Added documentation comments using Doxygen, changed the file structure back to the original, and wrote the getter functions for the individual entries.

This commit is contained in:
rapturate
2026-05-28 12:33:46 -04:00
parent 5673f32e69
commit 9ce211c572
7 changed files with 192 additions and 101 deletions

View File

@@ -1,34 +0,0 @@
#pragma once
#include <regex>
#include <string>
struct Entry {
std::string ip;
std::string timestamp;
std::string request;
std::string status;
std::string bytes;
std::string referer;
std::string os;
std::string browser;
};
class parsed_logs {
private:
Entry entry;
std::vector<Entry> logs;
std::string url;
public:
parsed_logs(std::string url);
void parse(const std::smatch& match);
std::vector<parsed_logs> parse_file(std::string log_url);
std::string get_ip();
std::string get_timestamp();
std::string get_request();
std::string get_status();
std::string get_bytes();
std::string get_referer();
std::string get_os();
std::string get_browser();
};

View File

@@ -0,0 +1,88 @@
#include <iostream>
#include <fstream>
#include <ostream>
#include <string>
#include <vector>
#include <regex>
#include "log_parsing.h"
parsed_logs::parsed_logs(std::string log_path) {
std::ifstream file(log_path);
if (!file.is_open()) {
std::cerr << "Error loading " << log_path << std::endl;
return;
}
std::regex log_pattern(R"((\S+)\s+-\s+-\s+\[([^\]]+)\]\s+\"([^\"]+)\"\s+(\d+)\s+(\d+|-)\s+\"([^\"]*)\"\s+\"([^\"]*)\")");
std::string line;
std::smatch match;
while (std::getline(file, line)) {
if (std::regex_search(line, match, log_pattern)) {
Entry current_entry;
current_entry.ip = match[1].str();
current_entry.timestamp = match[2].str();
current_entry.request = match[3].str();
current_entry.status = match[4].str();
current_entry.bytes = match[5].str();
current_entry.referer = match[6].str();
std::string raw_ua = match[7].str();
if (raw_ua.find("Windows NT 10.0") != std::string::npos) current_entry.os = "Windows 10/11";
else if (raw_ua.find("Windows NT 6.1") != std::string::npos) current_entry.os = "Windows 7";
else if (raw_ua.find("iPhone") != std::string::npos) current_entry.os = "iOS (iPhone)";
else if (raw_ua.find("iPad") != std::string::npos) current_entry.os = "iOS (iPad)";
else if (raw_ua.find("Android") != std::string::npos) current_entry.os = "Android";
else if (raw_ua.find("Macintosh") != std::string::npos) current_entry.os = "macOS";
else if (raw_ua.find("Linux") != std::string::npos) current_entry.os = "Linux";
else current_entry.os = "Unknown OS";
if (raw_ua.find("Edg/") != std::string::npos) current_entry.browser = "Microsoft Edge";
else if (raw_ua.find("OPR/") != std::string::npos) current_entry.browser = "Opera";
else if (raw_ua.find("Chrome/") != std::string::npos) current_entry.browser = "Google Chrome";
else if (raw_ua.find("Safari/") != std::string::npos) current_entry.browser = "Apple Safari";
else if (raw_ua.find("Firefox/") != std::string::npos) current_entry.browser = "Mozilla Firefox";
else if (raw_ua.find("curl/") != std::string::npos) current_entry.browser = "curl (CLI Tool)";
else current_entry.browser = "Unknown Browser/Bot";
logs.push_back(current_entry);
}
}
file.close();
}
std::string parsed_logs::entryx_ip(int x){
return logs[x].ip;
}
std::string parsed_logs::entryx_timestamp(int x){
return logs[x].timestamp;
}
std::string parsed_logs::entryx_request(int x){
return logs[x].request;
}
std::string parsed_logs::entryx_status(int x){
return logs[x].status;
}
std::string parsed_logs::entryx_bytes(int x){
return logs[x].bytes;
}
std::string parsed_logs::entryx_referer(int x){
return logs[x].referer;
}
std::string parsed_logs::entryx_os(int x){
return logs[x].os;
}
std::string parsed_logs::entryx_browser(int x){
return logs[x].browser;
}

