Add helpers to extract CONTRACT_CODE entries efficiently from BL.

Author: graydon

Diff for: src/bucket/BucketSnapshot.cpp

@@ -313,6 +313,43 @@ LiveBucketSnapshot::scanForEviction(
     return Loop::INCOMPLETE;
 }
 
+// Scans contract code entries in the bucket.
+Loop
+LiveBucketSnapshot::scanForContractCode(
+    std::function<Loop(BucketEntry const&)> callback) const
+{
+    ZoneScoped;
+    if (isEmpty())
+    {
+        return Loop::INCOMPLETE;
+    }
+
+    auto range = mBucket->getContractCodeRange();
+    if (!range)
+    {
+        return Loop::INCOMPLETE;
+    }
+
+    auto& stream = getStream();
+    stream.seek(range->first);
+
+    BucketEntry be;
+    while (stream.pos() < range->second && stream.readOne(be))
+    {
+
+        if (((be.type() == LIVEENTRY || be.type() == INITENTRY) &&
+             be.liveEntry().data.type() == CONTRACT_CODE) ||
+            (be.type() == DEADENTRY && be.deadEntry().type() == CONTRACT_CODE))
+        {
+            if (callback(be) == Loop::COMPLETE)
+            {
+                return Loop::COMPLETE;
+            }
+        }
+    }
+    return Loop::INCOMPLETE;
+}
+
 template <class BucketT>
 XDRInputFileStream&
 BucketSnapshotBase<BucketT>::getStream() const

Diff for: src/bucket/BucketSnapshot.h

@@ -87,6 +87,10 @@ class LiveBucketSnapshot : public BucketSnapshotBase<LiveBucket>
                          std::list<EvictionResultEntry>& evictableKeys,
                          SearchableLiveBucketListSnapshot const& bl,
                          uint32_t ledgerVers) const;
+
+    // Scans contract code entries in the bucket.
+    Loop
+    scanForContractCode(std::function<Loop(BucketEntry const&)> callback) const;
 };
 
 class HotArchiveBucketSnapshot : public BucketSnapshotBase<HotArchiveBucket>

Diff for: src/bucket/InMemoryIndex.cpp

@@ -59,6 +59,8 @@ InMemoryIndex::InMemoryIndex(BucketManager const& bm,
     std::streamoff lastOffset = 0;
     std::optional<std::streamoff> firstOffer;
     std::optional<std::streamoff> lastOffer;
+    std::optional<std::streamoff> firstContractCode;
+    std::optional<std::streamoff> lastContractCode;
 
     while (in && in.readOne(be))
     {
@@ -108,6 +110,16 @@ InMemoryIndex::InMemoryIndex(BucketManager const& bm,
             lastOffer = lastOffset;
         }
 
+        // Populate contractCodeRange
+        if (!firstContractCode && lk.type() == CONTRACT_CODE)
+        {
+            firstContractCode = lastOffset;
+        }
+        if (!lastContractCode && lk.type() > CONTRACT_CODE)
+        {
+            lastContractCode = lastOffset;
+        }
+
         lastOffset = in.pos();
     }
 
@@ -129,5 +141,24 @@ InMemoryIndex::InMemoryIndex(BucketManager const& bm,
     {
         mOfferRange = std::nullopt;
     }
+
+    if (firstContractCode)
+    {
+        if (lastContractCode)
+        {
+            mContractCodeRange = {*firstContractCode, *lastContractCode};
+        }
+        // If we didn't see any entries after contract code, then the upper
+        // bound is EOF
+        else
+        {
+            mContractCodeRange = {*firstContractCode,
+                                  std::numeric_limits<std::streamoff>::max()};
+        }
+    }
+    else
+    {
+        mContractCodeRange = std::nullopt;
+    }
 }
 }

Diff for: src/bucket/InMemoryIndex.h

@@ -61,6 +61,7 @@ class InMemoryIndex
     AssetPoolIDMap mAssetPoolIDMap;
     BucketEntryCounters mCounters{};
     std::optional<std::pair<std::streamoff, std::streamoff>> mOfferRange;
+    std::optional<std::pair<std::streamoff, std::streamoff>> mContractCodeRange;
 
   public:
     using IterT = InMemoryBucketState::IterT;
@@ -103,6 +104,12 @@ class InMemoryIndex
         return mOfferRange;
     }
 
