From 548340782943d709e4cab5c2acc455db2e68fdb4 Mon Sep 17 00:00:00 2001 From: Adam Miller Date: Thu, 6 Jun 2024 14:32:45 -0700 Subject: [PATCH] Make logging of metadata record optional, but default. --- .../modules/extractor/ExtractorYoutubeDL.java | 17 +++++++++++++++++ 1 file changed, 17 insertions(+) diff --git a/contrib/src/main/java/org/archive/modules/extractor/ExtractorYoutubeDL.java b/contrib/src/main/java/org/archive/modules/extractor/ExtractorYoutubeDL.java index af9ecd0b..89015b64 100644 --- a/contrib/src/main/java/org/archive/modules/extractor/ExtractorYoutubeDL.java +++ b/contrib/src/main/java/org/archive/modules/extractor/ExtractorYoutubeDL.java @@ -27,6 +27,7 @@ import java.io.FileInputStream; import java.io.IOException; import java.io.InputStream; import java.io.InputStreamReader; +import java.io.OutputStream; import java.io.RandomAccessFile; import java.io.Reader; import java.io.UnsupportedEncodingException; @@ -187,6 +188,19 @@ public class ExtractorYoutubeDL extends Extractor this.controller = controller; } + { + setLogMetadataRecord(true); + } + public boolean getLogMetadataRecord() { + return (Boolean) kp.get("logMetadataRecord"); + } + /** + * Whether or not to create a crawl.log entry for any WARC Metadata Records written. + */ + public void setLogMetadataRecord(boolean logMetadataRecord) { + kp.put("logMetadataRecord",logMetadataRecord); + } + @Override public void start() { if (!isRunning) { @@ -649,6 +663,9 @@ public class ExtractorYoutubeDL extends Extractor */ @Override public void postWrite(WARCRecordInfo recordInfo, CrawlURI curi) { + if(!this.getLogMetadataRecord()) + return; + CrawlURI pseudoCuri = null; try { pseudoCuri = curi.createCrawlURI(recordInfo.getUrl(), LinkContext.EMBED_MISC, Hop.INFERRED);