From 8acae852a0dba1db846ff0a98c5972c4753db046 Mon Sep 17 00:00:00 2001
From: Michael Peter Christen <mc@yacy.net>
Date: Wed, 25 Jun 2014 11:51:11 +0200
Subject: [PATCH] write <em>-tagged texts also into the bold_txt field

---
 source/net/yacy/document/parser/html/ContentScraper.java | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/source/net/yacy/document/parser/html/ContentScraper.java b/source/net/yacy/document/parser/html/ContentScraper.java
index 896ad89f0..a41a9cc4d 100644
--- a/source/net/yacy/document/parser/html/ContentScraper.java
+++ b/source/net/yacy/document/parser/html/ContentScraper.java
@@ -109,6 +109,7 @@ public class ContentScraper extends AbstractScraper implements Scraper {
         h6(TagType.pair),
         title(TagType.pair),
         b(TagType.pair),
+        em(TagType.pair),
         strong(TagType.pair),
         u(TagType.pair),
         i(TagType.pair),
@@ -563,6 +564,9 @@ public class ContentScraper extends AbstractScraper implements Scraper {
         } else if ((tag.name.equalsIgnoreCase("strong")) && (tag.content.length() < 1024)) {
             h = cleanLine(CharacterCoding.html2unicode(stripAllTags(tag.content.getChars())));
             if (h.length() > 0) this.bold.inc(h);
+        } else if ((tag.name.equalsIgnoreCase("em")) && (tag.content.length() < 1024)) {
+            h = cleanLine(CharacterCoding.html2unicode(stripAllTags(tag.content.getChars())));
+            if (h.length() > 0) this.bold.inc(h);
         } else if ((tag.name.equalsIgnoreCase("i")) && (tag.content.length() < 1024)) {
             h = cleanLine(CharacterCoding.html2unicode(stripAllTags(tag.content.getChars())));
             if (h.length() > 0) this.italic.inc(h);