From 7312ad597ebb616b61e52f4d68e3252d2c40b3b6 Mon Sep 17 00:00:00 2001 From: Noah Levitt Date: Fri, 7 Nov 2014 13:21:42 -0800 Subject: [PATCH] so that it can appear in the crawl log, add contentSize to CrawlURI extraInfo, for http transactions (the only case in heritrix currently where contentLength and contentSize differ) --- .../java/org/archive/crawler/io/UriProcessingFormatter.java | 4 +--- .../src/main/java/org/archive/modules/fetcher/FetchHTTP.java | 4 ++++ 2 files changed, 5 insertions(+), 3 deletions(-) diff --git a/engine/src/main/java/org/archive/crawler/io/UriProcessingFormatter.java b/engine/src/main/java/org/archive/crawler/io/UriProcessingFormatter.java index 02cb9a67..5c12e4ae 100644 --- a/engine/src/main/java/org/archive/crawler/io/UriProcessingFormatter.java +++ b/engine/src/main/java/org/archive/crawler/io/UriProcessingFormatter.java @@ -83,14 +83,12 @@ extends Formatter implements Preformatter, CoreAttributeConstants { } else if (curi.getContentSize() > 0) { length = Long.toString(curi.getContentSize()); } - mime = curi.getContentType(); } else { if (curi.getContentSize() > 0) { length = Long.toString(curi.getContentSize()); } - mime = curi.getContentType(); } - mime = MimetypeUtils.truncate(mime); + mime = MimetypeUtils.truncate(curi.getContentType()); long time = System.currentTimeMillis(); diff --git a/modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java b/modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java index 249334df..fed4e501 100644 --- a/modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java +++ b/modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java @@ -962,6 +962,10 @@ public class FetchHTTP extends Processor implements Lifecycle { protected void setSizes(CrawlURI curi, Recorder rec) { // set reporting size curi.setContentSize(rec.getRecordedInput().getSize()); + + // add contentSize to extraInfo so it's available to log in the crawl log + curi.addExtraInfo("contentSize", rec.getRecordedInput().getSize()); + // special handling for 304-not modified if (curi.getFetchStatus() == HttpStatus.SC_NOT_MODIFIED && curi.getFetchHistory() != null) {