From 9e60765713bba4bc985a6d3b8a64072504e82ce0 Mon Sep 17 00:00:00 2001
From: Adriaan van Rossum <git@adriaan.io>
Date: Thu, 14 Nov 2019 14:42:47 +0700
Subject: [PATCH] Add unique page views to our scripts

https://github.com/simpleanalytics/roadmap/issues/12
---
 dist/default.js   |  2 +-
 dist/external.js  |  2 +-
 minify.js         | 27 ++++++++++++++++++++++-----
 package-lock.json |  5 +++++
 package.json      |  4 +++-
 src/default.js    | 46 ++++++++++++++++++++++++++++++----------------
 src/external.js   | 33 ++++++++++++++++++++++++++-------
 7 files changed, 88 insertions(+), 31 deletions(-)

diff --git a/dist/default.js b/dist/default.js
index e33117c..97e343b 100644
--- a/dist/default.js
+++ b/dist/default.js
@@ -1,3 +1,3 @@
 /* Simple Analytics - Privacy friend analytics (docs.simpleanalytics.com/script) */
 
-!function(s){if(s){var r,n,e="https://",c=e+"api.simpleanalytics.io/post",t=s.console;try{function h(e){t&&t.warn&&t.warn("Simple Analytics: "+e)}var u,l=s.navigator,p=s.location,d=s.document,f=l.userAgent,a=s.dispatchEvent,m="Not sending requests ",i=d.querySelector('script[src="https://cdn.simpleanalytics.io/hello.js"]'),v=i?i.getAttribute("data-mode"):null;if(-1<f.search(/(bot|spider|crawl)/gi))return h(m+"because user agent is a robot");function o(e){var t=p.protocol+"//"+p.hostname+p.pathname;if("hash"===v&&p.hash&&(t+=p.hash.split("?")[0]),u!==t){if(u=t,"visibilityState"in d&&"prerender"===d.visibilityState)return h(m+"when prerender");if("doNotTrack"in l&&"1"===l.doNotTrack)return h(m+"when doNotTrack is enabled");if("localhost"===p.hostname||"file:"===p.protocol)return h(m+"from localhost");var r=p.search.match(/[?&](utm_source|source|ref)=([^?&]+)/gi),n=r?r.map(function(e){return e.split("=")[1]}):[],a={url:t};f&&(a.ua=f),n&&n[0]&&(a.urlReferrer=n[0]),d.referrer&&!e&&(a.referrer=d.referrer),s.innerWidth&&(a.width=s.innerWidth);try{a.timezone=Intl.DateTimeFormat().resolvedOptions().timeZone}catch(o){}var i=new XMLHttpRequest;i.open("POST",c,!0),i.setRequestHeader("Content-Type","text/plain; charset=UTF-8"),i.send(JSON.stringify(a))}}var g=s.history;if((g?g.pushState:null)&&Event&&a){g.pushState=(n=g[r="pushState"],function(){var e=n.apply(this,arguments),t=new Event(r);return t.arguments=arguments,a(t),e}),s.addEventListener("pushState",function(){o(!0)})}"hash"===v&&"onhashchange"in s&&(s.onhashchange=function(){o(!0)}),o()}catch(w){t&&t.error&&t.error(w);var y=c+".gif";w&&w.message&&(y=y+"?error="+encodeURIComponent(w.message)),(new Image).src=y}}}(window);
\ No newline at end of file
+!function(h){if(h){function p(e){t&&t.warn&&t.warn("Simple Analytics: "+e)}var n,r,e="https://",u=e+"api.simpleanalytics.io/post",t=h.console;try{function a(e,t){return e&&e.getAttribute("data-"+t)}var l,f=h.navigator,g=h.location,d=h.document,m=f.userAgent,i=h.dispatchEvent,y=h.performance,v="Not sending requests ",o=d.querySelector('script[src="https://cdn.simpleanalytics.io/hello.js"]'),T=a(o,"mode"),w="true"===a(o,"skip-dnt");if(-1<m.search(/(bot|spider|crawl)/gi))return p(v+"because user agent is a robot");function s(e){var t=g.protocol+"//"+g.hostname+g.pathname;if("hash"===T&&g.hash&&(t+=g.hash.split("?")[0]),l!==t){if(l=t,!w&&"doNotTrack"in f&&"1"===f.doNotTrack)return p(v+"when doNotTrack is enabled");if("localhost"===g.hostname||"file:"===g.protocol)return p(v+"from localhost");var n=g.search.match(/[?&](utm_source|source|ref)=([^?&]+)/gi),r=n?n.map(function(e){return e.split("=")[1]}):[],a={url:t};m&&(a.ua=m),r&&r[0]&&(a.urlReferrer=r[0]),d.referrer&&!e&&(a.referrer=d.referrer),h.innerWidth&&(a.width=h.innerWidth);try{var i=y&&y.getEntriesByType&&y.getEntriesByType("navigation")[0]&&y.getEntriesByType("navigation")[0].type?-1<["reload","back_forward"].indexOf(y.getEntriesByType("navigation")[0].type):null,o="boolean"==typeof i?i:y&&y.navigation&&y.navigation.type&&-1<[y.navigation.TYPE_RELOAD,y.navigation.TYPE_BACK_FORWARD].indexOf(y.navigation.type);a.unique=!e&&!o&&(d.referrer&&d.referrer.split("/")[2]!==g.hostname)}catch(c){}try{a.timezone=Intl.DateTimeFormat().resolvedOptions().timeZone}catch(c){}var s=new XMLHttpRequest;s.open("POST",u,!0),s.setRequestHeader("Content-Type","text/plain; charset=UTF-8"),s.send(JSON.stringify(a))}}var c=h.history;if((c?c.pushState:null)&&Event&&i){c.pushState=(r=c[n="pushState"],function(){var e=r.apply(this,arguments),t=new Event(n);return t.arguments=arguments,i(t),e}),h.addEventListener("pushState",function(){s(!0)})}"hash"===T&&"onhashchange"in h&&(h.onhashchange=function(){s(!0)}),s()}catch(S){p(S.message);var E=u+".gif";S.message&&(E=E+"?error="+encodeURIComponent(S.message)),(new Image).src=E}}}(window);
\ No newline at end of file
diff --git a/dist/external.js b/dist/external.js
index eb163cb..3041a40 100644
--- a/dist/external.js
+++ b/dist/external.js
@@ -1,3 +1,3 @@
 /* Simple Analytics - Privacy friend analytics (docs.simpleanalytics.com/script) */
 
