ACE/examples/Web_Crawler/Iterators.cpp

   1 #include "ace/Truncate.h"
   2 #include "Options.h"
   3 #include "Iterators.h"
   4
   5 URL_Iterator::~URL_Iterator ()
   6 {
   7 }
   8
   9 int
  10 URL_Iterator::destroy ()
  11 {
  12   // Commit suicide.
  13   delete this;
  14   return 0;
  15 }
  16
  17 HTML_Body_Iterator::HTML_Body_Iterator (URL &url)
  18   : url_ (url)
  19 {
  20 }
  21
  22 int
  23 HTML_Body_Iterator::next (ACE_CString &url)
  24 {
  25   size_t len = BUFSIZ;
  26   const char *buf;
  27   ACE_CString buffer;
  28   int href_index = 0;
  29
  30   for (buf = this->url_.stream ().recv (len);
  31        buf != 0;
  32        buf = this->url_.stream ().recv (len))
  33     {
  34       buffer.set (buf, BUFSIZ, 1);
  35
  36       href_index = ACE_Utils::truncate_cast<int> (buffer.find ("HREF"));
  37
  38       if (href_index < 0)
  39         href_index = ACE_Utils::truncate_cast<int> (buffer.find ("href"));
  40
  41       // Grep fpr " and grab the string until end-"
  42       if ( href_index > 0)
  43         {
  44           // Get back to buffer start location.
  45           this->url_.stream ().seek (-1 * static_cast<ACE_OFF_T> (len),
  46                                      SEEK_CUR);
  47
  48           int start_index =
  49             ACE_Utils::truncate_cast<int> (
  50               buffer.find ('\"', href_index));
  51
  52           if (start_index <= 0)
  53             break;
  54
  55           start_index += href_index;
  56
  57           int end_index =
  58             ACE_Utils::truncate_cast<int> (
  59               buffer.find ('\"', start_index + 1));
  60
  61           if (end_index <= 0)
  62             break;
  63
  64           end_index += start_index + 1;
  65
  66           ssize_t url_len = end_index - (start_index + 1);
  67
  68           ACE_CString temp = buffer.substring (start_index + 1,
  69                                                url_len);
  70           url.set (temp.c_str (), len, 1);
  71
  72           this->url_.stream ().seek (end_index + 1);
  73
  74           return url_len;
  75         }
  76     }
  77   return 0;
  78 }
  79
  80 HTTP_Header_Iterator::HTTP_Header_Iterator (URL &url)
  81   : url_ (url),
  82     end_of_header_ (0)
  83 {
  84 }
  85
  86 int
  87 HTTP_Header_Iterator::next (ACE_CString &line)
  88 {
  89   if (this->end_of_header_)
  90     return 0;
  91   else
  92     {
  93       for (char c;
  94            (c = this->url_.stream ().get_char ()) != (char)EOF;
  95            )
  96         {
  97           // Check to see if we're at the end of the header line.
  98           if (c == '\r' && this->url_.stream ().peek_char (0) == '\n')
  99             {
 100               line.set (this->url_.stream ().recv (),
 101                         this->url_.stream ().recv_len () - 1,
 102                         1);
 103
 104               // Check to see if we're at the end of the header.
 105               if (this->url_.stream ().peek_char (1) == '\r'
 106                   && this->url_.stream ().peek_char (2) == '\n')
 107                 {
 108                   this->end_of_header_ = 1;
 109                   // We're at the end of the header section.
 110                   this->url_.stream ().seek (3);
 111                 }
 112               else
 113                 // We're at the end of the line.
 114                 this->url_.stream ().seek (1);
 115
 116               return 1;
 117             }
 118           // Handle broken Web servers that use '\n' instead of
 119           // '\r\n'.
 120           else if (c == '\n')
 121             {
 122               line.set (this->url_.stream ().recv (),
 123                         (this->url_.stream ().recv_len ()),
 124                         1);
 125
 126               // Check to see if we're at the end of the header.
 127               if (this->url_.stream ().peek_char (0) == '\n')
 128                 {
 129                   // We're at the end of the header section.
 130                   this->url_.stream ().seek (1);
 131                   this->end_of_header_ = 1;
 132                 }
 133
 134               return 1;
 135             }
 136         }
 137
 138     }
 139   return 0;
 140 }
 141
 142 URL_Download_Iterator::URL_Download_Iterator (URL &url)
 143   : url_ (url)
 144 {
 145 }
 146
 147 int
 148 URL_Download_Iterator::next (ACE_CString &buffer)
 149 {
 150   size_t len = BUFSIZ;
 151
 152   const char *buf = this->url_.stream ().recv (len);
 153
 154
 155   if (buf == 0)
 156     return 0;
 157   else
 158     {
 159       buffer.set (buf, len, 1);
 160       return 1;
 161     }
 162 }