Fetch all hashes (#45)

* [core] close files after calculating their MD5 hash

FileInputStream was never closed, so eventually ran into
ToManyFilesOpen.

Will come back to look at this again with IO.bracket for better
guarantee that FIS is closed.

Signed-off-by: Paul Campbell <pcampbell@kemitix.net>

* [aws-lib] Fetch all MD5 hashes under prefix

Initial request only returns the first 1000.

aws-lib/src/main/scala/net/kemitix/s3thorp/aws/lib/S3ClientLogging.scala

@@ -1,14 +1,14 @@
 package net.kemitix.s3thorp.aws.lib
 
 import cats.effect.IO
-import com.amazonaws.services.s3.model.{CopyObjectResult, DeleteObjectsResult, ListObjectsV2Result, PutObjectResult}
+import com.amazonaws.services.s3.model.{CopyObjectResult, DeleteObjectsResult, ListObjectsV2Result, PutObjectResult, S3ObjectSummary}
 import net.kemitix.s3thorp.domain.{Bucket, LocalFile, RemoteKey}
 
 object S3ClientLogging {
 
   def logListObjectsStart(bucket: Bucket,
                           prefix: RemoteKey)
-                         (implicit info: Int => String => Unit): ListObjectsV2Result => IO[ListObjectsV2Result] =
+                         (implicit info: Int => String => Unit): Stream[S3ObjectSummary] => IO[Stream[S3ObjectSummary]] =
     in => IO {
       info(3)(s"Fetch S3 Summary: ${bucket.name}:${prefix.key}")
       in
@@ -16,7 +16,7 @@ object S3ClientLogging {
 
   def logListObjectsFinish(bucket: Bucket,
                            prefix: RemoteKey)
-                          (implicit info: Int => String => Unit): ListObjectsV2Result => IO[Unit] =
+                          (implicit info: Int => String => Unit): Stream[S3ObjectSummary] => IO[Unit] =
     _ => IO {
       info(2)(s"Fetched S3 Summary: ${bucket.name}:${prefix.key}")
     }

aws-lib/src/main/scala/net/kemitix/s3thorp/aws/lib/S3ClientObjectLister.scala

@@ -2,7 +2,7 @@ package net.kemitix.s3thorp.aws.lib
 
 import cats.effect.IO
 import com.amazonaws.services.s3.AmazonS3
-import com.amazonaws.services.s3.model.ListObjectsV2Request
+import com.amazonaws.services.s3.model.{ListObjectsV2Request, S3ObjectSummary}
 import net.kemitix.s3thorp.aws.lib.S3ClientLogging.{logListObjectsFinish, logListObjectsStart}
 import net.kemitix.s3thorp.aws.lib.S3ObjectsByHash.byHash
 import net.kemitix.s3thorp.aws.lib.S3ObjectsByKey.byKey
@@ -15,17 +15,43 @@ class S3ClientObjectLister(amazonS3: AmazonS3) {
   def listObjects(bucket: Bucket,
                   prefix: RemoteKey)
                  (implicit info: Int => String => Unit): IO[S3ObjectsData] = {
-    val request = new ListObjectsV2Request()
-        .withBucketName(bucket.name)
-        .withPrefix(prefix.key)
-    IO {
-      amazonS3.listObjectsV2(request)
-    }.bracket(
-      logListObjectsStart(bucket, prefix))(
-      logListObjectsFinish(bucket,prefix))
-      .map(_.getObjectSummaries)
-      .map(_.asScala)
-      .map(_.toStream)
+
+    type Token = String
+    type Batch = (Stream[S3ObjectSummary], Option[Token])
+
+    val requestInitial = new ListObjectsV2Request()
+      .withBucketName(bucket.name)
+      .withPrefix(prefix.key)
+
+    val requestMore = (token:Token) => new ListObjectsV2Request()
+      .withBucketName(bucket.name)
+      .withPrefix(prefix.key)
+      .withContinuationToken(token)
+
+    def fetchBatch: ListObjectsV2Request => IO[Batch] =
+      request => IO{
+        val result = amazonS3.listObjectsV2(request)
+        val more: Option[Token] =
+          if (result.isTruncated) Some(result.getNextContinuationToken)
+          else None
+        (result.getObjectSummaries.asScala.toStream, more)
+      }
+
+    def fetchAll: ListObjectsV2Request => IO[Stream[S3ObjectSummary]] =
+      request =>
+          for {
+            batch <- fetchBatch(request)
+            (summaries, more) = batch
+            rest <- more match {
+              case None => IO{Stream()}
+              case Some(token) => fetchAll(requestMore(token))
+            }
+          } yield summaries ++ rest
+
+    fetchAll(requestInitial)
+      .bracket(
+        logListObjectsStart(bucket, prefix))(
+        logListObjectsFinish(bucket,prefix))
       .map(os => S3ObjectsData(byHash(os), byKey(os)))
   }
 

core/src/main/scala/net.kemitix.s3thorp.core/MD5HashGenerator.scala

@@ -24,6 +24,7 @@ object MD5HashGenerator {
     fis read buffer
     val hash = md5PartBody(buffer)
     info(5)(s"md5:generated:${hash.hash}")
+    fis.close
     hash
   }