scrape.js: Use GraphQL API

Use GitHub GraphQL API instead of REST API

Closes #111

lib/queries/github_search_org.graphql

@@ -0,0 +1,9 @@
+query($query: String!) {
+  search(type: USER, query: $query, first: 1) {
+    nodes {
+      ...on Organization {
+          login
+      }
+    }
+  }
+}

lib/queries/github_user_info.graphql

@@ -0,0 +1,6 @@
+query($user: String!) {
+  user(login: $user) {
+    login
+    updatedAt
+  }
+}

lib/queries/index.js

@@ -1,3 +1,5 @@
 const { loadQuery } = require('../utils')
 
 module.exports.GITHUB_REPO_INFO_QUERY = loadQuery('github_repo_info')
+module.exports.GITHUB_SEARCH_ORG_QUERY = loadQuery('github_search_org')
+module.exports.GITHUB_USER_INFO_QUERY = loadQuery('github_user_info')

lib/scrape.js

@@ -8,7 +8,11 @@ const validUsername = require('valid-github-username')
 const wdk = require('wikidata-sdk')
 const cheerio = require('cheerio')
 
-const { GITHUB_REPO_INFO_QUERY } = require('./queries')
+const {
+  GITHUB_REPO_INFO_QUERY,
+  GITHUB_SEARCH_ORG_QUERY,
+  GITHUB_USER_INFO_QUERY,
+} = require('./queries')
 const { getLatestCommitMessage } = require('./utils')
 
 const GH_BASE = 'https://github.com'
@@ -137,6 +141,7 @@ async function fetchRepositoryInfo(org) {
     ;({ data, errors } = await client.query(GITHUB_REPO_INFO_QUERY, { org }))
   } catch (error) {
     console.warn(`GitHub query for org ${org} fails, error: ${error}`)
+    return []
   }
 
   if (data && data.organization) {
@@ -209,12 +214,34 @@ async function checkGitHubUserExists(user) {
 }
 
 async function searchGitHubOrgs(query) {
+  let results = []
+  // use REST API to fetch
   const res = await fetch(
-    `${GH_API_BASE}/search/users?q=${query}%20type:org`,
+    `${GH_API_BASE}/search/users?q=${query}`,
     GH_API_OPTIONS
   )
+  console.log('query=', `${GH_API_BASE}/search/users?q=${query}`)
   const { items } = await res.json()
-  return items || []
+  results = results.concat(items)
+  // use GraphQL API to fetch
+  let data, errors
+  try {
+    ;({ data, errors } = await client.query(GITHUB_SEARCH_ORG_QUERY, { query }))
+  } catch (error) {
+    console.warn(`GitHub query ${query} fails, error: ${error}`)
+  }
+
+  if (data && data.search && data.search.nodes) {
+    results = results.concat(data.search.nodes)
+  } else {
+    const errorMessage =
+      errors && errors.length ? errors[0].message : 'unknown error'
+    console.warn(
+      `Cannot query ${query} from GitHub, error message: ${errorMessage}`
+    )
+  }
+  console.log('results', results)
+  return unique(results)
 }
 
 async function getGitHubUserHistory(user, from, to) {
@@ -262,12 +289,34 @@ function findMatches(input, pattern) {
 }
 
 async function getGitHubUser(user) {
-  const res = await fetch(`${GH_API_BASE}/users/${user}`, GH_API_OPTIONS)
-  let response = await res.json()
-  if (response && response.message) {
-    response = undefined
+  let data, errors
+  try {
+    ;({ data, errors } = await client.query(GITHUB_USER_INFO_QUERY, { user }))
+  } catch (error) {
+    console.warn(`GitHub query for user ${user} fails, error: ${error}`)
+  }
+
+  if (data && data.user) {
+    return data.user
+  } else {
+    const errorMessage =
+      errors && errors.length ? errors[0].message : 'unknown error'
+    console.warn(
+      `Cannot fetch user ${user} via GitHub GraphQL,`,
+      `error message: ${errorMessage}, resorting to GitHub REST API hit`
+    )
+    const res = await fetch(`${GH_API_BASE}/users/${user}`, GH_API_OPTIONS)
+    const response = await res.json()
+    if (response && response.message) {
+      console.warn(
+        `Cannot fetch user ${user} via GitHub REST API,`,
+        `error message: ${response.message}`
+      )
+      return undefined
+    } else {
+      return response
+    }
   }
-  return response
 }
 
 async function findOrganization({
@@ -307,10 +356,14 @@ async function findOrganization({
   )
 
   const removePattern = /the|project|\([a-zA-Z]+\)/gi
-  const searchQuery = name.replace(removePattern, '').trim()
+  const searchQuery = name.replace(removePattern, '').trim() + ' type:org'
   const searchResults = await searchGitHubOrgs(searchQuery)
 
-  if (searchResults.length > 0 && searchResults[0].score > MIN_SEARCH_SCORE) {
+  if (searchResults.length > 0) {
+    if (searchResults[0].score && searchResults[0].score <= MIN_SEARCH_SCORE) {
+      // GitHub REST API returns a list of matches with confidence score
+      return null
+    }
     return searchResults[0].login
   }
 
@@ -658,6 +711,10 @@ async function fetchDates() {
   return res.json()
 }
 
+function unique(arr) {
+  return Array.from(new Set(arr))
+}
+
 ;(async () => {
   const { competition_open_starts } = await fetchProgram()
   COMPETITION_OPEN = new Date(competition_open_starts)