Add command line options and the ability to sort into sets or lists.

02-filter-a-log-file/chains_parser.js

@@ -2,6 +2,7 @@ import fs from "fs";
 import assert from "assert";
 import UAParser from "ua-parser-js";
 import readline from "readline";
+import { program } from "commander";
 
 class Parser {
   constructor() {
@@ -195,12 +196,26 @@ const chain_to_set = (requests) => {
   return path.values();
 }
 
-const sort_request_chains = (by_ip) => {
+const chain_to_list = (requests) => {
+  const path = [];
+  let seen;
+
+  for(let r of requests) {
+    if(r.url != seen) {
+      path.push(r.url);
+      seen = r.url;
+    }
+  }
+
+  return path.values();
+}
+
+const sort_request_chains = (by_ip, as_set) => {
   let ip_chains = {};
   let seen;
 
   for(let [ip, requests] of Object.entries(by_ip)) {
-    const chain = chain_to_set(requests);
+    const chain = as_set ? chain_to_set(requests) : chain_to_list(requests);
 
     const ref = requests[0].refer ? `[${requests[0].refer}]` : "";
     const url_set = [ref, ...chain].join(" ");
@@ -214,11 +229,29 @@ const sort_request_chains = (by_ip) => {
   return chains_sorted;
 }
 
-const [by_ip, stats] = await parse_logs(process.argv[2]);
-const chains_sorted = sort_request_chains(by_ip);
+const output_results = (min, chains_sorted) => {
+  for(let [url, count] of chains_sorted) {
+    if(count >= min) {
+      console.log(count, url);
+    }
+  }
 
-for(let [url, count] of chains_sorted) {
-  console.log(count, url);
+  console.log(stats);
 }
 
-console.log(stats);
+program
+  .option("--no-set", "Use a Set instead of a list for chains.")
+  .option("--min <Number>", "The lowest count to print. Stop at this.", 1)
+  .requiredOption("--input <String>", "Input file.")
+  .description("Processes different web server logs to determine request chain frequency.")
+  .version(0.1);
+
+program.parse();
+const OPTS = program.opts();
+OPTS.min = parseInt(OPTS.min);
+
+assert(!isNaN(OPTS.min), `min must be a number, you have ${OPTS.min}`);
+
+const [by_ip, stats] = await parse_logs(OPTS.input);
+const chains_sorted = sort_request_chains(by_ip, OPTS.set);
+output_results(OPTS.min, chains_sorted);