Merge origin/master into ak/t5gemma2 and resolve conflicts

TAG=agy

CONV=7b651079-7501-4d53-a9a3-6058ec11ea33

Signed-off-by: Akhilesh Kumar <akhilbussiness@gmail.com>

Author: akh64bit

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "vllm-bart-plugin"
-version = "0.2.0"
+version = "0.3.4"
 description = "BART model plugin for vLLM"
 readme = "README.md"
 requires-python = ">=3.10"
@@ -26,7 +26,7 @@ classifiers = [
 ]
 
 dependencies = [
-    "vllm>=0.14.0",
+    "vllm>=0.13.0,<=0.18",
     "torch>=2.9.0",
     "transformers>=4.56.0,<5",
 ]

scripts/bump_version.py

@@ -0,0 +1,116 @@
+#!/usr/bin/env python3
+"""Simple script to bump the package version in setup.py and pyproject.toml."""
+
+import argparse
+import re
+import sys
+from pathlib import Path
+
+
+def get_current_version(content: str) -> str | None:
+    """Extract version from file content."""
+    match = re.search(r'version\s*=\s*["\']([^"\']+)["\']', content)
+    return match.group(1) if match else None
+
+
+def bump_version(version: str, part: str) -> str:
+    """Bump the specified part of the version."""
+    parts = version.split(".")
+    if len(parts) != 3:
+        raise ValueError(f"Expected semantic version (x.y.z), got: {version}")
+
+    major, minor, patch = map(int, parts)
+
+    if part == "major":
+        major += 1
+        minor = 0
+        patch = 0
+    elif part == "minor":
+        minor += 1
+        patch = 0
+    elif part == "patch":
+        patch += 1
+    else:
+        raise ValueError(f"Unknown version part: {part}")
+
+    return f"{major}.{minor}.{patch}"
+
+
+def update_file(filepath: Path, old_version: str, new_version: str) -> bool:
+    """Update version in a file. Returns True if file was modified."""
+    if not filepath.exists():
+        return False
+
+    content = filepath.read_text()
+    new_content = re.sub(
+        rf'(version\s*=\s*["\']){re.escape(old_version)}(["\'])',
+        rf"\g<1>{new_version}\g<2>",
+        content,
+    )
+
+    if content != new_content:
+        filepath.write_text(new_content)
+        return True
+    return False
+
+
+def main():
+    parser = argparse.ArgumentParser(description="Bump package version")
+    parser.add_argument(
+        "part",
+        choices=["major", "minor", "patch"],
+        nargs="?",
+        default="patch",
+        help="Version part to bump (default: patch)",
+    )
+    parser.add_argument(
+        "--set",
+        dest="set_version",
+        metavar="VERSION",
+        help="Set a specific version instead of bumping",
+    )
+    parser.add_argument(
+        "--dry-run",
+        action="store_true",
+        help="Show what would be changed without modifying files",
+    )
+    args = parser.parse_args()
+
+    root = Path(__file__).parent.parent
+    files = [root / "setup.py", root / "pyproject.toml"]
+
+    # Get current version from pyproject.toml
+    pyproject = root / "pyproject.toml"
+    if not pyproject.exists():
+        print("Error: pyproject.toml not found", file=sys.stderr)
+        sys.exit(1)
+
+    current_version = get_current_version(pyproject.read_text())
+    if not current_version:
+        print("Error: Could not find version in pyproject.toml", file=sys.stderr)
+        sys.exit(1)
+
+    # Determine new version
+    if args.set_version:
+        new_version = args.set_version
+    else:
+        new_version = bump_version(current_version, args.part)
+
+    print(f"Version: {current_version} -> {new_version}")
+
+    if args.dry_run:
+        print("Dry run - no files modified")
+        return
+
+    # Update files
+    for filepath in files:
+        if update_file(filepath, current_version, new_version):
+            print(f"Updated: {filepath.name}")
+        else:
+            print(f"Skipped: {filepath.name} (not found or no changes)")
+
+    print(f"\nDone! Don't forget to commit and tag: git tag v{new_version}")
+
+
+if __name__ == "__main__":
+    main()

setup.py

@@ -4,14 +4,14 @@
 
 setup(
     name="vllm-bart-plugin",
-    version="0.2.0",
+    version="0.3.4",
     description="BART model plugin for vLLM",
     author="Nicolò Lucchesi",
     author_email="nick.lucche@redhat.com",
     packages=find_packages(),
     python_requires=">=3.10",
     install_requires=[
-        "vllm>=0.13.0",
+        "vllm>=0.13.0,<=0.18",
         "torch>=2.9.0",
         "transformers >= 4.56.0, < 5",
     ],

tests/test_vllm_018_compat.py

@@ -0,0 +1,47 @@
+"""Regression tests for vLLM 0.18 compatibility in the BART processor."""
+
+import torch
+
+
+def test_text_data_parser_handles_v018_empty_inputs():
+    from vllm_bart_plugin.bart import TextDataParser
+
+    parser = TextDataParser()
+
+    assert parser._parse_text_data("") is None
+    assert parser._parse_text_data([]) is None
+
+
+def test_create_encoder_prompt_uses_placeholder_token():
+    from vllm_bart_plugin.bart import BartMultiModalProcessor
+
+    processor = BartMultiModalProcessor.__new__(BartMultiModalProcessor)
+
+    assert processor.create_encoder_prompt("<s>decoder text", {"texts": ["encoder text"]}) == [0]
+
+
+def test_call_hf_processor_accepts_pretokenized_decoder_prompt():
+    from vllm_bart_plugin.bart import BartMultiModalProcessor
+
+    class FakeTokenizer:
+        def __call__(self, text, return_tensors="pt", **kwargs):
+            if text == "encoder text":
+                return {"input_ids": torch.tensor([[11, 12, 13]])}
+            return {"input_ids": torch.tensor([[21, 22]])}
+
+    class FakeInfo:
+        def get_tokenizer(self):
+            return FakeTokenizer()
+
+    processor = BartMultiModalProcessor.__new__(BartMultiModalProcessor)
+    processor.info = FakeInfo()
+
+    out = processor._call_hf_processor(
+        [7, 8, 9],
+        {"texts": ["encoder text"]},
+        {},
+        {},
+    )
+
+    assert torch.equal(out["encoder_input_ids"], torch.tensor([[11, 12, 13]]))
+    assert torch.equal(out["input_ids"], torch.tensor([[7, 8, 9]]))