From b0c32097aac459296bbdfc09c71c08826bb30d66 Mon Sep 17 00:00:00 2001
From: Pavel Ivanov <mr.pavel.ivanov@gmail.com>
Date: Sun, 5 May 2024 18:21:14 +0200
Subject: [PATCH] fix: fixed sorting when some lines are empty

---
 Cargo.lock   |  4 ++--
 Cargo.toml   |  2 +-
 src/app.rs   | 28 +++++++++++++++++++++++++++-
 src/index.rs |  2 ++
 src/input.rs | 21 +++++++++++++++++++--
 5 files changed, 51 insertions(+), 6 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index e913395f..1a8c0742 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -611,7 +611,7 @@ checksum = "edd0f118536f44f5ccd48bcb8b111bdc3de888b58c74639dfb034a357d0f206d"
 
 [[package]]
 name = "encstr"
-version = "0.29.2-alpha.3"
+version = "0.29.2-alpha.4"
 
 [[package]]
 name = "enum-map"
@@ -768,7 +768,7 @@ checksum = "7f24254aa9a54b5c858eaee2f5bccdb46aaf0e486a595ed5fd8f86ba55232a70"
 
 [[package]]
 name = "hl"
-version = "0.29.2-alpha.3"
+version = "0.29.2-alpha.4"
 dependencies = [
  "atoi",
  "bincode",
diff --git a/Cargo.toml b/Cargo.toml
index 88f6638b..2c613031 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -4,7 +4,7 @@ members = [".", "crate/encstr"]
 [workspace.package]
 repository = "https://github.com/pamburus/hl"
 authors = ["Pavel Ivanov <mr.pavel.ivanov@gmail.com>"]
-version = "0.29.2-alpha.3"
+version = "0.29.2-alpha.4"
 edition = "2021"
 license = "MIT"
 
diff --git a/src/app.rs b/src/app.rs
index 8a59914f..fdf8a735 100644
--- a/src/app.rs
+++ b/src/app.rs
@@ -107,6 +107,11 @@ impl Options {
         Self { raw, ..self }
     }
 
+    #[cfg(test)]
+    fn with_sort(self, sort: bool) -> Self {
+        Self { sort, ..self }
+    }
+
     #[cfg(test)]
     fn with_input_info(self, input_info: Option<InputInfo>) -> Self {
         Self { input_info, ..self }
@@ -1178,8 +1183,29 @@ mod tests {
         assert_eq!(std::str::from_utf8(&output).unwrap(), format!("{}\n\n{}\n", L1, L2),);
     }
 
+    #[test]
+    fn test_sort_with_blank_lines() {
+        let input = input(concat!(
+            r#"{"level":"debug","ts":"2024-01-25T19:10:20.435369+01:00","msg":"m2"}"#,
+            "\n\r\n",
+            r#"{"level":"debug","ts":"2024-01-25T19:09:16.860711+01:00","msg":"m1"}"#,
+            "\n",
+        ));
+
+        let mut output = Vec::new();
+        let app = App::new(options().with_sort(true));
+        app.run(vec![input], &mut output).unwrap();
+        assert_eq!(
+            std::str::from_utf8(&output).unwrap(),
+            concat!(
+                "2024-01-25 18:09:16.860 |DBG| m1\n",
+                "2024-01-25 18:10:20.435 |DBG| m2\n",
+            ),
+        );
+    }
+
     fn input<S: Into<String>>(s: S) -> InputHolder {
-        InputHolder::new(InputReference::File("-".into()), Some(Box::new(Cursor::new(s.into()))))
+        InputHolder::new(InputReference::Stdin, Some(Box::new(Cursor::new(s.into()))))
     }
 
     fn options() -> Options {
diff --git a/src/index.rs b/src/index.rs
index 39bff9e7..b0176fb5 100644
--- a/src/index.rs
+++ b/src/index.rs
@@ -442,6 +442,8 @@ impl Indexer {
                         }
                     }
                 }
+            } else {
+                stat.add_invalid();
             }
             lines.push((ts.or(prev_ts), i as u32, offset));
             offset += data_len as u32 + 1;
diff --git a/src/input.rs b/src/input.rs
index 10cb7706..10f229be 100644
--- a/src/input.rs
+++ b/src/input.rs
@@ -141,7 +141,7 @@ impl InputHolder {
 
     pub fn open(self) -> io::Result<Input> {
         match self.reference {
-            InputReference::Stdin => Ok(Input::new(self.reference, Box::new(stdin()))),
+            InputReference::Stdin => Ok(Input::new(self.reference.clone(), self.stdin())),
             InputReference::File(path) => match self.stream {
                 Some(stream) => Input::open_stream(&path, stream),
                 None => Input::open(&path),
@@ -151,13 +151,19 @@ impl InputHolder {
 
     pub fn index(self, indexer: &Indexer) -> Result<IndexedInput> {
         match self.reference {
-            InputReference::Stdin => IndexedInput::open_sequential(self.reference.clone(), Box::new(stdin()), indexer),
+            InputReference::Stdin => IndexedInput::open_sequential(self.reference.clone(), self.stdin(), indexer),
             InputReference::File(path) => match self.stream {
                 Some(stream) => IndexedInput::open_stream(&path, stream, indexer),
                 None => IndexedInput::open(&path, indexer),
             },
         }
     }
+
+    fn stdin(self) -> InputStream {
+        self.stream
+            .map(|s| Box::new(ReadSeekToRead(s)) as InputStream)
+            .unwrap_or_else(|| Box::new(stdin()))
+    }
 }
 
 pub struct Input {
@@ -515,6 +521,17 @@ pub trait ReadSeek: Read + Seek {}
 
 impl<T: Read + Seek> ReadSeek for T {}
 
+pub struct ReadSeekToRead<T>(T);
+
+impl<T> Read for ReadSeekToRead<T>
+where
+    T: ReadSeek,
+{
+    fn read(&mut self, buf: &mut [u8]) -> io::Result<usize> {
+        self.0.read(buf)
+    }
+}
+
 trait AsInputStream {
     fn as_input_stream(self) -> InputStream;
 }