Implement squash filter for list of commits (#948)

Author: christian-schilling

josh-proxy/src/lib.rs

@@ -393,6 +393,7 @@ fn split_changes(
                         &repo.find_commit(changes[i].1)?,
                         &vec![&parent],
                         &new_tree,
+                        None,
                     )?;
                     changes[i].1 = new_commit;
                     new_bases.push(new_commit);

src/bin/josh-filter.rs

@@ -37,9 +37,24 @@ fn make_app() -> clap::Command {
         )
         .arg(
             clap::Arg::new("squash")
+                .help("Produce a history that contains only commits pointed to by references matching the given pattern")
+                .long("squash")
+        )
+        .arg(
+            clap::Arg::new("author")
+                .help("Author to use for commits with rewritten message")
+                .long("author")
+        )
+        .arg(
+            clap::Arg::new("email")
+                .help("Author email to use for commits with rewritten message")
+                .long("email")
+        )
+        .arg(
+            clap::Arg::new("single")
                 .action(clap::ArgAction::SetTrue)
-                .help("Only output one commit, without history")
-                .long("squash"),
+                .help("Produce a history that contains only one single commit")
+                .long("single"),
         )
         .arg(
             clap::Arg::new("discover")
@@ -138,10 +153,6 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
 
     let mut filterobj = josh::filter::parse(&specstr)?;
 
-    if args.get_flag("squash") {
-        filterobj = josh::filter::chain(josh::filter::parse(":SQUASH")?, filterobj);
-    }
-
     if args.get_flag("print-filter") {
         let filterobj = if args.get_flag("reverse") {
             josh::filter::invert(filterobj)?
@@ -162,6 +173,38 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
     let transaction = josh::cache::Transaction::new(repo, None);
     let repo = transaction.repo();
 
+    let input_ref = args.get_one::<String>("input").unwrap();
+
+    let mut refs = vec![];
+    let mut ids = vec![];
+
+    let reference = repo.resolve_reference_from_short_name(input_ref).unwrap();
+    let input_ref = reference.name().unwrap().to_string();
+    refs.push((input_ref.clone(), reference.target().unwrap()));
+
+    if args.get_flag("single") {
+        filterobj = josh::filter::chain(josh::filter::squash(None), filterobj);
+    }
+
+    if let Some(pattern) = args.get_one::<String>("squash") {
+        let pattern = pattern.to_string();
+        for reference in repo.references_glob(&pattern).unwrap() {
+            let reference = reference?;
+            if let Some(target) = reference.target() {
+                ids.push((target, reference.name().unwrap().to_string()));
+                refs.push((reference.name().unwrap().to_string(), target));
+            }
+        }
+        filterobj = josh::filter::chain(
+            josh::filter::squash(Some((
+                args.get_one::<String>("author").unwrap(),
+                args.get_one::<String>("email").unwrap(),
+                &ids,
+            ))),
+            filterobj,
+        );
+    };
+
     let odb = repo.odb()?;
     let mp = if args.get_flag("pack") {
         let mempack = odb.add_new_mempack_backend(1000)?;
@@ -188,10 +231,8 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
         }
     });
 
-    let input_ref = args.get_one::<String>("input").unwrap();
-
     if args.get_flag("discover") {
-        let r = repo.revparse_single(input_ref)?;
+        let r = repo.revparse_single(&input_ref)?;
         let hs = josh::housekeeping::find_all_workspaces_and_subdirectories(&r.peel_to_tree()?)?;
         for i in hs {
             if i.contains(":workspace=") {
@@ -210,23 +251,10 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
 
     let update_target = args.get_one::<String>("update").unwrap();
 
-    let src = input_ref;
     let target = update_target;
 
     let reverse = args.get_flag("reverse");
 
-    let t = if reverse {
-        "refs/JOSH_TMP".to_owned()
-    } else {
-        target.to_string()
-    };
-    let src_r = repo
-        .revparse_ext(src)?
-        .1
-        .ok_or(josh::josh_error("reference not found"))?;
-
-    let src = src_r.name().unwrap().to_string();
-
     let check_permissions = args.get_flag("check-permission");
     let mut permissions_filter = josh::filter::empty();
     if check_permissions {
@@ -264,28 +292,31 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
         permissions_filter = josh::filter::empty();
     }
 
