From 562a6298244056f7fecd64f8dcf4d2181fd9a4e9 Mon Sep 17 00:00:00 2001
From: "google-labs-jules[bot]"
 <161369871+google-labs-jules[bot]@users.noreply.github.com>
Date: Wed, 8 Apr 2026 23:42:39 +0000
Subject: [PATCH] Add test for LZ77 backreference boundary conditions

Added tests to `test_compression_lz77_edge_cases.cpp` and
`test_compression_lz77.cpp` to explicitly cover the exact
boundary condition where the requested backreference distance
exceeds the available output history by exactly 1 byte.

Co-authored-by: segin <480709+segin@users.noreply.github.com>
---
 run_lz77_tests.sh                          |  1 +
 tests/test_compression_lz77.cpp            | 22 ++++++++++++++++++
 tests/test_compression_lz77_edge_cases.cpp | 26 ++++++++++++++++++++++
 3 files changed, 49 insertions(+)
 create mode 100755 run_lz77_tests.sh

diff --git a/run_lz77_tests.sh b/run_lz77_tests.sh
new file mode 100755
index 00000000..d48f1533
--- /dev/null
+++ b/run_lz77_tests.sh
@@ -0,0 +1 @@
+make -j$(nproc) check TESTS="test_compression_lz77 test_compression_lz77_edge_cases"
diff --git a/tests/test_compression_lz77.cpp b/tests/test_compression_lz77.cpp
index 0d00495a..5565ed41 100644
--- a/tests/test_compression_lz77.cpp
+++ b/tests/test_compression_lz77.cpp
@@ -184,6 +184,28 @@ int edge_case_tests() {
         ASSERT_EQ(output[1], 0x42);
     }
 
+
+    // 7. Exact Boundary Distance (Distance == Output Size + 1)
+    {
+        // Block 1: 8 Literals ('A'...'H') -> Output "ABCDEFGH" (size 8)
+        // Block 2: Ref (Dist 9, Len 3)
+        // Flags: 0 (all literals)
+        // Data: 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H'
+        // Flags: 1 (Ref at bit 0)
+        // Data: Dist 9, Len 3. Dist 9 -> 0x009.
+        // b1 = (9 >> 4) = 0.
+        // b2 = ((9 & 0xF) << 4) | 0 = 0x90.
+        std::vector<uint8_t> input = {
+            0x00, 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', // Block 1
+            0x01, 0x00, 0x90                               // Block 2
+        };
+        auto output = decompressor.decompress(input.data(), input.size());
+
+        ASSERT_EQ(output.size(), 11);
+        std::string s(output.begin(), output.end());
+        ASSERT_EQ(s, "ABCDEFGHABC");
+    }
+
     std::cout << "[EDGE] Passed.\n";
     return 0;
 }
diff --git a/tests/test_compression_lz77_edge_cases.cpp b/tests/test_compression_lz77_edge_cases.cpp
index 34cbce46..07a0388f 100644
--- a/tests/test_compression_lz77_edge_cases.cpp
+++ b/tests/test_compression_lz77_edge_cases.cpp
@@ -118,6 +118,32 @@ int main() {
         std::cout << "Passed.\n";
     }
 
+
+    // Test 7: Exact Boundary Backreference (Distance == Output Size + 1)
+    // We need some initial data to be valid and a full block to reset flags.
+    // Block 1: 8 Literals [0x00, 'A'...'H'] -> Output "ABCDEFGH" (size 8)
+    // Block 2: Flags [0x01] (1st item Ref)
+    //          Ref bytes [0x00, 0x90] -> Dist 9, Len 3.
+    //
+    // Distance 9 == Output Size 8 + 1.
+    // Implementation clamps distance to output.size() = 8.
+    // Start = 8 - 8 = 0.
+    // Copies output[0]...output[2] -> "ABC".
+    // Result: "ABCDEFGHABC".
+    {
+        std::cout << "  Test 7: Exact Boundary Backreference... ";
+        std::vector<uint8_t> input = {
+            0x00, 'A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', // Block 1
+            0x01, 0x00, 0x90                               // Block 2
+        };
+        auto output = decompressor.decompress(input.data(), input.size());
+
+        ASSERT_EQ(output.size(), 11);
+        std::string s(output.begin(), output.end());
+        ASSERT_EQ(s, "ABCDEFGHABC");
+        std::cout << "Passed.\n";
+    }
+
     std::cout << "[EDGE] All tests passed.\n";
     return 0;
 }