libra: add conventional commit message check

common/Cargo.toml

@@ -21,3 +21,4 @@ config = { workspace = true }
 envsubst = "0.2.1"
 rand = { workspace = true }
 serde_json = { workspace = true }
+regex.workspace = true

common/src/utils.rs

@@ -1,5 +1,6 @@
 use idgenerator::IdInstance;
 use rand::{distributions::Alphanumeric, thread_rng, Rng};
+use regex::Regex;
 use serde_json::{json, Value};
 
 pub const ZERO_ID: &str = match std::str::from_utf8(&[b'0'; 40]) {
@@ -44,3 +45,127 @@ pub fn generate_rich_text(content: &str) -> String {
 pub fn mr_ref_name(mr_link: &str) -> String {
     format!("refs/heads/{}", mr_link)
 }
+
+/// Format commit message with GPG signature<br>
+/// There must be a `blank line`(\n) before `message`, or remote unpack failed.<br>
+/// If there is `GPG signature`,
+/// `blank line` should be placed between `signature` and `message`
+pub fn format_commit_msg(msg: &str, gpg_sig: Option<&str>) -> String {
+    match gpg_sig {
+        None => {
+            format!("\n{}", msg)
+        }
+        Some(gpg) => {
+            format!("{}\n\n{}", gpg, msg)
+        }
+    }
+}
+
+/// parse commit message
+pub fn parse_commit_msg(msg_gpg: &str) -> (String, Option<String>) {
+    const GPG_SIG_START: &str = "gpgsig -----BEGIN PGP SIGNATURE-----";
+    const GPG_SIG_END: &str = "-----END PGP SIGNATURE-----";
+    let gpg_start = msg_gpg.find(GPG_SIG_START);
+    let gpg_end = msg_gpg.find(GPG_SIG_END).map(|end| end + GPG_SIG_END.len());
+    let gpg_sig = match (gpg_start, gpg_end) {
+        (Some(start), Some(end)) => {
+            if start < end {
+                Some(msg_gpg[start..end].to_string())
+            } else {
+                None
+            }
+        }
+        _ => None,
+    };
+    match gpg_sig {
+        Some(gpg) => {
+            // Skip the leading '\n\n' (blank lines).
+            // Some commit messages may use '\n \n\n' or similar patterns.
+            // To handle such cases, remove all leading blank lines from the message.
+            let msg = msg_gpg[gpg_end.unwrap()..].trim_start().to_string();
+            (msg, Some(gpg))
+        }
+        None => {
+            assert!(msg_gpg.starts_with('\n'), "commit message format error");
+            let msg = msg_gpg[1..].to_string(); // skip the leading '\n' (blank line)
+            (msg, None)
+        }
+    }
+}
+
+// check if the commit message is conventional commit
+// ref: https://www.conventionalcommits.org/en/v1.0.0/
+pub fn check_conventional_commits_message(msg: &str) -> bool {
+    let first_line = msg.lines().next().unwrap_or_default();
+    #[allow(unused_variables)]
+    let body_footer = msg.lines().skip(1).collect::<Vec<_>>().join("\n");
+
+    let unicode_pattern = r"\p{L}\p{N}\p{P}\p{S}\p{Z}";
+    // type only support characters&numbers, others fields support all unicode characters
+    let regex_str = format!(
+        r"^(?P<type>[\p{{L}}\p{{N}}_-]+)(?:\((?P<scope>[{unicode}]+)\))?!?: (?P<description>[{unicode}]+)$",
+        unicode = unicode_pattern
+    );
+
+    let re = Regex::new(&regex_str).unwrap();
+    const RECOMMENDED_TYPES: [&str; 8] = [
+        "build", "chore", "ci", "docs", "feat", "fix", "perf", "refactor",
+    ];
+
+    if let Some(captures) = re.captures(first_line) {
+        let commit_type = captures.name("type").map(|m| m.as_str().to_string());
+        #[allow(unused_variables)]
+        let scope = captures.name("scope").map(|m| m.as_str().to_string());
+        let description = captures.name("description").map(|m| m.as_str().to_string());
+        if commit_type.is_none() || description.is_none() {
+            return false;
+        }
+
+        let commit_type = commit_type.unwrap();
+        if !RECOMMENDED_TYPES.contains(&commit_type.to_lowercase().as_str()) {
+            println!("`{}` is not a recommended commit type, refer to https://www.conventionalcommits.org/en/v1.0.0/ for more information", commit_type);
+        }
+
+        // println!("{}({}): {}\n{}", commit_type, scope.unwrap_or("None".to_string()), description.unwrap(), body_footer);
+
+        return true;
+    }
+    false
+}
+
+#[cfg(test)]
+mod test {
+    use super::*;
+
+    #[test]
+    fn test_check_conventional_commits() {
+        // successfull cases
+        let msg = "feat: add new feature";
+        assert!(check_conventional_commits_message(msg));
+
+        let msg = "fix(common crate): bug fix";
+        assert!(check_conventional_commits_message(msg));
+
+        let msg = "chore(范围)!: 依存関係を更新する";
+        assert!(check_conventional_commits_message(msg));
+
+        let msg = "se_lf-ty9pe(scope)!: Description\n\n여기 시체가 있어요\n\nвот нога";
+        assert!(check_conventional_commits_message(msg));
+
+        let msg = "feat(scope)!: Description\n\n\nbody one\n\nbody two\n\nfooter";
+        assert!(check_conventional_commits_message(msg));
+
+        // failed casesmsg
+        let msg = "feat:add new feature"; // missing ' ' before ':'
+        assert!(!check_conventional_commits_message(msg));
+
+        let msg = "fix(common crate)bug fix"; // missing ':'
+        assert!(!check_conventional_commits_message(msg));
+
+        let msg = "類@型(common): add new feature"; // unssupported characters in type
+        assert!(!check_conventional_commits_message(msg));
+
+        let msg = "()(common): add new feature"; // unssupported characters in type
+        assert!(!check_conventional_commits_message(msg));
+    }
+}

