diff --git a/code/processes/export-task-process/java/nu/marginalia/extractor/AtagExporter.java b/code/processes/export-task-process/java/nu/marginalia/extractor/AtagExporter.java
index cee07513..9c66b882 100644
--- a/code/processes/export-task-process/java/nu/marginalia/extractor/AtagExporter.java
+++ b/code/processes/export-task-process/java/nu/marginalia/extractor/AtagExporter.java
@@ -89,6 +89,8 @@ public class AtagExporter implements ExporterIf {
                 continue;
             if (null == doc.documentBody)
                 continue;
+            if (!doc.contentType.toLowerCase().startsWith("text/html"))
+                continue;
 
             var baseUrl = new EdgeUrl(doc.url);
             var parsed = Jsoup.parse(doc.documentBody);
diff --git a/code/processes/export-task-process/java/nu/marginalia/extractor/FeedExporter.java b/code/processes/export-task-process/java/nu/marginalia/extractor/FeedExporter.java
index 050b11f6..abe5c708 100644
--- a/code/processes/export-task-process/java/nu/marginalia/extractor/FeedExporter.java
+++ b/code/processes/export-task-process/java/nu/marginalia/extractor/FeedExporter.java
@@ -83,6 +83,8 @@ public class FeedExporter implements ExporterIf {
                 continue;
             if (null == doc.documentBody)
                 continue;
+            if (!doc.contentType.toLowerCase().startsWith("text/html"))
+                continue;
 
             var baseUrl = new EdgeUrl(doc.url);
             var parsed = Jsoup.parse(doc.documentBody);
diff --git a/code/processes/export-task-process/java/nu/marginalia/extractor/TermFrequencyExporter.java b/code/processes/export-task-process/java/nu/marginalia/extractor/TermFrequencyExporter.java
index d5d8184c..3255edf2 100644
--- a/code/processes/export-task-process/java/nu/marginalia/extractor/TermFrequencyExporter.java
+++ b/code/processes/export-task-process/java/nu/marginalia/extractor/TermFrequencyExporter.java
@@ -111,7 +111,7 @@ public class TermFrequencyExporter implements ExporterIf {
 
                 if (!(stream.next() instanceof CrawledDocument doc)) continue;
                 if (doc.documentBody == null) continue;
-                if (!doc.contentType.startsWith("text/html"))
+                if (!doc.contentType.toLowerCase().startsWith("text/html"))
                     continue;
 
                 docCount.incrementAndGet();