cascache.py: Allow CASCache.list_refs() to handle globs

This commit ensures that CASCache.list_refs(), and
ArtifactCache.list_artifacts(), can both handle glob expressions.

2 files changed, 24 insertions, 7 deletions

diff --git a/buildstream/_artifactcache.py b/buildstream/_artifactcache.py
index b3ca01618..bc0032bec 100644
--- a/buildstream/_artifactcache.py
+++ b/buildstream/_artifactcache.py
@@ -514,11 +514,14 @@ class ArtifactCache():
     #
     # List artifacts in this cache in LRU order.
     #
+    # Args:
+    #     glob (str): An option glob expression to be used to list artifacts satisfying the glob
+    #
     # Returns:
     #     ([str]) - A list of artifact names as generated in LRU order
     #
-    def list_artifacts(self):
-        return self.cas.list_refs()
+    def list_artifacts(self, *, glob=None):
+        return self.cas.list_refs(glob=glob)
 
     # remove():
     #
diff --git a/buildstream/_cas/cascache.py b/buildstream/_cas/cascache.py
index 560587055..792bf3eb9 100644
--- a/buildstream/_cas/cascache.py
+++ b/buildstream/_cas/cascache.py
@@ -24,6 +24,7 @@ import stat
 import errno
 import uuid
 import contextlib
+from fnmatch import fnmatch
 
 import grpc
 
@@ -472,22 +473,35 @@ class CASCache():
     #
     # List refs in Least Recently Modified (LRM) order.
     #
+    # Args:
+    #     glob (str) - An optional glob expression to be used to list refs satisfying the glob
+    #
     # Returns:
     #     (list) - A list of refs in LRM order
     #
-    def list_refs(self):
+    def list_refs(self, *, glob=None):
         # string of: /path/to/repo/refs/heads
         ref_heads = os.path.join(self.casdir, 'refs', 'heads')
+        path = ref_heads
+
+        if glob is not None:
+            globdir = os.path.dirname(glob)
+            if not any(c in "*?[" for c in globdir):
+                # path prefix contains no globbing characters so
+                # append the glob to optimise the os.walk()
+                path = os.path.join(ref_heads, globdir)
 
         refs = []
         mtimes = []
 
-        for root, _, files in os.walk(ref_heads):
+        for root, _, files in os.walk(path):
             for filename in files:
                 ref_path = os.path.join(root, filename)
-                refs.append(os.path.relpath(ref_path, ref_heads))
-                # Obtain the mtime (the time a file was last modified)
-                mtimes.append(os.path.getmtime(ref_path))
+                relative_path = os.path.relpath(ref_path, ref_heads)  # Relative to refs head
+                if not glob or fnmatch(relative_path, glob):
+                    refs.append(relative_path)
+                    # Obtain the mtime (the time a file was last modified)
+                    mtimes.append(os.path.getmtime(ref_path))
 
         # NOTE: Sorted will sort from earliest to latest, thus the
         # first ref of this list will be the file modified earliest.