-!function(i,e){if(i){var n,r,s=i.navigator,o=i.location,c=o.hostname,t=i.document,a=i.console,u="//"+e,l="https://"+e;try{function p(e,t){return e&&e.getAttribute("data-"+t)}function h(e){a&&a.warn&&a.warn("Simple Analytics: "+e)}var f,m=s.userAgent,d=i.dispatchEvent,g="Not sending requests ",v=t.querySelector('script[src$="'+u+'/app.js"]'),w=p(v,"mode"),y="true"===p(v,"skip-dnt"),b=p(v,"sa-global")||"sa";if("localhost"===c)return h(g+"from localhost");if(/(bot|spider|crawl)/i.test(m))return h(g+"because user agent is a robot");function E(e){var t=o.search.match(new RegExp("[?&]("+e+")=([^?&]+)","gi")),n=t?t.map(function(e){return e.split("=")[1]}):[];if(n&&n[0])return n[0]}function S(e){var t=o.protocol+"//"+c+o.pathname;if("hash"===w&&o.hash&&(t+=o.hash.split("?")[0]),f!==t){if(f=t,!y&&"doNotTrack"in s&&"1"===s.doNotTrack)return h(g+"when doNotTrack is enabled");var n={url:t};m&&(n.ua=m),T&&(n.urlReferrer=T),R&&!e&&(n.referrer=R),i.innerWidth&&(n.width=i.innerWidth);try{n.timezone=Intl.DateTimeFormat().resolvedOptions().timeZone}catch(a){}var r=new XMLHttpRequest;r.open("POST",u+"/api",!0),r.setRequestHeader("Content-Type","text/plain; charset=UTF-8"),r.send(JSON.stringify(n))}}var T=E("utm_source|source|ref"),q=E("utm_campaign|campaign"),R=t.referrer.replace(/^https?:\/\/((m|l|w{2,3}([0-9]+)?)\.)?([^?#]+)(.*)$/,"$4").replace(/^([^/]+)\/$/,"$1")||null,$=i.history;if(($?$.pushState:null)&&Event&&d){$.pushState=(r=$[n="pushState"],function(){var e=r.apply(this,arguments),t=new Event(n);return t.arguments=arguments,d(t),e}),i.addEventListener("pushState",function(){S(!0)})}"hash"===w&&"onhashchange"in i&&(i.onhashchange=S),S();var k=/\.(.+)/.exec(e)[1];if(c!==k&&!new RegExp("."+k+"$","i").test(c))return h("Events via this script only work on "+k+" domains");function x(){A=!0;var a=t.createElement("iframe");a.setAttribute("src",u+"/iframe.html"),a.style.display="none",a.onload=function(){var t=a.contentWindow,n=T||R;try{if(N)for(var e=0;e<N.length;e++)t.postMessage({event:N[e][0],ref:n,campaign:q},l)}catch(r){}i[b]=function(e){t.postMessage({event:e,ref:n,campaign:q},l)}},t.body.appendChild(a)}var N=i[b]&&i[b].q?i[b].q:[],A=!1;i[b]=function(){A||x(),N.push([].slice.call(arguments))},!A&&N.length&&x()}catch(I){a&&a.error&&a.error(I);var C=u+"/image.gif";I&&I.message&&(C=C+"?error="+encodeURIComponent(I.message)),(new Image).src=C}}}(window,"<!--# echo var="http_host" default="" -->");
\ No newline at end of file
