outerbounds 0.3.176rc4__py3-none-any.whl → 0.3.176rc6__py3-none-any.whl

outerbounds/apps/app_cli.py

@@ -594,7 +594,7 @@ def list(ctx, project, branch, name, tags, format, auth_type):
     "--tag",
     "tags",
     multiple=True,
-    type=KVPairType,
+    type=KVDictType,
     help="Filter apps to delete by tag. Format KEY=VALUE. Example --tag foo=bar --tag x=y. If multiple tags are provided, the app must match all of them.",
 )
 @click.pass_context

outerbounds/apps/code_package/code_packager.py

@@ -7,6 +7,7 @@ from io import BytesIO
 from typing import List, Tuple, Dict, Any, Optional, Callable, Union
 
 from metaflow.datastore.content_addressed_store import ContentAddressedStore
+from metaflow.util import to_unicode
 from metaflow.metaflow_config import (
     DATASTORE_SYSROOT_S3,
     DATASTORE_SYSROOT_AZURE,
@@ -18,6 +19,58 @@ from metaflow.metaflow_config import (
 CODE_PACKAGE_PREFIX = "apps-code-packages"
 
 
+# this is os.walk(follow_symlinks=True) with cycle detection
+def walk_without_cycles(top_root):
+    seen = set()
+
+    def _recurse(root):
+        for parent, dirs, files in os.walk(root):
+            for d in dirs:
+                path = os.path.join(parent, d)
+                if os.path.islink(path):
+                    # Breaking loops: never follow the same symlink twice
+                    #
+                    # NOTE: this also means that links to sibling links are
+                    # not followed. In this case:
+                    #
+                    #   x -> y
+                    #   y -> oo
+                    #   oo/real_file
+                    #
+                    # real_file is only included twice, not three times
+                    reallink = os.path.realpath(path)
+                    if reallink not in seen:
+                        seen.add(reallink)
+                        for x in _recurse(path):
+                            yield x
+            yield parent, files
+
+    for x in _recurse(top_root):
+        yield x
+
+
+def symlink_friendly_walk(root, exclude_hidden=True, suffixes=None):
+    if suffixes is None:
+        suffixes = []
+    root = to_unicode(root)  # handle files/folder with non ascii chars
+    prefixlen = len("%s/" % os.path.dirname(root))
+    for (
+        path,
+        files,
+    ) in walk_without_cycles(root):
+        if exclude_hidden and "/." in path:
+            continue
+        # path = path[2:] # strip the ./ prefix
+        # if path and (path[0] == '.' or './' in path):
+        #    continue
+        for fname in files:
+            if (fname[0] == "." and fname in suffixes) or (
+                fname[0] != "." and any(fname.endswith(suffix) for suffix in suffixes)
+            ):
+                p = os.path.join(path, fname)
+                yield p, p[prefixlen:]
+
+
 class CodePackager:
     """
     A datastore-agnostic class for packaging code.
@@ -384,11 +437,13 @@ class CodePackager:
 
     @staticmethod
     def directory_walker(
-        root, exclude_hidden=True, suffixes=None, follow_symlinks=True
+        root,
+        exclude_hidden=True,
+        suffixes=None,
     ) -> List[Tuple[str, str]]:
         """
         Walk a directory and yield tuples of (file_path, relative_arcname) for files
-        that match the given suffix filters.
+        that match the given suffix filters. It will follow symlinks, but not cycles.
 
         This function is similar to MetaflowPackage._walk and handles symlinks safely.
 
@@ -400,8 +455,6 @@ class CodePackager:
             Whether to exclude hidden files and directories (those starting with '.')
         suffixes : List[str], optional
             List of file suffixes to include (e.g. ['.py', '.txt'])
-        follow_symlinks : bool, default True
-            Whether to follow symlinks (with cycle detection)
 
