index: make focus index calculate-churn a command

Other organizations would conceivably be interested in this information, so surface it better.

focus/commands/src/cli/main.rs

@@ -248,6 +248,7 @@ fn feature_name_for(subcommand: &Subcommand) -> String {
         Subcommand::Upgrade { .. } => "upgrade".to_string(),
         Subcommand::Index { subcommand } => match subcommand {
             IndexSubcommand::Clear { .. } => "index-clear".to_string(),
+            IndexSubcommand::CalculateChurn { .. } => "index-calculate-churn".to_string(),
             IndexSubcommand::Fetch { .. } => "index-fetch".to_string(),
             IndexSubcommand::Get { .. } => "index-get".to_string(),
             IndexSubcommand::Generate { .. } => "index-generate".to_string(),
@@ -494,6 +495,18 @@ enum IndexSubcommand {
         sparse_repo: PathBuf,
     },
 
+    /// Calculate statistics on cache invalidation for the projects in the
+    /// repository.
+    CalculateChurn {
+        /// Path to the sparse repository.
+        #[clap(long, parse(from_os_str), default_value = ".")]
+        sparse_repo: PathBuf,
+
+        /// The number of commits backwards to examine.
+        #[clap(long, default_value = "1000")]
+        num_commits: usize,
+    },
+
     /// Fetch the pre-computed index for the repository.
     Fetch {
         /// Path to the sparse repository.
@@ -1036,6 +1049,15 @@ fn run_subcommand(app: Arc<App>, tracker: &Tracker, options: FocusOpts) -> Resul
                 Ok(ExitCode(0))
             }
 
+            IndexSubcommand::CalculateChurn {
+                sparse_repo,
+                num_commits,
+            } => {
+                let sparse_repo_path = paths::find_repo_root_from(app.clone(), sparse_repo)?;
+                focus_operations::index::print_churn_stats(app, sparse_repo_path, num_commits)?;
+                Ok(ExitCode(0))
+            }
+
             IndexSubcommand::Fetch {
                 sparse_repo,
                 force,

focus/commands/examples/calc_invalidation_rate.rs → focus/internals/src/lib/index/churn.rs

@@ -1,12 +1,6 @@
 // Copyright 2022 Twitter, Inc.
 // SPDX-License-Identifier: Apache-2.0
 
-//! Run with
-//!
-//! ```bash
-//! cargo run --example calc_invalidation_rate -- ~/workspace/path/to/repo 10000
-//! ```
-
 use std::{
     collections::HashMap,
     ops::Deref,
@@ -17,23 +11,14 @@ use std::{
     },
 };
 
-use clap::Parser;
-use focus_internals::{
+use crate::{
     index::{content_hash, ContentHash, DependencyKey, HashContext},
     model::{repo::Repo, selection::Project},
     target::{Target, TargetSet},
 };
 use focus_util::{app::App, git_helper};
 use rayon::iter::{IntoParallelRefIterator, ParallelIterator};
 
-#[derive(Parser, Debug)]
-struct Opts {
-    sparse_repo_path: PathBuf,
-
-    /// The number of commits backward from the `HEAD` commit to sample.
-    num_commits: usize,
-}
-
 struct RepoPool {
     repo_path: PathBuf,
     repos: Arc<Mutex<Vec<git2::Repository>>>,
@@ -90,13 +75,12 @@ fn average(values: impl IntoIterator<Item = f64>) -> f64 {
     sum / len
 }
 
-fn main() -> anyhow::Result<()> {
-    let Opts {
-        sparse_repo_path,
-        num_commits,
-    } = Opts::parse();
-
-    let app = Arc::new(App::new(false, None, None, None)?);
+/// Calculate and print statistics on target/project churn.
+pub fn print_churn_stats(
+    app: Arc<App>,
+    sparse_repo_path: PathBuf,
+    num_commits: usize,
+) -> anyhow::Result<()> {
     let repo = Repo::open(&sparse_repo_path, app)?;
     let selections = repo.selection_manager()?;
     let all_projects: HashMap<String, Project> = selections

focus/internals/src/lib/index/mod.rs

@@ -7,10 +7,12 @@
 
 #![warn(missing_docs)]
 
+mod churn;
 mod content_hash;
 mod dependency_graph;
 mod object_database;
 
+pub use churn::print_churn_stats;
 pub use content_hash::{content_hash, ContentHash, HashContext};
 pub use dependency_graph::{
     get_files_to_materialize, update_object_database_from_resolution, DependencyKey,

focus/operations/src/index.rs

@@ -16,7 +16,7 @@ use focus_util::paths::assert_focused_repo;
 use tracing::{debug, debug_span, info};
 
 use focus_internals::index::{
-    content_hash, get_files_to_materialize, ContentHash, DependencyKey, HashContext,
+    self, content_hash, get_files_to_materialize, ContentHash, DependencyKey, HashContext,
     ObjectDatabase, PathsToMaterializeResult, RocksDBCache, RocksDBMemoizationCacheExt,
     FUNCTION_ID,
 };
@@ -400,6 +400,14 @@ pub fn push(
     Ok(ExitCode(0))
 }
 
+pub fn print_churn_stats(
+    app: Arc<App>,
+    sparse_repo: PathBuf,
+    num_commits: usize,
+) -> anyhow::Result<()> {
+    index::print_churn_stats(app, sparse_repo, num_commits)
+}
+
 #[cfg(test)]
 mod tests {
     use focus_testing::ScratchGitRepo;