---

logrecord.C

Go to the documentation of this file.

// $Id: logrecord_C-source.html,v 1.1 2001/04/16 17:43:02 dvermeir Exp $

#include <string>
#include <ctype.h> // isxdigit, isalnum
#include "logrecord.h"
#include "configuration.h"
#include "date.h"

// Retrieve and store relevant information from a line from a log file.
bool
LogRecord::parse_line(const string& line,const Configuration& conf) {
/* A parse succeeds if all components can be parsed.
   We only parse the parts that are needed according to
   conf.criteria();
*/
const set<Configuration::CRITERIUM>& criteria(conf.criteria());

if (criteria.count(Configuration::DATE))
  if (!parse_date(line))
    return false;

if (criteria.count(Configuration::PATH))
  if (!parse_path(line))
    return false;

if (criteria.count(Configuration::DOMAIN))
  if (!parse_domain(line))
    return false;

return true;
}

// Retrieve and store date information from a line from a log file.
bool
LogRecord::parse_date(const string& line) {
/* Select date part: just after first '[' up to first folloing ' '
   e.g.

        spider1.tiscalinet.be - - [02/Nov/2000:10:20:36 +0100] ..

   will select

        02/Nov/2000:10:20:36
*/
string::size_type date_start(line.find_first_of('['));
if (date_start==string::npos)
  return false;
string::size_type date_end(line.find_first_of(' ',date_start));
if (date_end==string::npos)
  return false;
string            date_string(line,date_start+1,date_end - date_start-1);
if (date_string.size()==0)
  return false;
/* Make date_string acceptable to the Date::Date(const char*) parser.

   1. replace '/' by '-'. In the example, this will yield 

        02-Nov-2000:10:20:36
*/
for (string::size_type i=0; i<date_string.size(); ++i)
  if (date_string[i]=='/')
    date_string[i] = '-';
/*
   2. Replace first ':' by space. In the example, this will yield 

        02-Nov-2000 10:20:36
*/
date_string.replace(date_string.find_first_of(':'),1," ");

/* Now parse using Date::Date(const string&). It will throw an exception
   if the date cannot be parsed. In that case, we catch the exception
   and return false.
*/
try {
  Date  d(date_string);
  date_.set(d.year(),d.month(),d.day(),d.hours());
  }
catch (exception& e) {
  cerr << "LogRecord::parse_date failed: " << e.what() << endl;
  return false;
  }
return true;
}


// Return value of hex digit.
static inline int 
hexdigit(char c) { // assert(isxdigit(c))
if (isdigit(c)) 
  return c - '0';
else if (isupper(c))
  return c - 'A' + 10;
else // must be lower case letter
  return c - 'a' +10;
}

// In-place decoding of encoded url, see RFC 1738.
void
www_decode(string& s) {
unsigned int i(0); // input index
unsigned int j(0); // output index
unsigned int len(s.size());
for (i=0; (i<len); ++i,++j)
  switch (s[i]) {
    // case '+': s[j] = ' '; break; // Apparently not used in the log file.
    case '%': // the following 2 characters must be hex digits: 0-9,a-f
      if ((i+2<len) && isxdigit(s[i+1]) && isxdigit(s[i+2])) {
        s[j] = hexdigit(s[i+1])* 16 + hexdigit(s[i+2]);
        i += 2;
        }
      else
        s[j] = s[i];
      break;
    default: s[j] = s[i]; break;
    }
s.resize(j);
}

// Retrieve and store path information from a line from a log file.
bool
LogRecord::parse_path(const string& line) {
/* The path part can be found between the first and second occurrences
   of `"' (double quote). An example is shown below.

    "GET /ssl/kiesoos.cgi?rolnr=58769\&stjcode=5L10021 HTTP/1.0" 

   This will result in 

     _path = <"","ssl","kiesoos.cgi">
*/
string::size_type path_start(line.find_first_of('\"'));
if (path_start==string::npos)
  return false;
/* We are not interested in the verb (e.g. GET), so we skip until after the first ' '.
   In the example, path_start would then point to.

    /ssl/kiesoos.cgi?rolnr=58769\&stjcode=5L10021 HTTP/1.0" 
*/
path_start = line.find_first_of(' ',path_start);
if (path_start==string::npos)
  return false;
++path_start; // After ' '.
/*  We only want the path until the first '?', '#' or ' '. In the example
    this should result in.

    path_string = /ssl/kiesoos.cgi
*/

string::size_type path_end = line.find_first_of("#? ",path_start);
if (path_end==string::npos)
  return false;

if (path_end==path_start) // cannot have an empty path
  return false;

string path_string(line,path_start,path_end - path_start);

// Decode to translate "%7E" back to '~' etc.
www_decode(path_string);

return path_.parse(path_string);
}

// Retrieve and store domain information from a line from a log file.
bool
LogRecord::parse_domain(const string& line) {
/* The domain informatin is easy to find: it's the first part of
   a line, followed by a ' ', as illustrated in the following example.

        igwe.vub.ac.be - - [12/Feb/2000:13:18:49 +0100] ...
        
*/
string::size_type domain_start(0);
string::size_type domain_end(line.find_first_of(' '));
if (domain_end==string::npos)
  return false;
string domain_string(line,domain_start,domain_end);

return domain_.parse(domain_string);
}

---

httpstats-stage04 [ 7 April, 2001]