vllm-cpu 0.11.0.post2__cp312-cp312-manylinux_2_17_x86_64.whl → 0.11.1__cp312-cp312-manylinux_2_17_x86_64.whl

vllm/__init__.py

@@ -1,121 +1,3 @@
-# =============================================================================
-# CPU Platform Fix - Injected into vllm/__init__.py during build
-# =============================================================================
-# This code creates a fake "vllm" package metadata entry with +cpu suffix
-# so that vLLM's platform detection (vllm_version_matches_substr("cpu"))
-# works correctly for pip-installed vllm-cpu packages.
-#
-# Why this is needed:
-# - vLLM's cpu_platform_plugin() calls vllm_version_matches_substr("cpu")
-# - This checks if "cpu" is in importlib.metadata.version("vllm")
-# - PyPI doesn't allow +cpu suffix (PEP 440 forbids local version identifiers)
-# - So we create a fake vllm dist-info with +cpu version at runtime
-#
-# This runs once on first import of vllm, creating the dist-info if needed.
-# =============================================================================
-
-def _ensure_cpu_platform_detection():
-    """Create vllm package alias for CPU platform detection if needed."""
-    import os
-    import sys
-
-    try:
-        # Check if we're a vllm-cpu variant package
-        from importlib.metadata import version, distributions
-
-        # Find our actual package (vllm-cpu, vllm-cpu-avx512, etc.)
-        cpu_version = None
-        cpu_package = None
-        cpu_dist_location = None
-        for dist in distributions():
-            name = dist.metadata.get('Name', '')
-            if name.startswith('vllm-cpu'):
-                cpu_version = dist.metadata.get('Version', '')
-                cpu_package = name
-                # Get the location of this distribution's metadata
-                # dist._path is the path to the .dist-info directory
-                try:
-                    cpu_dist_location = str(dist._path.parent)
-                except Exception:
-                    pass
-                break
-
-        if not cpu_version:
-            # Not a vllm-cpu package, skip
-            return
-
-        # Check if "vllm" already returns a cpu version
-        try:
-            vllm_ver = version("vllm")
-            if "cpu" in vllm_ver.lower():
-                # Already has cpu in version, nothing to do
-                return
-        except Exception:
-            pass  # vllm package doesn't exist yet, we'll create it
-
-        # Find site-packages directory - prefer the location where vllm-cpu is installed
-        site_packages = cpu_dist_location
-        if not site_packages or not os.path.isdir(site_packages):
-            # Fallback: search sys.path for site-packages
-            for path in sys.path:
-                if 'site-packages' in path and os.path.isdir(path):
-                    site_packages = path
-                    break
-
-        if not site_packages:
-            return
-
-        # Create fake vllm dist-info
-        vllm_dist = os.path.join(site_packages, "vllm-0.0.0.dist-info")
-        if os.path.exists(vllm_dist):
-            return  # Already exists
-
-        try:
-            os.makedirs(vllm_dist, exist_ok=True)
-
-            # Ensure version has +cpu suffix
-            if "cpu" not in cpu_version.lower():
-                cpu_version = f"{cpu_version}+cpu"
-
-            # Write METADATA file
-            metadata_path = os.path.join(vllm_dist, "METADATA")
-            with open(metadata_path, "w") as f:
-                f.write(f"Metadata-Version: 2.1\n")
-                f.write(f"Name: vllm\n")
-                f.write(f"Version: {cpu_version}\n")
-                f.write(f"Summary: vLLM CPU package alias for platform detection\n")
-
-            # Write minimal RECORD file (required for proper dist-info)
-            record_path = os.path.join(vllm_dist, "RECORD")
-            with open(record_path, "w") as f:
-                f.write("vllm-0.0.0.dist-info/METADATA,,\n")
-                f.write("vllm-0.0.0.dist-info/RECORD,,\n")
-
-            # Invalidate importlib.metadata cache so it picks up the new dist-info
-            try:
-                from importlib import invalidate_caches
-                invalidate_caches()
-            except Exception:
-                pass
-
-        except (OSError, IOError):
-            # Can't write to site-packages (e.g., system Python, no permissions)
-            # This is fine - user can use the manual workaround or run in Docker
-            pass
-
-    except Exception:
-        # Don't break vllm import if anything goes wrong
-        pass
-
-# Run the fix on import
-_ensure_cpu_platform_detection()
-
-# Clean up namespace
-del _ensure_cpu_platform_detection
-
-# =============================================================================
-# End of CPU Platform Fix
-# =============================================================================
 # SPDX-License-Identifier: Apache-2.0
 # SPDX-FileCopyrightText: Copyright contributors to the vLLM project
 """vLLM: a high-throughput and memory-efficient inference engine for LLMs"""
@@ -139,7 +21,7 @@ MODULE_ATTRS = {
     "AsyncLLMEngine": ".engine.async_llm_engine:AsyncLLMEngine",
     "LLMEngine": ".engine.llm_engine:LLMEngine",
     "LLM": ".entrypoints.llm:LLM",
-    "initialize_ray_cluster": ".executor.ray_utils:initialize_ray_cluster",
+    "initialize_ray_cluster": ".v1.executor.ray_utils:initialize_ray_cluster",
     "PromptType": ".inputs:PromptType",
     "TextPrompt": ".inputs:TextPrompt",
     "TokensPrompt": ".inputs:TokensPrompt",
@@ -163,17 +45,23 @@ if typing.TYPE_CHECKING:
     from vllm.engine.async_llm_engine import AsyncLLMEngine
     from vllm.engine.llm_engine import LLMEngine
     from vllm.entrypoints.llm import LLM
-    from vllm.executor.ray_utils import initialize_ray_cluster
     from vllm.inputs import PromptType, TextPrompt, TokensPrompt
     from vllm.model_executor.models import ModelRegistry
-    from vllm.outputs import (ClassificationOutput,
-                              ClassificationRequestOutput, CompletionOutput,
-                              EmbeddingOutput, EmbeddingRequestOutput,
-                              PoolingOutput, PoolingRequestOutput,
-                              RequestOutput, ScoringOutput,
-                              ScoringRequestOutput)
+    from vllm.outputs import (
+        ClassificationOutput,
+        ClassificationRequestOutput,
+        CompletionOutput,
+        EmbeddingOutput,
+        EmbeddingRequestOutput,
+        PoolingOutput,
+        PoolingRequestOutput,
+        RequestOutput,
+        ScoringOutput,
+        ScoringRequestOutput,
+    )
     from vllm.pooling_params import PoolingParams
     from vllm.sampling_params import SamplingParams
+    from vllm.v1.executor.ray_utils import initialize_ray_cluster
 
     from ._bc_linter import bc_linter_include, bc_linter_skip
 else:
@@ -186,8 +74,7 @@ else:
             module = import_module(module_name, __package__)
             return getattr(module, attr_name)
         else:
-            raise AttributeError(
-                f'module {__package__} has no attribute {name}')
+            raise AttributeError(f"module {__package__} has no attribute {name}")
 
 
 __all__ = [