httpstats-final

00011                                                                   {
00012 /* A parse succeeds if all components can be parsed.
00013    We only parse the parts that are needed according to
00014    conf.criteria();
00015 */
00016 const set<Configuration::CRITERIUM>& criteria(conf.criteria());
00017 
00018 if (criteria.count(Configuration::DATE))
00019   if (!parse_date(line))
00020     return false;
00021 
00022 if (criteria.count(Configuration::PATH))
00023   if (!parse_path(line))
00024     return false;
00025 
00026 if (criteria.count(Configuration::DOMAIN))
00027   if (!parse_domain(line))
00028     return false;
00029 
00030 return true;
00031 }

00019 { return date_; }

00020 { return path_; }

00021 { return domain_; }

00035                                         {
00036 /* Select date part: just after first '[' up to first folloing ' '
00037    e.g.
00038 
00039         spider1.tiscalinet.be - - [02/Nov/2000:10:20:36 +0100] ..
00040 
00041    will select
00042 
00043         02/Nov/2000:10:20:36
00044 */
00045 string::size_type date_start(line.find_first_of('['));
00046 if (date_start==string::npos)
00047   return false;
00048 string::size_type date_end(line.find_first_of(' ',date_start));
00049 if (date_end==string::npos)
00050   return false;
00051 string            date_string(line,date_start+1,date_end - date_start-1);
00052 if (date_string.size()==0)
00053   return false;
00054 /* Make date_string acceptable to the Date::Date(const char*) parser.
00055 
00056    1. replace '/' by '-'. In the example, this will yield 
00057 
00058         02-Nov-2000:10:20:36
00059 */
00060 for (string::size_type i=0; i<date_string.size(); ++i)
00061   if (date_string[i]=='/')
00062     date_string[i] = '-';
00063 /*
00064    2. Replace first ':' by space. In the example, this will yield 
00065 
00066         02-Nov-2000 10:20:36
00067 */
00068 date_string.replace(date_string.find_first_of(':'),1," ");
00069 
00070 /* Now parse using Date::Date(const string&). It will throw an exception
00071    if the date cannot be parsed. In that case, we catch the exception
00072    and return false.
00073 */
00074 try {
00075   Date  d(date_string);
00076   date_.set(d.year(),d.month(),d.day(),d.hours());
00077   }
00078 catch (exception& e) {
00079   cerr << "LogRecord::parse_date failed: " << e.what() << endl;
00080   return false;
00081   }
00082 return true;
00083 }

00123                                         {
00124 /* The path part can be found between the first and second occurrences
00125    of `"' (double quote). An example is shown below.
00126 
00127     "GET /ssl/kiesoos.cgi?rolnr=58769\&stjcode=5L10021 HTTP/1.0" 
00128 
00129    This will result in 
00130 
00131      _path = <"","ssl","kiesoos.cgi">
00132 */
00133 string::size_type path_start(line.find_first_of('\"'));
00134 if (path_start==string::npos)
00135   return false;
00136 /* We are not interested in the verb (e.g. GET), so we skip until after the first ' '.
00137    In the example, path_start would then point to.
00138 
00139     /ssl/kiesoos.cgi?rolnr=58769\&stjcode=5L10021 HTTP/1.0" 
00140 */
00141 path_start = line.find_first_of(' ',path_start);
00142 if (path_start==string::npos)
00143   return false;
00144 ++path_start; // After ' '.
00145 /*  We only want the path until the first '?', '#' or ' '. In the example
00146     this should result in.
00147 
00148     path_string = /ssl/kiesoos.cgi
00149 */
00150 
00151 string::size_type path_end = line.find_first_of("#? ",path_start);
00152 if (path_end==string::npos)
00153   return false;
00154 
00155 if (path_end==path_start) // cannot have an empty path
00156   return false;
00157 
00158 string path_string(line,path_start,path_end - path_start);
00159 
00160 // Decode to translate "%7E" back to '~' etc.
00161 www_decode(path_string);
00162 
00163 return path_.parse(path_string);
00164 }

00168                                           {
00169 /* The domain informatin is easy to find: it's the first part of
00170    a line, followed by a ' ', as illustrated in the following example.
00171 
00172         igwe.vub.ac.be - - [12/Feb/2000:13:18:49 +0100] ...
00173         
00174 */
00175 string::size_type domain_start(0);
00176 string::size_type domain_end(line.find_first_of(' '));
00177 if (domain_end==string::npos)
00178   return false;
00179 string domain_string(line,domain_start,domain_end);
00180 
00181 return domain_.parse(domain_string);
00182 }

Public Methods
	LogRecord ()
bool	parse_line (const string &line, const Configuration &conf)
	fill in date_, domain_, path_ from line.
const DatePattern&	date () const
const Path&	path () const
const Domain&	domain () const
Private Methods
bool	parse_date (const string &line)
	Auxiliary function for parse_line().
bool	parse_path (const string &line)
	Auxiliary function for parse_line().
bool	parse_domain (const string &line)
	Auxiliary function for parse_line().
Private Attributes
DatePattern	date_
Path	path_
Domain	domain_

LogRecord Class Reference

Public Methods

Private Methods

Private Attributes

Constructor & Destructor Documentation

Member Function Documentation

Member Data Documentation