-    let old_oid = if let Ok(id) = transaction.repo().refname_to_id(&t) {
+    let old_oid = if let Ok(id) = transaction.repo().refname_to_id(&target) {
         id
     } else {
         git2::Oid::zero()
     };
-    let mut updated_refs = josh::filter_refs(
-        &transaction,
-        filterobj,
-        &[(src.clone(), src_r.target().unwrap())],
-        permissions_filter,
-    )?;
-    updated_refs[0].0 = t;
-    josh::update_refs(&transaction, &mut updated_refs, "");
-    if args.get_one::<String>("update").map(|v| v.as_str()) != Some("FILTERED_HEAD")
-        && updated_refs.len() == 1
-        && updated_refs[0].1 == old_oid
-    {
-        println!(
-            "Warning: reference {} wasn't updated",
-            args.get_one::<String>("update").unwrap()
-        );
+
+    let mut updated_refs = josh::filter_refs(&transaction, filterobj, &refs, permissions_filter)?;
+    for i in 0..updated_refs.len() {
+        if updated_refs[i].0 == input_ref {
+            if reverse {
+                updated_refs[i].0 = "refs/JOSH_TMP".to_string();
+            } else {
+                updated_refs[i].0 = target.to_string();
+            }
+        } else {
+            updated_refs[i].0 =
+                updated_refs[i]
+                    .0
+                    .replacen("refs/heads/", "refs/heads/filtered/", 1);
+            updated_refs[i].0 = updated_refs[i]
+                .0
+                .replacen("refs/tags/", "refs/tags/filtered/", 1);
+        }
     }
+    josh::update_refs(&transaction, &mut updated_refs, "");
 
     #[cfg(feature = "search")]
     if let Some(searchstring) = args.get_one::<String>("search") {
@@ -324,7 +355,7 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
     if reverse {
         let new = repo.revparse_single(target).unwrap().id();
         let old = repo.revparse_single("JOSH_TMP").unwrap().id();
-        let unfiltered_old = repo.revparse_single(input_ref).unwrap().id();
+        let unfiltered_old = repo.revparse_single(&input_ref).unwrap().id();
 
         match josh::history::unapply_filter(
             &transaction,
@@ -337,7 +368,7 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
             &mut None,
         ) {
             Ok(rewritten) => {
-                repo.reference(&src, rewritten, true, "unapply_filter")?;
+                repo.reference(&input_ref, rewritten, true, "unapply_filter")?;
             }
             Err(JoshError(msg)) => {
                 println!("{}", msg);
@@ -346,6 +377,17 @@ fn run_filter(args: Vec<String>) -> josh::JoshResult<i32> {
         }
     }
 
+    if !reverse
+        && args.get_one::<String>("update") != Some(&"FILTERED_HEAD".to_string())
+        && updated_refs.len() == 1
+        && updated_refs[0].1 == old_oid
+    {
+        println!(
+            "Warning: reference {} wasn't updated",
+            args.get_one::<String>("update").unwrap()
+        );
+    }
+
     if let Some(gql_query) = args.get_one::<String>("graphql") {
         let context = josh::graphql::context(transaction.try_clone()?, transaction.try_clone()?);
         *context.allow_refs.lock()? = true;

src/cache.rs

@@ -33,10 +33,10 @@ pub fn print_stats() {
         let name = String::from_utf8(name.to_vec()).unwrap();
         let t = db.open_tree(&name).unwrap();
         if !t.is_empty() {
-            let name = if name.contains("SUBTRACT") || name.starts_with('_') {
-                name.clone()
+            let name = if let Ok(filter) = filter::parse(&name) {
+                filter::pretty(filter, 4)
             } else {
-                filter::pretty(filter::parse(&name).unwrap(), 4)
+                name.clone()
             };
             v.push((t.len(), name));
         }

src/filter/mod.rs

@@ -61,6 +61,18 @@ pub fn empty() -> Filter {
     to_filter(Op::Empty)
 }
 
+pub fn squash(ids: Option<(&str, &str, &[(git2::Oid, String)])>) -> Filter {
+    if let Some((author, email, ids)) = ids {
+        to_filter(Op::Squash(Some(
+            ids.iter()
+                .map(|(x, y)| (*x, (y.clone(), author.to_string(), email.to_string())))
+                .collect(),
+        )))
+    } else {
+        to_filter(Op::Squash(None))
+    }
+}
+
 fn to_filter(op: Op) -> Filter {
     let s = format!("{:?}", op);
     let f = Filter(
@@ -85,7 +97,7 @@ enum Op {
     Empty,
     Fold,
     Paths,
-    Squash,
+    Squash(Option<std::collections::HashMap<git2::Oid, (String, String, String)>>),
     Linear,
 
