Add filter version config field

josh-sync.example.toml

@@ -9,12 +9,12 @@ path = "library/stdarch"
 # Note that this option is mutually exclusive with `path`
 #filter = ...
 
-# Optionally, you can specify a set of commands executed after a successful pull.
-# If the executed command changes the local git state (performs some modifications to files that
-# were already tracked), then a new commit with the given message will be created.
-#[[post-pull]]
-#cmd = ["cargo", "fmt"]
-#commit-message = "reformat"
+# Optionally, you can specify the filter version, which tells josh-sync which postprocessing to
+# apply to the specified filter.
+# If not specified, it defaults to version 1.
+# Version 1 - default "legacy" behavior of Josh, which preserves empty merge commits
+# Version 2 - newer behavior that skips empty merge commits
+#filter-version = 1
 
 # Optionally, you can specify a subtree filter.
 # This will be applied to the local `HEAD` during the round-trip check.
@@ -28,3 +28,10 @@ path = "library/stdarch"
 # E.g., if the `filter` is ":/compiler/rustc_public:prefix=rustc_public",
 # the `subtree-filter` should be:
 #subtree-filter = ":/rustc_public:prefix=rustc_public"
+
+# Optionally, you can specify a set of commands executed after a successful pull.
+# If the executed command changes the local git state (performs some modifications to files that
+# were already tracked), then a new commit with the given message will be created.
+#[[post-pull]]
+#cmd = ["cargo", "fmt"]
+#commit-message = "reformat"

src/bin/rustc_josh_sync.rs

@@ -3,7 +3,7 @@ use clap::Parser;
 use rustc_josh_sync::SyncContext;
 use rustc_josh_sync::config::{JoshConfig, load_config};
 use rustc_josh_sync::josh::{JoshProxy, try_install_josh_proxy};
-use rustc_josh_sync::sync::{DEFAULT_UPSTREAM_REPO, GitSync, RustcPullError};
+use rustc_josh_sync::sync::{DEFAULT_UPSTREAM_REPO, FilterVersion, GitSync, RustcPullError};
 use rustc_josh_sync::utils::{get_current_head_sha, prompt};
 use std::path::{Path, PathBuf};
 
@@ -90,6 +90,7 @@ fn main() -> anyhow::Result<()> {
                 filter: None,
                 post_pull: vec![],
                 subtree_filter: None,
+                filter_version: FilterVersion::latest(),
             };
             config
                 .write(Path::new(DEFAULT_CONFIG_PATH))

src/config.rs

@@ -1,3 +1,4 @@
+use crate::sync::FilterVersion;
 use anyhow::Context;
 use std::path::Path;
 
@@ -20,6 +21,70 @@ pub struct JoshConfig {
     /// Optional subtree filter applied to the local `HEAD` during round-trip check.
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub subtree_filter: Option<String>,
+    /// Optional filter version that determines which post-processing will be applied to the
+    /// specified filter.
+    ///
+    /// This exists for backwards compatibility with repositories using an older filter syntax.
+    #[serde(
+        default = "default_filter_version",
+        skip_serializing_if = "skip_serializing_filter_version",
+        with = "filter_version"
+    )]
+    pub filter_version: FilterVersion,
+}
+
+impl JoshConfig {
+    pub fn full_repo_name(&self) -> String {
+        format!("{}/{}", self.org, self.repo)
+    }
+
+    pub fn write(&self, path: &Path) -> anyhow::Result<()> {
+        let config = toml::to_string_pretty(self).context("cannot serialize config")?;
+        std::fs::write(path, config).context("cannot write config")?;
+        Ok(())
+    }
+}
+
+mod filter_version {
+    use crate::sync::FilterVersion;
+    use serde::de::{Error, Unexpected};
+    use serde::{Deserialize, Deserializer, Serialize, Serializer};
+
+    pub fn serialize<S: Serializer>(
+        version: &FilterVersion,
+        serializer: S,
+    ) -> Result<S::Ok, S::Error> {
+        let num: u32 = match version {
+            FilterVersion::Version1 => 1,
+            FilterVersion::Version2 => 2,
+        };
+        num.serialize(serializer)
+    }
+
+    pub fn deserialize<'de, D: Deserializer<'de>>(
+        deserializer: D,
+    ) -> Result<FilterVersion, D::Error> {
+        let num = u32::deserialize(deserializer)?;
+        match num {
+            1 => Ok(FilterVersion::Version1),
+            2 => Ok(FilterVersion::Version2),
+            v => Err(D::Error::invalid_value(
+                Unexpected::Unsigned(v as u64),
+                &"1 or 2",
+            )),
+        }
+    }
+}
+
+fn default_filter_version() -> FilterVersion {
+    FilterVersion::Version1
+}
+
+fn skip_serializing_filter_version(version: &FilterVersion) -> bool {
+    match version {
+        FilterVersion::Version1 => true,
+        FilterVersion::Version2 => false,
+    }
 }
 
 /// Execute an operation after a pull, and if something changes in the local git state,
@@ -36,30 +101,6 @@ pub struct PostPullOperation {
     pub commit_message: String,
 }
 
-impl JoshConfig {
-    pub fn full_repo_name(&self) -> String {
-        format!("{}/{}", self.org, self.repo)
-    }
-
-    pub fn construct_josh_filter(&self) -> String {
-        let filter = match (&self.path, &self.filter) {
-            (Some(path), None) => format!(":/{path}"),
-            (None, Some(filter)) => filter.clone(),
-            _ => unreachable!("Config contains both path and a filter"),
-        };
-
-        let filter = convert_rev_syntax(&filter);
-        let filter = wrap_compat(&filter);
-        filter
-    }
-
-    pub fn write(&self, path: &Path) -> anyhow::Result<()> {
-        let config = toml::to_string_pretty(self).context("cannot serialize config")?;
-        std::fs::write(path, config).context("cannot write config")?;
-        Ok(())
-    }
-}
-
 fn default_org() -> String {
     String::from("rust-lang")
 }
@@ -78,138 +119,3 @@ pub fn load_config(path: &Path) -> anyhow::Result<JoshConfig> {
 