         Returns
         -------
@@ -410,61 +463,12 @@ class CodePackager:
             - file_path is the full path to the file
             - relative_arcname is the path to use within the archive
         """
-        if suffixes is None:
-            suffixes = []
-
-        # Convert root to unicode to handle files/folders with non-ascii chars
-        root = str(root)
-
-        # Calculate the prefix length to strip from paths
-        prefixlen = len(os.path.dirname(root)) + 1  # +1 for the trailing slash
-
-        # Use a set to track visited symlinks to avoid cycles
-        seen = set()
-
-        def _walk_without_cycles(walk_root):
-            for parent, dirs, files in os.walk(walk_root, followlinks=follow_symlinks):
-                # If not following symlinks, we're done
-                if not follow_symlinks:
-                    yield parent, files
-                    continue
-
-                # When following symlinks, we need to check for cycles
-                for d_idx in range(
-                    len(dirs) - 1, -1, -1
-                ):  # Iterate backwards to safely remove
-                    d = dirs[d_idx]
-                    path = os.path.join(parent, d)
-                    if os.path.islink(path):
-                        # Break cycles by never following the same symlink twice
-                        reallink = os.path.realpath(path)
-                        if reallink in seen:
-                            # Remove from dirs to avoid following it
-                            dirs.pop(d_idx)
-                        else:
-                            seen.add(reallink)
-
-                yield parent, files
-
-        # Build the list of path tuples
-        result = []
-        for path, files in _walk_without_cycles(root):
-            # Skip hidden directories if requested
-            if exclude_hidden and "/." in path:
-                continue
-
-            for fname in files:
-                # Skip hidden files if requested, unless they have a specified suffix
-                if (
-                    (fname[0] == "." and fname in suffixes)
-                    or (fname[0] != "." or not exclude_hidden)
-                    and any(fname.endswith(suffix) for suffix in suffixes)
-                ):
-                    file_path = os.path.join(path, fname)
-                    rel_path = file_path[prefixlen:]
-                    result.append((file_path, rel_path))
-
-        return result
+        files = []
+        for file_path, rel_path in symlink_friendly_walk(
+            root, exclude_hidden, suffixes
+        ):
+            files.append((file_path, rel_path))
+        return files
 
     @staticmethod
     def default_package_create(
@@ -515,7 +519,6 @@ class CodePackager:
                         path,
                         exclude_hidden=True,
                         suffixes=suffixes,
-                        follow_symlinks=True,
                     ):
                         tar.add(
                             file_path,
@@ -546,13 +549,12 @@ class CodePackager:
         suffixes: Optional[List[str]] = None,
         exclude_hidden: bool = True,
         metadata: Optional[Dict[str, Any]] = None,
-        follow_symlinks: bool = True,
     ) -> bytes:
         """
         Package a directory and all of its contents that match the given suffixes.
 
         This is a convenience method that works similarly to MetaflowPackage._walk
-        to package a directory for deployment.
+        to package a directory for deployment. Will default follow_symlinks.
 
         Parameters
         ----------
@@ -564,9 +566,6 @@ class CodePackager:
             Whether to exclude hidden files and directories
         metadata : Dict[str, Any], optional
             Metadata to include in the package
-        follow_symlinks : bool, default True
-            Whether to follow symlinks when walking the directory
-
         Returns
         -------
         bytes
@@ -602,7 +601,6 @@ class CodePackager:
                 directory_path,
                 exclude_hidden=exclude_hidden,
                 suffixes=suffixes,
-                follow_symlinks=follow_symlinks,
             ):
                 # Remove debug print statement
                 tar.add(file_path, arcname=rel_path, recursive=False, filter=no_mtime)

{outerbounds-0.3.176rc4.dist-info → outerbounds-0.3.176rc6.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: outerbounds
-Version: 0.3.176rc4
+Version: 0.3.176rc6
 Summary: More Data Science, Less Administration
 License: Proprietary
 Keywords: data science,machine learning,MLOps
@@ -29,8 +29,8 @@ Requires-Dist: google-cloud-secret-manager (>=2.20.0,<3.0.0) ; extra == "gcp"
 Requires-Dist: google-cloud-storage (>=2.14.0,<3.0.0) ; extra == "gcp"
 Requires-Dist: metaflow-checkpoint (==0.2.1)
 Requires-Dist: ob-metaflow (==2.15.14.1)
-Requires-Dist: ob-metaflow-extensions (==1.1.163rc6)
-Requires-Dist: ob-metaflow-stubs (==6.0.3.176rc4)
+Requires-Dist: ob-metaflow-extensions (==1.1.163rc7)
+Requires-Dist: ob-metaflow-stubs (==6.0.3.176rc6)
 Requires-Dist: opentelemetry-distro (>=0.41b0) ; extra == "otel"
 Requires-Dist: opentelemetry-exporter-otlp-proto-http (>=1.20.0) ; extra == "otel"
 Requires-Dist: opentelemetry-instrumentation-requests (>=0.41b0) ; extra == "otel"

