GH-40096: [C++][Parquet] Expand ParquetVersion enum values

cpp/src/parquet/metadata.cc

@@ -83,6 +83,14 @@ std::string ParquetVersionToString(ParquetVersion::type ver) {
       return "2.4";
     case ParquetVersion::PARQUET_2_6:
       return "2.6";
+    case ParquetVersion::PARQUET_2_7:
+      return "2.7";
+    case ParquetVersion::PARQUET_2_8:
+      return "2.8";
+    case ParquetVersion::PARQUET_2_9:
+      return "2.9";
+    case ParquetVersion::PARQUET_2_10:
+      return "2.10";
   }
 
   // This should be unreachable

cpp/src/parquet/metadata_test.cc

@@ -745,5 +745,15 @@ TEST(ApplicationVersion, FullWithSpaces) {
   ASSERT_EQ("cd", version.version.build_info);
 }
 
+TEST(ParquetVersionToString, AllVersions) {
+  ASSERT_EQ("1.0", ParquetVersionToString(ParquetVersion::PARQUET_1_0));
+  ASSERT_EQ("2.4", ParquetVersionToString(ParquetVersion::PARQUET_2_4));
+  ASSERT_EQ("2.6", ParquetVersionToString(ParquetVersion::PARQUET_2_6));
+  ASSERT_EQ("2.7", ParquetVersionToString(ParquetVersion::PARQUET_2_7));
+  ASSERT_EQ("2.8", ParquetVersionToString(ParquetVersion::PARQUET_2_8));
+  ASSERT_EQ("2.9", ParquetVersionToString(ParquetVersion::PARQUET_2_9));
+  ASSERT_EQ("2.10", ParquetVersionToString(ParquetVersion::PARQUET_2_10));
+}
+
 }  // namespace metadata
 }  // namespace parquet

cpp/src/parquet/type_fwd.h

@@ -54,11 +54,43 @@ struct ParquetVersion {
     /// Note: Parquet format 2.6.0 was released in September 2018.
     PARQUET_2_6,
 
+    /// Enable Parquet format 2.7 and earlier features when writing
+    ///
+    /// This enables bloom filters and encryption in addition to the
+    /// PARQUET_2_6 features.
+    ///
+    /// Note: Parquet format 2.7.0 was released in June 2019.
+    PARQUET_2_7,
+
+    /// Enable Parquet format 2.8 and earlier features when writing
+    ///
+    /// This enables BYTE_STREAM_SPLIT encoding in addition to the
+    /// PARQUET_2_7 features.
+    ///
+    /// Note: Parquet format 2.8.0 was released in February 2020.
+    PARQUET_2_8,
+
+    /// Enable Parquet format 2.9 and earlier features when writing
+    ///
+    /// This enables interoperable LZ4 codec in addition to the
+    /// PARQUET_2_8 features.
+    ///
+    /// Note: Parquet format 2.9.0 was released in January 2021.
+    PARQUET_2_9,
+
+    /// Enable Parquet format 2.10 and earlier features when writing
+    ///
+    /// This enables Float16 logical type in addition to the
+    /// PARQUET_2_9 features.
+    ///
+    /// Note: Parquet format 2.10.0 was released in October 2022.
+    PARQUET_2_10,
+
     /// Enable latest Parquet format 2.x features
     ///
     /// This value is equal to the greatest 2.x version supported by
     /// this library.
-    PARQUET_2_LATEST = PARQUET_2_6
+    PARQUET_2_LATEST = PARQUET_2_10
   };
 };
 

python/pyarrow/includes/libparquet.pxd

@@ -144,6 +144,10 @@ cdef extern from "parquet/api/schema.h" namespace "parquet" nogil:
         ParquetVersion_V1" parquet::ParquetVersion::PARQUET_1_0"
         ParquetVersion_V2_4" parquet::ParquetVersion::PARQUET_2_4"
         ParquetVersion_V2_6" parquet::ParquetVersion::PARQUET_2_6"
+        ParquetVersion_V2_7" parquet::ParquetVersion::PARQUET_2_7"
+        ParquetVersion_V2_8" parquet::ParquetVersion::PARQUET_2_8"
+        ParquetVersion_V2_9" parquet::ParquetVersion::PARQUET_2_9"
+        ParquetVersion_V2_10" parquet::ParquetVersion::PARQUET_2_10"
 
     enum ParquetSortOrder" parquet::SortOrder::type":
         ParquetSortOrder_SIGNED" parquet::SortOrder::SIGNED"

r/R/enums.R

@@ -135,7 +135,8 @@ FileType <- enum("FileType",
 #' @export
 #' @rdname enums
 ParquetVersionType <- enum("ParquetVersionType",
-  PARQUET_1_0 = 0L, PARQUET_2_4 = 2L, PARQUET_2_6 = 3L
+  PARQUET_1_0 = 0L, PARQUET_2_4 = 2L, PARQUET_2_6 = 3L,
+  PARQUET_2_7 = 4L, PARQUET_2_8 = 5L, PARQUET_2_9 = 6L, PARQUET_2_10 = 7L
 )
 
 #' @export

r/R/parquet.R

@@ -234,7 +234,11 @@ valid_parquet_version <- c(
   "1.0" = ParquetVersionType$PARQUET_1_0,
   "2.4" = ParquetVersionType$PARQUET_2_4,
   "2.6" = ParquetVersionType$PARQUET_2_6,
-  "latest" = ParquetVersionType$PARQUET_2_6
+  "2.7" = ParquetVersionType$PARQUET_2_7,
+  "2.8" = ParquetVersionType$PARQUET_2_8,
+  "2.9" = ParquetVersionType$PARQUET_2_9,
+  "2.10" = ParquetVersionType$PARQUET_2_10,
+  "latest" = ParquetVersionType$PARQUET_2_10
 )
 
 make_valid_parquet_version <- function(version, valid_versions = valid_parquet_version) {

r/tests/testthat/test-parquet.R

@@ -142,9 +142,25 @@ test_that("make_valid_parquet_version()", {
     make_valid_parquet_version("2.6"),
     ParquetVersionType$PARQUET_2_6
   )
+  expect_equal(
+    make_valid_parquet_version("2.7"),
+    ParquetVersionType$PARQUET_2_7
+  )
+  expect_equal(
+    make_valid_parquet_version("2.8"),
+    ParquetVersionType$PARQUET_2_8
+  )
+  expect_equal(
+    make_valid_parquet_version("2.9"),
+    ParquetVersionType$PARQUET_2_9
+  )
+  expect_equal(
+    make_valid_parquet_version("2.10"),
+    ParquetVersionType$PARQUET_2_10
+  )
   expect_equal(
     make_valid_parquet_version("latest"),
-    ParquetVersionType$PARQUET_2_6
+    ParquetVersionType$PARQUET_2_10
   )
 
   expect_equal(make_valid_parquet_version(1), ParquetVersionType$PARQUET_1_0)