+!function(s,e){if(s){function c(e){t&&t.warn&&t.warn("Simple Analytics: "+e)}var n,a,p=s.navigator,l=s.location,u=l.hostname,h=s.document,t=s.console,f=s.performance,g="//"+e,i="https://"+e;try{function r(e,t){return e&&e.getAttribute("data-"+t)}var m,v=p.userAgent,o=s.dispatchEvent,d="Not sending requests ",y=h.querySelector('script[src$="'+g+'/app.js"]'),w=r(y,"mode"),E="true"===r(y,"skip-dnt"),T=r(y,"sa-global")||"sa";if("localhost"===u)return c(d+"from localhost");if(/(bot|spider|crawl)/i.test(v))return c(d+"because user agent is a robot");function b(e){var t=l.search.match(new RegExp("[?&]("+e+")=([^?&]+)","gi")),n=t?t.map(function(e){return e.split("=")[1]}):[];if(n&&n[0])return n[0]}function R(e){var t=l.protocol+"//"+u+l.pathname;if("hash"===w&&l.hash&&(t+=l.hash.split("?")[0]),m!==t){if(m=t,!E&&"doNotTrack"in p&&"1"===p.doNotTrack)return c(d+"when doNotTrack is enabled");var n={url:t};v&&(n.ua=v),S&&(n.urlReferrer=S),x&&!e&&(n.referrer=x),s.innerWidth&&(n.width=s.innerWidth);try{var a=f&&f.getEntriesByType&&f.getEntriesByType("navigation")[0]&&f.getEntriesByType("navigation")[0].type?-1<["reload","back_forward"].indexOf(f.getEntriesByType("navigation")[0].type):null,r="boolean"==typeof a?a:f&&f.navigation&&f.navigation.type&&-1<[f.navigation.TYPE_RELOAD,f.navigation.TYPE_BACK_FORWARD].indexOf(f.navigation.type);n.unique=!e&&!r&&(h.referrer&&h.referrer.split("/")[2]!==l.hostname)}catch(o){}try{n.timezone=Intl.DateTimeFormat().resolvedOptions().timeZone}catch(o){}var i=new XMLHttpRequest;i.open("POST",g+"/api",!0),i.setRequestHeader("Content-Type","text/plain; charset=UTF-8"),i.send(JSON.stringify(n))}}var S=b("utm_source|source|ref"),q=b("utm_campaign|campaign"),x=h.referrer.replace(/^https?:\/\/((m|l|w{2,3}([0-9]+)?)\.)?([^?#]+)(.*)$/,"$4").replace(/^([^/]+)\/$/,"$1")||null,A=s.history;if((A?A.pushState:null)&&Event&&o){A.pushState=(a=A[n="pushState"],function(){var e=a.apply(this,arguments),t=new Event(n);return t.arguments=arguments,o(t),e}),s.addEventListener("pushState",function(){R(!0)})}"hash"===w&&"onhashchange"in s&&(s.onhashchange=R),R();var O=/\.(.+)/.exec(e)[1];if(u!==O&&!new RegExp("."+O+"$","i").test(u))return c("Events via this script only work on "+O+" domains");function k(){_=!0;var r=h.createElement("iframe");r.setAttribute("src",g+"/iframe.html"),r.style.display="none",r.onload=function(){var t=r.contentWindow,n=S||x;try{if($)for(var e=0;e<$.length;e++)t.postMessage({event:$[e][0],ref:n,campaign:q},i)}catch(a){}s[T]=function(e){t.postMessage({event:e,ref:n,campaign:q},i)}},h.body.appendChild(r)}var $=s[T]&&s[T].q?s[T].q:[],_=!1;s[T]=function(){_||k(),$.push([].slice.call(arguments))},!_&&$.length&&k()}catch(N){c(N.message);var B=g+"/image.gif";N.message&&(B=B+"?error="+encodeURIComponent(N.message)),(new Image).src=B}}}(window,"<!--# echo var="http_host" default="" -->");
\ No newline at end of file
diff --git a/minify.js b/minify.js
index 094427c..94b314f 100644
--- a/minify.js
+++ b/minify.js
@@ -1,30 +1,35 @@
 const UglifyJS = require('uglify-js')
 const fs = require('fs')
+const esprima = require('esprima')
 
