support pause interrupt: (#825)

- add new interrupt reason / exit code
- add isCrawlPaused() which checks redis <id>:paused key
- exit gracefully, upload WACZ file when paused

fixes #824

src/crawler.ts

@@ -1497,6 +1497,10 @@ self.__bx_behaviors.selectMainBehavior();
       }
     }
 
+    if (await this.crawlState.isCrawlPaused()) {
+      interrupt = InterruptReason.CrawlPaused;
+    }
+
     if (interrupt) {
       this.uploadAndDeleteLocal = true;
       this.gracefulFinishOnInterrupt(interrupt);
@@ -1859,12 +1863,9 @@ self.__bx_behaviors.selectMainBehavior();
       if (isFinished || (await this.crawlState.isCrawlCanceled())) {
         return;
       }
-      // if stopped, won't get anymore data
-      if (await this.crawlState.isCrawlStopped()) {
-        // possibly restarted after committing, so assume done here!
-        if ((await this.crawlState.numDone()) > 0) {
-          return;
-        }
+      // possibly restarted after committing, so assume done here!
+      if ((await this.crawlState.numDone()) > 0) {
+        return;
       }
       // fail crawl otherwise
       logger.fatal("No WARC Files, assuming crawl failed");

src/util/constants.ts

@@ -88,4 +88,5 @@ export enum InterruptReason {
   DiskUtilization = 4,
   BrowserCrashed = 5,
   SignalInterrupted = 6,
+  CrawlPaused = 7,
 }

src/util/state.ts

@@ -521,6 +521,14 @@ return inx;
     return false;
   }
 
+  async isCrawlPaused() {
+    if ((await this.redis.get(`${this.key}:paused`)) === "1") {
+      return true;
+    }
+
+    return false;
+  }
+
   async isCrawlCanceled() {
     return (await this.redis.get(`${this.key}:canceled`)) === "1";
   }