From c5b2973739901f8cd4bc26f3cfc8232249eb7968 Mon Sep 17 00:00:00 2001
From: Kirawi <67773714+kirawi@users.noreply.github.com>
Date: Fri, 2 Jul 2021 10:54:50 -0400
Subject: [PATCH] `:reload` (#374)

* reloading functionality

* fn with_newline_eof()

* fmt

* wip

* wip

* wip

* wip

* moved to core, added simd feature for encoding_rs

* wip

* rm

* .gitignore

* wip

* local wip

* wip

* wip

* no features

* wip

* nit

* remove simd

* doc

* clippy

* clippy

* address comments

* add indentation & line ending change
---
 Cargo.lock                 | 35 +++++++++++++++++++
 helix-core/Cargo.toml      |  5 +++
 helix-core/src/diff.rs     | 70 ++++++++++++++++++++++++++++++++++++++
 helix-core/src/lib.rs      |  1 +
 helix-term/src/commands.rs | 32 +++++++++++++++++
 helix-view/src/document.rs | 61 ++++++++++++++++++++++++++++-----
 6 files changed, 196 insertions(+), 8 deletions(-)
 create mode 100644 helix-core/src/diff.rs

diff --git a/Cargo.lock b/Cargo.lock
index 473ae8c8..a377e2f4 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -317,10 +317,12 @@ dependencies = [
  "etcetera",
  "helix-syntax",
  "once_cell",
+ "quickcheck",
  "regex",
  "ropey",
  "rust-embed",
  "serde",
+ "similar",
  "smallvec",
  "tendril",
  "toml",
@@ -692,6 +694,15 @@ dependencies = [
  "unicase",
 ]
 
+[[package]]
+name = "quickcheck"
+version = "1.0.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "588f6378e4dd99458b60ec275b4477add41ce4fa9f64dcba6f15adccb19b50d6"
+dependencies = [
+ "rand",
+]
+
 [[package]]
 name = "quote"
 version = "1.0.9"
@@ -701,6 +712,24 @@ dependencies = [
  "proc-macro2",
 ]
 
+[[package]]
+name = "rand"
+version = "0.8.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "2e7573632e6454cf6b99d7aac4ccca54be06da05aca2ef7423d22d27d4d4bcd8"
+dependencies = [
+ "rand_core",
+]
+
+[[package]]
+name = "rand_core"
+version = "0.6.3"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "d34f1408f55294453790c48b2f1ebbb1c5b4b7563eb1f418bcfcfdbb06ebb4e7"
+dependencies = [
+ "getrandom",
+]
+
 [[package]]
 name = "redox_syscall"
 version = "0.2.9"
@@ -872,6 +901,12 @@ dependencies = [
  "libc",
 ]
 
+[[package]]
+name = "similar"
+version = "1.3.0"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "1ad1d488a557b235fc46dae55512ffbfc429d2482b08b4d9435ab07384ca8aec"
+
 [[package]]
 name = "slab"
 version = "0.4.3"
diff --git a/helix-core/Cargo.toml b/helix-core/Cargo.toml
index 726e90cc..80d559a9 100644
--- a/helix-core/Cargo.toml
+++ b/helix-core/Cargo.toml
@@ -31,5 +31,10 @@ regex = "1"
 serde = { version = "1.0", features = ["derive"] }
 toml = "0.5"
 
