scrape.js: Detect if GCI name is GitHub username

This detects when a student's GCI display name is their GitHub
username. To ensure we don't get the wrong account, this will first
verify that the user has been active since the start of GCI. If they
have, we then check if the user has created an issue or PR to the
organization's GitHub org. If so, we consider them the correct user.

Closes coala#32

lib/scrape.js

@@ -17,6 +17,17 @@ const CHAT_IMAGES = {
   OTHER: 'static/images/chat.png',
 }
 
+const GITHUB_OPTIONS = {
+  headers: process.env.GITHUB_TOKEN
+    ? { Authorization: `token ${process.env.GITHUB_TOKEN}` }
+    : {},
+}
+
+async function fetchProgram() {
+  const res = await fetch('https://codein.withgoogle.com/api/program/2017/')
+  return await res.json()
+}
+
 async function fetchOrgs() {
   const res = await fetch(
     'https://codein.withgoogle.com/api/program/2017/organization/?status=2'
@@ -34,17 +45,40 @@ async function fetchLeaders(id) {
 }
 
 async function searchGitHubOrgs(query) {
-  const token = process.env.GITHUB_TOKEN
   const res = await fetch(
     `${GITHUB_API_BASE}/search/users?q=${query}%20type:org`,
-    {
-      headers: token ? { Authorization: `token ${token}` } : {},
-    }
+    GITHUB_OPTIONS
   )
   const { items } = await res.json()
   return items || []
 }
 
+async function getGitHubUserCommits(user, from, to) {
+  const commitPattern = /<a href="(?:[a-zA-Z1-9/-]+)">([a-zA-Z1-9/-]+)<\/a>/g
+  const res = await fetch(
+    `https://github.com/users/${user}/created_commits?from=${from}&to=${to}`
+  )
+  const body = await res.text()
+
+  const commits = []
+  let match = commitPattern.exec(body)
+  while (match) {
+    commits.push(match[1])
+    match = commitPattern.exec(body)
+  }
+
+  return commits
+}
+
+async function getGitHubUser(user) {
+  const res = await fetch(`${GITHUB_API_BASE}/users/${user}`, GITHUB_OPTIONS)
+  let response = await res.json()
+  if (response && response.message) {
+    response = undefined
+  }
+  return response
+}
+
 async function findOrganization({
   name,
   description,
@@ -92,6 +126,37 @@ async function findOrganization({
   return null
 }
 
+async function findGitHubUser({ display_name }, org) {
+  if (!org) return
+
+  display_name = display_name.replace(/ /g, '')
+
+  const displayNamePattern = /^[a-zA-Z0-9-]{1,39}$/
+
+  const displayNameMatches = displayNamePattern.exec(display_name)
+  if (!displayNameMatches) return
+
+  const user = await getGitHubUser(display_name)
+  if (!user) return
+
+  const login = user.login
+
+  const { competition_open_starts } = await fetchProgram()
+
+  const updatedTime = new Date(user.updated_at)
+  const openTime = new Date(competition_open_starts)
+
+  if (updatedTime.getTime() - openTime.getTime() < 0) return
+
+  const nov = await getGitHubUserCommits(login, '2017-11-28', '2017-11-30')
+  const dec = await getGitHubUserCommits(login, '2017-12-01', '2017-12-31')
+  const jan = await getGitHubUserCommits(login, '2018-01-01', '2018-01-17')
+  const orgs = [...nov, ...dec, ...jan].map(repo => repo.split('/')[0])
+  if (orgs.includes(org)) {
+    return user.login
+  }
+}
+
 async function fetchOrgsWithData() {
   const orgs = await fetchOrgs()
   const fetchingLeaders = orgs.map(org => fetchLeaders(org.id))
@@ -101,17 +166,30 @@ async function fetchOrgsWithData() {
   const orgGitHub = await Promise.all(fetchingGitHub)
   const orgChats = await Promise.all(fetchingChat)
 
-  return orgs.map((org, index) =>
-    Object.assign(org, {
-      leaders: orgLeaders[index],
+  const fetchingAll = orgs.map(async (org, index) => {
+    const fetchingUsers = orgLeaders[index].map(user =>
+      findGitHubUser(user, orgGitHub[index])
+    )
+    const orgUsers = await Promise.all(fetchingUsers)
+
+    const leaders = orgLeaders[index].map((user, index) =>
+      Object.assign(user, {
+        github_account: orgUsers[index],
+      })
+    )
+
+    return Object.assign(org, {
+      leaders: leaders,
       github: orgGitHub[index],
       chat: {
         url: orgChats[index].url,
         platform: chattie.CHAT[orgChats[index].type],
         image: CHAT_IMAGES[chattie.CHAT[orgChats[index].type]],
       },
     })
-  )
+  })
+
+  return await Promise.all(fetchingAll)
 }
 
 async function fetchDates() {

templates/main.html

@@ -28,7 +28,8 @@ <h1>Google Code-in 2017 Current Leaders</h1>
     </div>
     <hr>
     <i>
-      The leading participants for each organization are listed randomly. 
+      The leading participants for each organization (and their GitHub accounts,
+      if applicable) are listed randomly.
     </i>
     <div class="orgs">
       {{#withLeader}}
@@ -67,7 +68,15 @@ <h3>
             <div class="org-leaderboard">
               <ul>
                 {{#leaders}}
-                  <li>{{display_name}}</li>
+                  <li>
+                    {{#github_account}}
+                      <a href="https://github.com/{{github_account}}">
+                    {{/github_account}}
+                    {{display_name}}
+                    {{#github_account}}
+                      </a>
+                    {{/github_account}}
+                  </li>
                 {{/leaders}}
               </ul>
             </div>