 const green = '\x1b[32m%s\x1b[0m'
 const yellow = '\x1b[33m%s\x1b[0m'
-// const red = '\x1b[31m%s\x1b[0m'
+const red = '\x1b[31m%s\x1b[0m'
 
 const files = {
   default: {
+    type: 'js',
     original: `${__dirname}/src/default.js`,
     minified: `${__dirname}/dist/default.js`,
     prepend: ['/* Simple Analytics - Privacy friend analytics (docs.simpleanalytics.com/script) */', ''],
     append: []
   },
   embed: {
+    type: 'js',
     original: `${__dirname}/src/embed.js`,
     minified: `${__dirname}/dist/embed.js`,
     prepend: ['/* Simple Analytics - Privacy friend analytics (docs.simpleanalytics.com/embed-graph-on-your-site) */', ''],
     append: []
   },
   external: {
+    type: 'js',
     original: `${__dirname}/src/external.js`,
     minified: `${__dirname}/dist/external.js`,
     prepend: ['/* Simple Analytics - Privacy friend analytics (docs.simpleanalytics.com/script) */', ''],
     append: []
   },
   iframe: {
+    type: 'html',
     original: `${__dirname}/src/iframe.js`,
     minified: `${__dirname}/dist/iframe.html`,
     prepend: ['<!DOCTYPE html>', '<title>SA</title>', '<script>'],
@@ -42,16 +47,28 @@ const options = {
 
 for (const file in files) {
   if (files.hasOwnProperty(file)) {
-    const { original, minified, prepend, append } = files[file]
+    const { type, original, minified, prepend, append } = files[file]
     const name = file.toUpperCase()
 
