the Unix "ls -l"
 * command. However, legacy systems are still in place, especially in big
 * companies. These often use very exotic listing styles.
 * Another problem are localized listings containing date strings. In some
 * cases these listings are ambiguous and cannot be distinguished.
 * Example for an ambiguous date: 04-05-06. All of the 6 permutations for
 * the location of year, month and day are valid dates.
 * Some servers send multiline listings where a single entry can span two
 * lines, this has to be detected as well, as far as possible.
 *
 * Some servers send MVS style listings which can consist of just the 
 * filename without any additional data. In order to prevent problems, this 
 * format is only parsed if the server is in fact recognizes as MVS server.
 *
 * Please see tests/dirparsertest.cpp for a list of supported formats and the
 * expected parser result.
 *
 * If adding data to the parser, it first decomposes the raw data into lines,
 * which then are processed further. Each line gets consecutively tested for
 * different formats, starting with the most common Unix style format.
 * Lines not containing a recognized format (e.g. a part of a multiline
 * entry) are rememberd and if the next line cannot be parsed either, they
 * get concatenated to be parsed again (and discarded if not recognized).
 */

class CLine;
class CToken;
class CControlSocket;
class CDirectoryListingParser
{
public:
  CDirectoryListingParser(CControlSocket* pControlSocket, const CServer& server);
  ~CDirectoryListingParser();

  CDirectoryListing Parse(const CServerPath &path);

  void AddData(char *pData, int len);
  void AddLine(const wxChar* pLine);

  void Reset();

  void SetTimezoneOffset(const wxTimeSpan& span) { m_timezoneOffset = span; }

  void SetServer(const CServer& server) { m_server = server; };

protected:
  CLine *GetLine(bool breakAtEnd = false);

  void ParseData(bool partial);

  bool ParseLine(CLine *pLine, const enum ServerType serverType, bool concatenated);

  bool ParseAsUnix(CLine *pLine, CDirentry &entry, bool expect_date);
  bool ParseAsDos(CLine *pLine, CDirentry &entry);
  bool ParseAsEplf(CLine *pLine, CDirentry &entry);
  bool ParseAsVms(CLine *pLine, CDirentry &entry);
  bool ParseAsIbm(CLine *pLine, CDirentry &entry);
  bool ParseOther(CLine *pLine, CDirentry &entry);
  bool ParseAsWfFtp(CLine *pLine, CDirentry &entry);
  bool ParseAsIBM_MVS(CLine *pLine, CDirentry &entry);
  bool ParseAsIBM_MVS_PDS(CLine *pLine, CDirentry &entry);
  bool ParseAsIBM_MVS_PDS2(CLine *pLine, CDirentry &entry);
  bool ParseAsIBM_MVS_Migrated(CLine *pLine, CDirentry &entry);
  bool ParseAsMlsd(CLine *pLine, CDirentry &entry);
  bool ParseAsOS9(CLine *pLine, CDirentry &entry);
  
  // Only call this if servertype set to ZVM since it conflicts
  // with other formats.
  bool ParseAsZVM(CLine *pLine, CDirentry &entry);

  // Only call this if servertype set to HPNONSTOP since it conflicts
  // with other formats.
  bool ParseAsHPNonstop(CLine *pLine, CDirentry &entry);

  // Date / time parsers
  bool ParseUnixDateTime(CLine *pLine, int &index, CDirentry &entry);
  bool ParseShortDate(CToken &token, CDirentry &entry, bool saneFieldOrder = false);
  bool ParseTime(CToken &token, CDirentry &entry);

  // Parse file sizes given like this: 123.4M
  bool ParseComplexFileSize(CToken& token, wxLongLong& size, int blocksize = -1);

  bool GetMonthFromName(const wxString& name, int &month);

  CControlSocket* m_pControlSocket;

  static std::map<wxString, int> m_MonthNamesMap;
  
  struct t_list
  {
    char *p;
    int len;
  };
  int m_currentOffset;

  std::list<t_list> m_DataList;
  std::list<CDirentry> m_entryList;

  CLine *m_prevLine;

  CServer m_server;

  bool m_fileListOnly;
  std::list<wxString> m_fileList;
  
  bool m_maybeMultilineVms;

  wxTimeSpan m_timezoneOffset;
};

#endif

本文转自 h2appy  51CTO博客，原文链接：http://blog.51cto.com/h2appy/122279，如需转载请自行联系原作者