diff --git a/engine/src/main/java/org/archive/crawler/migrate/H1toH3.map b/engine/src/main/java/org/archive/crawler/migrate/H1toH3.map index 15119159..915fc746 100644 --- a/engine/src/main/java/org/archive/crawler/migrate/H1toH3.map +++ b/engine/src/main/java/org/archive/crawler/migrate/H1toH3.map @@ -95,27 +95,27 @@ metadata.operatorFrom //controller/*[@name="http-headers"]/*[@name="from"] CONTA //controller/*[@name="extract-processors"]/*[@name="ExtractorHTML"]/*[@name="extract-only-form-gets"] true //controller/*[@name="extract-processors"]/*[@name="ExtractorHTML"]/*[@name="extract-value-attributes"] true //controller/*[@name="extract-processors"]/*[@name="ExtractorHTML"]/*[@name="ignore-unexpected-html"] true -//controller/*[@name="extract-processors"]/*[@name="ExtractorCSS"]/*[@name="enabled"] true -//controller/*[@name="extract-processors"]/*[@name="ExtractorJS"]/*[@name="enabled"] true -//controller/*[@name="extract-processors"]/*[@name="ExtractorSWF"]/*[@name="enabled"] true -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="enabled"] true -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="compress"] true -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="prefix"] IAH -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="suffix"] ${HOSTNAME} -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="max-size-bytes"] 100000000 -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="path"]/string arcs -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="pool-max-active"] 5 -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="pool-max-wait"] 300000 -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="total-bytes-to-write"] 0 -//controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="skip-identical-digests"] false -//controller/*[@name="post-processors"]/*[@name="Updater"]/*[@name="enabled"] true -//controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="enabled"] true -//controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="override-logger"] false -//controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="seed-redirects-new-seed"] true -//controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="preference-depth-hops"] -1 -//controller/*[@name="post-processors"]/*[@name="Scheduler"]/*[@name="enabled"] true -//controller/*[@name="loggers"]/*[@name="crawl-statistics"]/*[@name="interval-seconds"] 20 -//controller/*[@name="checkpoint-copy-bdbje-logs"] true -//controller/*[@name="recover-retain-failures"] false -//controller/*[@name="recover-scope-includes"] true -//controller/*[@name="recover-scope-enqueues"] true \ No newline at end of file +? //controller/*[@name="extract-processors"]/*[@name="ExtractorCSS"]/*[@name="enabled"] true +? //controller/*[@name="extract-processors"]/*[@name="ExtractorJS"]/*[@name="enabled"] true +? //controller/*[@name="extract-processors"]/*[@name="ExtractorSWF"]/*[@name="enabled"] true +? //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="enabled"] true +warcWriter.compress //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="compress"] true +warcWriter.prefix //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="prefix"] IAH +warcWriter.suffix //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="suffix"] ${HOSTNAME} +warcWriter.maxFileSizeBytes //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="max-size-bytes"] 100000000 +? warcWriter.storePaths.(list) //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="path"]/string arcs +warcWriter.poolMaxActive //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="pool-max-active"] 5 +warcWriter.poolMaxWaitMs //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="pool-max-wait"] 300000 +warcWriter.maxTotalBytesToWrite //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="total-bytes-to-write"] 0 +warcWriter.skipIdenticalDigests //controller/*[@name="write-processors"]/*[@name="Archiver"]/*[@name="skip-identical-digests"] false +? //controller/*[@name="post-processors"]/*[@name="Updater"]/*[@name="enabled"] true ++ //controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="enabled"] true ++ linkScoper //controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="override-logger"] false +candidates.seedsRedirectNewSeeds //controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="seed-redirects-new-seed"] true +preparer.preferenceDepthHops //controller/*[@name="post-processors"]/*[@name="LinksScoper"]/*[@name="preference-depth-hops"] -1 +? //controller/*[@name="post-processors"]/*[@name="Scheduler"]/*[@name="enabled"] true +statisticsTracker.intervalSeconds //controller/*[@name="loggers"]/*[@name="crawl-statistics"]/*[@name="interval-seconds"] 20 +? //controller/*[@name="checkpoint-copy-bdbje-logs"] true +? //controller/*[@name="recover-retain-failures"] false +? //controller/*[@name="recover-scope-includes"] true +? //controller/*[@name="recover-scope-enqueues"] true \ No newline at end of file