fix(parquet): bundle snappy-java so Snappy-compressed files import

CHANGELOG.md

@@ -8,6 +8,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ## [Unreleased]
 
 ### Fixed
+- Parquet import reads Snappy-compressed files, the default of Spark, parquet-rs and DuckDB, instead of failing for lack of `snappy-java`. ([#645](https://github.com/dfa1/vortex-java/issues/645))
 - A decimal column's values are narrowed and compressed under `vortex.decimal_byte_parts`, as Rust does, instead of stored as plain 64-bit integers: TPC-H's decimal columns are 2.5 to 16 times smaller. ([#644](https://github.com/dfa1/vortex-java/issues/644))
 - A nullable timestamp column holding nulls is written as `vortex.datetimeparts` with the validity on its days part, as Rust writes it, instead of `vortex.ext` over a masked i64. ([#627](https://github.com/dfa1/vortex-java/issues/627))
 - A timestamp column picks the encodings Rust picks: its `vortex.datetimeparts` parts are narrowed first, so runny days with nulls run-length encode (9% smaller files), and its zone-stats bounds stay plain storage when splitting them is no smaller. ([#634](https://github.com/dfa1/vortex-java/issues/634))

cli/pom.xml

@@ -54,6 +54,11 @@
             <groupId>com.github.luben</groupId>
             <artifactId>zstd-jni</artifactId>
         </dependency>
+        <dependency>
+            <!-- Same for SNAPPY-compressed Parquet files. -->
+            <groupId>org.xerial.snappy</groupId>
+            <artifactId>snappy-java</artifactId>
+        </dependency>
         <dependency>
             <!-- ImportCommand calls {@code options.writeOptions().allowedCascading()} via the
                  csv / parquet importers' API surface; that transitively reaches

integration/pom.xml

@@ -41,6 +41,12 @@
             <artifactId>zstd-jni</artifactId>
             <scope>test</scope>
         </dependency>
+        <!-- ... and snappy-java for SNAPPY-compressed pages -->
+        <dependency>
+            <groupId>org.xerial.snappy</groupId>
+            <artifactId>snappy-java</artifactId>
+            <scope>test</scope>
+        </dependency>
         <dependency>
             <groupId>io.github.dfa1.vortex</groupId>
             <artifactId>vortex-writer</artifactId>

parquet/pom.xml

@@ -37,6 +37,11 @@
             <groupId>com.github.luben</groupId>
             <artifactId>zstd-jni</artifactId>
         </dependency>
+        <dependency>
+            <!-- Required by hardwood-core for SNAPPY-compressed Parquet files (optional upstream). -->
+            <groupId>org.xerial.snappy</groupId>
+            <artifactId>snappy-java</artifactId>
+        </dependency>
         <!-- testing -->
         <!-- reader declares the zstd bindings optional; pull it in so the test can decode ZSTD chunks -->
         <dependency>

parquet/src/test/java/io/github/dfa1/vortex/parquet/ParquetImporterTest.java

@@ -721,6 +721,29 @@ void projection_unknownColumn_throws(@TempDir Path tmp) throws Exception {
                     .isInstanceOf(IllegalArgumentException.class)
                     .hasMessageContaining("does_not_exist");
         }
+
+        @Test
+        void snappyCompressedFile_isImported(@TempDir Path tmp) throws Exception {
+            // Given — pages compressed with SNAPPY, the codec of Spark, parquet-rs and DuckDB (#645)
+            Path vortex = tmp.resolve("out.vortex");
+            Path parquet = Path.of(ParquetImporterTest.class
+                    .getResource("/fixtures/snappy_compressed.parquet").toURI());
+
+            // When
+            ParquetImporter.importParquet(parquet, vortex);
+
+            // Then
+            try (VortexReader reader = VortexReader.open(vortex);
+                 ScanIterator iter = reader.scan(ScanOptions.all())) {
+                long rows = 0;
+                while (iter.hasNext()) {
+                    try (Chunk c = iter.next()) {
+                        rows += c.rowCount();
+                    }
+                }
+                assertThat(rows).isEqualTo(1000L);
+            }
+        }
     }
 
     private static Path fixture() throws Exception {

pom.xml

@@ -73,6 +73,7 @@
         <jmh.version>1.37</jmh.version>
         <!-- testing -->
         <zstd-jni.version>1.5.7-24</zstd-jni.version>
+        <snappy-java.version>1.1.10.8</snappy-java.version>
         <h2.version>2.5.252</h2.version>
         <junit.version>6.1.3</junit.version>
         <jazzer.version>0.30.0</jazzer.version>
@@ -251,6 +252,11 @@
                 <artifactId>zstd-jni</artifactId>
                 <version>${zstd-jni.version}</version>
             </dependency>
+            <dependency>
+                <groupId>org.xerial.snappy</groupId>
+                <artifactId>snappy-java</artifactId>
+                <version>${snappy-java.version}</version>
+            </dependency>
 
             <dependency>
                 <groupId>org.openjdk.jmh</groupId>