Add helpers to extract CONTRACT_CODE entries efficiently from BL.

Author: graydon

src/bucket/BucketBase.cpp

@@ -51,6 +51,12 @@ BucketBase::getOfferRange() const
     return getIndex().getOfferRange();
 }
 
+std::optional<std::pair<std::streamoff, std::streamoff>>
+BucketBase::getContractCodeRange() const
+{
+    return getIndex().getContractCodeRange();
+}
+
 void
 BucketBase::setIndex(std::unique_ptr<BucketIndex const>&& index)
 {

src/bucket/BucketBase.h

@@ -93,6 +93,11 @@ class BucketBase : public NonMovableOrCopyable
     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;
+
     // Sets index, throws if index is already set
     void setIndex(std::unique_ptr<BucketIndex const>&& index);
 

src/bucket/BucketIndex.h

@@ -130,6 +130,11 @@ class BucketIndex : public NonMovableOrCopyable
     virtual std::optional<std::pair<std::streamoff, std::streamoff>>
     getOfferRange() const = 0;
 
+    // Returns lower bound and upper bound for contract code entry positions in
+    // the given bucket, or std::nullopt if no contract code entries exist
+    virtual std::optional<std::pair<std::streamoff, std::streamoff>>
+    getContractCodeRange() const = 0;
+
     // Returns page size for index. InidividualIndex returns 0 for page size
     virtual std::streamoff getPageSize() const = 0;
 

src/bucket/BucketIndexImpl.cpp

@@ -598,6 +598,20 @@ BucketIndexImpl<IndexT>::getOfferRange() const
     return getOffsetBounds(lowerBound, upperBound);
 }
 
+template <class IndexT>
+std::optional<std::pair<std::streamoff, std::streamoff>>
+BucketIndexImpl<IndexT>::getContractCodeRange() const
+{
+    // 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 getOffsetBounds(lowerBound, upperBound);
+}
+
 #ifdef BUILD_TESTS
 template <class IndexT>
 bool

src/bucket/BucketIndexImpl.h

@@ -100,6 +100,9 @@ template <class IndexT> class BucketIndexImpl : public BucketIndex
     virtual std::optional<std::pair<std::streamoff, std::streamoff>>
     getOfferRange() const override;
 
+    virtual std::optional<std::pair<std::streamoff, std::streamoff>>
+    getContractCodeRange() const override;
+
     virtual std::streamoff
     getPageSize() const override
     {

src/bucket/BucketSnapshot.cpp

@@ -282,6 +282,41 @@ LiveBucketSnapshot::scanForEviction(
     return Loop::INCOMPLETE;
 }
 
+// Scans contract code entries in the bucket.
+Loop
+LiveBucketSnapshot::scanForContractCode(
+    std::function<Loop(LedgerEntry 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() != DEADENTRY &&
+            be.liveEntry().data.type() == CONTRACT_CODE)
+        {
+            if (callback(be.liveEntry()) == Loop::COMPLETE)
+            {
+                return Loop::COMPLETE;
+            }
+        }
+    }
+    return Loop::INCOMPLETE;
+}
+
 template <class BucketT>
 XDRInputFileStream&
 BucketSnapshotBase<BucketT>::getStream() const

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(LedgerEntry const&)> callback) const;
 };
 
 class HotArchiveBucketSnapshot : public BucketSnapshotBase<HotArchiveBucket>

src/bucket/SearchableBucketList.cpp

@@ -62,6 +62,18 @@ SearchableLiveBucketListSnapshot::scanForEviction(
     return result;
 }
 
+void
+SearchableLiveBucketListSnapshot::scanForContractCode(
+    std::function<Loop(LedgerEntry 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(

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(LedgerEntry const&)> callback) const;
+
     friend SearchableSnapshotConstPtr
     BucketSnapshotManager::copySearchableLiveBucketListSnapshot() const;
 };

src/main/ApplicationUtils.cpp

@@ -1011,4 +1011,27 @@ getStellarCoreMajorReleaseVersion(std::string const& vstr)
     return std::nullopt;
 }
 
+int
+listContracts(Config const& cfg)
+{
+    VirtualClock clock(VirtualClock::REAL_TIME);
+    auto config = cfg;
+    config.setNoListen();
+    auto app = Application::create(clock, config, /* newDB */ false);
+
+    // Start app to ensure BucketManager is initialized
+    app->start();
+
+    auto snap = app->getBucketManager()
+                    .getBucketSnapshotManager()
+                    .copySearchableLiveBucketListSnapshot();
+    snap->scanForContractCode([](LedgerEntry const& entry) {
+        auto hash = sha256(entry.data.contractCode().code);
+        std::cout << binToHex(hash) << std::endl;
+        return Loop::INCOMPLETE; // Continue scanning
+    });
+
+    return 0;
+}
+
 }

src/main/ApplicationUtils.h

@@ -64,4 +64,6 @@ void setAuthenticatedLedgerHashPair(Application::pointer app,
                                     std::string startHash);
 std::optional<uint32_t>
 getStellarCoreMajorReleaseVersion(std::string const& vstr);
+
+int listContracts(Config const& cfg);
 }

src/main/CommandLine.cpp

@@ -1471,6 +1471,18 @@ runReportLastHistoryCheckpoint(CommandLineArgs const& args)
         });
 }
 
+int
+runListContracts(CommandLineArgs const& args)
+{
+    CommandLine::ConfigOption configOption;
+
+    return runWithHelp(args, {configurationParser(configOption)},
+                       [&] { return listContracts(configOption.getConfig()); });
+
+
+    });
+}
+
 int
 run(CommandLineArgs const& args)
 {
@@ -1945,6 +1957,9 @@ handleCommandLine(int argc, char* const* argv)
          {"test", "execute test suite", runTest},
          {"apply-load", "run apply time load test", runApplyLoad},
 #endif
+         {"list-contracts",
+          "List sha256 hashes of all contract code entries in the bucket list",
+          runListContracts},
          {"version", "print version information", runVersion}}};
 
     auto adjustedCommandLine = commandLine.adjustCommandLine({argc, argv});