[
https://issues.apache.org/jira/browse/NUTCH-2042?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Sebastian Nagel updated NUTCH-2042:
-----------------------------------
Attachment: NUTCH-2042-2x-v1.patch
NUTCH-2042-trunk-v1.patch
Trivial patches for trunk and 2.x, also simplifying getting the chunk String.
Of course, could use o.a.tika.parser.html.HtmlEncodingDetector (or parse-tika
;)) instead.
> parse-html increase chunk size used to detect charset
> -----------------------------------------------------
>
> Key: NUTCH-2042
> URL: https://issues.apache.org/jira/browse/NUTCH-2042
> Project: Nutch
> Issue Type: Bug
> Components: parser
> Affects Versions: 2.3, 1.10
> Reporter: Sebastian Nagel
> Priority: Minor
> Fix For: 2.4, 1.11
>
> Attachments: NUTCH-2042-2x-v1.patch, NUTCH-2042-trunk-v1.patch
>
>
> The chunk used to detect the encoding of a document is set to 2000 bytes.
> Although it is definitely best practice to "define" the character set on top,
> 2000 bytes are sometimes not enough: 20 longer <link> elements pointing to
> javascript and css libs may "hide" the <meta> element containing content type
> and encoding. Same problem has been observed in TIKA-357 and solved by
> increasing the buffer size to 8 kB.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