{outerbounds-0.3.176rc4.dist-info → outerbounds-0.3.176rc6.dist-info}/RECORD

@@ -40,13 +40,13 @@ outerbounds/_vendor/yaml/scanner.py,sha256=ZcI8IngR56PaQ0m27WU2vxCqmDCuRjz-hr7pi
 outerbounds/_vendor/yaml/serializer.py,sha256=8wFZRy9SsQSktF_f9OOroroqsh4qVUe53ry07P9UgCc,4368
 outerbounds/_vendor/yaml/tokens.py,sha256=JBSu38wihGr4l73JwbfMA7Ks1-X84g8-NskTz7KwPmA,2578
 outerbounds/apps/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
-outerbounds/apps/app_cli.py,sha256=zbzFOYpRkgFVtBhcZGl0sn8BkL6w0uc9PNMx25CECfA,23456
+outerbounds/apps/app_cli.py,sha256=hIyDEWDAwx5kVgSgjm0_p5GT-3LdKi51q45jwwktN_A,23456
 outerbounds/apps/app_config.py,sha256=KBmW9grhiuG9XZG-R0GZkM-024cjj6ztGzOX_2wZW34,11291
 outerbounds/apps/artifacts.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
 outerbounds/apps/capsule.py,sha256=C0xxsGNmsBiwTdybS38jYfLn8VzMp0mQLI6pssha1UY,16965
 outerbounds/apps/cli_to_config.py,sha256=hV6rfPgCiAX03O363GkvdjSIJBt3-oSbL6F2sTUucFE,3195
 outerbounds/apps/code_package/__init__.py,sha256=8McF7pgx8ghvjRnazp2Qktlxi9yYwNiwESSQrk-2oW8,68
-outerbounds/apps/code_package/code_packager.py,sha256=SQDBXKwizzpag5GpwoZpvvkyPOodRSQwk2ecAAfO0HI,23316
+outerbounds/apps/code_package/code_packager.py,sha256=RWvM5BKjgLhu7icsO_n5SSYC57dwyST0dWpoWF88ovU,22881
 outerbounds/apps/code_package/examples.py,sha256=aF8qKIJxCVv_ugcShQjqUsXKKKMsm1oMkQIl8w3QKuw,4016
 outerbounds/apps/config_schema.yaml,sha256=R1sSKFroqCSkcJScdAKU2vCvhgdFt_PJNZzecHB1oe0,8629
 outerbounds/apps/dependencies.py,sha256=SqvdFQdFZZW0wXX_CHMHCrfE0TwaRkTvGCRbQ2Mx3q0,3935
@@ -72,7 +72,7 @@ outerbounds/utils/metaflowconfig.py,sha256=l2vJbgPkLISU-XPGZFaC8ZKmYFyJemlD6bwB-
 outerbounds/utils/schema.py,sha256=lMUr9kNgn9wy-sO_t_Tlxmbt63yLeN4b0xQXbDUDj4A,2331
 outerbounds/utils/utils.py,sha256=4Z8cszNob_8kDYCLNTrP-wWads_S_MdL3Uj3ju4mEsk,501
 outerbounds/vendor.py,sha256=gRLRJNXtZBeUpPEog0LOeIsl6GosaFFbCxUvR4bW6IQ,5093
-outerbounds-0.3.176rc4.dist-info/METADATA,sha256=bF2r5YAzE_L92TOuSGmgpNuFvqQb73q-boJXB_cF6Gs,1846
-outerbounds-0.3.176rc4.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
-outerbounds-0.3.176rc4.dist-info/entry_points.txt,sha256=7ye0281PKlvqxu15rjw60zKg2pMsXI49_A8BmGqIqBw,47
-outerbounds-0.3.176rc4.dist-info/RECORD,,
+outerbounds-0.3.176rc6.dist-info/METADATA,sha256=hfETPwYSaPJHpV1DZB2KKpZJxj9kI8pExlBeTdL50h8,1846
+outerbounds-0.3.176rc6.dist-info/WHEEL,sha256=sP946D7jFCHeNz5Iq4fL4Lu-PrWrFsgfLXbbkciIZwg,88
+outerbounds-0.3.176rc6.dist-info/entry_points.txt,sha256=7ye0281PKlvqxu15rjw60zKg2pMsXI49_A8BmGqIqBw,47
+outerbounds-0.3.176rc6.dist-info/RECORD,,