rebase on branch line_ending_detection

1 files changed, 74 insertions, 3 deletions

diff --git a/helix-view/src/document.rs b/helix-view/src/document.rs
index a1c4b407..8b735b9d 100644
--- a/helix-view/src/document.rs
+++ b/helix-view/src/document.rs
@@ -8,7 +8,7 @@ use helix_core::{
     chars::{char_is_linebreak, char_is_whitespace},
     history::History,
     syntax::{LanguageConfiguration, LOADER},
-    ChangeSet, Diagnostic, Rope, Selection, State, Syntax, Transaction,
+    ChangeSet, Diagnostic, History, Rope, RopeSlice, RopeGraphemes, Selection, State, Syntax, Transaction,
 };
 
 use crate::{DocumentId, ViewId};
@@ -22,12 +22,28 @@ pub enum Mode {
     Insert,
 }
 
+
 #[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)]
 pub enum IndentStyle {
     Tabs,
     Spaces(u8),
 }
 
+/// Represents one of the valid Unicode line endings.
+/// Also acts as an index into `LINE_ENDINGS`.
+#[derive(PartialEq, Copy, Clone, Debug)]
+pub enum LineEnding {
+    None = 0, // No line ending
+    CRLF = 1, // CarriageReturn followed by LineFeed
+    LF = 2,   // U+000A -- LineFeed
+    VT = 3,   // U+000B -- VerticalTab
+    FF = 4,   // U+000C -- FormFeed
+    CR = 5,   // U+000D -- CarriageReturn
+    NEL = 6,  // U+0085 -- NextLine
+    LS = 7,   // U+2028 -- Line Separator
+    PS = 8,   // U+2029 -- ParagraphSeparator
+}
+
 pub struct Document {
     // rope + selection
     pub(crate) id: DocumentId,
@@ -61,6 +77,7 @@ pub struct Document {
 
     diagnostics: Vec<Diagnostic>,
     language_server: Option<Arc<helix_lsp::Client>>,
+    line_ending: LineEnding
 }
 
 use std::fmt;
@@ -146,11 +163,61 @@ pub fn canonicalize_path(path: &Path) -> std::io::Result<PathBuf> {
     std::env::current_dir().map(|current_dir| normalize_path(&current_dir.join(path)))
 }
 
+pub fn auto_detect_line_ending(doc: &Rope) -> LineEnding {
+    // based on https://github.com/cessen/led/blob/27572c8838a1c664ee378a19358604063881cc1d/src/editor/mod.rs#L88-L162
+
+        let mut ending = LineEnding::None;
+        for line in doc.lines().take(1) { // check first line only - unsure how sound this is
+            // Get the line ending
+            ending = if line.len_chars() == 1 {
+                let g = RopeGraphemes::new(line.slice((line.len_chars() - 1)..))
+                    .last()
+                    .unwrap();
+                rope_slice_to_line_ending(&g)
+            } else if line.len_chars() > 1 {
+                let g = RopeGraphemes::new(line.slice((line.len_chars() - 2)..))
+                    .last()
+                    .unwrap();
+                rope_slice_to_line_ending(&g)
+            } else {
+                LineEnding::None
+            };
+        }
+        ending
+}
+
+pub fn rope_slice_to_line_ending(g: &RopeSlice) -> LineEnding {
+    if let Some(text) = g.as_str() {
+        str_to_line_ending(text)
+    } else if g == "\u{000D}\u{000A}" {
+        LineEnding::CRLF
+    } else {
+        // Not a line ending
+        LineEnding::None
+    }
+}
+
+pub fn str_to_line_ending(g: &str) -> LineEnding {
+    match g {
+        "\u{000D}\u{000A}" => LineEnding::CRLF,
+        "\u{000A}" => LineEnding::LF,
+        "\u{000B}" => LineEnding::VT,
+        "\u{000C}" => LineEnding::FF,
+        "\u{000D}" => LineEnding::CR,
+        "\u{0085}" => LineEnding::NEL,
+        "\u{2028}" => LineEnding::LS,
+        "\u{2029}" => LineEnding::PS,
+
+        // Not a line ending
+        _ => LineEnding::None,
+    }
+}
+
 use helix_lsp::lsp;
 use url::Url;
 
 impl Document {
-    pub fn new(text: Rope) -> Self {
+    pub fn new(text: Rope, line_ending: LineEnding) -> Self {
         let changes = ChangeSet::new(&text);
         let old_state = None;
 
@@ -171,6 +238,7 @@ impl Document {
             history: Cell::new(History::default()),
             last_saved_revision: 0,
             language_server: None,
+            line_ending: line_ending
         }
     }
 
@@ -190,7 +258,10 @@ impl Document {
             doc
         };
 
-        let mut doc = Self::new(doc);
+        // search for line endings
+        let line_ending = auto_detect_line_ending(&doc);
+
+        let mut doc = Self::new(doc, line_ending);
         // set the path and try detecting the language
         doc.set_path(&path)?;
         doc.detect_indent_style();