githubmirror/lapce

lapce-proxy/src/buffer.rs
303 строки10 KB
Dongdong Zhou
xi-rope switch to crates.io
01 ноя 2022, 22:30
01 ноя 2022, 22:30417aaab
use std::{
    borrow::Cow,
    ffi::OsString,
    fs,
    fs::File,
    io::{Read, Write},
    path::{Path, PathBuf},
    time::SystemTime,
};

use anyhow::{anyhow, Result};
use lapce_core::{
    buffer::rope_text::CharIndicesJoin, encoding::offset_utf8_to_utf16,
};
use lapce_rpc::buffer::BufferId;
use lapce_xi_rope::{interval::IntervalBounds, rope::Rope, RopeDelta};
use lsp_types::*;

#[derive(Clone)]
pub struct Buffer {
    pub language_id: &'static str,
    pub id: BufferId,
    pub rope: Rope,
    pub path: PathBuf,
    pub rev: u64,
    pub mod_time: Option<SystemTime>,
}

impl Buffer {
    pub fn new(id: BufferId, path: PathBuf) -> Buffer {
        let rope = Rope::from(load_file(&path).unwrap_or_default());
        let rev = if rope.is_empty() { 0 } else { 1 };
        let language_id = language_id_from_path(&path).unwrap_or("");
        let mod_time = get_mod_time(&path);
        Buffer {
            id,
            rope,
            path,
            language_id,
            rev,
            mod_time,
        }
    }

    pub fn save(&mut self, rev: u64) -> Result<()> {
        if self.rev != rev {
            return Err(anyhow!("not the right rev"));
        }
        let tmp_extension = self.path.extension().map_or_else(
            || OsString::from("swp"),
            |ext| {
                let mut ext = ext.to_os_string();
                ext.push(".swp");
                ext
            },
        );
        let tmp_path = &self.path.with_extension(tmp_extension);

        let mut f = File::create(tmp_path)?;
        for chunk in self.rope.iter_chunks(..self.rope.len()) {
            f.write_all(chunk.as_bytes())?;
        }

        if let Ok(metadata) = fs::metadata(&self.path) {
            let perm = metadata.permissions();
            fs::set_permissions(tmp_path, perm)?;
        }

        fs::rename(tmp_path, &self.path)?;
        self.mod_time = get_mod_time(&self.path);
        Ok(())
    }

    pub fn update(
        &mut self,
        delta: &RopeDelta,
        rev: u64,
    ) -> Option<TextDocumentContentChangeEvent> {
        if self.rev + 1 != rev {
            return None;
        }
        self.rev += 1;
        let content_change = get_document_content_changes(delta, self);
        self.rope = delta.apply(&self.rope);
        Some(
            content_change.unwrap_or_else(|| TextDocumentContentChangeEvent {
                range: None,
                range_length: None,
                text: self.get_document(),
            }),
        )
    }

    pub fn get_document(&self) -> String {
        self.rope.to_string()
    }

    pub fn offset_of_line(&self, offset: usize) -> usize {
        self.rope.offset_of_line(offset)
    }

    pub fn line_of_offset(&self, offset: usize) -> usize {
        self.rope.line_of_offset(offset)
    }

    pub fn offset_to_line_col(&self, offset: usize) -> (usize, usize) {
        let line = self.line_of_offset(offset);
        (line, offset - self.offset_of_line(line))
    }

    /// Converts a UTF8 offset to a UTF16 LSP position  
    pub fn offset_to_position(&self, offset: usize) -> Position {
        let (line, col) = self.offset_to_line_col(offset);
        // Get the offset of line to make the conversion cheaper, rather than working
        // from the very start of the document to `offset`
        let line_offset = self.offset_of_line(line);
        let utf16_col =
            offset_utf8_to_utf16(self.char_indices_iter(line_offset..), col);

        Position {
            line: line as u32,
            character: utf16_col as u32,
        }
    }

    pub fn slice_to_cow<T: IntervalBounds>(&self, range: T) -> Cow<str> {
        self.rope.slice_to_cow(range)
    }

    /// Iterate over (utf8_offset, char) values in the given range  
    /// This uses `iter_chunks` and so does not allocate, compared to `slice_to_cow` which can
    pub fn char_indices_iter<T: IntervalBounds>(
        &self,
        range: T,
    ) -> impl Iterator<Item = (usize, char)> + '_ {
        CharIndicesJoin::new(self.rope.iter_chunks(range).map(str::char_indices))
    }

    pub fn len(&self) -> usize {
        self.rope.len()
    }

    pub fn is_empty(&self) -> bool {
        self.len() == 0
    }
}

pub fn load_file(path: &Path) -> Result<String> {
    Ok(read_path_to_string_lossy(path)?)
}

pub fn read_path_to_string_lossy<P: AsRef<Path>>(
    path: P,
) -> Result<String, std::io::Error> {
    let path = path.as_ref();

    let mut file = File::open(path)?;
    // Read the file in as bytes
    let mut buffer = Vec::new();
    file.read_to_end(&mut buffer)?;

