From 6a3a2cd053875e235d236b1bc8ca48b4bd8f7a11 Mon Sep 17 00:00:00 2001
From: Doug Martin <dmartin@concord.org>
Date: Wed, 22 May 2024 15:56:46 -0400
Subject: [PATCH] feat: Hash usernames using SHA1 when hiding names
 [PT-184362179]

---
 query-creator/create-query/app.js       |  4 +--
 query-creator/create-query/steps/aws.js | 33 +++++++++++++++----------
 2 files changed, 22 insertions(+), 15 deletions(-)

diff --git a/query-creator/create-query/app.js b/query-creator/create-query/app.js
index f502201..d74cf42 100644
--- a/query-creator/create-query/app.js
+++ b/query-creator/create-query/app.js
@@ -139,8 +139,8 @@ const learnersReport = async (params, body, tokenServiceEnv, debugSQL, reportSer
   const doLearnerLogReporting= async () => {
     // generate the sql for the query
     const sql = narrowLearners
-      ? aws.generateNarrowLogSQL(queryIdsPerRunnable, authDomain, reportServiceSource) // hideNames not needed here as no learner info is output
-      : aws.generateLearnerLogSQL(queryIdsPerRunnable, authDomain, reportServiceSource, hideNames);
+      ? aws.generateNarrowLogSQL(queryIdsPerRunnable, hideNames)
+      : aws.generateLearnerLogSQL(queryIdsPerRunnable, hideNames);
 
     if (debugSQL) {
       sqlOutput.push(sql);
diff --git a/query-creator/create-query/steps/aws.js b/query-creator/create-query/steps/aws.js
index 6fc51e0..ac60cb0 100644
--- a/query-creator/create-query/steps/aws.js
+++ b/query-creator/create-query/steps/aws.js
@@ -543,25 +543,31 @@ exports.generateSQL = (runnableInfo, usageReport, authDomain, sourceKey, hideNam
     GROUP BY l.run_remote_endpoint )`
 */
 
+const getLogCols = (hideNames) => {
+  return ["id", "session", "username", "application", "activity", "event", "event_value", "time", "parameters", "extras", "run_remote_endpoint", "timestamp"]
+    .map(col => `"log"."${col}"`)
+    .map(col => col === `"log"."username"` && hideNames ? `to_hex(sha1(cast("log"."username" as varbinary))) as username` : col)
+}
+
+const getLearnerCols = (hideNames) => {
+  return ["learner_id", "run_remote_endpoint", "class_id", "runnable_url", "student_id", "class", "school", "user_id", "offering_id", "permission_forms", "username", "student_name", "teachers", "last_run", "query_id"]
+    .map(col => `"learner"."${col}"`)
+    .map(col => col === `"learner"."username"` && hideNames ? `to_hex(sha1(cast("learner"."username" as varbinary))) as username` : col)
+    .map(col => col === `"learner"."student_name"` && hideNames ? `"learner"."student_id" as student_name` : col)
+}
+
 /*
 Generates a very wide row including all fields from the log and learner.
 */
-exports.generateLearnerLogSQL = (queryIdsPerRunnable, authDomain, sourceKey, hideNames) => {
+exports.generateLearnerLogSQL = (queryIdsPerRunnable, hideNames) => {
   const logDb = process.env.LOG_ATHENA_DB_NAME;
   const runnableUrls = Object.keys(queryIdsPerRunnable);
   const queryIds = Object.values(queryIdsPerRunnable);
 
-  const logCols = [
-    "id", "session", "username", "application", "activity", "event", "event_value", "time", "parameters", "extras", "run_remote_endpoint", "timestamp"
-  ].map(col => `"log"."${col}"`)
-
-  const learnerCols = [
-    "learner_id", "run_remote_endpoint", "class_id", "runnable_url", "student_id", "class", "school", "user_id", "offering_id", "permission_forms", "username", "student_name", "teachers", "last_run", "query_id"
-  ]
-  .map(col => `"learner"."${col}"`)
-  .map(col => col === `"learner"."student_name"` && hideNames ? `"learner"."student_id" as student_name` : col).join(", ")
+  const logCols = getLogCols(hideNames)
+  const learnerCols = getLearnerCols(hideNames)
 
-  const cols = logCols.concat(learnerCols)
+  const cols = logCols.concat(learnerCols).join(", ")
 
   return `
   -- name ${runnableUrls.join(", ")}
@@ -615,17 +621,18 @@ exports.generateUserLogSQL = (usernames, activities, start_date, end_date) => {
 /*
 Generates a smaller row of event details only, no portal info.
 */
-exports.generateNarrowLogSQL = (queryIdsPerRunnable, authDomain, sourceKey) => {
+exports.generateNarrowLogSQL = (queryIdsPerRunnable, hideNames) => {
   const logDb = process.env.LOG_ATHENA_DB_NAME;
   const runnableUrls = Object.keys(queryIdsPerRunnable);
   const queryIds = Object.values(queryIdsPerRunnable);
+  const logCols = getLogCols(hideNames).join(", ");
 
   return `
   -- name ${runnableUrls.join(", ")}
   -- type learner event log ⎯ [qids: ${queryIds.join(", ")}]
   -- reportType narrow-learner-event-log
 
-  SELECT log.*
+  SELECT ${logCols}
   FROM "${logDb}"."logs_by_time" log
   INNER JOIN "report-service"."learners" learner
   ON