mathcraft-ocr 0.2.6__tar.gz → 0.2.7__tar.gz

{mathcraft_ocr-0.2.6/mathcraft_ocr.egg-info → mathcraft_ocr-0.2.7}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mathcraft-ocr
-Version: 0.2.6
+Version: 0.2.7
 Summary: ONNX-only OCR runtime for mathematical documents
 Author: SakuraMathcraft
 License-Expression: GPL-3.0-only
@@ -215,4 +215,4 @@ python -m build --no-isolation --outdir .\release_assets\mathcraft-ocr-package\d
 
 ## License
 
-MIT. See `LICENSE`.
+GPL-3.0-only. See `LICENSE`.

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/README_MATHCRAFT_OCR.md

@@ -169,4 +169,4 @@ python -m build --no-isolation --outdir .\release_assets\mathcraft-ocr-package\d
 
 ## License
 
-MIT. See `LICENSE`.
+GPL-3.0-only. See `LICENSE`.

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/__init__.py

@@ -2,7 +2,7 @@
 
 from __future__ import annotations
 
-__version__ = "0.2.6"
+__version__ = "0.2.7"
 
 __all__ = [
     "DoctorReport",

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/adapters/common.py

@@ -33,13 +33,21 @@ def create_session(model_path: str | Path, provider_info: ProviderInfo):
         raise RuntimeError(
             f"requested ONNX GPU provider {active}, but session providers are {actual}"
         )
+    disable_fallback = getattr(session, "disable_fallback", None)
+    if not callable(disable_fallback):
+        raise RuntimeError("ONNX Runtime session cannot disable provider fallback")
+    disable_fallback()
     return session
 
 
 @lru_cache(maxsize=16)
 def _create_session_cached(model_path: str, providers: tuple[str, ...]):
     ort = _ort()
-    return ort.InferenceSession(model_path, providers=list(providers))
+    return ort.InferenceSession(
+        model_path,
+        providers=list(providers),
+        enable_fallback=False,
+    )
 
 
 def clear_session_cache() -> None:

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/adapters/text_recognizer.py

@@ -57,11 +57,21 @@ def _create_pp_text_recognizer(model_dir: Path, provider_info) -> TextRecognizer
     active_provider = str(getattr(provider_info, "active_provider", "") or "")
     use_cuda = active_provider in {"CUDAExecutionProvider", "TensorrtExecutionProvider"}
     use_dml = active_provider == "DmlExecutionProvider"
-    return _create_pp_text_recognizer_cached(str(model_dir), use_cuda, use_dml)
+    return _create_pp_text_recognizer_cached(
+        str(model_dir),
+        active_provider,
+        use_cuda,
+        use_dml,
+    )
 
 
 @lru_cache(maxsize=8)
-def _create_pp_text_recognizer_cached(model_dir: str, use_cuda: bool, use_dml: bool) -> TextRecognizer:
+def _create_pp_text_recognizer_cached(
+    model_dir: str,
+    active_provider: str,
+    use_cuda: bool,
+    use_dml: bool,
+) -> TextRecognizer:
     model_dir = Path(model_dir)
     model_candidates = sorted(model_dir.glob("**/*rec*.onnx"))
     if not model_candidates:
@@ -111,7 +121,26 @@ def _create_pp_text_recognizer_cached(model_dir: str, use_cuda: bool, use_dml: b
             },
         },
     })
-    return TextRecognizer(config)
+    recognizer = TextRecognizer(config)
+    _enforce_strict_provider(recognizer, active_provider)
+    return recognizer
+
+
+def _enforce_strict_provider(recognizer: TextRecognizer, active_provider: str) -> None:
+    engine = getattr(recognizer, "session", None)
+    session = getattr(engine, "session", None)
+    if session is None:
+        raise RuntimeError("RapidOCR did not expose its ONNX Runtime session")
+    actual = list(session.get_providers() or [])
+    if not active_provider or not actual or actual[0] != active_provider:
+        raise RuntimeError(
+            f"requested ONNX provider {active_provider or '<none>'}, "
+            f"but RapidOCR session providers are {actual}"
+        )
+    disable_fallback = getattr(session, "disable_fallback", None)
+    if not callable(disable_fallback):
+        raise RuntimeError("RapidOCR ONNX Runtime session cannot disable provider fallback")
+    disable_fallback()
 
 
 def clear_text_recognizer_cache() -> None:

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/error_patterns.py

@@ -17,6 +17,13 @@ CUDA_RUNTIME_ERROR_MARKERS = (
     "cufft",
 )
 
+GPU_PROVIDER_ERROR_MARKERS = (
+    "gpu provider was requested",
+    "requested onnx gpu provider",
+    "requested onnx provider dmlexecutionprovider",
+    "requested onnx provider tensorrtexecutionprovider",
+)
+
 ONNXRUNTIME_INSTALL_ERROR_MARKERS = (
     "failed to import onnxruntime",
     "failed to query onnx providers",
@@ -31,6 +38,11 @@ def looks_like_cuda_runtime_error(detail: object) -> bool:
     return any(marker in text for marker in CUDA_RUNTIME_ERROR_MARKERS)
 
 
+def looks_like_gpu_provider_error(detail: object) -> bool:
+    text = str(detail or "").lower()
+    return any(marker in text for marker in GPU_PROVIDER_ERROR_MARKERS)
+
+
 def looks_like_onnxruntime_install_error(detail: object) -> bool:
     text = str(detail or "").lower()
     return any(marker in text for marker in ONNXRUNTIME_INSTALL_ERROR_MARKERS)
@@ -38,7 +50,9 @@ def looks_like_onnxruntime_install_error(detail: object) -> bool:
 
