sparse: make use of matchers for faster path listing

Summary:
The current code iterates over all files in the manifest, filtering against a prefix.

But a manifest supports using a matcher directly, and efficient implementations like the treemanifest will prune the tree to a much smaller subset rapidly based on the path in a matcher. Switching to using a matcher dramatically improves --cwd-list performance in fbsource, when treemanifests are available.

Reviewed By: quark-zju

Differential Revision: D7056650

fbshipit-source-id: 2bf62ea93680323a49c9282266118805881d7b02

hgext/fbsparse.py

@@ -1214,18 +1214,22 @@ def _cwdlist(repo):
             _("the current working directory should begin "
               "with the root %s") % root)
 
+    matcher = matchmod.match(
+        repo.root, repo.getcwd(),
+        patterns=['path:' + cwd])
+    files = mf.matches(matcher)
+
     sparsematch = repo.sparsematch(ctx.rev())
     checkedoutentries = set()
     allentries = set()
     cwdlength = len(cwd)
 
-    for filepath in mf:
-        if filepath.startswith(cwd):
-            entryname = filepath[cwdlength:].partition(pycompat.ossep)[0]
+    for filepath in files:
+        entryname = filepath[cwdlength:].partition(pycompat.ossep)[0]
 
-            allentries.add(entryname)
-            if sparsematch(filepath):
-                checkedoutentries.add(entryname)
+        allentries.add(entryname)
+        if sparsematch(filepath):
+            checkedoutentries.add(entryname)
 
     ui = repo.ui
     for entry in sorted(allentries):

tests/test-fb-hgext-treemanifest-sparse.t

@@ -0,0 +1,52 @@
+test interaction between sparse and treemanifest (sparse file listing)
+
+  $ cat >> $HGRCPATH <<EOF
+  > [extensions]
+  > sparse=$TESTDIR/../hgext/fbsparse.py
+  > treemanifest=
+  > [treemanifest]
+  > treeonly = True
+  > [remotefilelog]
+  > reponame = master
+  > cachepath = $PWD/hgcache
+  > EOF
+
+Setup the repository
+
+  $ hg init myrepo
+  $ cd myrepo
+  $ touch show
+  $ touch hide
+  $ mkdir -p subdir/foo/spam subdir/bar/ham hiddensub/foo hiddensub/bar
+  $ touch subdir/foo/spam/show
+  $ touch subdir/bar/ham/hide
+  $ touch hiddensub/foo/spam
+  $ touch hiddensub/bar/ham
+  $ hg add .
+  adding hiddensub/bar/ham
+  adding hiddensub/foo/spam
+  adding hide
+  adding show
+  adding subdir/bar/ham/hide
+  adding subdir/foo/spam/show
+  $ hg commit -m 'Init'
+  $ hg sparse --include show
+  $ hg sparse --exclude hide
+  $ hg sparse --include subdir
+  $ hg sparse --exclude subdir/foo
+
+Test --cwd-list
+
+  $ hg sparse --cwd-list
+  - hiddensub
+  - hide
+    show
+    subdir
+  $ cd subdir
+  $ hg sparse --cwd-list
+    bar
+  - foo
+  $ hg sparse --include foo
+  $ hg sparse --cwd-list
+    bar
+  - foo