libra/Cargo.toml

@@ -17,6 +17,7 @@ chrono = { workspace = true }
 clap = { workspace = true, features = ["derive"] }
 color-backtrace = "0.6.1"
 colored = { workspace = true }
+common = { workspace = true }
 flate2 = { workspace = true } # add features = ["zlib"] if slow
 futures = { workspace = true }
 futures-util = { workspace = true }

libra/src/command/branch.rs

@@ -234,13 +234,15 @@ mod tests {
         let commit_args = CommitArgs {
             message: "first".to_string(),
             allow_empty: true,
+            conventional: false,
         };
         commit::execute(commit_args).await;
         let first_commit_id = Branch::find_branch("master", None).await.unwrap().commit;
 
         let commit_args = CommitArgs {
             message: "second".to_string(),
             allow_empty: true,
+            conventional: false,
         };
         commit::execute(commit_args).await;
         let second_commit_id = Branch::find_branch("master", None).await.unwrap().commit;
@@ -318,6 +320,7 @@ mod tests {
         let args = CommitArgs {
             message: "first".to_string(),
             allow_empty: true,
+            conventional: false,
         };
         commit::execute(args).await;
         let hash = Head::current_commit().await.unwrap();
@@ -349,6 +352,7 @@ mod tests {
         let args = CommitArgs {
             message: "first".to_string(),
             allow_empty: true,
+            conventional: false,
         };
         commit::execute(args).await;
 

libra/src/command/commit.rs

@@ -6,14 +6,15 @@ use crate::internal::head::Head;
 use crate::utils::client_storage::ClientStorage;
 use crate::utils::path;
 use crate::utils::util;
-use mercury::internal::index::Index;
 use clap::Parser;
+use common::utils::{check_conventional_commits_message, format_commit_msg};
 use mercury::hash::SHA1;
+use mercury::internal::index::Index;
 use mercury::internal::object::commit::Commit;
 use mercury::internal::object::tree::{Tree, TreeItem, TreeItemMode};
 use mercury::internal::object::ObjectTrait;
 
-use super::{format_commit_msg, save_object};
+use super::save_object;
 
 #[derive(Parser, Debug)]
 pub struct CommitArgs {
@@ -22,6 +23,9 @@ pub struct CommitArgs {
 
     #[arg(long)]
     pub allow_empty: bool,
+
+    #[arg(long, requires("message"))]
+    pub conventional: bool,
 }
 
 pub async fn execute(args: CommitArgs) {
@@ -30,7 +34,12 @@ pub async fn execute(args: CommitArgs) {
     let storage = ClientStorage::init(path::objects());
     let tracked_entries = index.tracked_entries(0);
     if tracked_entries.is_empty() && !args.allow_empty {
-        panic!("fatal: no changes added to commit, use --allow-empty to override");
+        println!("fatal: no changes added to commit, use --allow-empty to override");
+        return;
+    }
+    if args.conventional && !check_conventional_commits_message(&args.message) {
+        println!("fatal: commit message does not follow conventional commits");
+        return;
     }
 
     /* Create tree */
@@ -39,7 +48,11 @@ pub async fn execute(args: CommitArgs) {
     /* Create & save commit objects */
     let parents_commit_ids = get_parents_ids().await;
     // There must be a `blank line`(\n) before `message`, or remote unpack failed
-    let commit = Commit::from_tree_id(tree.id, parents_commit_ids, &format_commit_msg(&args.message, None));
+    let commit = Commit::from_tree_id(
+        tree.id,
+        parents_commit_ids,
+        &format_commit_msg(&args.message, None),
+    );
 
