1 files changed, 92 insertions, 0 deletions
diff --git a/app/backends/common.py b/app/backends/common.py
index a12a366..7974b58 100644
--- a/app/backends/common.py
+++ b/app/backends/common.py
@@ -11,6 +11,7 @@ run.
 
 import os
 import re
+import shutil
 import sys
 import time
 import urllib.parse
@@ -743,3 +744,94 @@ def pip_uninstall(packages: List[str], *, emit=None,
     """
     from backends import envs
     return envs.pip_uninstall(packages, emit=emit, env_dir=env_dir)
+
+
+# ----------------------------------------------------------------------
+# HuggingFace hub cache
+#
+# Every pip-based backend's model weights are fetched by its server into
+# the standard hub cache on first start (and pre-downloaded by the
+# Install action via the venv's hf CLI). These helpers read and delete
+# the same directories ``from_pretrained`` writes, honoring the same
+# environment overrides, so per-model (un)installs land exactly where a
+# server start would look. The cache is shared across backends: a repo
+# two backends host exists once, and its deletion affects both — the
+# same convention every backend here accepts.
+# ----------------------------------------------------------------------
+
+def hf_cache_dir() -> Path:
+    """The HF hub cache dir servers fetch model weights into.
+
+    Resolution mirrors huggingface_hub.constants: HF_HUB_CACHE beats
+    HUGGINGFACE_HUB_CACHE beats HF_HOME/hub beats ~/.cache/huggingface/hub.
+    """
+    override = os.environ.get("HF_HUB_CACHE") or os.environ.get(
+        "HUGGINGFACE_HUB_CACHE")
+    if override:
+        return Path(override)
+    home = os.environ.get("HF_HOME")
+    if home:
+        return Path(home) / "hub"
+    return Path.home() / ".cache" / "huggingface" / "hub"
+
+
+def hf_repo_dir(repo_id: str) -> Path:
+    """The cache directory HF keeps REPO_ID's weights in."""
+    return hf_cache_dir() / ("models--" + repo_id.replace("/", "--"))
+
+
+def hf_tree_has_file(path: Path) -> bool:
+    """True when any file or symlink exists under PATH (recursively)."""
+    try:
+        for item in path.iterdir():
+            # Snapshot files are symlinks into blobs/; count them even
+            # when temporarily broken (presence is what the loader
+            # checks).
+            if item.is_symlink() or item.is_file():
+                return True
+            if item.is_dir() and hf_tree_has_file(item):
+                return True
+    except OSError:
+        return False
+    return False
+
+
+def hf_delete_model_weights(repo_ids: List[str]) -> int:
+    """Delete the cached HF weight dirs of REPO_IDS; returns how many
+    were present and removed.
+
+    Best-effort rmtree of each ``models--<org>--<name>`` directory; only
+    those directories are ever touched — the rest of the HF cache may be
+    shared with unrelated tools. Prints its progress, which streams into
+    the task view under curses too.
+    """
+    removed = 0
+    for repo_id in repo_ids:
+        directory = hf_repo_dir(repo_id)
+        if not directory.is_dir():
+            continue
+        print(f"[INFO] Removing cached {repo_id} weights...")
+        shutil.rmtree(directory, ignore_errors=True)
+        if directory.exists():
+            print(f"[WARNING] Could not fully remove {directory}")
+            continue
+        removed += 1
+    if removed:
+        print(f"[OK] Deleted cached weights for {removed} "
+              f"{'model' if removed == 1 else 'models'}.")
+    return removed
+
+
+def hf_download_prefix(env_dir: Path) -> Optional[List[str]]:
+    """The venv ENV_DIR's hf CLI argv prefix (None when absent).
+
+    The hf CLI (or its older ``huggingface-cli`` name) is what a server
+    start uses to fetch weights, so the install action pre-downloads
+    through it too — resumable, streamed, cancelable.
+    """
+    from backends import envs
+    for name in ("hf", "huggingface-cli"):
+        candidate = envs.env_script(name, env_dir)
+        if candidate.is_file():
+            return [str(candidate)]
+    return None