Changing the patching strategy for contentUrl, atLocation.location an…

…d url (#5249) * Patch url to the new location for unknown files * Patch url to the new location for unknown files when matching the prefix * Patch location and url to the stripped contentUrl when it matches the prefix --------- Co-authored-by: Simon Dumas <[email protected]>
BlueBrain · Nov 26, 2024 · 2d6f506 · 2d6f506
1 parent 9fbf56f
commit 2d6f506
Show file tree

Hide file tree

Showing 4 changed files with 83 additions and 11 deletions.
diff --git a/ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/config/FileProcessingConfig.scala b/ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/config/FileProcessingConfig.scala
@@ -11,7 +11,7 @@ final case class FileProcessingConfig(
     importBucket: String,
     targetBucket: String,
     prefix: Option[Path],
-    locationPrefixToStripOpt: Option[Uri],
+    locationPrefixToStrip: Option[Uri],
     skipFileEvents: Boolean,
     mediaTypeDetector: MediaTypeDetectorConfig
 )

diff --git a/ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/resources/DistributionPatcher.scala b/ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/resources/DistributionPatcher.scala
@@ -69,10 +69,29 @@ final class DistributionPatcher(
   }
 
   private def stripLocationOnUnknownFile(json: Json): Json = {
+
+    def strip(original: String, locationPrefixToStrip: String) = original.replaceFirst(locationPrefixToStrip, "file://")
+
     locationPrefixToStripOpt.fold(json) { locationPrefixToStrip =>
-      root.atLocation.location.string.modify { location =>
-        location.replaceFirst(locationPrefixToStrip.toString, "file://")
-      }(json)
+      val locationPrefixToStripString = locationPrefixToStrip.toString
+      val matchingContentUrl          = root.contentUrl.string.getOption(json).filter(_.startsWith(locationPrefixToStripString))
+
+      matchingContentUrl match {
+        case Some(value) =>
+          // If there is a contentUrl matching the prefix it had the priority, we strip the path and apply it to contentUrl, location and url
+          val patchedValue = strip(value, locationPrefixToStripString)
+          setContentUrl(patchedValue).andThen(setLocation(patchedValue)).andThen(setUrl(patchedValue))(json)
+        case None        =>
+          // We fallback to a location matching the prefix, we patch the location and the url
+          val matchedLocation =
+            root.atLocation.location.string.getOption(json).filter(_.startsWith(locationPrefixToStripString))
+          matchedLocation match {
+            case Some(value) =>
+              val patchedValue = strip(value, locationPrefixToStripString)
+              setLocation(patchedValue).andThen(setUrl(patchedValue))(json)
+            case None        => json
+          }
+      }
     }
   }
 
@@ -95,6 +114,7 @@ final class DistributionPatcher(
   private def setContentUrl(newContentUrl: String)                = root.contentUrl.string.replace(newContentUrl)
   private def setLocation(newLocation: String)                    = (json: Json) =>
     json.deepMerge(Json.obj("atLocation" := Json.obj("location" := newLocation)))
+  private def setUrl(newUrl: String)                              = (json: Json) => json.deepMerge(Json.obj("url" := newUrl))
   private def setContentSize(newSize: Long)                       = (json: Json) =>
     json.deepMerge(Json.obj("contentSize" := Json.obj("unitCode" := "bytes", "value" := newSize)))
   private def setEncodingFormat(contentType: Option[ContentType]) = (json: Json) =>

diff --git a/ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/resources/SourcePatcher.scala b/ship/src/main/scala/ch/epfl/bluebrain/nexus/ship/resources/SourcePatcher.scala
@@ -64,7 +64,7 @@ object SourcePatcher {
         projectMapper,
         iriPatcher,
         targetBase,
-        config.files.locationPrefixToStripOpt,
+        config.files.locationPrefixToStrip,
         fetchFileAttributes
       )
     new SourcePatcher(distributionPatcher, iriPatcher)

diff --git a/ship/src/test/scala/ch/epfl/bluebrain/nexus/ship/resources/DistributionPatcherSuite.scala b/ship/src/test/scala/ch/epfl/bluebrain/nexus/ship/resources/DistributionPatcherSuite.scala
@@ -260,7 +260,7 @@ class DistributionPatcherSuite extends NexusSuite {
       .map(distrubutionDigest)
       .assertEquals(jobj"""{
                             "algorithm": "SHA-256",
-                            "value": "${digest}"
+                            "value": "$digest"
                           }""")
   }
 
@@ -279,20 +279,72 @@ class DistributionPatcherSuite extends NexusSuite {
     patcher.patchAll(input).assertEquals(expected)
   }
 
-  test("Patch and strip the distribution location when it matches the given prefix") {
+  test("Patch and strip the distribution contentUrl when it matches the prefix and set the value to location and url") {
     val input =
       json"""{
         "distribution": {
+          "contentUrl": "file:///location_to_strip/project/a/b/c/d/file.txt"
+        }
+      }"""
+
+    val expected = json"""{
+        "distribution": {
+          "atLocation": {
+            "location": "file:///project/a/b/c/d/file.txt"
+          },
+          "contentUrl": "file:///project/a/b/c/d/file.txt",
+          "url": "file:///project/a/b/c/d/file.txt"
+        }
+      }"""
+
+    patcher.patchAll(input).assertEquals(expected)
+  }
+
+  test("Do not patch the distribution contentUrl when it does not match the prefix") {
+    val input =
+      json"""{
+        "distribution": {
+          "contentUrl": "file:///some/other/location/project/a/b/c/d/file.txt"
+        }
+      }"""
+
+    patcher.patchAll(input).assertEquals(input)
+  }
+
+  test("Patch and strip the distribution location when it matches the prefix, setting the url to the same value") {
+    val input =
+      json"""{
+        "distribution": {
+          "url": "XXX",
           "atLocation": {
             "location": "file:///location_to_strip/project/a/b/c/d/file.txt"
           }
         }
       }"""
 
-    patcher
-      .patchAll(input)
-      .map(distributionLocation)
-      .assertEquals("file:///project/a/b/c/d/file.txt")
+    val expected = json"""{
+        "distribution": {
+          "url": "file:///project/a/b/c/d/file.txt",
+          "atLocation": {
+            "location": "file:///project/a/b/c/d/file.txt"
+          }
+        }
+      }"""
+
+    patcher.patchAll(input).assertEquals(expected)
+  }
+
+  test("Do not patch the location or the url if the distribution location when it does not match the prefix") {
+    val input =
+      json"""{
+        "distribution": {
+          "atLocation": {
+            "location": "file:///some/other/location/project/a/b/c/d/file.txt"
+          }
+        }
+      }"""
+
+    patcher.patchAll(input).assertEquals(input)
   }
 
   private def distributionContentSize(json: Json): JsonObject = {