From a9d36bb268728b4b9e2654905558897da5db2da7 Mon Sep 17 00:00:00 2001
From: achingbrain <alex@achingbrain.net>
Date: Thu, 16 Mar 2023 14:03:51 +0100
Subject: [PATCH] fix: test for file under/over reads

When we read too much or too little data, it normally means that the
UnixFS metadata in the root node is incorrect, so throw an error.
---
 .../src/resolvers/unixfs-v1/content/file.ts   |  16 +-
 .../test/exporter-esoteric.spec.ts            | 150 ++++++++++++++++++
 2 files changed, 164 insertions(+), 2 deletions(-)

diff --git a/packages/ipfs-unixfs-exporter/src/resolvers/unixfs-v1/content/file.ts b/packages/ipfs-unixfs-exporter/src/resolvers/unixfs-v1/content/file.ts
index 21ff5005..c5f80b51 100644
--- a/packages/ipfs-unixfs-exporter/src/resolvers/unixfs-v1/content/file.ts
+++ b/packages/ipfs-unixfs-exporter/src/resolvers/unixfs-v1/content/file.ts
@@ -137,15 +137,27 @@ const fileContent: UnixfsV1Resolver = (cid, node, unixfs, path, resolve, depth,
       return
     }
 
+    let read = 0n
     const queue = pushable()
 
     void walkDAG(blockstore, node, queue, 0n, offset, offset + length, options)
+      .then(() => {
+        const wanted = length - offset
+
+        if (read < wanted) {
+          throw errCode(new Error('Traversed entire DAG but did not read enough bytes'), 'ERR_UNDER_READ')
+        }
+
+        if (read > wanted) {
+          throw errCode(new Error('Read too many bytes - the file size reported by the UnixFS data in the root node may be incorrect'), 'ERR_OVER_READ')
+        }
+
+        queue.end()
+      })
       .catch(err => {
         queue.end(err)
       })
 
-    let read = 0n
-
     for await (const buf of queue) {
       if (buf == null) {
         continue
diff --git a/packages/ipfs-unixfs-exporter/test/exporter-esoteric.spec.ts b/packages/ipfs-unixfs-exporter/test/exporter-esoteric.spec.ts
index ad10c2d4..a99de23d 100644
--- a/packages/ipfs-unixfs-exporter/test/exporter-esoteric.spec.ts
+++ b/packages/ipfs-unixfs-exporter/test/exporter-esoteric.spec.ts
@@ -211,4 +211,154 @@ describe('exporter esoteric DAGs', () => {
     const data = uint8ArrayConcat(await all(exported.content()))
     expect(data).to.deep.equal(buf)
   })
+
+  it('errors on DAG with blocksizes that are too large', async () => {
+    const leaves = await Promise.all([
+      randomBytes(5),
+      randomBytes(3),
+      randomBytes(6)
+    ].map(async buf => {
+      return {
+        cid: await storeBlock(buf, raw.code),
+        buf
+      }
+    }))
+
+    const unixfs = new UnixFS({
+      type: 'file',
+      blockSizes: [
+        BigInt(leaves[0].buf.byteLength),
+        BigInt(leaves[1].buf.byteLength + 5), // this is wrong
+        BigInt(leaves[2].buf.byteLength)
+      ]
+    })
+
+    const rootNode = {
+      Data: unixfs.marshal(),
+      Links: [{
+        Name: '',
+        Hash: leaves[0].cid,
+        Tsize: leaves[0].buf.byteLength
+      }, {
+        Name: '',
+        Hash: leaves[1].cid,
+        Tsize: leaves[1].buf.byteLength
+      }, {
+        Name: '',
+        Hash: leaves[2].cid,
+        Tsize: leaves[2].buf.byteLength
+      }]
+    }
+
+    const rootBuf = dagPb.encode(rootNode)
+    const rootCid = await storeBlock(rootBuf, dagPb.code)
+    const exported = await exporter(rootCid, block)
+
+    if (exported.type !== 'file') {
+      throw new Error('Unexpected type')
+    }
+
+    await expect(all(exported.content())).to.eventually.be.rejected
+      .with.property('code', 'ERR_UNDER_READ')
+  })
+
+  it('errors on DAG with blocksizes that are too small', async () => {
+    const leaves = await Promise.all([
+      randomBytes(5),
+      randomBytes(3),
+      randomBytes(6)
+    ].map(async buf => {
+      return {
+        cid: await storeBlock(buf, raw.code),
+        buf
+      }
+    }))
+
+    const unixfs = new UnixFS({
+      type: 'file',
+      blockSizes: [
+        BigInt(leaves[0].buf.byteLength),
+        BigInt(leaves[1].buf.byteLength - 2), // this is wrong
+        BigInt(leaves[2].buf.byteLength)
+      ]
+    })
+
+    const rootNode = {
+      Data: unixfs.marshal(),
+      Links: [{
+        Name: '',
+        Hash: leaves[0].cid,
+        Tsize: leaves[0].buf.byteLength
+      }, {
+        Name: '',
+        Hash: leaves[1].cid,
+        Tsize: leaves[1].buf.byteLength
+      }, {
+        Name: '',
+        Hash: leaves[2].cid,
+        Tsize: leaves[2].buf.byteLength
+      }]
+    }
+
+    const rootBuf = dagPb.encode(rootNode)
+    const rootCid = await storeBlock(rootBuf, dagPb.code)
+    const exported = await exporter(rootCid, block)
+
+    if (exported.type !== 'file') {
+      throw new Error('Unexpected type')
+    }
+
+    await expect(all(exported.content())).to.eventually.be.rejected
+      .with.property('code', 'ERR_OVER_READ')
+  })
+
+  it('errors on DAG with incorrect number of blocksizes', async () => {
+    const leaves = await Promise.all([
+      randomBytes(5),
+      randomBytes(3),
+      randomBytes(6)
+    ].map(async buf => {
+      return {
+        cid: await storeBlock(buf, raw.code),
+        buf
+      }
+    }))
+
+    const unixfs = new UnixFS({
+      type: 'file',
+      blockSizes: [
+        BigInt(leaves[0].buf.byteLength),
+        // BigInt(leaves[1].buf.byteLength), // this is wrong
+        BigInt(leaves[2].buf.byteLength)
+      ]
+    })
+
+    const rootNode = {
+      Data: unixfs.marshal(),
+      Links: [{
+        Name: '',
+        Hash: leaves[0].cid,
+        Tsize: leaves[0].buf.byteLength
+      }, {
+        Name: '',
+        Hash: leaves[1].cid,
+        Tsize: leaves[1].buf.byteLength
+      }, {
+        Name: '',
+        Hash: leaves[2].cid,
+        Tsize: leaves[2].buf.byteLength
+      }]
+    }
+
+    const rootBuf = dagPb.encode(rootNode)
+    const rootCid = await storeBlock(rootBuf, dagPb.code)
+    const exported = await exporter(rootCid, block)
+
+    if (exported.type !== 'file') {
+      throw new Error('Unexpected type')
+    }
+
+    await expect(all(exported.content())).to.eventually.be.rejected
+      .with.property('code', 'ERR_NOT_UNIXFS')
+  })
 })