From 4080ea26892c6155eb9239b94102b32d7237678e Mon Sep 17 00:00:00 2001
From: Bryan Newbold <bnewbold@archive.org>
Date: Tue, 24 Jul 2018 12:16:00 -0700
Subject: more PDF mimetypes; fix return refactor

---
 scalding/src/main/scala/sandcrawler/CdxBackfillJob.scala | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

(limited to 'scalding')

diff --git a/scalding/src/main/scala/sandcrawler/CdxBackfillJob.scala b/scalding/src/main/scala/sandcrawler/CdxBackfillJob.scala
index eb168ac..f98b6e9 100644
--- a/scalding/src/main/scala/sandcrawler/CdxBackfillJob.scala
+++ b/scalding/src/main/scala/sandcrawler/CdxBackfillJob.scala
@@ -96,6 +96,10 @@ object CdxBackfillJob {
     val normalMime = Map(
       "application/pdf" -> "application/pdf",
       "application/x-pdf" -> "application/pdf",
+      "('application/pdf'" -> "application/pdf",
+      "image/pdf" -> "application/pdf",
+      "text/pdf" -> "application/pdf",
+      "\"application/pdf\"" -> "application/pdf",
       "application/postscript" -> "application/postscript",
       "text/html" -> "text/html",
       "text/xml" -> "text/xml",
@@ -105,8 +109,7 @@ object CdxBackfillJob {
     val lower = raw.toLowerCase()
     normalMime.foreach { case (key, value) =>
       if (lower.startsWith(key)) {
-        lower = value
-        break
+        return value
       }
     }
     lower
-- 
cgit v1.2.3