git: respect selective pull

Summary:
Some git repos (ex. pytorch) has thousands of branches that won't scale well.
Change no-argument pull to respect selective pull to mitigate it.

Reviewed By: DurhamG

Differential Revision: D33485186

fbshipit-source-id: b735704850847c49b0bfeb94972e4bb778e726a4

eden/scm/edenscm/mercurial/bookmarks.py

@@ -1055,6 +1055,14 @@ class remotenames(dict):
                     self._node2hoists.setdefault(node[0], []).append(name)
         return self._node2hoists
 
+    def get(self, name):
+        """Resolve a remote bookmark. Return None if the bookmark does not exist"""
+        nodes = self["bookmarks"].get(name)
+        if nodes:
+            return nodes[0]
+        else:
+            return None
+
     @property
     def changecount(self):
         return self._changecount

eden/scm/edenscm/mercurial/commands/__init__.py

@@ -4584,6 +4584,8 @@ def pull(ui, repo, source="default", **opts):
         if source == "default":
             source = "origin"
         refspecs = (opts.get("bookmark") or []) + (opts.get("rev") or [])
+        if not refspecs:
+            refspecs = git.defaultpullrefspecs(repo, source)
         ret = git.pull(repo, source, refspecs)
         if ret == 0 and opts.get("update"):
             # Figure out the node to checkout

eden/scm/edenscm/mercurial/git.py

@@ -15,9 +15,9 @@ import subprocess
 import bindings
 from edenscm import tracing
 
-from . import error, util
+from . import bookmarks as bookmod, error, util
 from .i18n import _
-from .node import hex, nullid
+from .node import bin, hex, nullid
 
 GIT_DIR_FILE = "gitdir"
 GIT_REQUIREMENT = "git"
@@ -68,7 +68,8 @@ def clone(ui, url, destpath=None, update=True):
             raise error.Abort(_("git clone was not successful"))
         initgit(repo, "git")
         if url:
-            pull(repo, "origin", [])
+            refspecs = defaultpullrefspecs(repo, "origin")
+            pull(repo, "origin", refspecs)
     except Exception:
         repo = None
         shutil.rmtree(destpath, ignore_errors=True)
@@ -186,6 +187,9 @@ def revparse(repo, revspec):
 
 def pull(repo, source, refspecs):
     """Run `git fetch` on the backing repo to perform a pull"""
+    if not refspecs:
+        # Nothing to pull
+        return 0
     ret = rungit(
         repo,
         ["fetch", "--no-write-fetch-head", "--no-tags", "--prune", source] + refspecs,
@@ -212,6 +216,41 @@ def push(repo, dest, pushnode, to, force=False):
     return ret
 
 
+def listremote(repo, source, patterns):
+    """List references of the remote peer
+    Return a dict of name to node.
+    """
+    out = callgit(repo, ["ls-remote", "--refs", source] + patterns)
+    refs = {}
+    for line in out.splitlines():
+        if b"\t" not in line:
+            continue
+        hexnode, name = line.split(b"\t", 1)
+        refs[name.decode("utf-8")] = bin(hexnode)
+    return refs
+
+
+def defaultpullrefspecs(repo, source):
+    """default refspecs for 'git fetch' respecting selective pull"""
+    names = bookmod.selectivepullbookmarknames(repo)
+
+    patterns = ["refs/heads/%s" % name for name in names]
+    listed = listremote(repo, source, patterns)
+
+    refspecs = []
+    for name in names:
+        refspec = "refs/heads/%s" % name
+        remotenode = listed.get(refspec)
+        if repo._remotenames.get("%s/%s" % (source, name)) == remotenode:
+            # not changed, skip
+            continue
+        if remotenode is None:
+            # removed remotely
+            refspec = ":refs/remotes/%s/heads/%s" % (source, name)
+        refspecs.append(refspec)
+    return refspecs
+
+
 def callgit(repo, args):
     """Run git command in the backing git repo, return its output"""
     gitdir = readgitdir(repo)

eden/scm/lib/configparser/src/builtin.rs

@@ -22,6 +22,7 @@ autopullpattern=
 disallowedto=^origin/
 disallowhint=please don't specify 'origin/' prefix in remote bookmark's name
 hoist=origin
+selectivepulldefault=main,master
 
 [smartlog]
 names=main,master

eden/scm/tests/test-git.t

@@ -201,11 +201,10 @@ Test clone with flags (--noupdate, --updaterev):
   $ hg log -r . -T '{node|short}\n'
   000000000000
   $ hg bookmarks --remote
-     origin/foo                5c9a5ee451a8
      origin/master             3f5848713286
   $ cd ..
 
-  $ hg clone "git+file://$TESTTMP/gitrepo" cloned1
+  $ hg clone "git+file://$TESTTMP/gitrepo" cloned1 --config remotenames.selectivepulldefault=foo,master
   From file:/*/$TESTTMP/gitrepo (glob)
    * [new branch]      foo        -> origin/foo
    * [new branch]      master     -> origin/master
@@ -214,10 +213,9 @@ Test clone with flags (--noupdate, --updaterev):
   5c9a5ee451a8 origin/foo alpha3
   $ cd ..
 
-  $ hg clone --updaterev origin/foo "git+file://$TESTTMP/gitrepo" cloned2
+  $ hg clone --updaterev origin/foo "git+file://$TESTTMP/gitrepo" cloned2 --config remotenames.selectivepulldefault=foo
   From file:/*/$TESTTMP/gitrepo (glob)
    * [new branch]      foo        -> origin/foo
-   * [new branch]      master     -> origin/master
   2 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg --cwd cloned2 log -r . -T '{node|short} {remotenames} {desc}\n'
   5c9a5ee451a8 origin/foo alpha3