This is an automated email from the ASF dual-hosted git repository.
nick pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/tika.git
commit 9a2c7d89e03ca7c0e821b69c394165297edfb9d4
Author: Nick Burch <n...@gagravarr.org>
AuthorDate: Thu Sep 6 09:28:14 2018 +0100
Mime magic for "MIME Encapsulation of Aggregate HTML Documents" (MHTML),
pulled out from rfc822 (may not be fully correct long-term...)
---
.../org/apache/tika/mime/tika-mimetypes.xml | 22 ++++++++++++++++++++--
1 file changed, 20 insertions(+), 2 deletions(-)
diff --git
a/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
b/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
index 007ec53..bd1adfa 100644
--- a/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
+++ b/tika-core/src/main/resources/org/apache/tika/mime/tika-mimetypes.xml
@@ -5980,9 +5980,28 @@
</magic>
<glob pattern="*.eml"/>
<glob pattern="*.mime"/>
+ <sub-class-of type="text/x-tika-text-based-message"/>
+ </mime-type>
+
+ <!-- TODO See TIKA-2723 for discussions on the mime type hierarchy -->
+ <!-- and best parser structure for these email-like formats -->
+ <mime-type type="multipart/related">
+ <acronym>MHTML</acronym>
+ <_comment>MIME Encapsulation of Aggregate HTML Documents</_comment>
+ <tika:link>http://tools.ietf.org/html/rfc2557</tika:link>
+ <alias type="application/x-mimearchive"/>
+ <alias type="message/rfc2557"/>
+ <!-- higher priority than message/rfc822 -->
+ <magic priority="60">
+ <match value="From: \x3cSaved by Windows Internet Explorer 8\x3e"
type="stringignorecase" offset="0"/>
+ <match value="From: \x22Saved by Internet Explorer 11\x22"
type="stringignorecase" offset="0"/>
+ <match value="MIME-Version: 1.0" type="string" offset="0">
+ <match value="\nContent-Type: multipart/related" type="string"
offset="16:512"/>
+ </match>
+ </magic>
<glob pattern="*.mht"/>
<glob pattern="*.mhtml"/>
- <sub-class-of type="text/x-tika-text-based-message"/>
+ <sub-class-of type="message/rfc822"/>
</mime-type>
<mime-type type="message/s-http"/>
@@ -6084,7 +6103,6 @@
<mime-type type="multipart/header-set"/>
<mime-type type="multipart/mixed"/>
<mime-type type="multipart/parallel"/>
- <mime-type type="multipart/related"/>
<mime-type type="multipart/report"/>
<mime-type type="multipart/signed"/>
<mime-type type="multipart/voice-message"/>
