From a63dce2c468f3ab252a41486704dfc50146e5a62 Mon Sep 17 00:00:00 2001
From: Axel Kappel <69117984+Kl4rry@users.noreply.github.com>
Date: Thu, 27 Jun 2024 21:35:59 +0200
Subject: [PATCH] add automatic language detection

---
 crates/ferrite-core/src/buffer.rs          | 13 ++++-
 crates/ferrite-core/src/language.rs        |  1 +
 crates/ferrite-core/src/language/detect.rs | 66 ++++++++++++++++++++++
 thing.h                                    | 14 +++++
 4 files changed, 92 insertions(+), 2 deletions(-)
 create mode 100644 crates/ferrite-core/src/language/detect.rs
 create mode 100644 thing.h

diff --git a/crates/ferrite-core/src/buffer.rs b/crates/ferrite-core/src/buffer.rs
index 9f2ba61..d3d9277 100644
--- a/crates/ferrite-core/src/buffer.rs
+++ b/crates/ferrite-core/src/buffer.rs
@@ -22,7 +22,10 @@ use super::{
     indent::Indentation,
     language::{get_language_from_path, syntax::Syntax},
 };
-use crate::{clipboard, event_loop_proxy::EventLoopProxy, keymap::LineMoveDir};
+use crate::{
+    clipboard, event_loop_proxy::EventLoopProxy, keymap::LineMoveDir,
+    language::detect::detect_language,
+};
 
 pub mod case;
 pub mod encoding;
@@ -212,7 +215,13 @@ impl Buffer {
             syntax.update_text(rope.clone());
         }
 
-        // TODO add parsing of shebang and other heuristics
+        if let Some(language) = detect_language(syntax.get_language_name().as_deref(), rope.clone())
+        {
+            if let Err(err) = syntax.set_language(language) {
+                tracing::error!("Error setting language: {err}");
+            }
+            syntax.update_text(rope.clone());
+        }
 
         let name = path.file_name().unwrap().to_string_lossy().into();
 
diff --git a/crates/ferrite-core/src/language.rs b/crates/ferrite-core/src/language.rs
index eb4a682..0cf961c 100644
--- a/crates/ferrite-core/src/language.rs
+++ b/crates/ferrite-core/src/language.rs
@@ -5,6 +5,7 @@ use tree_sitter::Language;
 
 use self::syntax::HighlightConfiguration;
 
+pub mod detect;
 pub mod syntax;
 
 #[derive(Clone)]
diff --git a/crates/ferrite-core/src/language/detect.rs b/crates/ferrite-core/src/language/detect.rs
new file mode 100644
index 0000000..80688c8
--- /dev/null
+++ b/crates/ferrite-core/src/language/detect.rs
@@ -0,0 +1,66 @@
+use ropey::{Rope, RopeSlice};
+
+pub fn detect_language(inital_guess: Option<&str>, content: Rope) -> Option<&'static str> {
+    tracing::info!("inital_guess: {inital_guess:?}");
+    if inital_guess == Some("c") {
+        let cpp_markers = [
+            "public",
+            "protected",
+            "private",
+            "std::",
+            "dynamic_cast",
+            "static_cast",
+            "reinterpret_cast",
+            "#include <iostream>",
+            "#include <vector>",
+            "#include <string>",
+            "class",
+            "throw",
+            "catch",
+            "try",
+            "nullptr",
+            "const&",
+            "final",
+        ];
+        if detect_markers(content.slice(..), &cpp_markers) > 3 {
+            return Some("cpp");
+        }
+    }
+
+    detect_shebang(content.slice(..))
+}
+
+fn detect_shebang(content: RopeSlice) -> Option<&'static str> {
+    let first_line = content
+        .slice(..content.len_chars().min(1000))
+        .get_line(0)?
+        .to_string();
+
+    let shebangs = [
+        ("python3", "python"),
+        ("python2", "python"),
+        ("python", "python"),
+        ("#!/bin/bash", "bash"),
+        ("#!/usr/bin/bash", "bash"),
+        ("#!/bin/sh", "bash"),
+        ("#!/usr/bin/env bash", "bash"),
+        ("zsh", "bash"),
+    ];
+
+    for (shebang, language) in shebangs {
+        if first_line.contains(shebang) {
+            return Some(language);
+        }
+    }
+
+    None
+}
+
+fn detect_markers(content: RopeSlice, markers: &[&str]) -> usize {
+    let start = content.slice(..content.len_chars().min(1000)).to_string();
+    let mut count = 0;
+    for marker in markers {
+        count += start.contains(marker) as usize;
+    }
+    count
+}
diff --git a/thing.h b/thing.h
new file mode 100644
index 0000000..11f6205
--- /dev/null
+++ b/thing.h
@@ -0,0 +1,14 @@
+
+
+#include <iostream>
+#include <vector>
+#include <string>
+
+class Thing {
+    public:
+    int thing;
+}
+
+int main() {
+    std::cout << "hello" << std::endl;
+}
\ No newline at end of file