diff --git a/engine/src/main/java/org/archive/crawler/postprocessor/DispositionProcessor.java b/engine/src/main/java/org/archive/crawler/postprocessor/DispositionProcessor.java index 49ae9e72..38f3284d 100644 --- a/engine/src/main/java/org/archive/crawler/postprocessor/DispositionProcessor.java +++ b/engine/src/main/java/org/archive/crawler/postprocessor/DispositionProcessor.java @@ -189,12 +189,6 @@ public class DispositionProcessor extends Processor { // Update robots info try { if ("/robots.txt".equals(curi.getUURI().getPath())) { - // Update server with robots info - // NOTE: in some cases the curi's status can be changed here -// server.updateRobots(metadata.getRobotsHonoringPolicy(), curi); - - server.updateRobots(curi); - // shortcut retries w/ DEEMED when ignore-all if (metadata.getRobotsPolicy() instanceof IgnoreRobotsPolicy) { if(curi.getFetchStatus() < 0 && curi.getFetchStatus()!=S_DEFERRED) { @@ -202,6 +196,10 @@ public class DispositionProcessor extends Processor { curi.setFetchStatus(S_DEEMED_NOT_FOUND); } } + + // Update server with robots info + // NOTE: in some cases the curi's status can be changed here + server.updateRobots(curi); } } catch (URIException e) { diff --git a/engine/src/main/java/org/archive/crawler/prefetch/PreconditionEnforcer.java b/engine/src/main/java/org/archive/crawler/prefetch/PreconditionEnforcer.java index d1de5341..1c2c7f19 100644 --- a/engine/src/main/java/org/archive/crawler/prefetch/PreconditionEnforcer.java +++ b/engine/src/main/java/org/archive/crawler/prefetch/PreconditionEnforcer.java @@ -198,7 +198,7 @@ public class PreconditionEnforcer extends Processor { * @param curi CrawlURI we're checking for any required preconditions. * @return True, if this curi has a precondition or processing * should be terminated for some other reason. False if - * we can precede to process this url. + * we can proceed to process this url. */ protected boolean considerRobotsPreconditions(CrawlURI curi) { // treat /robots.txt fetches specially 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 51ee0442..605eab5c 100644 --- a/modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java +++ b/modules/src/main/java/org/archive/modules/fetcher/FetchHTTP.java @@ -620,6 +620,10 @@ public class FetchHTTP extends Processor implements Lifecycle { } }; } + + // Save method into curi too. Midfetch filters may want to leverage + // info in here. + curi.setHttpMethod(method); HostConfiguration customConfigOrNull = configureMethod(curi, method); @@ -793,9 +797,6 @@ public class FetchHTTP extends Processor implements Lifecycle { curi.setFetchStatus(method.getStatusCode()); Header ct = method.getResponseHeader("content-type"); curi.setContentType((ct == null) ? null : ct.getValue()); - // Save method into curi too. Midfetch filters may want to leverage - // info in here. - curi.setHttpMethod(method); } /** diff --git a/modules/src/main/java/org/archive/modules/net/CrawlServer.java b/modules/src/main/java/org/archive/modules/net/CrawlServer.java index 4038aaee..ad535d78 100644 --- a/modules/src/main/java/org/archive/modules/net/CrawlServer.java +++ b/modules/src/main/java/org/archive/modules/net/CrawlServer.java @@ -142,7 +142,7 @@ public class CrawlServer implements Serializable, FetchStats.HasFetchStats, Iden robotsFetched = System.currentTimeMillis(); boolean gotSomething = curi.getFetchType() == HTTP_GET - && (curi.getFetchStatus() > 0 || curi.getFetchStatus() == S_DEEMED_NOT_FOUND ); + && (curi.getFetchStatus() > 0 || curi.getFetchStatus() == S_DEEMED_NOT_FOUND); if (!gotSomething && curi.getFetchAttempts() < MIN_ROBOTS_RETRIES) {