ACE/examples/Web_Crawler/Iterators.h

   1 /* -*- C++ -*- */
   2
   3 //=============================================================================
   4 /**
   5  *  @file    Iterators.h
   6  *
   7  *  @author Douglas C. Schmidt <d.schmidt@vanderbilt.edu>
   8  */
   9 //=============================================================================
  10
  11
  12 #ifndef _ITERATORS_H
  13 #define _ITERATORS_H
  14
  15 #include "URL.h"
  16
  17 #if !defined (ACE_LACKS_PRAGMA_ONCE)
  18 #pragma once
  19 #endif /* ACE_LACKS_PRAGMA_ONCE */
  20
  21 /**
  22  * @class URL_Iterator
  23  *
  24  * @brief An abstract base class that defines an iterator.
  25  *
  26  * Subclasses of this base class can define what strings
  27  * to return from <next>.  This class decouples higher-level
  28  * software from the details of whatever type of URL header or
  29  * body we're iterating over.
  30  */
  31 class URL_Iterator
  32 {
  33 public:
  34   /// "virtual" destructor.
  35   virtual int destroy ();
  36
  37   // = Iterator methods.
  38   /// Pass back the next <string> that hasn't been seen yet.  Returns 0
  39   /// when all items have been seen, else 1.
  40   virtual int next (ACE_CString &string) = 0;
  41
  42 protected:
  43   /// C++ destructor.
  44   virtual ~URL_Iterator ();
  45 };
  46
  47 /**
  48  * @class HTML_Body_Iterator
  49  *
  50  * @brief An iterator that returns URLs embedded in HTML files.
  51  */
  52 class HTML_Body_Iterator : public URL_Iterator
  53 {
  54 public:
  55   /// Constructor.
  56   HTML_Body_Iterator (URL &url);
  57
  58   // = Iterator methods.
  59   /**
  60    * Pass back the next <url> that hasn't been seen in the
  61    * memory-mapped file.  Returns 0 when all items have been seen,
  62    * else 1.
  63    */
  64   virtual int next (ACE_CString &url);
  65
  66 private:
  67   /// HTTP URL that we're iterating over.
  68   URL &url_;
  69 };
  70
  71 /**
  72  * @class HTTP_Header_Iterator
  73  *
  74  * @brief An iterator that iterates over the HTTP header.
  75  */
  76 class HTTP_Header_Iterator : public URL_Iterator
  77 {
  78 public:
  79   /// Constructor.
  80   HTTP_Header_Iterator (URL &url);
  81
  82   // = Iterator methods.
  83   /**
  84    * Pass back the next <line> that hasn't been seen in the
  85    * memory-mapped file header.  Returns 0 when we've reached the end
  86    * of the header.  seen, else 1.
  87    */
  88   virtual int next (ACE_CString &line);
  89
  90 private:
  91   /// HTTP URL that we're iterating over.
  92   URL &url_;
  93
  94   /// We've found the end of the header, which means this iterator is
  95   /// finished.
  96   int end_of_header_;
  97 };
  98
  99 /**
 100  * @class URL_Download_Iterator
 101  *
 102  * @brief An iterator that iterates over the contents of an entire URL,
 103  * i.e., both header and body, and returns it in <BUFSIZ>
 104  * <buffer>s.
 105  */
 106 class URL_Download_Iterator : public URL_Iterator
 107 {
 108 public:
 109   /// Constructor.
 110   URL_Download_Iterator (URL &url);
 111
 112   // = Iterator methods.
 113   /**
 114    * Pass back the next <buffer> data from the stream, where
 115    * <buffer.size> <= <BUFSIZ> .  Returns 0 when we've reached the end
 116    * of the header, else 1.
 117    */
 118   virtual int next (ACE_CString &buffer);
 119
 120 private:
 121   /// HTTP URL that we're iterating over.
 122   URL &url_;
 123 };
 124
 125 #endif /* _ITERATORS_H */