main, GitCommitReader: Don't use --stat with partial clones

The --stat (or --shortstat) parameter would cause us to download lots
of data from origin for relatively little benefit. Walking the full
repository this is also fairly inefficient.

We warn when this happens so the user knows not to rely on the change
statistic, or to make a full clone if needed.

Discussed in GitHub #13.

src/gitcommitreader.rs

@@ -56,23 +56,30 @@ pub struct GitCommitReader
 
 impl GitCommitReader
 {
-    pub fn new(repo_path: std::path::PathBuf, repo_name: &str, since: DateTime<Utc>) -> Result<GitCommitReader>
+    pub fn new(repo_path: std::path::PathBuf, repo_name: &str, since: DateTime<Utc>, use_stat: bool) -> Result<GitCommitReader>
     {
         let repo_path = repo_path.canonicalize().unwrap();
-        let stdout = Command::new("git")
-            .arg("-C")
-            .arg(&repo_path)
-            .arg("log")
-            .arg("--branches")
-            .arg("--remotes")
-            .arg("--pretty=format:%H__sep__%aD__sep__%aN__sep__%aE__sep__%cD__sep__%cN__sep__%cE")
-            .arg("--reverse")
-            .arg("--since")
-            .arg(since.to_rfc2822())
-            .arg("--date-order")
-            .arg("--shortstat")
-            .arg("HEAD")
-            .stdout(Stdio::piped())
+        let mut cmd;
+
+        cmd = Command::new("git");
+        cmd.arg("-C")
+           .arg(&repo_path)
+           .arg("log")
+           .arg("--branches")
+           .arg("--remotes")
+           .arg("--pretty=format:%H__sep__%aD__sep__%aN__sep__%aE__sep__%cD__sep__%cN__sep__%cE")
+           .arg("--reverse")
+           .arg("--since")
+           .arg(since.to_rfc2822())
+           .arg("--date-order")
+           .arg("HEAD");
+
+        if use_stat
+        {
+            cmd.arg("--shortstat");
+        }
+
+        let stdout = cmd.stdout(Stdio::piped())
             .spawn().chain_err(|| "Could not spawn git")?
             .stdout.chain_err(|| "Could not read git output")?;
         let reader = BufReader::new(stdout);

src/main.rs

@@ -36,6 +36,7 @@ mod projectmeta;
 mod statuslogger;
 
 use std::path::PathBuf;
+use std::process::Command;
 use structopt::StructOpt;
 use errors::*;
 use crate::commitdb::CommitDb;
@@ -169,12 +170,33 @@ fn run_ingest(db_path: PathBuf, repo_tree_paths: Vec<PathBuf>, _meta: &ProjectMe
             .file_name().unwrap()
             .to_string_lossy()
             .into_owned();
-        let gcr = GitCommitReader::new(path.clone(),
-                                       &repo_name,
-                                       cdb.get_last_author_time(&repo_name))?;
 
         sl.begin_repo(&repo_name);
 
+        // Check for promisor for origin remote; we interpret its presence
+        // as a preference for remote storage. If found, we turn off --stat
+        // collection, since that would cause git to fetch all the remote
+        // blobs (slowly).
+        //
+        // This will break change counts. Author and commit counts will still
+        // work.
+
+        let mut cmd;
+        cmd = Command::new("git");
+        cmd.arg("-C").arg(&path).arg("config").arg("remote.origin.promisor");
+        let output = cmd.output().unwrap();
+        let has_promisor = std::str::from_utf8(&output.stdout).unwrap().trim() == "true";
+
+        if has_promisor
+        {
+            sl.log_warning("origin has a promisor; change details omitted.");
+        }
+
+        let gcr = GitCommitReader::new(path.clone(),
+                                       &repo_name,
+                                       cdb.get_last_author_time(&repo_name),
+                                       !has_promisor)?;
+
         for commit in gcr
         {
             cdb.insert_raw_commit(&commit)?;