     const contents = fs.readFileSync(original, 'utf8')
     const { code, warnings } = UglifyJS.minify(contents, options)
     for (const warning of (warnings || [])) console.warn(yellow, `[MINIFY][${name}] ${warning}`)
     const nginxCode = code.replace('simpleanalytics.example.com', '<!--# echo var="http_host" default="" -->')
-    const lines = [...prepend, nginxCode, ...append]
-    fs.writeFileSync(minified, lines.join('\n'))
-    const bytes = new TextEncoder('utf-8').encode(lines.join('\n')).length
+    const lines = [...prepend, nginxCode, ...append].join('\n')
+
+    const validate = type === 'js' ? lines.replace('<!--# echo var="http_host" default="" -->', '') : code
+
+    try {
+      esprima.parseScript(validate)
+    } catch (error) {
+      const { index, lineNumber, description } = error
+      console.log(red, `[MINIFY][${name}][ERROR] ${original.split('/').pop()} ${description} at line ${lineNumber} position ${index}`)
+      continue
+    }
+
+    fs.writeFileSync(minified, lines)
+    const bytes = new TextEncoder('utf-8').encode(lines).length
+
     console.log(green, `[MINIFY][${name}] Minified ${original.split('/').pop()} into ${bytes} bytes`)
   }
 }
diff --git a/package-lock.json b/package-lock.json
index b2860dc..da7aeac 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -499,6 +499,11 @@
       "integrity": "sha1-G2HAViGQqN/2rjuyzwIAyhMLhtQ=",
       "dev": true
     },
+    "esprima": {
+      "version": "4.0.1",
+      "resolved": "https://registry.npmjs.org/esprima/-/esprima-4.0.1.tgz",
+      "integrity": "sha512-eGuFFw7Upda+g4p+QHvnW0RyTX/SVeJBDM/gCtMARO0cLuT2HcEKnTPvhjV6aGeqrCB/sbNop0Kszm0jsaWU4A=="
+    },
     "execa": {
       "version": "0.7.0",
       "resolved": "https://registry.npmjs.org/execa/-/execa-0.7.0.tgz",
diff --git a/package.json b/package.json
index ebef3c7..4eba0d5 100644
--- a/package.json
+++ b/package.json
@@ -18,7 +18,9 @@
     "url": "https://github.com/simpleanalytics/scripts/issues"
   },
   "homepage": "https://github.com/simpleanalytics/scripts#readme",
-  "dependencies": {},
+  "dependencies": {
+    "esprima": "^4.0.1"
+  },
   "devDependencies": {
     "nodemon": "^1.19.1",
     "uglify-js": "^3.6.0"
diff --git a/src/default.js b/src/default.js
index a208697..ee95c94 100644
--- a/src/default.js
+++ b/src/default.js
@@ -1,9 +1,5 @@
 /* eslint-env browser */
 
-// We use https://jscompress.com to compress our script so you can test
-// if this script is indeed the same script as the minified version we
-// published on https://cdn.simpleanalytics.io/hello.js
-
 (function(window, hostname, path, cdn) {
   if (!window) return;
 
@@ -13,22 +9,25 @@
   var cdnUrl = cdn ? protocol + cdn : (apiUrl + '.js');
   var con = window.console;
 
+  // A simple log function so the user knows why a request is not being send
+  var warn = function(message) {
+    if (con && con.warn) con.warn('Simple Analytics: ' + message);
+  }
+
   try {
     var nav = window.navigator;
     var loc = window.location;
     var doc = window.document;
     var userAgent = nav.userAgent;
     var dis = window.dispatchEvent;
+    var perf = window.performance;
     var lastSendUrl;
     var notSending = 'Not sending requests ';
 
     var script = doc.querySelector('script[src="' + cdnUrl + '"]');
-    var mode = script ? script.getAttribute('data-mode') : null;
-
-    // A simple log function so the user knows why a request is not being send
-    var warn = function(message) {
-      if (con && con.warn) con.warn('Simple Analytics: ' + message);
-    }
+    var attr = function(script, attribute) { return script && script.getAttribute('data-' + attribute) }
+    var mode = attr(script, 'mode')
+    var skipDNT = attr(script, 'skip-dnt') === 'true'
 
     // We do advanced bot detection in our API, but this line filters already most bots
     if (userAgent.search(/(bot|spider|crawl)/ig) > -1) return warn(notSending + 'because user agent is a robot');
@@ -44,11 +43,8 @@
       if (lastSendUrl === url) return;
       lastSendUrl = url;
 
-      // Skip prerender requests
-      if ('visibilityState' in doc && doc.visibilityState === 'prerender') return warn(notSending + 'when prerender');
-
       // Don't track when Do Not Track is set to true
-      if ('doNotTrack' in nav && nav.doNotTrack === '1') return warn(notSending + 'when doNotTrack is enabled');
+      if (!skipDNT && 'doNotTrack' in nav && nav.doNotTrack === '1') return warn(notSending + 'when doNotTrack is enabled');
 
       // Don't track when localhost
       if (loc.hostname === 'localhost' || loc.protocol === 'file:') return warn(notSending + 'from localhost');
@@ -63,6 +59,24 @@
       if (doc.referrer && !isPushState) data.referrer = doc.referrer;
       if (window.innerWidth) data.width = window.innerWidth;
 
