feat(arrow_csv): add header validation option (#10144)

# Which issue does this PR close?

<!--
We generally require a GitHub issue to be filed for all bug fixes and
enhancements and this helps us generate change logs for our releases.
You can link an issue to this PR using the GitHub syntax.
-->

- Closes #10143.

# Rationale for this change

<!--
Why are you proposing this change? If this is already explained clearly
in the issue then this section is not needed.
Explaining clearly why changes are proposed helps reviewers understand
your changes and offer better suggestions for fixes.
-->
Explained in the issue.

# What changes are included in this PR?

<!--
There is no need to duplicate the description in the issue here but it
is sometimes worth providing a summary of the individual changes in this
PR.
-->
- Adds a new `ReaderBuilder` and `Format` method
.with_header_validation(bool) to enable CSV header validation
- Implements header row validation, which verifies each value in the
first CSV row against the schema to ensure the field names match the
header columns.

# Are these changes tested?

<!--
We typically require tests for all PRs in order to:
1. Prevent the code from being accidentally broken by subsequent changes
2. Serve as another way to document the expected behavior of the code

If tests are not included in your PR, please explain why (for example,
are they covered by existing tests)?

If this PR claims a performance improvement, please include evidence
such as benchmark results.
-->
Corresponding tests added.

# Are there any user-facing changes?

<!--
If there are user-facing changes then we may require documentation to be
updated before approving the PR.

If there are any breaking changes to public APIs, please call them out.
-->

The `ReaderBuilder` struct gets a new method added,
`.with_header_validation(bool)`.

There is no breaking change, since the validation behavior is disabled
by default.

Author: XiNiHa

arrow-csv/src/reader/mod.rs

@@ -273,6 +273,7 @@ impl InferredDataType {
 #[derive(Debug, Clone, Default)]
 pub struct Format {
     header: bool,
+    header_validation: bool,
     delimiter: Option<u8>,
     escape: Option<u8>,
     quote: Option<u8>,
@@ -291,6 +292,16 @@ impl Format {
         self
     }
 
+    /// Specify whether to validate the CSV header against the schema, defaults to `false`
+    ///
+    /// When `true`, the first row gets validated against the schema before any data is read
+    ///
+    /// Only applies when [`Self::with_header`] is set to `true`
+    pub fn with_header_validation(mut self, validate_header: bool) -> Self {
+        self.header_validation = validate_header;
+        self
+    }
+
     /// Specify a custom delimiter character, defaults to comma `','`
     pub fn with_delimiter(mut self, delimiter: u8) -> Self {
         self.delimiter = Some(delimiter);
@@ -610,6 +621,9 @@ pub struct Decoder {
     /// Rows to skip
     to_skip: usize,
 
+    /// Whether to validate the first skipped row against the schema
+    header_validation: bool,
+
     /// Current line number
     line_number: usize,
 
@@ -635,6 +649,20 @@ impl Decoder {
     /// network sources such as object storage
     pub fn decode(&mut self, buf: &[u8]) -> Result<usize, ArrowError> {
         if self.to_skip != 0 {
+            if self.header_validation {
+                let (skipped, bytes) = self.record_decoder.decode(buf, 1)?;
+
+                if skipped == 0 {
+                    return Ok(bytes);
+                }
+
+                let rows = self.record_decoder.flush()?;
+                validate_header(&rows, self.schema.fields())?;
+                self.header_validation = false;
+                self.to_skip -= 1;
+                return Ok(bytes);
+            }
+
             // Skip in units of `to_read` to avoid over-allocating buffers
             let to_skip = self.to_skip.min(self.batch_size);
             let (skipped, bytes) = self.record_decoder.decode(buf, to_skip)?;
@@ -678,6 +706,24 @@ impl Decoder {
     }
 }
 
