From 7c2bfb82d02b75ca851bc616231b55c8ad499196 Mon Sep 17 00:00:00 2001
From: Sreedhar M B <sreedharmb@users.noreply.github.com>
Date: Sat, 16 Sep 2017 15:38:45 +0530
Subject: [PATCH 1/3] Add files via upload

---
 crawler.js   | 69 +++++++++++++++++++++++++++++++++++++++++++++++-----
 package.json |  4 ++-
 2 files changed, 66 insertions(+), 7 deletions(-)

diff --git a/crawler.js b/crawler.js
index b248dbc..46cfaaf 100644
--- a/crawler.js
+++ b/crawler.js
@@ -1,9 +1,64 @@
 /**
- * Created by tushar on 13/09/17.
+ * Created by Sreedhar M B on 15/09/17.
  */
 
 'use strict'
 
+const request = require('request');
+const cheerio = require('cheerio');
+const fs = require('fs');
+
+const siteURL = "http://localhost:8080";
+
+let linksMap = {};
+let linksVisitedMap = {};
+let strLexElem;
+let siteVisitCount = 0;
+
+const getSiteDetails = (nextSiteHash, resolveFunc, callback) => {
+  let nextSite = siteURL + nextSiteHash;
+  siteVisitCount++;
+  request(nextSite, function(error, response, body) {
+    if(error) {
+      console.log("Error: " + error);
+    }
+
+    console.log(nextSite);
+    console.log("Status code: " + response.statusCode);
+    let $ = cheerio.load(body);
+
+    $('a.link').each(function( index ) {
+      let linkElement = $(this).attr('href');
+      if(!linksVisitedMap[linkElement]) {
+        linksMap[linkElement] = linkElement;
+      }
+    });
+
+    $('h1').each(function( index ) {
+      let strElem = $(this).text().trim();
+      strLexElem = strLexElem ? (strElem < strLexElem) ? strElem : strLexElem : strElem;
+    });
+
+
+    while(Object.keys(linksMap).length) {
+      let nextPageHash = linksMap[Object.keys(linksMap)[0]];
+      getSiteDetails(nextPageHash, resolveFunc, (strLexElem) => {
+        console.log(strLexElem);
+        console.log(siteVisitCount);
+        siteVisitCount--;
+        if(siteVisitCount == 1) {
+          resolveFunc(strLexElem);
+        }
+      });
+      linksVisitedMap[nextPageHash] = nextPageHash;
+      delete linksMap[nextPageHash];
+    }
+
+    callback(strLexElem);
+  });
+};
+
+
 /**
  * Crawls a website using a start {url}, and returns the lexicographically smallest string.
  * @param url
@@ -11,8 +66,10 @@
  */
 module.exports = url =>
   new Promise((resolve, reject) => {
-    /**
-     * TODO: Write your high performance code here.
-     */
-    reject(new Error('NotImplemented'))
-  })
+    getSiteDetails('', resolve, (strLexElem) => {
+      console.log(strLexElem);
+    });
+
+
+    // reject(new Error('NotImplemented'))
+  });
diff --git a/package.json b/package.json
index 4795255..aa340f7 100644
--- a/package.json
+++ b/package.json
@@ -10,10 +10,12 @@
   "author": "",
   "license": "ISC",
   "dependencies": {
+    "cheerio": "^1.0.0-rc.2",
     "express": "^4.15.4",
     "express-rate-limit": "^2.9.0",
     "mocha": "^3.5.3",
     "nodemon": "^1.12.0",
-    "pug": "^2.0.0-rc.4"
+    "pug": "^2.0.0-rc.4",
+    "request": "^2.81.0"
   }
 }

From 6bff68d273b2b7a3b44b9835bdeb1409dcfab3f2 Mon Sep 17 00:00:00 2001
From: Sreedhar M B <sreedharmb@users.noreply.github.com>
Date: Sat, 16 Sep 2017 16:05:38 +0530
Subject: [PATCH 2/3] Updated

---
 crawler.js | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)

diff --git a/crawler.js b/crawler.js
index 46cfaaf..d3dacc3 100644
--- a/crawler.js
+++ b/crawler.js
@@ -8,7 +8,7 @@ const request = require('request');
 const cheerio = require('cheerio');
 const fs = require('fs');
 
-const siteURL = "http://localhost:8080";
+let siteURL = "http://localhost:8080";
 
 let linksMap = {};
 let linksVisitedMap = {};
@@ -66,6 +66,7 @@ const getSiteDetails = (nextSiteHash, resolveFunc, callback) => {
  */
 module.exports = url =>
   new Promise((resolve, reject) => {
+    siteURL = url;
     getSiteDetails('', resolve, (strLexElem) => {
       console.log(strLexElem);
     });

From 9147b95382f9a3b398e9eeb81403329ddb482af3 Mon Sep 17 00:00:00 2001
From: Sreedhar M B <sreedharmb@users.noreply.github.com>
Date: Sat, 16 Sep 2017 16:21:17 +0530
Subject: [PATCH 3/3] Remove Console logs

---
 crawler.js | 10 +++++-----
 1 file changed, 5 insertions(+), 5 deletions(-)

diff --git a/crawler.js b/crawler.js
index d3dacc3..8cc9932 100644
--- a/crawler.js
+++ b/crawler.js
@@ -23,8 +23,8 @@ const getSiteDetails = (nextSiteHash, resolveFunc, callback) => {
       console.log("Error: " + error);
     }
 
-    console.log(nextSite);
-    console.log("Status code: " + response.statusCode);
+    // console.log(nextSite);
+    // console.log("Status code: " + response.statusCode);
     let $ = cheerio.load(body);
 
     $('a.link').each(function( index ) {
@@ -43,8 +43,8 @@ const getSiteDetails = (nextSiteHash, resolveFunc, callback) => {
     while(Object.keys(linksMap).length) {
       let nextPageHash = linksMap[Object.keys(linksMap)[0]];
       getSiteDetails(nextPageHash, resolveFunc, (strLexElem) => {
-        console.log(strLexElem);
-        console.log(siteVisitCount);
+        // console.log(strLexElem);
+        // console.log(siteVisitCount);
         siteVisitCount--;
         if(siteVisitCount == 1) {
           resolveFunc(strLexElem);
@@ -68,7 +68,7 @@ module.exports = url =>
   new Promise((resolve, reject) => {
     siteURL = url;
     getSiteDetails('', resolve, (strLexElem) => {
-      console.log(strLexElem);
+      // console.log(strLexElem);
     });