improve copy from #382

Author: hualxie

.aitk/configs/checks.json

@@ -1,6 +1,6 @@
 {
     "configCheck": 167,
-    "copyCheck": 182,
+    "copyCheck": 184,
     "extensionCheck": 2,
     "gitignoreCheck": 44,
     "inferenceModelCheck": 25,
@@ -12,5 +12,6 @@
     "pathCheck": 1423,
     "requirementsCheck": 37,
     "templateCheck": 3,
-    "venvRequirementsCheck": 17
+    "venvRequirementsCheck": 17,
+    "winmlCopyCheck": 39
 }

.aitk/scripts/project_processor.py

@@ -4,18 +4,24 @@
 from pathlib import Path
 from typing import Dict, List, Optional
 
-import yaml
 from model_lab import RuntimeEnum
 from sanitize.constants import ArchitectureEnum, EPNames, IconEnum, ModelStatusEnum
-from sanitize.copy_config import CopyConfig
+from sanitize.copy_config import Copy, CopyConfig
 from sanitize.generator_amd import generator_amd
 from sanitize.generator_dml import generator_dml
 from sanitize.generator_intel import generator_intel
 from sanitize.generator_qnn import generator_qnn
 from sanitize.generator_trtrtx import generator_trtrtx
 from sanitize.model_info import ModelInfo, ModelList
 from sanitize.project_config import ModelInfoProject, ModelProjectConfig, WorkflowItem
-from sanitize.utils import GlobalVars, isLLM_by_id, open_ex
+from sanitize.utils import (
+    GlobalVars,
+    WINML_COPY_EXEMPT_IDS,
+    isLLM_by_id,
+    iter_aitk_info_yml,
+    open_ex,
+    winml_copy_src_for,
+)
 
 def fetch_pipeline_tags(model_link: str) -> Optional[List[str]]:
     """Fetch pipeline_tag from HuggingFace API for a given model link.
@@ -209,22 +215,9 @@ def project_processor():
 
     all_ids = set()
     all_summary = AllModelSummary()
-    for yml_file in root_dir.rglob("info.yml"):
+    for yml_file, yaml_object in iter_aitk_info_yml(root_dir):
         # if "DEBUG_ID" in str(yml_file):
         #    pass
-        # read yml file as yaml object
-        with yml_file.open("r", encoding="utf-8") as file:
-            try:
-                yaml_content = file.read()
-                yaml_object = yaml.safe_load(yaml_content)
-            except yaml.YAMLError as e:
-                print(f"Error reading {yml_file}: {e}")
-                continue
-            aitk = yaml_object.get("aitk", [])
-            if not aitk:
-                if yml_file.parent.name == "aitk":
-                    raise KeyError(f"aitk not found in {yml_file}")
-                continue
         print(f"Process aitk for {yml_file}")
         # model info
         modelInfo = convert_yaml_to_model_info(root_dir, yml_file, yaml_object)
@@ -236,10 +229,24 @@ def project_processor():
             raise KeyError(f"same id found in {yml_file}")
         all_ids.add(modelInfo.id.lower())
         modelList.models.append(modelInfo)
-        # copy pre
+        # copy pre — auto-ensure winml.py copy entry (unless exempt), then run pre-phase copies
         copyConfigFile = yml_file.parent / "_copy.json.config"
-        if copyConfigFile.exists():
-            copyConfig = CopyConfig.Read(copyConfigFile.as_posix())
+        copyConfig: CopyConfig | None = (
+            CopyConfig.Read(copyConfigFile.as_posix()) if copyConfigFile.exists() else None
+        )
+        if modelInfo.id not in WINML_COPY_EXEMPT_IDS:
+            desired_src = winml_copy_src_for(modelInfo.id)
+            if copyConfig is None:
+                copyConfig = CopyConfig()
+                copyConfig._file = str(copyConfigFile)
+                copyConfig._fileContent = None
+            existing = next((c for c in copyConfig.copies if c.dst == "winml.py"), None)
+            if existing is None:
+                copyConfig.copies.append(Copy(src=desired_src, dst="winml.py"))
+            elif existing.src != desired_src:
+                existing.src = desired_src
+            GlobalVars.winmlCopyCheck += 1
+        if copyConfig is not None:
             copyConfig.process(yml_file.parent.as_posix(), pre=True)
             copyConfig.writeIfChanged()
         # model summary
@@ -258,4 +265,4 @@ def project_processor():
 
 
 if __name__ == "__main__":
-    project_processor()
+    project_processor()

.aitk/scripts/sanitize/main.py

@@ -63,7 +63,7 @@ def main():
             version = model.version
             modelVerDir = modelDir if model.relativePath else os.path.join(modelDir, str(version))
 
-            # process copy
+            # process copy (post phase — .json.config copies; winml.py is auto-ensured earlier in project_processor)
             copyConfigFile = os.path.join(modelVerDir, "_copy.json.config")
             if os.path.exists(copyConfigFile):
                 copyConfig = CopyConfig.Read(copyConfigFile)

