mirror of
https://github.com/internetarchive/heritrix3.git
synced 2026-11-09 15:36:40 +00:00
Fix for [HER-1291] accurate scoping not possible at log-recovery time; URIs may be recrawled against expectations
* FrontierJournal.java
refactor to always operate on CrawlURI instances (with hops-path and via info), and to write longer info on success/failure lines
This commit is contained in:
1 parent
a530af6e5c
commit
ab6942a60d
1 file changed
+10
-22
@@ -88,10 +88,13 @@ public class FrontierJournal extends CrawlerJournal implements Checkpointable {
|
||||
timestamp_interval = 10000;
|
||||
}
|
||||
|
||||
|
||||
public synchronized void added(CrawlURI curi) {
|
||||
public void added(CrawlURI curi) {
|
||||
writeLongUriLine(F_ADD, curi);
|
||||
}
|
||||
|
||||
public synchronized void writeLongUriLine(String tag, CrawlURI curi) {
|
||||
accumulatingBuffer.length(0);
|
||||
this.accumulatingBuffer.append(F_ADD).
|
||||
this.accumulatingBuffer.append(tag).
|
||||
append(curi.toString()).
|
||||
append(" ").
|
||||
append(curi.getPathFromSeed()).
|
||||
@@ -101,15 +104,7 @@ public class FrontierJournal extends CrawlerJournal implements Checkpointable {
|
||||
}
|
||||
|
||||
public void finishedSuccess(CrawlURI curi) {
|
||||
finishedSuccess(curi.toString());
|
||||
}
|
||||
|
||||
public void finishedSuccess(UURI uuri) {
|
||||
finishedSuccess(uuri.toString());
|
||||
}
|
||||
|
||||
protected void finishedSuccess(String uuri) {
|
||||
writeLine(F_SUCCESS, uuri);
|
||||
writeLongUriLine(F_SUCCESS, curi);
|
||||
}
|
||||
|
||||
public void emitted(CrawlURI curi) {
|
||||
@@ -118,20 +113,12 @@ public class FrontierJournal extends CrawlerJournal implements Checkpointable {
|
||||
}
|
||||
|
||||
public void finishedFailure(CrawlURI curi) {
|
||||
finishedFailure(curi.toString());
|
||||
writeLongUriLine(F_FAILURE,curi);
|
||||
}
|
||||
|
||||
public void finishedDisregard(CrawlURI curi) {
|
||||
writeLine(F_DISREGARD, curi.toString());
|
||||
}
|
||||
|
||||
public void finishedFailure(UURI uuri) {
|
||||
finishedFailure(uuri.toString());
|
||||
}
|
||||
|
||||
public void finishedFailure(String u) {
|
||||
writeLine(F_FAILURE, u);
|
||||
}
|
||||
|
||||
public void rescheduled(CrawlURI curi) {
|
||||
writeLine(F_RESCHEDULE, curi.toString());
|
||||
@@ -234,7 +221,8 @@ public class FrontierJournal extends CrawlerJournal implements Checkpointable {
|
||||
}
|
||||
frontier.considerIncluded(u);
|
||||
if (newJournal != null) {
|
||||
newJournal.writeLine(lineType, u.toString());
|
||||
// write same line as read
|
||||
newJournal.writeLine(read);
|
||||
}
|
||||
} catch (URIException e) {
|
||||
e.printStackTrace();
|
||||
|
||||
Reference in new issue
Block a user