chore: handle offsets when data is on internal and leaf nodes

Author: achingbrain

package.json

@@ -59,7 +59,7 @@
     "bs58": "^4.0.1",
     "cids": "~0.5.3",
     "deep-extend": "~0.6.0",
-    "ipfs-unixfs": "~0.1.14",
+    "ipfs-unixfs": "~0.1.15",
     "ipld": "~0.17.2",
     "ipld-dag-pb": "~0.14.4",
     "left-pad": "^1.3.0",

src/exporter/file.js

@@ -139,6 +139,12 @@ function streamBytes (dag, node, fileSize, offset, length) {
 function extractDataFromBlock (block, streamPosition, begin, end) {
   const blockLength = block.length
 
+  if (begin >= streamPosition + blockLength) {
+    // If begin is after the start of the block, return an empty block
+    // This can happen when internal nodes contain data
+    return Buffer.alloc(0)
+  }
+
   if (end - streamPosition < blockLength) {
     // If the end byte is in the current block, truncate the block to the end byte
     block = block.slice(0, end - streamPosition)

test/exporter.js

@@ -713,6 +713,27 @@ module.exports = (repo) => {
         }
       ], done)
     })
+
+    it('exports file with data on internal and leaf nodes with an offset that only fetches data from leaf nodes', function (done) {
+      waterfall([
+        (cb) => createAndPersistNode(ipld, 'raw', [0x04, 0x05, 0x06, 0x07], [], cb),
+        (leaf, cb) => createAndPersistNode(ipld, 'file', [0x00, 0x01, 0x02, 0x03], [
+          leaf
+        ], cb),
+        (file, cb) => {
+          pull(
+            exporter(file.multihash, ipld, {
+              offset: 4
+            }),
+            pull.asyncMap((file, cb) => readFile(file, cb)),
+            pull.through(buffer => {
+              expect(buffer).to.deep.equal(Buffer.from([0x04, 0x05, 0x06, 0x07]))
+            }),
+            pull.collect(cb)
+          )
+        }
+      ], done)
+    })
   })
 }