From 701a64cdded5823313d6631d3e6dc206e065c9fa Mon Sep 17 00:00:00 2001
From: David M Penner <david.penner@emu.edu>
Date: Tue, 24 Mar 2020 17:48:10 -0400
Subject: [PATCH] Capture dml_write_exceptions from parse_html()

---
 classes/robot/crawler.php | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)

diff --git a/classes/robot/crawler.php b/classes/robot/crawler.php
index 0e8c0829..6fce7faa 100644
--- a/classes/robot/crawler.php
+++ b/classes/robot/crawler.php
@@ -619,7 +619,15 @@ public function crawl($node, $verbose = false) {
                 // Look for new links on this page from the html.
                 // Insert new links into tool_crawler_edge, and into tool_crawler_url table.
                 // Find the course, cm, and context of where we are for the main scraped URL.
-                $this->parse_html($result, $result->externalurl, $verbose);
+                try {
+                    $this->parse_html($result, $result->externalurl, $verbose);
+                } catch (\dml_write_exception $e) {
+                    mtrace("Database write error while processing page '{$result->url}'");
+                    if ($verbose) {
+                        mtrace("Exception: <" . get_class($e) . ">: \"" . $e->getMessage() . "\" in {$e->getFile()} at line {$e->getLine()}");
+                        mtrace("Trace:\n{$e->getTraceAsString()}");
+                    }
+                }
             } else {
                 if ($verbose) {
                     echo "NOT html\n";