From 39e4242dc736f4e8ded7b14afe2a01d53c95ae6f Mon Sep 17 00:00:00 2001
From: EarthlingDavey <15802017+EarthlingDavey@users.noreply.github.com>
Date: Thu, 5 Dec 2024 12:41:58 +0000
Subject: [PATCH] Allow retry of failed S3 sync.

---
 conf/node/controllers/httrack.js | 17 +++++++++++------
 conf/node/controllers/main.js    | 30 ++++++++++++++++++------------
 2 files changed, 29 insertions(+), 18 deletions(-)

diff --git a/conf/node/controllers/httrack.js b/conf/node/controllers/httrack.js
index 0a573c1..ef524bb 100644
--- a/conf/node/controllers/httrack.js
+++ b/conf/node/controllers/httrack.js
@@ -141,6 +141,7 @@ export const getHttrackProgress = async (dest) => {
   }
 
   const files = {
+    index: `${dest}/index.html`,
     log: `${dest}/hts-cache/new.txt`,
     lock: `${dest}/hts-in_progress.lock`,
   };
@@ -148,7 +149,7 @@ export const getHttrackProgress = async (dest) => {
   const response = {
     requestCount: 0,
     rate: 0,
-    complete: !fs.existsSync(files.lock),
+    complete: fs.existsSync(files.index) && !fs.existsSync(files.lock),
   };
 
   if (!fs.existsSync(files.log)) {
@@ -167,7 +168,11 @@ export const getHttrackProgress = async (dest) => {
   // The first line is a header, so we subtract 1.
   response.requestCount = lineCount - 1;
 
-  // If the time is 23:59:55 - 00:00:05, await 5 seconds.
+  // If the current time is between 00:00:00 and 00:00:05, wait for 5 seconds to avoid spanning requests across two days.
+  // This prevents the recent requests in `/hts-cache/new.txt` from spanning 2 days.
+  if (new Date().toTimeString().split(" ")[0] < "00:00:05") {
+    await new Promise((resolve) => setTimeout(resolve, 5000));
+  }
 
   // Get the last 20 lines from the file. Or all of them if there are less than 20.
   const lastLines = execSync(
@@ -229,10 +234,10 @@ export const waitForHttrackComplete = async (
     fs.existsSync(`${dest}/hts-in_progress.lock`)
   ) {
     if (iterations < 10 || iterations % logFrequency === 0) {
-      const elapsedTime = new Date(iterations * intervalSeconds * 1000).toISOString().substring(11, 19);
-      console.log(
-        `Waiting for httrack to complete ... ${elapsedTime} elapsed`,
-      );
+      const elapsedTime = new Date(iterations * intervalSeconds * 1000)
+        .toISOString()
+        .substring(11, 19);
+      console.log(`Waiting for httrack to complete ... ${elapsedTime} elapsed`);
     }
     await new Promise((resolve) => setTimeout(resolve, intervalSeconds * 1000));
   }
diff --git a/conf/node/controllers/main.js b/conf/node/controllers/main.js
index bbe1f3d..dded6da 100644
--- a/conf/node/controllers/main.js
+++ b/conf/node/controllers/main.js
@@ -5,6 +5,7 @@ import {
   getSnapshotPaths,
   getHttrackArgs,
   runHttrack,
+  getHttrackProgress,
   waitForHttrackComplete,
 } from "./httrack.js";
 import { sync } from "./s3.js";
@@ -20,21 +21,26 @@ import { sync } from "./s3.js";
 export const main = async ({ url, agency, depth }) => {
   const paths = getSnapshotPaths({ host: url.host, agency });
 
-  const httrackArgs = getHttrackArgs({
-    url,
-    dest: paths.fs,
-    agency,
-    jwt,
-    depth,
-  });
+  const { complete } = getHttrackProgress(paths.fs);
 
-  runHttrack(httrackArgs);
+  // If the snapshot is already complete, skip httrack
+  if (!complete) {
+    const httrackArgs = getHttrackArgs({
+      url,
+      dest: paths.fs,
+      agency,
+      jwt,
+      depth,
+    });
 
-  const { timedOut } = await waitForHttrackComplete(paths.fs);
+    runHttrack(httrackArgs);
 
-  if (timedOut) {
-    console.error("Httrack timed out", { url: url.href, agency, depth });
-    return;
+    const { timedOut } = await waitForHttrackComplete(paths.fs);
+
+    if (timedOut) {
+      console.error("Httrack timed out", { url: url.href, agency, depth });
+      return;
+    }
   }
 
   // Remove sensitive files - before syncing to S3