Use insta for DataFrame tests (#15165)

* Add insta for df testing

* Do not use `get_plan_string`

* Switch from `assert_batches_eq`

* Switch from `assert_batches_sorted_eq`

* Switch from `assert_eq`

* Format toml

---------

Co-authored-by: Andrew Lamb <andrew@nerdnetworks.org>

Author: blaginin

Cargo.lock

@@ -155,6 +155,7 @@ parquet = { version = "54.2.1", default-features = false, features = [
 pbjson = { version = "0.7.0" }
 pbjson-types = "0.7"
 # Should match arrow-flight's version of prost.
+insta = { version = "1.41.1", features = ["glob", "filters"] }
 prost = "0.13.1"
 rand = "0.8.5"
 recursive = "0.1.1"

Cargo.toml

@@ -67,7 +67,7 @@ url = { workspace = true }
 [dev-dependencies]
 assert_cmd = "2.0"
 ctor = { workspace = true }
-insta = { version = "1.41.1", features = ["glob", "filters"] }
+insta = { workspace = true }
 insta-cmd = "0.6.0"
 predicates = "3.0"
 rstest = { workspace = true }

datafusion-cli/Cargo.toml

@@ -17,6 +17,9 @@
 
 //! Utility functions to make testing DataFusion based crates easier
 
+use crate::arrow::util::pretty::pretty_format_batches_with_options;
+use crate::format::DEFAULT_FORMAT_OPTIONS;
+use arrow::array::RecordBatch;
 use std::{error::Error, path::PathBuf};
 
 /// Compares formatted output of a record batch with an expected
@@ -73,6 +76,31 @@ macro_rules! assert_batches_eq {
     };
 }
 
+pub fn batches_to_string(batches: &[RecordBatch]) -> String {
+    let actual = pretty_format_batches_with_options(batches, &DEFAULT_FORMAT_OPTIONS)
+        .unwrap()
+        .to_string();
+
+    actual.trim().to_string()
+}
+
+pub fn batches_to_sort_string(batches: &[RecordBatch]) -> String {
+    let actual_lines =
+        pretty_format_batches_with_options(batches, &DEFAULT_FORMAT_OPTIONS)
+            .unwrap()
+            .to_string();
+
+    let mut actual_lines: Vec<&str> = actual_lines.trim().lines().collect();
+
+    // sort except for header + footer
+    let num_lines = actual_lines.len();
+    if num_lines > 3 {
+        actual_lines.as_mut_slice()[2..num_lines - 1].sort_unstable()
+    }
+
+    actual_lines.join("\n")
+}
+
 /// Compares formatted output of a record batch with an expected
 /// vector of strings in a way that order does not matter.
 /// This is a macro so errors appear on the correct line

datafusion/common/src/test_util.rs

@@ -152,6 +152,7 @@ datafusion-functions-window-common = { workspace = true }
 datafusion-physical-optimizer = { workspace = true }
 doc-comment = { workspace = true }
 env_logger = { workspace = true }
+insta = { workspace = true }
 paste = "^1.0"
 rand = { workspace = true, features = ["small_rng"] }
 rand_distr = "0.4.3"