Prevent preview binary or large file (#939)

* Prevent preview binary or large file (#847)

* fix wrong method name

* fix add use trait

* update lock file

* rename MAX_PREVIEW_SIZE from MAX_BYTE_PREVIEW

* read small bytes to determine cotent type

* [WIP] add preview struct to represent calcurated preveiw

* Refactor content type detection

- Remove unwraps
- Reuse a single read buffer to avoid 1kb reallocations between previews

* Refactor preview rendering so we don't construct docs when not necessary

* Replace unwarap whit Preview::NotFound

* Use index access to hide unwrap

Co-authored-by: Blaž Hrastnik <blaz@mxxn.io>

* fix Get and unwarp equivalent to referce of Index acess

* better preview implementation

* Rename Preview enum and vairant

Co-authored-by: Gokul Soumya <gokulps15@gmail.com>

* fixup! Rename Preview enum and vairant

* simplify long match

* Center text, add docs, fix formatting, refactor

Co-authored-by: Blaž Hrastnik <blaz@mxxn.io>
Co-authored-by: Gokul Soumya <gokulps15@gmail.com>

3 files changed, 115 insertions, 20 deletions

diff --git a/Cargo.lock b/Cargo.lock
index 0ce8ee8f..e036828a 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -111,6 +111,15 @@ dependencies = [
 ]
 
 [[package]]
+name = "content_inspector"
+version = "0.2.4"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "b7bda66e858c683005a53a9a60c69a4aca7eeaa45d124526e389f7aec8e62f38"
+dependencies = [
+ "memchr",
+]
+
+[[package]]
 name = "crossbeam-utils"
 version = "0.8.5"
 source = "registry+https://github.com/rust-lang/crates.io-index"
@@ -415,6 +424,7 @@ version = "0.5.0"
 dependencies = [
  "anyhow",
  "chrono",
+ "content_inspector",
  "crossterm",
  "fern",
  "futures-util",
diff --git a/helix-term/Cargo.toml b/helix-term/Cargo.toml
index 45b4eb2c..a0079feb 100644
--- a/helix-term/Cargo.toml
+++ b/helix-term/Cargo.toml
@@ -46,6 +46,8 @@ fuzzy-matcher = "0.3"
 ignore = "0.4"
 # markdown doc rendering
 pulldown-cmark = { version = "0.8", default-features = false }
+# file type detection
+content_inspector = "0.2.4"
 
 # config
 toml = "0.5"
diff --git a/helix-term/src/ui/picker.rs b/helix-term/src/ui/picker.rs
index 7fc6af0f..291f1f85 100644
--- a/helix-term/src/ui/picker.rs
+++ b/helix-term/src/ui/picker.rs
@@ -12,7 +12,12 @@ use fuzzy_matcher::skim::SkimMatcherV2 as Matcher;
 use fuzzy_matcher::FuzzyMatcher;
 use tui::widgets::Widget;
 
-use std::{borrow::Cow, collections::HashMap, path::PathBuf};
+use std::{
+    borrow::Cow,
+    collections::HashMap,
+    io::Read,
+    path::{Path, PathBuf},
+};
 
 use crate::ui::{Prompt, PromptEvent};
 use helix_core::Position;
@@ -23,18 +28,58 @@ use helix_view::{
 };
 
 pub const MIN_SCREEN_WIDTH_FOR_PREVIEW: u16 = 80;
+/// Biggest file size to preview in bytes
+pub const MAX_FILE_SIZE_FOR_PREVIEW: u64 = 10 * 1024 * 1024;
 
-/// File path and line number (used to align and highlight a line)
+/// File path and range of lines (used to align and highlight lines)
 type FileLocation = (PathBuf, Option<(usize, usize)>);
 
 pub struct FilePicker<T> {
     picker: Picker<T>,
     /// Caches paths to documents
-    preview_cache: HashMap<PathBuf, Document>,
+    preview_cache: HashMap<PathBuf, CachedPreview>,
+    read_buffer: Vec<u8>,
     /// Given an item in the picker, return the file path and line number to display.
     file_fn: Box<dyn Fn(&Editor, &T) -> Option<FileLocation>>,
 }
 
+pub enum CachedPreview {
+    Document(Document),
+    Binary,
+    LargeFile,
+    NotFound,
+}
+
+// We don't store this enum in the cache so as to avoid lifetime constraints
+// from borrowing a document already opened in the editor.
+pub enum Preview<'picker, 'editor> {
+    Cached(&'picker CachedPreview),
+    EditorDocument(&'editor Document),
+}
+
+impl Preview<'_, '_> {
+    fn document(&self) -> Option<&Document> {
+        match self {
+            Preview::EditorDocument(doc) => Some(doc),
+            Preview::Cached(CachedPreview::Document(doc)) => Some(doc),
+            _ => None,
+        }
+    }
+
+    /// Alternate text to show for the preview.
+    fn placeholder(&self) -> &str {
+        match *self {
+            Self::EditorDocument(_) => "<File preview>",
+            Self::Cached(preview) => match preview {
+                CachedPreview::Document(_) => "<File preview>",
+                CachedPreview::Binary => "<Binary file>",
+                CachedPreview::LargeFile => "<File too large to preview>",
+                CachedPreview::NotFound => "<File not found>",
+            },
+        }
+    }
+}
+
 impl<T> FilePicker<T> {
     pub fn new(
         options: Vec<T>,
@@ -45,6 +90,7 @@ impl<T> FilePicker<T> {
         Self {
             picker: Picker::new(false, options, format_fn, callback_fn),
             preview_cache: HashMap::new(),
+            read_buffer: Vec::with_capacity(1024),
             file_fn: Box::new(preview_fn),
         }
     }
@@ -60,14 +106,45 @@ impl<T> FilePicker<T> {
             })
     }
 
