From d4f45cf05e3af5d57cf402665b669ab246b5a667 Mon Sep 17 00:00:00 2001 From: Michael Peter Christen Date: Tue, 24 Jan 2017 15:56:29 +0100 Subject: [PATCH] added dc.date.modified and dc.date.created to date parser --- source/net/yacy/document/parser/html/ContentScraper.java | 8 ++++++++ 1 file changed, 8 insertions(+) diff --git a/source/net/yacy/document/parser/html/ContentScraper.java b/source/net/yacy/document/parser/html/ContentScraper.java index 15f3bace2..ce13b72e6 100644 --- a/source/net/yacy/document/parser/html/ContentScraper.java +++ b/source/net/yacy/document/parser/html/ContentScraper.java @@ -1128,6 +1128,14 @@ public class ContentScraper extends AbstractScraper implements Scraper { content = this.metas.get("date"); if (content != null) try {return ISO8601Formatter.FORMATTER.parse(content, this.timezoneOffset).getTime();} catch (ParseException e) {} + // + content = this.metas.get("dc.date.modified"); + if (content != null) try {return ISO8601Formatter.FORMATTER.parse(content, this.timezoneOffset).getTime();} catch (ParseException e) {} + + // + content = this.metas.get("dc.date.created"); + if (content != null) try {return ISO8601Formatter.FORMATTER.parse(content, this.timezoneOffset).getTime();} catch (ParseException e) {} + // content = this.metas.get("dc.date"); if (content != null) try {return ISO8601Formatter.FORMATTER.parse(content, this.timezoneOffset).getTime();} catch (ParseException e) {}