     // TODO  default signature created in `from_tree_id`, wait `git config` to set correct user info
 
@@ -161,6 +174,23 @@ mod test {
     };
 
     use super::*;
+    #[test]
+    fn test_parse_args() {
+        let args = CommitArgs::try_parse_from(["commit", "-m", "init"]);
+        assert!(args.is_ok());
+
+        let args = CommitArgs::try_parse_from(["commit", "-m", "init", "--allow-empty"]);
+        assert!(args.is_ok());
+
+        let args = CommitArgs::try_parse_from(["commit", "--conventional", "-m", "init"]);
+        assert!(args.is_ok());
+
+        let args = CommitArgs::try_parse_from(["commit", "--conventional"]);
+        assert!(args.is_err(), "conventional should require message");
+
+        let args = CommitArgs::try_parse_from(["commit"]);
+        assert!(args.is_err(), "message is required");
+    }
 
     #[tokio::test]
     async fn test_create_tree() {
@@ -191,6 +221,7 @@ mod test {
         let args = CommitArgs {
             message: "init".to_string(),
             allow_empty: false,
+            conventional: false,
         };
         execute(args).await;
     }
@@ -203,6 +234,7 @@ mod test {
             let args = CommitArgs {
                 message: "init".to_string(),
                 allow_empty: true,
+                conventional: false,
             };
             execute(args).await;
 
@@ -219,7 +251,7 @@ mod test {
             let branch = Branch::find_branch(&branch_name, None).await.unwrap();
             assert_eq!(branch.commit, commit.id);
         }
-        
+
         // create a new commit
         {
             // create `a.txt` `bb/b.txt` `bb/c.txt`
@@ -239,6 +271,7 @@ mod test {
             let args = CommitArgs {
                 message: "add some files".to_string(),
                 allow_empty: false,
+                conventional: false,
             };
             execute(args).await;
 

libra/src/command/log.rs

@@ -16,7 +16,7 @@ use std::str::FromStr;
 use mercury::hash::SHA1;
 use mercury::internal::object::commit::Commit;
 
-use super::parse_commit_msg;
+use common::utils::parse_commit_msg;
 #[derive(Parser, Debug)]
 pub struct LogArgs {
     /// Limit the number of output

libra/src/command/mod.rs

@@ -86,53 +86,6 @@ pub fn ask_basic_auth() -> BasicAuth {
     BasicAuth { username, password }
 }
 
-/// Format commit message with GPG signature<br>
-/// There must be a `blank line`(\n) before `message`, or remote unpack failed.<br>
-/// If there is `GPG signature`,
-/// `blank line` should be placed between `signature` and `message`
-pub fn format_commit_msg(msg: &str, gpg_sig: Option<&str>) -> String {
-    match gpg_sig {
-        None => {
-            format!("\n{}", msg)
-        }
-        Some(gpg) => {
-            format!("{}\n\n{}", gpg, msg)
-        }
-    }
-}
-
-/// parse commit message
-pub fn parse_commit_msg(msg_gpg: &str) -> (String, Option<String>) {
-    const GPG_SIG_START: &str = "gpgsig -----BEGIN PGP SIGNATURE-----";
-    const GPG_SIG_END: &str = "-----END PGP SIGNATURE-----";
-    let gpg_start = msg_gpg.find(GPG_SIG_START);
-    let gpg_end = msg_gpg.find(GPG_SIG_END).map(|end| end + GPG_SIG_END.len());
-    let gpg_sig = match (gpg_start, gpg_end) {
-        (Some(start), Some(end)) => {
-            if start < end {
-                Some(msg_gpg[start..end].to_string())
-            } else {
-                None
-            }
-        }
-        _ => None,
-    };
-    match gpg_sig {
-        Some(gpg) => {
-            // Skip the leading '\n\n' (blank lines).
-            // Some commit messages may use '\n \n\n' or similar patterns.
-            // To handle such cases, remove all leading blank lines from the message.
-            let msg = msg_gpg[gpg_end.unwrap()..].trim_start().to_string();
-            (msg, Some(gpg))
-        }
-        None => {
-            assert!(msg_gpg.starts_with('\n'), "commit message format error");
-            let msg = msg_gpg[1..].to_string(); // skip the leading '\n' (blank line)
-            (msg, None)
-        }
-    }
-}
-
 /// Calculate the hash of a file blob
 /// - for `lfs` file: calculate hash of the pointer data
 pub fn calc_file_blob_hash(path: impl AsRef<Path>) -> io::Result<SHA1> {
@@ -174,6 +127,7 @@ pub async fn get_target_commit(branch_or_commit: &str) -> Result<SHA1, Box<dyn s
 
 #[cfg(test)]
 mod test {
+    use common::utils::{format_commit_msg, parse_commit_msg};
     use mercury::internal::object::commit::Commit;
 
     use super::*;