+fn validate_header(rows: &StringRecords<'_>, fields: &Fields) -> Result<(), ArrowError> {
+    let header = rows.iter().next().ok_or_else(|| {
+        ArrowError::CsvError("CSV header validation failed: no header row found".to_string())
+    })?;
+
+    for (idx, field) in fields.iter().enumerate() {
+        let actual = header.get(idx);
+        let expected = field.name();
+        if actual != expected {
+            return Err(ArrowError::CsvError(format!(
+                "CSV header does not match schema at column {idx}: expected {expected:?} but found {actual:?}"
+            )));
+        }
+    }
+
+    Ok(())
+}
+
 /// Parses a slice of [`StringRecords`] into a [RecordBatch]
 fn parse(
     rows: &StringRecords<'_>,
@@ -1154,6 +1200,14 @@ impl ReaderBuilder {
         self
     }
 
+    /// Set whether to validate the CSV header against the schema
+    ///
+    /// This option only applies when [`Self::with_header`] is set to `true`, and defaults to `false`
+    pub fn with_header_validation(mut self, validate_header: bool) -> Self {
+        self.format.header_validation = validate_header;
+        self
+    }
+
     /// Overrides the [Format] of this [ReaderBuilder]
     pub fn with_format(mut self, format: Format) -> Self {
         self.format = format;
@@ -1261,6 +1315,7 @@ impl ReaderBuilder {
         Decoder {
             schema: self.schema,
             to_skip: start,
+            header_validation: self.format.header && self.format.header_validation,
             record_decoder,
             line_number: start,
             end,
@@ -2351,6 +2406,86 @@ mod tests {
         }
     }
 
+    #[test]
+    fn test_header_validation() {
+        let schema = Arc::new(Schema::new(vec![
+            Field::new("a", DataType::Int32, false),
+            Field::new("b", DataType::Int32, false),
+        ]));
+
+        let csv = "a,c\n1,2\n";
+        let err = ReaderBuilder::new(schema.clone())
+            .with_header(true)
+            .with_header_validation(true)
+            .build_buffered(Cursor::new(csv.as_bytes()))
+            .unwrap()
+            .next()
+            .unwrap()
+            .unwrap_err()
+            .to_string();
+        assert_eq!(
+            err,
+            "Csv error: CSV header does not match schema at column 1: expected \"b\" but found \"c\""
+        );
+
+        let batch = ReaderBuilder::new(schema)
+            .with_header(true)
+            .with_header_validation(false)
+            .build_buffered(Cursor::new(csv.as_bytes()))
+            .unwrap()
+            .next()
+            .unwrap()
+            .unwrap();
+        assert_eq!(batch.num_rows(), 1);
+    }
+
+    #[test]
+    fn test_header_validation_with_buffered_reader() {
+        let schema = Arc::new(Schema::new(vec![
+            Field::new("a", DataType::Int32, false),
+            Field::new("b", DataType::Int32, false),
+        ]));
+
+        let csv = "a,b\n1,2\n";
+        let buffered = std::io::BufReader::with_capacity(1, Cursor::new(csv.as_bytes()));
+        let batch = ReaderBuilder::new(schema)
+            .with_header(true)
+            .with_header_validation(true)
+            .build_buffered(buffered)
+            .unwrap()
+            .next()
+            .unwrap()
+            .unwrap();
+
+        assert_eq!(batch.num_rows(), 1);
+        let a = batch.column(0).as_primitive::<Int32Type>();
+        assert_eq!(a.value(0), 1);
+    }
+
+    #[test]
+    fn test_header_validation_with_truncated_rows() {
+        let schema = Arc::new(Schema::new(vec![
+            Field::new("a", DataType::Int32, true),
+            Field::new("b", DataType::Int32, true),
+        ]));
+
+        let csv = "a\n1\n";
+        let err = ReaderBuilder::new(schema.clone())
+            .with_header(true)
+            .with_header_validation(true)
+            .with_truncated_rows(true)
+            .build_buffered(Cursor::new(csv.as_bytes()))
+            .unwrap()
+            .next()
+            .unwrap()
+            .unwrap_err()
+            .to_string();
+        assert_eq!(
+            err,
+            "Csv error: CSV header does not match schema at column 1: expected \"b\" but found \"\"",
+        )
+    }
+
     #[test]
     fn test_null_boolean() {
         let csv = "true,false\nFalse,True\n,True\nFalse,";