.aitk/scripts/sanitize/utils.py

@@ -6,14 +6,41 @@
 import json
 import os
 from contextlib import contextmanager
-from typing import Any
+from pathlib import Path
+from typing import Any, Iterator, Tuple
 
 import pydash
+import yaml
 from model_lab import RuntimeEnum
 
 from .constants import EPNames, OliveDeviceTypes, OlivePropertyNames
 
 
+def iter_aitk_info_yml(root_dir: Path) -> Iterator[Tuple[Path, dict]]:
+    """Yield (yml_file, yaml_object) for each info.yml under root_dir that has
+    a top-level `aitk` key.
+
+    Files that fail to parse are skipped with a printed warning. An info.yml
+    sitting under a folder literally named `aitk` but missing the `aitk` key
+    raises KeyError, matching the invariant enforced by project_processor.
+    """
+    for yml_file in root_dir.rglob("info.yml"):
+        try:
+            with yml_file.open("r", encoding="utf-8") as f:
+                yaml_object = yaml.safe_load(f.read())
+        except yaml.YAMLError as e:
+            print(f"Error reading {yml_file}: {e}")
+            continue
+        if not isinstance(yaml_object, dict):
+            continue
+        aitk = yaml_object.get("aitk")
+        if not aitk:
+            if yml_file.parent.name == "aitk":
+                raise KeyError(f"aitk not found in {yml_file}")
+            continue
+        yield yml_file, yaml_object
+
+
 class GlobalVars:
     errorList = []
     verbose = False
@@ -34,6 +61,7 @@ class GlobalVars:
     copyCheck = 0
     licenseCheck = 0
     venvRequirementsCheck = set()
+    winmlCopyCheck = 0
 
     oliveCheck = 0
     RuntimeToEPName = {
@@ -209,6 +237,23 @@ def isLLM_by_id(id: str) -> bool:
     return any(check in id for check in check_list)
 
 
+# Projects that are themselves canonical winml.py sources and should not have
+# a winml.py copy entry auto-added to their _copy.json.config.
+WINML_COPY_EXEMPT_IDS = {
+    "huggingface/Intel/bert-base-uncased-mrpc",
+    "huggingface/deepseek-ai/DeepSeek-R1-Distill-Qwen-1.5B",
+}
+
+# Canonical winml.py sources. Paths are relative to the project's aitk folder
+# (matches the CopyConfig `src` format).
+WINML_SRC_LLM = "../../deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B/aitk/winml.py"
+WINML_SRC_NON_LLM = "../../intel-bert-base-uncased-mrpc/aitk/winml.py"
+
+
+def winml_copy_src_for(model_id: str) -> str:
+    return WINML_SRC_LLM if isLLM_by_id(model_id) else WINML_SRC_NON_LLM
+
+
 # TODO align with Skylight\vscode\ai-mlstudio\src\model-lab\utilities\runtimeUtils.ts
 def get_execute_runtime(runtime: RuntimeEnum) -> RuntimeEnum:
     if runtime in [RuntimeEnum.IntelAny, RuntimeEnum.IntelCPU, RuntimeEnum.IntelGPU, RuntimeEnum.IntelNPU]:
@@ -231,4 +276,4 @@ def get_eval_runtime(runtime: RuntimeEnum, isLLM: bool) -> RuntimeEnum:
 def get_eval_in_execute_runtime(runtime: RuntimeEnum) -> RuntimeEnum:
     if runtime == RuntimeEnum.QNN or runtime == RuntimeEnum.QNNGPU:
         return RuntimeEnum.QNN
-    raise ValueError(f"Unsupported runtime for eval in execute: {runtime}")
+    raise ValueError(f"Unsupported runtime for eval in execute: {runtime}")

OFA-Sys-chinese-clip-vit-base-patch16/aitk/_copy.json.config

@@ -0,0 +1,8 @@
+{
+    "copies": [
+        {
+            "src": "../../intel-bert-base-uncased-mrpc/aitk/winml.py",
+            "dst": "winml.py"
+        }
+    ]
+}

Qwen-Qwen2.5-0.5B-Instruct/aitk/_copy.json.config

@@ -35,7 +35,7 @@
             "dst": "requirements.txt"
         },
         {
-            "src": "../../intel-bert-base-uncased-mrpc/aitk/winml.py",
+            "src": "../../deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B/aitk/winml.py",
             "dst": "winml.py"
         }
     ]

Qwen-Qwen2.5-0.5B/aitk/_copy.json.config

@@ -6,7 +6,7 @@
             "replacements": []
         },
         {
-            "src": "../../intel-bert-base-uncased-mrpc/aitk/winml.py",
+            "src": "../../deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B/aitk/winml.py",
             "dst": "winml.py"
         }
     ]

Qwen-Qwen2.5-1.5B-Instruct/aitk/_copy.json.config

@@ -62,7 +62,7 @@
             ]
         },
         {
-            "src": "../../intel-bert-base-uncased-mrpc/aitk/winml.py",
+            "src": "../../deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B/aitk/winml.py",
             "dst": "winml.py"
         }
     ]

Qwen-Qwen2.5-14B-Instruct/aitk/_copy.json.config

@@ -39,7 +39,7 @@
             "dst": "requirements.txt"
         },
         {
-            "src": "../../intel-bert-base-uncased-mrpc/aitk/winml.py",
+            "src": "../../deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B/aitk/winml.py",
             "dst": "winml.py"
         },
         {

Qwen-Qwen2.5-3B-Instruct/aitk/_copy.json.config

@@ -25,7 +25,7 @@
             "dst": "requirements.txt"
         },
         {
-            "src": "../../intel-bert-base-uncased-mrpc/aitk/winml.py",
+            "src": "../../deepseek-ai-DeepSeek-R1-Distill-Qwen-1.5B/aitk/winml.py",
             "dst": "winml.py"
         },
         {