     RegexReplace(regex::Regex, String),
@@ -236,7 +248,18 @@ fn spec2(op: &Op) -> String {
         #[cfg(feature = "search")]
         Op::Index => ":INDEX".to_string(),
         Op::Fold => ":FOLD".to_string(),
-        Op::Squash => ":SQUASH".to_string(),
+        Op::Squash(None) => ":SQUASH".to_string(),
+        Op::Squash(Some(hs)) => {
+            let mut v = hs
+                .iter()
+                .map(|(x, y)| format!("{}:{}:{}:{}", x, y.0, y.1, y.2))
+                .collect::<Vec<String>>();
+            v.sort();
+            let s = v.join(",");
+            let s = git2::Oid::hash_object(git2::ObjectType::Blob, s.as_bytes())
+                .expect("hash_object filter");
+            format!(":SQUASH={}", s)
+        }
         Op::Linear => ":linear".to_string(),
         Op::Subdir(path) => format!(":/{}", parse::quote(&path.to_string_lossy())),
         Op::File(path) => format!("::{}", parse::quote(&path.to_string_lossy())),
@@ -341,8 +364,15 @@ fn apply_to_commit2(
                 Ok(Some(git2::Oid::zero()))
             };
         }
-        Op::Squash => {
-            return Some(history::rewrite_commit(repo, commit, &[], &commit.tree()?)).transpose()
+        Op::Squash(None) => {
+            return Some(history::rewrite_commit(
+                repo,
+                commit,
+                &[],
+                &commit.tree()?,
+                None,
+            ))
+            .transpose()
         }
         _ => {
             if let Some(oid) = transaction.get(filter, commit.id()) {
@@ -354,6 +384,27 @@ fn apply_to_commit2(
     rs_tracing::trace_scoped!("apply_to_commit", "spec": spec(filter), "commit": commit.id().to_string());
 
     let filtered_tree = match &to_op(filter) {
+        Op::Squash(Some(ids)) => {
+            if let Some(_) = ids.get(&commit.id()) {
+                commit.tree()?
+            } else {
+                for parent in commit.parents() {
+                    return Ok(
+                        if let Some(fparent) = transaction.get(filter, parent.id()) {
+                            Some(history::drop_commit(
+                                commit,
+                                vec![fparent],
+                                transaction,
+                                filter,
+                            )?)
+                        } else {
+                            None
+                        },
+                    );
+                }
+                tree::empty(repo)
+            }
+        }
         Op::Linear => {
             let p: Vec<_> = commit.parent_ids().collect();
             if p.is_empty() {
@@ -370,6 +421,7 @@ fn apply_to_commit2(
                 commit.tree()?,
                 transaction,
                 filter,
+                None,
             ))
             .transpose();
         }
@@ -452,6 +504,7 @@ fn apply_to_commit2(
                 filtered_tree,
                 transaction,
                 filter,
+                None,
             ))
             .transpose();
         }
@@ -528,12 +581,18 @@ fn apply_to_commit2(
 
     let filtered_parent_ids = some_or!(filtered_parent_ids, { return Ok(None) });
 
+    let message = match to_op(filter) {
+        Op::Squash(Some(ids)) => ids.get(&commit.id()).map(|x| x.clone()),
+        _ => None,
+    };
+
     Some(history::create_filtered_commit(
         commit,
         filtered_parent_ids,
         filtered_tree,
         transaction,
         filter,
+        message,
     ))
     .transpose()
 }
@@ -557,7 +616,8 @@ fn apply2<'a>(
         Op::Nop => Ok(tree),
         Op::Empty => return Ok(tree::empty(repo)),
         Op::Fold => Ok(tree),
-        Op::Squash => Ok(tree),
+        Op::Squash(None) => Ok(tree),
+        Op::Squash(Some(_)) => Err(josh_error("not applicable to tree")),
         Op::Linear => Ok(tree),
 
         Op::RegexReplace(regex, replacement) => {

src/filter/parse.rs

@@ -33,7 +33,8 @@ fn make_op(args: &[&str]) -> JoshResult<Op> {
             Where `path` is path to the directory where workspace.josh file is located
             "#
         ))),
-        ["SQUASH"] => Ok(Op::Squash),
+        ["SQUASH"] => Ok(Op::Squash(None)),
+        ["SQUASH", _ids @ ..] => Err(josh_error("SQUASH with ids can't be parsed")),
         ["linear"] => Ok(Op::Linear),
         ["PATHS"] => Ok(Op::Paths),
         #[cfg(feature = "search")]