+    std::optional<std::pair<std::streamoff, std::streamoff>>
+    getContractCodeRange() const
+    {
+        return mContractCodeRange;
+    }
+
 #ifdef BUILD_TESTS
     bool
     operator==(InMemoryIndex const& in) const

Diff for: src/bucket/LiveBucket.cpp

@@ -316,6 +316,12 @@ LiveBucket::getOfferRange() const
     return getIndex().getOfferRange();
 }
 
+std::optional<std::pair<std::streamoff, std::streamoff>>
+LiveBucket::getContractCodeRange() const
+{
+    return getIndex().getContractCodeRange();
+}
+
 std::vector<BucketEntry>
 LiveBucket::convertToBucketEntry(bool useInit,
                                  std::vector<LedgerEntry> const& initEntries,

Diff for: src/bucket/LiveBucket.h

@@ -90,6 +90,11 @@ class LiveBucket : public BucketBase<LiveBucket, LiveBucketIndex>,
     std::optional<std::pair<std::streamoff, std::streamoff>>
     getOfferRange() const;
 
+    // Returns [lowerBound, upperBound) of file offsets for all contract code
+    // entries in the bucket, or std::nullopt if no contract code exists
+    std::optional<std::pair<std::streamoff, std::streamoff>>
+    getContractCodeRange() const;
+
     // Create a fresh bucket from given vectors of init (created) and live
     // (updated) LedgerEntries, and dead LedgerEntryKeys. The bucket will
     // be sorted, hashed, and adopted in the provided BucketManager.

Diff for: src/bucket/LiveBucketIndex.cpp

@@ -183,6 +183,27 @@ LiveBucketIndex::getOfferRange() const
     return mInMemoryIndex->getOfferRange();
 }
 
+std::optional<std::pair<std::streamoff, std::streamoff>>
+LiveBucketIndex::getContractCodeRange() const
+{
+    if (mDiskIndex)
+    {
+        // Get the smallest and largest possible contract code keys
+        LedgerKey lowerBound(CONTRACT_CODE);
+        lowerBound.contractCode().hash.fill(
+            std::numeric_limits<uint8_t>::min());
+
+        LedgerKey upperBound(CONTRACT_CODE);
+        upperBound.contractCode().hash.fill(
+            std::numeric_limits<uint8_t>::max());
+
+        return mDiskIndex->getOffsetBounds(lowerBound, upperBound);
+    }
+
+    releaseAssertOrThrow(mInMemoryIndex);
+    return mInMemoryIndex->getContractCodeRange();
+}
+
 uint32_t
 LiveBucketIndex::getPageSize() const
 {

Diff for: src/bucket/LiveBucketIndex.h

@@ -97,6 +97,9 @@ class LiveBucketIndex : public NonMovableOrCopyable
     std::optional<std::pair<std::streamoff, std::streamoff>>
     getOfferRange() const;
 
+    std::optional<std::pair<std::streamoff, std::streamoff>>
+    getContractCodeRange() const;
+
     BucketEntryCounters const& getBucketEntryCounters() const;
     uint32_t getPageSize() const;
 

Diff for: src/bucket/SearchableBucketList.cpp

@@ -64,6 +64,18 @@ SearchableLiveBucketListSnapshot::scanForEviction(
     return result;
 }
 
+void
+SearchableLiveBucketListSnapshot::scanForContractCode(
+    std::function<Loop(BucketEntry const&)> callback) const
+{
+    ZoneScoped;
+    releaseAssert(mSnapshot);
+    auto f = [&callback](auto const& b) {
+        return b.scanForContractCode(callback);
+    };
+    loopAllBuckets(f, *mSnapshot);
+}
+
 template <class BucketT>
 std::optional<std::vector<typename BucketT::LoadT>>
 SearchableBucketListSnapshotBase<BucketT>::loadKeysInternal(

Diff for: src/bucket/SearchableBucketList.h

@@ -35,6 +35,9 @@ class SearchableLiveBucketListSnapshot
         std::shared_ptr<EvictionStatistics> stats,
         StateArchivalSettings const& sas, uint32_t ledgerVers) const;
 
+    void
+    scanForContractCode(std::function<Loop(BucketEntry const&)> callback) const;
+
     friend SearchableSnapshotConstPtr
     BucketSnapshotManager::copySearchableLiveBucketListSnapshot() const;
 };