+similar = "1.3"
+
 etcetera = "0.3"
 rust-embed = { version = "5.9.0", optional = true }
+
+[dev-dependencies]
+quickcheck = { version = "1", default-features = false }
diff --git a/helix-core/src/diff.rs b/helix-core/src/diff.rs
new file mode 100644
index 00000000..9c1fc999
--- /dev/null
+++ b/helix-core/src/diff.rs
@@ -0,0 +1,70 @@
+use ropey::Rope;
+
+use crate::{Change, Transaction};
+
+/// Compares `old` and `new` to generate a [`Transaction`] describing
+/// the steps required to get from `old` to `new`.
+pub fn compare_ropes(old: &Rope, new: &Rope) -> Transaction {
+    // `similar` only works on contiguous data, so a `Rope` has
+    // to be temporarily converted into a `String`.
+    let old_converted = old.to_string();
+    let new_converted = new.to_string();
+
+    // A timeout is set so after 1 seconds, the algorithm will start
+    // approximating. This is especially important for big `Rope`s or
+    // `Rope`s that are extremely dissimilar to each other.
+    //
+    // Note: Ignore the clippy warning, as the trait bounds of
+    // `Transaction::change()` require an iterator implementing
+    // `ExactIterator`.
+    let mut config = similar::TextDiff::configure();
+    config.timeout(std::time::Duration::from_secs(1));
+
+    let diff = config.diff_chars(&old_converted, &new_converted);
+
+    // The current position of the change needs to be tracked to
+    // construct the `Change`s.
+    let mut pos = 0;
+    let changes: Vec<Change> = diff
+        .ops()
+        .iter()
+        .map(|op| op.as_tag_tuple())
+        .filter_map(|(tag, old_range, new_range)| {
+            // `old_pos..pos` is equivalent to `start..end` for where
+            // the change should be applied.
+            let old_pos = pos;
+            pos += old_range.end - old_range.start;
+
+            match tag {
+                // Semantically, inserts and replacements are the same thing.
+                similar::DiffTag::Insert | similar::DiffTag::Replace => {
+                    // This is the text from the `new` rope that should be
+                    // inserted into `old`.
+                    let text: &str = {
+                        let start = new.char_to_byte(new_range.start);
+                        let end = new.char_to_byte(new_range.end);
+                        &new_converted[start..end]
+                    };
+                    Some((old_pos, pos, Some(text.into())))
+                }
+                similar::DiffTag::Delete => Some((old_pos, pos, None)),
+                similar::DiffTag::Equal => None,
+            }
+        })
+        .collect();
+    Transaction::change(old, changes.into_iter())
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    quickcheck::quickcheck! {
+        fn test_compare_ropes(a: String, b: String) -> bool {
+            let mut old = Rope::from(a);
+            let new = Rope::from(b);
+            compare_ropes(&old, &new).apply(&mut old);
+            old.to_string() == new.to_string()
+        }
+    }
+}
diff --git a/helix-core/src/lib.rs b/helix-core/src/lib.rs
index dfbbd748..c2bb8c55 100644
--- a/helix-core/src/lib.rs
+++ b/helix-core/src/lib.rs
@@ -2,6 +2,7 @@ pub mod auto_pairs;
 pub mod chars;
 pub mod comment;
 pub mod diagnostic;
+pub mod diff;
 pub mod graphemes;
 pub mod history;
 pub mod indent;
diff --git a/helix-term/src/commands.rs b/helix-term/src/commands.rs
index a3799e7e..860d8e22 100644
--- a/helix-term/src/commands.rs
+++ b/helix-term/src/commands.rs
@@ -1521,6 +1521,24 @@ mod cmd {
         }
     }
 
+    /// Sets the [`Document`]'s encoding..
+    fn set_encoding(cx: &mut compositor::Context, args: &[&str], _: PromptEvent) {
+        let (_, doc) = current!(cx.editor);
+        if let Some(label) = args.first() {
+            doc.set_encoding(label)
+                .unwrap_or_else(|e| cx.editor.set_error(e.to_string()));
+        } else {
+            let encoding = doc.encoding().name().to_string();
+            cx.editor.set_status(encoding)
+        }
+    }
+
+    /// Reload the [`Document`] from its source file.
+    fn reload(cx: &mut compositor::Context, _args: &[&str], _: PromptEvent) {
+        let (view, doc) = current!(cx.editor);
+        doc.reload(view.id).unwrap();
+    }
+
     pub const TYPABLE_COMMAND_LIST: &[TypableCommand] = &[
         TypableCommand {
             name: "quit",
@@ -1704,6 +1722,20 @@ mod cmd {
             fun: show_current_directory,
             completer: None,
         },
+        TypableCommand {
+            name: "encoding",
+            alias: None,
+            doc: "Set encoding based on `https://encoding.spec.whatwg.org`",
+            fun: set_encoding,
+            completer: None,
+        },
+        TypableCommand {
+            name: "reload",
+            alias: None,
+            doc: "Discard changes and reload from the source file.",
+            fun: reload,
+            completer: None,
+        }
     ];
 
