Merge remote-tracking branch 'oss/main' into unshred-variant

Author: scovich

arrow-arith/src/lib.rs

@@ -21,7 +21,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 pub mod aggregate;
 #[doc(hidden)] // Kernels to be removed in a future release

arrow-array/src/builder/fixed_size_binary_dictionary_builder.rs

@@ -252,6 +252,28 @@ where
         }
     }
 
+    /// Append a value multiple times to the array.
+    /// This is the same as [`Self::append`] but allows to append the same value multiple times without doing multiple lookups.
+    ///
+    /// Returns an error if the new index would overflow the key type.
+    pub fn append_n(
+        &mut self,
+        value: impl AsRef<[u8]>,
+        count: usize,
+    ) -> Result<K::Native, ArrowError> {
+        if self.byte_width != value.as_ref().len() as i32 {
+            Err(ArrowError::InvalidArgumentError(format!(
+                "Invalid input length passed to FixedSizeBinaryBuilder. Expected {} got {}",
+                self.byte_width,
+                value.as_ref().len()
+            )))
+        } else {
+            let key = self.get_or_insert_key(value)?;
+            self.keys_builder.append_value_n(key, count);
+            Ok(key)
+        }
+    }
+
     /// Appends a null slot into the builder
     #[inline]
     pub fn append_null(&mut self) {
@@ -401,6 +423,39 @@ mod tests {
         assert_eq!(ava.value(1), values[1].as_bytes());
     }
 
+    #[test]
+    fn test_fixed_size_dictionary_builder_append_n() {
+        let values = ["abc", "def"];
+        let mut b = FixedSizeBinaryDictionaryBuilder::<Int8Type>::new(3);
+        assert_eq!(b.append_n(values[0], 2).unwrap(), 0);
+        assert_eq!(b.append_n(values[1], 3).unwrap(), 1);
+        assert_eq!(b.append_n(values[0], 2).unwrap(), 0);
+        let array = b.finish();
+
+        assert_eq!(
+            array.keys(),
+            &Int8Array::from(vec![
+                Some(0),
+                Some(0),
+                Some(1),
+                Some(1),
+                Some(1),
+                Some(0),
+                Some(0),
+            ]),
+        );
+
+        // Values are polymorphic and so require a downcast.
+        let ava = array
+            .values()
+            .as_any()
+            .downcast_ref::<FixedSizeBinaryArray>()
+            .unwrap();
+
+        assert_eq!(ava.value(0), values[0].as_bytes());
+        assert_eq!(ava.value(1), values[1].as_bytes());
+    }
+
     #[test]
     fn test_fixed_size_dictionary_builder_wrong_size() {
         let mut b = FixedSizeBinaryDictionaryBuilder::<Int8Type>::new(3);
@@ -414,6 +469,11 @@ mod tests {
             err,
             "Invalid argument error: Invalid input length passed to FixedSizeBinaryBuilder. Expected 3 got 0"
         );
+        let err = b.append_n("a", 3).unwrap_err().to_string();
+        assert_eq!(
+            err,
+            "Invalid argument error: Invalid input length passed to FixedSizeBinaryBuilder. Expected 3 got 1"
+        );
     }
 
     #[test]

arrow-array/src/lib.rs

@@ -225,7 +225,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![deny(rustdoc::broken_intra_doc_links)]
 #![warn(missing_docs)]
 

arrow-avro/src/lib.rs

@@ -222,7 +222,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 #![allow(unused)] // Temporary
 

arrow-buffer/src/lib.rs

@@ -21,7 +21,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 
 pub mod alloc;

arrow-cast/src/lib.rs

@@ -21,7 +21,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 pub mod cast;
 pub use cast::*;

arrow-csv/src/lib.rs

@@ -21,7 +21,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 
 pub mod reader;

arrow-data/src/lib.rs

@@ -23,7 +23,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 mod data;
 pub use data::*;

arrow-flight/src/lib.rs

@@ -41,7 +41,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![allow(rustdoc::invalid_html_tags)]
 #![warn(missing_docs)]
 // The unused_crate_dependencies lint does not work well for crates defining additional examples/bin targets

arrow-integration-test/src/lib.rs

@@ -25,7 +25,7 @@
     html_logo_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_white-bg.svg",
     html_favicon_url = "https://arrow.apache.org/img/arrow-logo_chevrons_black-txt_transparent-bg.svg"
 )]
-#![cfg_attr(docsrs, feature(doc_auto_cfg))]
+#![cfg_attr(docsrs, feature(doc_cfg))]
 #![warn(missing_docs)]
 use arrow_buffer::{IntervalDayTime, IntervalMonthDayNano, ScalarBuffer};
 use hex::decode;