    // Parse the file contents as utf8, replacing non-utf8 data with the
    // replacement character
    let contents = String::from_utf8_lossy(&buffer);

    Ok(contents.to_string())
}

pub fn language_id_from_path(path: &Path) -> Option<&'static str> {
    // recommended language_id values
    // https://microsoft.github.io/language-server-protocol/specifications/lsp/3.17/specification/#textDocumentItem
    Some(match path.extension() {
        Some(ext) => {
            match ext.to_str()? {
                "C" | "H" => "cpp",
                "M" => "objective-c",
                // stop case-sensitive matching
                ext => match ext.to_lowercase().as_str() {
                    "bat" => "bat",
                    "clj" | "cljs" | "cljc" | "edn" => "clojure",
                    "coffee" => "coffeescript",
                    "c" | "h" => "c",
                    "cpp" | "hpp" | "cxx" | "hxx" | "c++" | "h++" | "cc" | "hh" => {
                        "cpp"
                    }
                    "cs" | "csx" => "csharp",
                    "css" => "css",
                    "d" | "di" | "dlang" => "dlang",
                    "diff" | "patch" => "diff",
                    "dart" => "dart",
                    "dockerfile" => "dockerfile",
                    "elm" => "elm",
                    "ex" | "exs" => "elixir",
                    "erl" | "hrl" => "erlang",
                    "fs" | "fsi" | "fsx" | "fsscript" => "fsharp",
                    "git-commit" | "git-rebase" => "git",
                    "go" => "go",
                    "groovy" | "gvy" | "gy" | "gsh" => "groovy",
                    "hbs" => "handlebars",
                    "htm" | "html" | "xhtml" => "html",
                    "ini" => "ini",
                    "java" | "class" => "java",
                    "js" => "javascript",
                    "jsx" => "javascriptreact",
                    "json" => "json",
                    "jl" => "julia",
                    "kt" | "kts" => "kotlin",
                    "less" => "less",
                    "lua" => "lua",
                    "makefile" | "gnumakefile" => "makefile",
                    "md" | "markdown" => "markdown",
                    "m" => "objective-c",
                    "mm" => "objective-cpp",
                    "plx" | "pl" | "pm" | "xs" | "t" | "pod" | "cgi" => "perl",
                    "p6" | "pm6" | "pod6" | "t6" | "raku" | "rakumod"
                    | "rakudoc" | "rakutest" => "perl6",
                    "php" | "phtml" | "pht" | "phps" => "php",
                    "ps1" | "ps1xml" | "psc1" | "psm1" | "psd1" | "pssc"
                    | "psrc" => "powershell",
                    "py" | "pyi" | "pyc" | "pyd" | "pyw" => "python",
                    "r" => "r",
                    "rb" => "ruby",
                    "rs" => "rust",
                    "scss" | "sass" => "scss",
                    "sc" | "scala" => "scala",
                    "sh" | "bash" | "zsh" => "shellscript",
                    "sql" => "sql",
                    "swift" => "swift",
                    "svelte" => "svelte",
                    "toml" => "toml",
                    "ts" => "typescript",
                    "tsx" => "typescriptreact",
                    "tex" => "tex",
                    "vb" => "vb",
                    "xml" => "xml",
                    "xsl" => "xsl",
                    "yml" | "yaml" => "yaml",
                    "zig" => "zig",
                    "vue" => "vue",
                    _ => return None,
                },
            }
        }
        // Handle paths without extension
        #[allow(clippy::match_single_binding)]
        None => match path.file_name()?.to_str()? {
            // case-insensitive matching
            filename => match filename.to_lowercase().as_str() {
                "dockerfile" => "dockerfile",
                "makefile" | "gnumakefile" => "makefile",
                _ => return None,
            },
        },
    })
}

fn get_document_content_changes(
    delta: &RopeDelta,
    buffer: &Buffer,
) -> Option<TextDocumentContentChangeEvent> {
    let (interval, _) = delta.summary();
    let (start, end) = interval.start_end();

    // TODO: Handle more trivial cases like typing when there's a selection or transpose
    if let Some(node) = delta.as_simple_insert() {
        let (start, end) = interval.start_end();
        let start = buffer.offset_to_position(start);

        let end = buffer.offset_to_position(end);

        Some(TextDocumentContentChangeEvent {
            range: Some(Range { start, end }),
            range_length: None,
            text: String::from(node),
        })
    }
    // Or a simple delete
    else if delta.is_simple_delete() {
        let end_position = buffer.offset_to_position(end);

        let start = buffer.offset_to_position(start);

        Some(TextDocumentContentChangeEvent {
            range: Some(Range {
                start,
                end: end_position,
            }),
            range_length: None,
            text: String::new(),
        })
    } else {
        None
    }
}

/// Returns the modification timestamp for the file at a given path,
/// if present.
pub fn get_mod_time<P: AsRef<Path>>(path: P) -> Option<SystemTime> {
    File::open(path)
        .and_then(|f| f.metadata())
        .and_then(|meta| meta.modified())
        .ok()
}