     pub static COMMANDS: Lazy<HashMap<&'static str, &'static TypableCommand>> = Lazy::new(|| {
diff --git a/helix-view/src/document.rs b/helix-view/src/document.rs
index 0f1f3a8f..f85ded11 100644
--- a/helix-view/src/document.rs
+++ b/helix-view/src/document.rs
@@ -307,6 +307,19 @@ pub async fn to_writer<'a, W: tokio::io::AsyncWriteExt + Unpin + ?Sized>(
     Ok(())
 }
 
+/// Inserts the final line ending into `rope` if it's missing. [Why?](https://stackoverflow.com/questions/729692/why-should-text-files-end-with-a-newline)
+pub fn with_line_ending(rope: &mut Rope) -> LineEnding {
+    // search for line endings
+    let line_ending = auto_detect_line_ending(rope).unwrap_or(DEFAULT_LINE_ENDING);
+
+    // add missing newline at the end of file
+    if rope.len_bytes() == 0 || !char_is_line_ending(rope.char(rope.len_chars() - 1)) {
+        rope.insert(rope.len_chars(), line_ending.as_str());
+    }
+
+    line_ending
+}
+
 /// Like std::mem::replace() except it allows the replacement value to be mapped from the
 /// original value.
 fn take_with<T, F>(mut_ref: &mut T, closure: F)
@@ -449,14 +462,7 @@ impl Document {
 
         let mut file = std::fs::File::open(&path).context(format!("unable to open {:?}", path))?;
         let (mut rope, encoding) = from_reader(&mut file, encoding)?;
-
-        // search for line endings
-        let line_ending = auto_detect_line_ending(&rope).unwrap_or(DEFAULT_LINE_ENDING);
-
-        // add missing newline at the end of file
-        if rope.len_bytes() == 0 || !char_is_line_ending(rope.char(rope.len_chars() - 1)) {
-            rope.insert(rope.len_chars(), line_ending.as_str());
-        }
+        let line_ending = with_line_ending(&mut rope);
 
         let mut doc = Self::from(rope, Some(encoding));
 
@@ -586,6 +592,45 @@ impl Document {
         }
     }
 
+    /// Reload the document from its path.
+    pub fn reload(&mut self, view_id: ViewId) -> Result<(), Error> {
+        let encoding = &self.encoding;
+        let path = self.path().filter(|path| path.exists());
+
+        // If there is no path or the path no longer exists.
+        if path.is_none() {
+            return Err(anyhow!("can't find file to reload from"));
+        }
+
+        let mut file = std::fs::File::open(path.unwrap())?;
+        let (mut rope, ..) = from_reader(&mut file, Some(encoding))?;
+        let line_ending = with_line_ending(&mut rope);
+
+        let transaction = helix_core::diff::compare_ropes(self.text(), &rope);
+        self.apply(&transaction, view_id);
+        self.append_changes_to_history(view_id);
+
+        // Detect indentation style and set line ending.
+        self.detect_indent_style();
+        self.line_ending = line_ending;
+
+        Ok(())
+    }
+
+    /// Sets the [`Document`]'s encoding with the encoding correspondent to `label`.
+    pub fn set_encoding(&mut self, label: &str) -> Result<(), Error> {
+        match encoding_rs::Encoding::for_label(label.as_bytes()) {
+            Some(encoding) => self.encoding = encoding,
+            None => return Err(anyhow::anyhow!("unknown encoding")),
+        }
+        Ok(())
+    }
+
+    /// Returns the [`Document`]'s current encoding.
+    pub fn encoding(&self) -> &'static encoding_rs::Encoding {
+        self.encoding
+    }
+
     fn detect_indent_style(&mut self) {
         // Build a histogram of the indentation *increases* between
         // subsequent lines, ignoring lines that are all whitespace.