display lossy commit msg when containing invalid utf8 (closes #150)

CHANGELOG.md

@@ -15,6 +15,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Fixed
 - clearer help headers ([#131](https://github.com/extrawurst/gitui/issues/131))
+- diisplay non-utf8 commit messages at least partially ([#150](https://github.com/extrawurst/gitui/issues/150))
 
 ## [0.7.0] - 2020-06-15
 

Cargo.lock

@@ -45,6 +45,7 @@ version = "0.7.0"
 dependencies = [
  "crossbeam-channel",
  "git2",
+ "invalidstring",
  "log",
  "rayon-core",
  "scopetime",
@@ -367,6 +368,10 @@ dependencies = [
  "unicode-normalization",
 ]
 
+[[package]]
+name = "invalidstring"
+version = "0.1.0"
+
 [[package]]
 name = "iovec"
 version = "0.1.4"

asyncgit/Cargo.toml

@@ -12,12 +12,13 @@ categories = ["concurrency","asynchronous"]
 keywords = ["git"]
 
 [dependencies]
+scopetime = { path = "../scopetime", version = "0.1" }
 git2 = { version = "0.13.6", default-features = false }
 rayon-core = "1.7"
 crossbeam-channel = "0.4"
 log = "0.4"
-scopetime = { path = "../scopetime", version = "0.1" }
 thiserror = "1.0"
 
 [dev-dependencies]
-tempfile = "3.1"
+tempfile = "3.1"
+invalidstring = { path = "../invalidstring", version = "0.1" }

asyncgit/src/sync/commits_info.rs

@@ -2,6 +2,7 @@ use super::utils::repo;
 use crate::error::Result;
 use git2::{Commit, Error, Oid};
 use scopetime::scope_time;
+use std::borrow::Cow;
 
 /// identifies a single commit
 #[derive(Debug, Copy, Clone, PartialEq, Eq, Hash)]
@@ -81,14 +82,13 @@ pub fn get_commits_info(
 }
 
 fn get_message(c: &Commit, message_length_limit: usize) -> String {
-    if let Some(msg) = c.message() {
-        limit_str(msg, message_length_limit)
-    } else {
-        String::from("<unknown>")
-    }
+    limit_str(
+        String::from_utf8_lossy(c.message_bytes()),
+        message_length_limit,
+    )
 }
 
-fn limit_str(s: &str, limit: usize) -> String {
+fn limit_str(s: Cow<'_, str>, limit: usize) -> String {
     if let Some(first) = s.lines().next() {
         first.chars().take(limit).collect::<String>()
     } else {
@@ -103,6 +103,7 @@ mod tests {
     use crate::error::Result;
     use crate::sync::{
         commit, stage_add_file, tests::repo_init_empty,
+        utils::get_head_repo,
     };
     use std::{fs::File, io::Write, path::Path};
 
@@ -130,4 +131,31 @@ mod tests {
 
         Ok(())
     }
+
+    #[test]
+    fn test_invalid_utf8() -> Result<()> {
+        let file_path = Path::new("foo");
+        let (_td, repo) = repo_init_empty().unwrap();
+        let root = repo.path().parent().unwrap();
+        let repo_path = root.as_os_str().to_str().unwrap();
+
+        File::create(&root.join(file_path))?.write_all(b"a")?;
+        stage_add_file(repo_path, file_path).unwrap();
+
+        let msg = invalidstring::invalid_utf8("test msg");
+        commit(repo_path, msg.as_str()).unwrap();
+
+        let res = get_commits_info(
+            repo_path,
+            &vec![get_head_repo(&repo).unwrap().into()],
+            50,
+        )
+        .unwrap();
+
+        assert_eq!(res.len(), 1);
+        dbg!(&res[0].message);
+        assert_eq!(res[0].message.starts_with("test msg"), true);
+
+        Ok(())
+    }
 }

invalidstring/Cargo.toml

@@ -0,0 +1,14 @@
+[package]
+name = "invalidstring"
+version = "0.1.0"
+authors = ["Stephan Dilly <dilly.stephan@gmail.com>"]
+edition = "2018"
+description = "just for testing invalid string data"
+homepage = "https://github.com/extrawurst/gitui"
+repository = "https://github.com/extrawurst/gitui"
+readme = "README.md"
+license = "MIT"
+categories = ["testing","utf8"]
+keywords = ["string"]
+
+[dependencies]

invalidstring/README.md

@@ -0,0 +1,5 @@
+# invalidstring
+
+*just for testing invalid string data*
+
+This crate is part of the [gitui](http://gitui.org) project. We need this to be a seperate crate so that `asyncgit` can remain forbidding `unsafe`.

invalidstring/src/lib.rs

@@ -0,0 +1,8 @@
+/// uses unsafe to postfix the string with invalid utf8 data
+pub fn invalid_utf8(prefix: &str) -> String {
+    let bytes = b"\xc3\x73";
+
+    unsafe {
+        format!("{}{}", prefix, std::str::from_utf8_unchecked(bytes))
+    }
+}