103
log_parsing/log_parsing.h Normal file
View File

@@ -0,0 +1,103 @@
/**
* @file log_parsing.h
* @author Lew Price (lewis.e.price@outlook.com)
* @brief A series of functions for parsing the access logs of an Apache2 webserver.
* @version 0.1
* @date 2026-05-28
*
* @copyright Copyright (c) 2026
*
*/
#pragma once
#include <regex>
#include <string>
struct Entry {
std::string ip;
std::string timestamp;
std::string request;
std::string status;
std::string bytes;
std::string referer;
std::string os;
std::string browser;
};
class parsed_logs {
private:
Entry entry;
std::vector<Entry> logs;
std::string url;
public:
/**
* @brief Construct a new parsed logs object
* Detailed explanation: This function takes in the absolute path to a .log file and parses the lines into individual "Entry"s. These Entrys are then stored into a vector that can then be accessed via getter functions.
* @param string
*/
parsed_logs(std::string);
/**
* @brief Getter function for a specific Entry's IP
*
* @param int
* @return std::string
*/
std::string entryx_ip(int);
/**
* @brief Getter function for a specific Entry's time stamp
*
* @param int
* @return std::string
*/
std::string entryx_timestamp(int);
/**
* @brief Getter function for a specific Entry's request
*
* @param int
* @return std::string
*/
std::string entryx_request(int);
/**
* @brief Getter function for a specific Entry's status
*
* @param int
* @return std::string
*/
std::string entryx_status(int);
/**
* @brief Getter function for a specific Entry's byte size
*
* @param int
* @return std::string
*/
std::string entryx_bytes(int);
/**
* @brief Getter function for a specific Entry's referer
*
* @param int
* @return std::string
*/
std::string entryx_referer(int);
/**
* @brief Getter function for a specific Entry's OS info
*
* @param int
* @return std::string
*/
std::string entryx_os(int);
/**
* @brief Getter function for a specific Entry's browser info
*
* @param int
* @return std::string
*/
std::string entryx_browser(int);
};

View File

@@ -1,5 +1,4 @@
#include "log_parsing/log_parsing.h"
int main(){ int main(){
parse logs;
} }

View File

@@ -1,65 +0,0 @@
#include <cstdlib>
#include <iostream>
#include <fstream>
#include <ostream>
#include <string>
#include <vector>
#include <regex>
#include "log_parsing.h"
parsed_logs::parse(const std::smatch& match) {
Entry entry;
entry.ip = match[1].str();
entry.timestamp = match[2].str();
entry.request = match[3].str();
entry.status = match[4].str();
entry.bytes = match[5].str();
entry.referer = match[6].str();
std::string raw_ua = match[7].str();
if (raw_ua.find("Windows NT 10.0") != std::string::npos) entry.os = "Windows 10/11";
else if (raw_ua.find("Windows NT 6.1") != std::string::npos) entry.os = "Windows 7";
else if (raw_ua.find("iPhone") != std::string::npos) entry.os = "iOS (iPhone)";
else if (raw_ua.find("iPad") != std::string::npos) entry.os = "iOS (iPad)";
else if (raw_ua.find("Android") != std::string::npos) entry.os = "Android";
else if (raw_ua.find("Macintosh") != std::string::npos) entry.os = "macOS";
else if (raw_ua.find("Linux") != std::string::npos) entry.os = "Linux";
else entry.os = "Unknown OS";
if (raw_ua.find("Edg/") != std::string::npos) entry.browser = "Microsoft Edge";
else if (raw_ua.find("OPR/") != std::string::npos) entry.browser = "Opera";
else if (raw_ua.find("Chrome/") != std::string::npos) entry.browser = "Google Chrome";
else if (raw_ua.find("Safari/") != std::string::npos) entry.browser = "Apple Safari";
else if (raw_ua.find("Firefox/") != std::string::npos) entry.browser = "Mozilla Firefox";
else if (raw_ua.find("curl/") != std::string::npos) entry.browser = "curl (CLI Tool)";
else entry.browser = "Unknown Browser/Bot";
}
std::vector<parse> parse_file(std::string log_url) {
std::ifstream file(log_url);
if(!file.is_open()) {
std::cerr << "Error loading " << log_url << std::endl;
EXIT_FAILURE;
}
std::vector<std::string> lines;
std::string line;
while(std::getline(file, line)){
lines.push_back(std::move(line));
}
file.close();
std::vector<parse> all_logs;
std::regex log_pattern(R"((\S+)\s+-\s+-\s+\[([^\]]+)\]\s+\"([^\"]+)\"\s+(\d+)\s+(\d+|-)\s+\"([^\"]*)\"\s+\"([^\"]*)\")");
for(const auto& log_line : lines) {
std::smatch match;
if(std::regex_search(log_line, match, log_pattern)) {
all_logs.emplace_back(match);
}
}
return all_logs;
}