+      // We put new code always in a try block to prevent huge issues
+      try {
+        // Check if back, forward or reload buttons are being use in modern browsers
+        var backModern = (perf && perf.getEntriesByType && perf.getEntriesByType('navigation')[0] && perf.getEntriesByType('navigation')[0].type)
+          ? ['reload', 'back_forward'].indexOf(perf.getEntriesByType('navigation')[0].type) > -1
+          : null
+
+        // Check if back, forward or reload buttons are being use in older browsers
+        var back = typeof backModern === 'boolean'
+          ? backModern
+          : perf && perf.navigation && perf.navigation.type && [perf.navigation.TYPE_RELOAD, perf.navigation.TYPE_BACK_FORWARD].indexOf(perf.navigation.type) > -1
+
+        // We set unique variable based on pushstate or back navigation, if no match we check the referrer
+        data.unique = isPushState || back ? false : doc.referrer && doc.referrer.split('/')[2] !== loc.hostname;
+      } catch (error) {
+        // nothing
+      }
+
       try {
         data.timezone = Intl.DateTimeFormat().resolvedOptions().timeZone
       } catch (error) {
@@ -106,9 +120,9 @@
 
     post();
   } catch (e) {
-    if (con && con.error) con.error(e);
+    warn(e.message);
     var url = apiUrl + '.gif';
-    if (e && e.message) url = url + '?error=' + encodeURIComponent(e.message);
+    if (e.message) url = url + '?error=' + encodeURIComponent(e.message);
     new Image().src = url;
   }
 })(window, 'api.simpleanalytics.io', '/post', 'cdn.simpleanalytics.io/hello.js');
diff --git a/src/external.js b/src/external.js
index 5f8e786..2e2f6e5 100644
--- a/src/external.js
+++ b/src/external.js
@@ -7,9 +7,15 @@
   var host = loc.hostname
   var doc = window.document
   var con = window.console
+  var perf = window.performance
   var uri = '//' + hostname
   var targetOrigin = 'https://' + hostname
 
+  // A simple log function so the user knows why a request is not being send
+  var warn = function(message) {
+    if (con && con.warn) con.warn('Simple Analytics: ' + message)
+  }
+
   try {
     var userAgent = nav.userAgent
     var dis = window.dispatchEvent
@@ -23,11 +29,6 @@
     var skipDNT = attr(script, 'skip-dnt') === 'true'
     var functionName = attr(script, 'sa-global') || 'sa'
 
-    // A simple log function so the user knows why a request is not being send
-    var warn = function(message) {
-      if (con && con.warn) con.warn('Simple Analytics: ' + message)
-    }
-
     // Don't track when host is localhost
     if (host === 'localhost') return warn(notSending + 'from localhost')
 
@@ -65,6 +66,24 @@
       if (cleanRef && !isPushState) data.referrer = cleanRef
       if (window.innerWidth) data.width = window.innerWidth
 
+      // We put new code always in a try block to prevent huge issues
+      try {
+        // Check if back, forward or reload buttons are being use in modern browsers
+        var backModern = (perf && perf.getEntriesByType && perf.getEntriesByType('navigation')[0] && perf.getEntriesByType('navigation')[0].type)
+          ? ['reload', 'back_forward'].indexOf(perf.getEntriesByType('navigation')[0].type) > -1
+          : null
+
+        // Check if back, forward or reload buttons are being use in older browsers
+        var back = typeof backModern === 'boolean'
+          ? backModern
+          : perf && perf.navigation && perf.navigation.type && [perf.navigation.TYPE_RELOAD, perf.navigation.TYPE_BACK_FORWARD].indexOf(perf.navigation.type) > -1
+
+        // We set unique variable based on pushstate or back navigation, if no match we check the referrer
+        data.unique = isPushState || back ? false : doc.referrer && doc.referrer.split('/')[2] !== loc.hostname;
+      } catch (error) {
+        // nothing
+      }
+
       try {
         data.timezone = Intl.DateTimeFormat().resolvedOptions().timeZone
       } catch (error) {
@@ -144,9 +163,9 @@
 
     if (!loading && queue.length) loadIframe()
   } catch (e) {
-    if (con && con.error) con.error(e)
+    warn(e.message)
     var url = uri + '/image.gif'
-    if (e && e.message) url = url + '?error=' + encodeURIComponent(e.message)
+    if (e.message) url = url + '?error=' + encodeURIComponent(e.message)
     new Image().src = url
   }
 })(window, 'simpleanalytics.example.com')