-    fn calculate_preview(&mut self, editor: &Editor) {
-        if let Some((path, _line)) = self.current_file(editor) {
-            if !self.preview_cache.contains_key(&path) && editor.document_by_path(&path).is_none() {
-                // TODO: enable syntax highlighting; blocked by async rendering
-                let doc = Document::open(&path, None, Some(&editor.theme), None).unwrap();
-                self.preview_cache.insert(path, doc);
-            }
+    /// Get (cached) preview for a given path. If a document corresponding
+    /// to the path is already open in the editor, it is used instead.
+    fn get_preview<'picker, 'editor>(
+        &'picker mut self,
+        path: &Path,
+        editor: &'editor Editor,
+    ) -> Preview<'picker, 'editor> {
+        if let Some(doc) = editor.document_by_path(path) {
+            return Preview::EditorDocument(doc);
+        }
+
+        if self.preview_cache.contains_key(path) {
+            return Preview::Cached(&self.preview_cache[path]);
         }
+
+        let data = std::fs::File::open(path).and_then(|file| {
+            let metadata = file.metadata()?;
+            // Read up to 1kb to detect the content type
+            let n = file.take(1024).read_to_end(&mut self.read_buffer)?;
+            let content_type = content_inspector::inspect(&self.read_buffer[..n]);
+            self.read_buffer.clear();
+            Ok((metadata, content_type))
+        });
+        let preview = data
+            .map(
+                |(metadata, content_type)| match (metadata.len(), content_type) {
+                    (_, content_inspector::ContentType::BINARY) => CachedPreview::Binary,
+                    (size, _) if size > MAX_FILE_SIZE_FOR_PREVIEW => CachedPreview::LargeFile,
+                    _ => {
+                        // TODO: enable syntax highlighting; blocked by async rendering
+                        Document::open(path, None, Some(&editor.theme), None)
+                            .map(CachedPreview::Document)
+                            .unwrap_or(CachedPreview::NotFound)
+                    }
+                },
+            )
+            .unwrap_or(CachedPreview::NotFound);
+        self.preview_cache.insert(path.to_owned(), preview);
+        Preview::Cached(&self.preview_cache[path])
     }
 }
 
@@ -79,12 +156,12 @@ impl<T: 'static> Component for FilePicker<T> {
         // |picker   | |         |
         // |         | |         |
         // +---------+ +---------+
-        self.calculate_preview(cx.editor);
         let render_preview = area.width > MIN_SCREEN_WIDTH_FOR_PREVIEW;
         let area = inner_rect(area);
         // -- Render the frame:
         // clear area
         let background = cx.editor.theme.get("ui.background");
+        let text = cx.editor.theme.get("ui.text");
         surface.clear_with(area, background);
 
         let picker_width = if render_preview {
@@ -113,17 +190,23 @@ impl<T: 'static> Component for FilePicker<T> {
             horizontal: 1,
         };
         let inner = inner.inner(&margin);
-
         block.render(preview_area, surface);
 
-        if let Some((doc, line)) = self.current_file(cx.editor).and_then(|(path, range)| {
-            cx.editor
-                .document_by_path(&path)
-                .or_else(|| self.preview_cache.get(&path))
-                .zip(Some(range))
-        }) {
+        if let Some((path, range)) = self.current_file(cx.editor) {
+            let preview = self.get_preview(&path, cx.editor);
+            let doc = match preview.document() {
+                Some(doc) => doc,
+                None => {
+                    let alt_text = preview.placeholder();
+                    let x = inner.x + inner.width.saturating_sub(alt_text.len() as u16) / 2;
+                    let y = inner.y + inner.height / 2;
+                    surface.set_stringn(x, y, alt_text, inner.width as usize, text);
+                    return;
+                }
+            };
+
             // align to middle
-            let first_line = line
+            let first_line = range
                 .map(|(start, end)| {
                     let height = end.saturating_sub(start) + 1;
                     let middle = start + (height.saturating_sub(1) / 2);
@@ -150,7 +233,7 @@ impl<T: 'static> Component for FilePicker<T> {
             );
 
             // highlight the line
-            if let Some((start, end)) = line {
+            if let Some((start, end)) = range {
                 let offset = start.saturating_sub(first_line) as u16;
                 surface.set_style(
                     Rect::new(