/* Filename:  websearch.cpp

   Author:  Br. David Carlson

   Date:  January 4, 2007

   This is a CGI program to handle a user's search for a certain target string.
   Using a browser pointed at websearch.html, the user fills in the desired target
   and then clicks on the submit button.  The uncgi program is used to put this
   string into an environment variable named WWW_Key.  Our particular version of
   uncgi also rejects strings that contain common hacker symbols.  If the user
   submits a string containing a % sign, for example, the input is rejected.
   Most non-alphanumeric symbols cause uncgi to display an error message and quit.
   In addition, when we use the GetValue function below, it does some filtering of
   user input as well.  When it finds an unacceptable character, it puts the empty
   string into the WWW variable.

   Note that websearch.html must be placed in some location under the web root for
   your web server.

   Assuming that WWW_Key gets a non-empty string, this program searchs through the
   text file named weblist for that target string.  The weblist file contains
   complete pathnames for various html files, one per line, like this example:

   /www/carlson/cs125/final.html
   /www/carlson/cs125/hw/homework2.html

   The weblist file must be located in the cgi-bin directory along with the
   compiled websearch program and webhead.html.  The weblist file can be created 
   with the separate makelist script.

   The output of the program is an HTML-marked up page with a list of the URLs
   corresponding to the lines of weblist that contained the target string.

   Note, to compile this program in Linux, use:
   g++ websearch.cpp stringhelp.cpp -o websearch -s
*/

#include "stringhelp.h"

// Next line should give the name of the text file to be searched:
#define DATAFILE    "weblist"

// Next line should give name of file containing opening HTML for search results:
#define HEAD       "webhead.html"

// The next line should give the start of all URLs for the html files in the above file:
#define URLSTART    "http://cis.stvincent.edu"

// The next line should give the number of chars to remove at the start of each pathname when making a URL:
#define REMOVECHARS 4


// Function prototypes:
void SearchFile(fstream & fs);
void PrintFile(StringType Filename);
void PrintURL(StringType Line, int Length);


int main(void)
   {
   fstream InFile;

   InFile.open(DATAFILE, ios::in);
   if (InFile.fail())
      {
      cout << "Error: Cannot open file named " << DATAFILE << " for input" << endl;
      exit(1);
      }

   SearchFile(InFile);
   InFile.close();

   return 0;
   }


/* Given:   fs   Text file stream already opened for input.
   Task:    To get the target string from the WWW_Key environment variable and
            do a sequential search of the fs file for all instances of this target
            string.  Each matching line of the file is printed for the user within
            an HTML page which is sent to standard output.
   Return:  Nothing.
*/
void SearchFile(fstream & fs)
   {
   int LineLength, KeyLength;
   StringType Key, Line;
   const char Quote = 34;  // 34 is ASCII code for double quote character.
   
   GetValue("WWW_Key", Key);   // Get the target key from the WWW environment variable set up by uncgi.
   KeyLength = strlen(Key);

   cout << "Content-type: text/html" << endl << endl;
   PrintFile(HEAD);
   cout << "<OL>" << endl;

   if (KeyLength == 0)
      {
      cout << "<LI>No keyword was supplied or keyword contained disallowed characters</LI>" << endl;
      cout << "</OL></BODY></HTML>" << endl;
      return;
      }

   LineLength = MyGetLine(fs, Line, StrMax);

   while (LineLength > 0)
      {
      if (SubstringPresent(Line, Key))
         {
         cout << "<LI><A HREF=" << Quote;
         PrintURL(Line, LineLength);
         cout << Quote << ">";
         PrintURL(Line, LineLength);
         cout << "</A></LI>" << endl;
         }
      LineLength = MyGetLine(fs, Line, StrMax);
      }

   cout << "</OL></BODY></HTML>" << endl;
   }


/* Given:   Filename    String containing name of a text file to output.
   Task:    To copy the contents of this file to standard output.
   Return:  Nothing.
*/
void PrintFile(StringType Filename)
   {
   fstream fs;
   StringType Line;

   fs.open(Filename, ios::in);
   if (fs.fail())
      exit(1);

   MyGetLine(fs, Line, StrMax);

   while (! fs.fail())
      {
      cout << Line << endl;
      MyGetLine(fs, Line, StrMax);
      }

   fs.close();
   }


/* Given:   Line    A string containing a complete pathname for a particular html file.
            Length  The number of chars in the string in Line.
   Task:    To print out the URL corresponding to this pathname.  This is done by printing
            URLSTART followed by all of Line except the first REMOVECHARS characters.
   Return:  Nothing.
*/ 
void PrintURL(StringType Line, int Length)
   {        
   int k;

   cout << URLSTART;
   
   for (k = REMOVECHARS; k < Length; k++)
      cout.put(Line[k]);
   }