     Ok(config)
 }
-
-/// Converts filters from old `:rev(sha:filter)` syntax to new
-/// `:rev(<=sha:filter)` syntax. Null SHAs (40 zeros) become `_`.
-/// Only touches SHAs inside `:rev(...)` blocks.
-fn convert_rev_syntax(input: &str) -> String {
-    let rev_block = regex::Regex::new(r":rev\([^)]*\)").unwrap();
-    let entry = regex::Regex::new(
-        r"(?x)
-        ([,(])                # delimiter before entry
-        (0{40}|[0-9a-f]{40})  # full SHA
-        :                     # colon separator
-    ",
-    )
-    .unwrap();
-
-    rev_block
-        .replace_all(input, |block: &regex::Captures| {
-            entry
-                .replace_all(&block[0], |caps: &regex::Captures| {
-                    let delim = &caps[1];
-                    let sha = &caps[2];
-                    if sha.chars().all(|c| c == '0') {
-                        format!("{delim}_:")
-                    } else {
-                        format!("{delim}<={sha}:")
-                    }
-                })
-                .into_owned()
-        })
-        .into_owned()
-}
-
-/// Wraps a filter with the backwards compatibility meta options for
-/// trivial merge preservation and CRLF normalization in gpgsig headers.
-///
-/// `:your/filter` becomes
-/// `:~(history="keep-trivial-merges",gpgsig="norm-lf")[:your/filter]`
-fn wrap_compat(filter: &str) -> String {
-    format!(":~(history=\"keep-trivial-merges\",gpgsig=\"norm-lf\")[{filter}]")
-}
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn no_rev_block_unchanged() {
-        assert_eq!(convert_rev_syntax(":/some/path"), ":/some/path");
-    }
-
-    #[test]
-    fn single_sha_gets_prefix() {
-        assert_eq!(
-            convert_rev_syntax(":rev(3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/some/path)"),
-            ":rev(<=3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/some/path)",
-        );
-    }
-
-    #[test]
-    fn null_sha_becomes_underscore() {
-        assert_eq!(
-            convert_rev_syntax(":rev(0000000000000000000000000000000000000000:/some/path)"),
-            ":rev(_:/some/path)",
-        );
-    }
-
-    #[test]
-    fn multiple_entries_in_rev_block() {
-        assert_eq!(
-            convert_rev_syntax(
-                ":rev(3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/p1,\
-                 e4c7a2d8f1b3e5a9d6c0f2b4a7e1d3c5f8a0b6e9:/p2,\
-                 0000000000000000000000000000000000000000:/p3)"
-            ),
-            ":rev(<=3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/p1,\
-             <=e4c7a2d8f1b3e5a9d6c0f2b4a7e1d3c5f8a0b6e9:/p2,\
-             _:/p3)",
-        );
-    }
-
-    #[test]
-    fn already_converted_syntax_unchanged() {
-        assert_eq!(
-            convert_rev_syntax(":rev(<=3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/some/path)"),
-            ":rev(<=3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/some/path)",
-        );
-    }
-
-    #[test]
-    fn underscore_syntax_unchanged() {
-        assert_eq!(
-            convert_rev_syntax(":rev(_:/some/path)"),
-            ":rev(_:/some/path)",
-        );
-    }
-
-    #[test]
-    fn sha_outside_rev_block_unchanged() {
-        assert_eq!(
-            convert_rev_syntax("3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/some/path"),
-            "3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/some/path",
-        );
-    }
-
-    #[test]
-    fn wrap_compat_simple_filter() {
-        assert_eq!(
-            wrap_compat(":/some/path"),
-            ":~(history=\"keep-trivial-merges\",gpgsig=\"norm-lf\")[:/some/path]",
-        );
-    }
-
-    #[test]
-    fn wrap_compat_rev_filter() {
-        assert_eq!(
-            wrap_compat(
-                ":rev(75dd959a3a40eb5b4574f8d2e23aa6efbeb33573:prefix=src/tools/miri):/src/tools/miri"
-            ),
-            ":~(history=\"keep-trivial-merges\",gpgsig=\"norm-lf\")\
-             [:rev(75dd959a3a40eb5b4574f8d2e23aa6efbeb33573:prefix=src/tools/miri):/src/tools/miri]",
-        );
-    }
-
-    #[test]
-    fn multiple_rev_blocks() {
-        assert_eq!(
-            convert_rev_syntax(
-                ":rev(3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/p1)\
-                 :rev(e4c7a2d8f1b3e5a9d6c0f2b4a7e1d3c5f8a0b6e9:/p2)"
-            ),
-            ":rev(<=3a1f5e2b9c8d4e7f6a0b1c2d3e4f5a6b7c8d9e0f:/p1)\
-             :rev(<=e4c7a2d8f1b3e5a9d6c0f2b4a7e1d3c5f8a0b6e9:/p2)",
-        );
-    }
-}

src/josh.rs

@@ -8,7 +8,7 @@ use std::time::Duration;
 
 const JOSH_PORT: u16 = 42042;
 /// Version of `josh-proxy` that should be downloaded for the user.
-const JOSH_VERSION: &str = "r26.06.11";
+const JOSH_VERSION: &str = "r26.07.19";
 
 pub struct JoshProxy {
     path: PathBuf,