 __all__ = [
     "CUDA_RUNTIME_ERROR_MARKERS",
+    "GPU_PROVIDER_ERROR_MARKERS",
     "ONNXRUNTIME_INSTALL_ERROR_MARKERS",
     "looks_like_cuda_runtime_error",
+    "looks_like_gpu_provider_error",
     "looks_like_onnxruntime_install_error",
 ]

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/providers.py

@@ -3,7 +3,6 @@
 from __future__ import annotations
 
 import importlib
-import os
 from dataclasses import dataclass
 
 from .errors import ProviderError
@@ -23,7 +22,6 @@ class ProviderInfo:
     device: str
     gpu_requested: bool
     gpu_runtime_ok: bool
-    cpu_fallback: bool
 
     @property
     def use_cuda(self) -> bool:
@@ -56,26 +54,17 @@ def detect_providers(prefer: str = "auto") -> ProviderInfo:
     except Exception as exc:  # pragma: no cover - runtime environment dependent
         raise ProviderError(f"failed to query ONNX providers: {exc}") from exc
 
-    force_cpu = os.environ.get("MATHCRAFT_FORCE_ORT_CPU", "").strip() == "1"
     gpu_candidates = tuple(name for name in GPU_PROVIDER_NAMES if name in available)
     gpu_visible = bool(gpu_candidates)
-    if force_cpu:
-        return ProviderInfo(
-            available_providers=available,
-            active_provider="CPUExecutionProvider" if "CPUExecutionProvider" in available else None,
-            device="cpu",
-            gpu_requested=(prefer_norm in {"auto", "gpu"}),
-            gpu_runtime_ok=False,
-            cpu_fallback=True,
-        )
     if prefer_norm == "cpu":
+        if "CPUExecutionProvider" not in available:
+            raise ProviderError(f"CPUExecutionProvider unavailable: {available}")
         return ProviderInfo(
             available_providers=available,
-            active_provider="CPUExecutionProvider" if "CPUExecutionProvider" in available else None,
+            active_provider="CPUExecutionProvider",
             device="cpu",
             gpu_requested=False,
             gpu_runtime_ok=False,
-            cpu_fallback=False,
         )
     if gpu_visible:
         return ProviderInfo(
@@ -84,13 +73,17 @@ def detect_providers(prefer: str = "auto") -> ProviderInfo:
             device="gpu",
             gpu_requested=True,
             gpu_runtime_ok=True,
-            cpu_fallback=False,
         )
+    if prefer_norm == "gpu":
+        raise ProviderError(
+            f"GPU provider was requested but none is available: {available}"
+        )
+    if "CPUExecutionProvider" not in available:
+        raise ProviderError(f"no supported ONNX execution provider is available: {available}")
     return ProviderInfo(
         available_providers=available,
-        active_provider="CPUExecutionProvider" if "CPUExecutionProvider" in available else None,
+        active_provider="CPUExecutionProvider",
         device="cpu",
-        gpu_requested=(prefer_norm in {"auto", "gpu"}),
+        gpu_requested=False,
         gpu_runtime_ok=False,
-        cpu_fallback=(prefer_norm == "gpu"),
     )

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/runtime.py

@@ -22,7 +22,7 @@ from .adapters.text_recognizer import (
 from .cache import ModelCacheState, inspect_manifest_roots, resolve_model_roots, resolve_user_models_dir
 from .doctor import DoctorReport, run_doctor
 from .downloader import download_model_archive
-from .error_patterns import looks_like_cuda_runtime_error
+from .error_patterns import looks_like_cuda_runtime_error, looks_like_gpu_provider_error
 from .errors import ModelCacheError
 from .formula_lines import (
     FormulaLineGroup,
@@ -207,7 +207,7 @@ class MathCraftRuntime:
     @staticmethod
     def _looks_like_broken_model_error(exc: Exception) -> bool:
         text = str(exc).lower()
-        if looks_like_cuda_runtime_error(text):
+        if looks_like_cuda_runtime_error(text) or looks_like_gpu_provider_error(text):
             return False
         needles = (
             "missing",

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/mathcraft_ocr/serialization.py

@@ -71,7 +71,6 @@ def provider_info_to_json(provider_info) -> dict:
         "device": provider_info.device,
         "gpu_requested": provider_info.gpu_requested,
         "gpu_runtime_ok": provider_info.gpu_runtime_ok,
-        "cpu_fallback": provider_info.cpu_fallback,
         "use_cuda": provider_info.use_cuda,
         "use_dml": provider_info.use_dml,
     }

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7/mathcraft_ocr.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: mathcraft-ocr
-Version: 0.2.6
+Version: 0.2.7
 Summary: ONNX-only OCR runtime for mathematical documents
 Author: SakuraMathcraft
 License-Expression: GPL-3.0-only
@@ -215,4 +215,4 @@ python -m build --no-isolation --outdir .\release_assets\mathcraft-ocr-package\d
 
 ## License
 
-MIT. See `LICENSE`.
+GPL-3.0-only. See `LICENSE`.

{mathcraft_ocr-0.2.6 → mathcraft_ocr-0.2.7}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "mathcraft-ocr"
-version = "0.2.6"
+version = "0.2.7"
 description = "ONNX-only OCR runtime for mathematical documents"
 readme = "README_MATHCRAFT_OCR.md"
 requires-python = ">=3.10"