From ab5888c69bd407028d5a6efdebe191fc6b25c15d Mon Sep 17 00:00:00 2001
From: Michael Macias <zaeleus@gmail.com>
Date: Tue, 26 Nov 2024 11:41:47 -0600
Subject: [PATCH] gff/io/reader/line: Skip blank lines

See _Generic Feature Format Version 3 (GFF3)_ (2020-08-18): "Blank lines
should be ignored by parsers..."
---
 noodles-gff/CHANGELOG.md                |  5 ++++
 noodles-gff/src/async/io/reader/line.rs | 37 +++++++++++++++++++++--
 noodles-gff/src/io/reader.rs            |  2 +-
 noodles-gff/src/io/reader/line.rs       | 39 +++++++++++++++++++++++--
 4 files changed, 78 insertions(+), 5 deletions(-)

diff --git a/noodles-gff/CHANGELOG.md b/noodles-gff/CHANGELOG.md
index 8f22fc9e4..f86d5034d 100644
--- a/noodles-gff/CHANGELOG.md
+++ b/noodles-gff/CHANGELOG.md
@@ -26,6 +26,11 @@
 
     This changes `Reader::records` to `Reader::record_bufs`.
 
+  * gff/io/reader/line: Skip blank lines.
+
+    See _Generic Feature Format Version 3 (GFF3)_ (2020-08-18): "Blank lines
+    should be ignored by parsers..."
+
   * gff/line: Hoist buffer to line.
 
     This moves the owned line buffer to `Line`. The structure is now a struct
diff --git a/noodles-gff/src/async/io/reader/line.rs b/noodles-gff/src/async/io/reader/line.rs
index 6d9119cff..d2b8b5ba8 100644
--- a/noodles-gff/src/async/io/reader/line.rs
+++ b/noodles-gff/src/async/io/reader/line.rs
@@ -6,7 +6,40 @@ pub(super) async fn read_line<R>(reader: &mut R, line: &mut Line) -> io::Result<
 where
     R: AsyncBufRead + Unpin,
 {
+    use crate::io::reader::line::is_blank;
+
     let buf = &mut line.0;
-    buf.clear();
-    super::read_line(reader, buf).await
+
+    loop {
+        buf.clear();
+
+        let n = super::read_line(reader, buf).await?;
+
+        if n == 0 || !is_blank(buf) {
+            return Ok(n);
+        }
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[tokio::test]
+    async fn test_read_line() -> io::Result<()> {
+        const DATA: &[u8] = b"\n#comment\n\t\n";
+
+        let mut line = Line::default();
+        let mut lines: Vec<String> = Vec::new();
+
+        let mut src = DATA;
+
+        while read_line(&mut src, &mut line).await? != 0 {
+            lines.push(line.as_ref().into());
+        }
+
+        assert_eq!(lines, [String::from("#comment")]);
+
+        Ok(())
+    }
 }
diff --git a/noodles-gff/src/io/reader.rs b/noodles-gff/src/io/reader.rs
index d0268c7a7..942863687 100644
--- a/noodles-gff/src/io/reader.rs
+++ b/noodles-gff/src/io/reader.rs
@@ -1,6 +1,6 @@
 //! GFF reader and iterators.
 
-mod line;
+pub(crate) mod line;
 mod line_bufs;
 mod record_bufs;
 
diff --git a/noodles-gff/src/io/reader/line.rs b/noodles-gff/src/io/reader/line.rs
index 8ed189bc5..b880b3a98 100644
--- a/noodles-gff/src/io/reader/line.rs
+++ b/noodles-gff/src/io/reader/line.rs
@@ -7,6 +7,41 @@ where
     R: BufRead,
 {
     let buf = &mut line.0;
-    buf.clear();
-    super::read_line(reader, buf)
+
+    loop {
+        buf.clear();
+
+        let n = super::read_line(reader, buf)?;
+
+        if n == 0 || !is_blank(buf) {
+            return Ok(n);
+        }
+    }
+}
+
+pub(crate) fn is_blank(s: &str) -> bool {
+    s.chars().all(char::is_whitespace)
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn test_read_line() -> io::Result<()> {
+        const DATA: &[u8] = b"\n#comment\n\t\n";
+
+        let mut line = Line::default();
+        let mut lines: Vec<String> = Vec::new();
+
+        let mut src = DATA;
+
+        while read_line(&mut src, &mut line)? != 0 {
+            lines.push(line.as_ref().into());
+        }
+
+        assert_eq!(lines, [String::from("#comment")]);
+
+        Ok(())
+    }
 }