verbatim-core 0.1.9__tar.gz

verbatim_core-0.1.9/.gitignore

@@ -0,0 +1,218 @@
+# Byte-compiled / optimized / DLL files
+__pycache__/
+*.py[cod]
+*$py.class
+
+# C extensions
+*.so
+
+# Distribution / packaging
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+share/python-wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+MANIFEST
+
+# PyInstaller
+#  Usually these files are written by a python script from a template
+#  before PyInstaller builds the exe, so as to inject date/other infos into it.
+*.manifest
+*.spec
+
+# Installer logs
+pip-log.txt
+pip-delete-this-directory.txt
+
+# Unit test / coverage reports
+htmlcov/
+.tox/
+.nox/
+.coverage
+.coverage.*
+.cache
+nosetests.xml
+coverage.xml
+*.cover
+*.py,cover
+.hypothesis/
+.pytest_cache/
+cover/
+
+# Translations
+*.mo
+*.pot
+
+# Django stuff:
+*.log
+local_settings.py
+db.sqlite3
+db.sqlite3-journal
+
+# Flask stuff:
+instance/
+.webassets-cache
+
+# Scrapy stuff:
+.scrapy
+
+# Sphinx documentation
+docs/_build/
+
+# PyBuilder
+.pybuilder/
+target/
+
+# Jupyter Notebook
+.ipynb_checkpoints
+
+# IPython
+profile_default/
+ipython_config.py
+
+# pyenv
+#   For a library or package, you might want to ignore these files since the code is
+#   intended to run in multiple environments; otherwise, check them in:
+# .python-version
+
+# pipenv
+#   According to pypa/pipenv#598, it is recommended to include Pipfile.lock in version control.
+#   However, in case of collaboration, if having platform-specific dependencies or dependencies
+#   having no cross-platform support, pipenv may install dependencies that don't work, or not
+#   install all needed dependencies.
+#Pipfile.lock
+
+# UV
+#   Similar to Pipfile.lock, it is generally recommended to include uv.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#uv.lock
+
+# poetry
+#   Similar to Pipfile.lock, it is generally recommended to include poetry.lock in version control.
+#   This is especially recommended for binary packages to ensure reproducibility, and is more
+#   commonly ignored for libraries.
+#   https://python-poetry.org/docs/basic-usage/#commit-your-poetrylock-file-to-version-control
+#poetry.lock
+
+# pdm
+#   Similar to Pipfile.lock, it is generally recommended to include pdm.lock in version control.
+#pdm.lock
+#   pdm stores project-wide configurations in .pdm.toml, but it is recommended to not include it
+#   in version control.
+#   https://pdm.fming.dev/latest/usage/project/#working-with-version-control
+.pdm.toml
+.pdm-python
+.pdm-build/
+
+# PEP 582; used by e.g. github.com/David-OConnor/pyflow and github.com/pdm-project/pdm
+__pypackages__/
+
+# Celery stuff
+celerybeat-schedule
+celerybeat.pid
+
+# SageMath parsed files
+*.sage.py
+
+# Environments
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+
+# Spyder project settings
+.spyderproject
+.spyproject
+
+# Rope project settings
+.ropeproject
+
+# mkdocs documentation
+/site
+
+# mypy
+.mypy_cache/
+.dmypy.json
+dmypy.json
+
+# Pyre type checker
+.pyre/
+
+# pytype static type analyzer
+.pytype/
+
+# Cython debug symbols
+cython_debug/
+
+# PyCharm
+#  JetBrains specific template is maintained in a separate JetBrains.gitignore that can
+#  be found at https://github.com/github/gitignore/blob/main/Global/JetBrains.gitignore
+#  and can be added to the global gitignore or merged into this file.  For a more nuclear
+#  option (not recommended) you can uncomment the following to ignore the entire idea folder.
+#.idea/
+
+# PyPI configuration file
+.pypirc
+
+documents/
+index/
+templates.json
+ui/
+.ruff_cache/
+
+data/
+.qodo
+
+# Frontend / Node.js
+frontend/node_modules/
+frontend/.env
+frontend/.env.local
+frontend/.env.development.local
+frontend/.env.test.local
+frontend/.env.production.local
+frontend/build/
+frontend/dist/
+frontend/.next/
+frontend/.nuxt/
+frontend/.vite/
+frontend/npm-debug.log*
+frontend/yarn-debug.log*
+frontend/yarn-error.log*
+frontend/pnpm-debug.log*
+frontend/lerna-debug.log*
+
+# IDEs and editors
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+
+# OS generated files
+.DS_Store
+.DS_Store?
+._*
+.Spotlight-V100
+.Trashes
+ehthumbs.db
+Thumbs.db
+
+
+mockup/
+research-copilot/

verbatim_core-0.1.9/PKG-INFO

@@ -0,0 +1,84 @@
+Metadata-Version: 2.4
+Name: verbatim-core
+Version: 0.1.9
+Summary: Lightweight verbatim span extraction -- the RAG-agnostic core of verbatim-rag
+Project-URL: Homepage, https://github.com/krlabsorg/verbatim-rag
+Project-URL: Bug Tracker, https://github.com/krlabsorg/verbatim-rag/issues
+Author-email: Adam Kovacs <kovacs@krlabs.eu>
+License: MIT
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Operating System :: OS Independent
+Classifier: Programming Language :: Python :: 3
+Requires-Python: >=3.10
+Requires-Dist: openai>=1.3.0
+Requires-Dist: pydantic>=2.0.0
+Provides-Extra: model
+Requires-Dist: scikit-learn==1.6.1; extra == 'model'
+Requires-Dist: torch>=2.6.0; extra == 'model'
+Requires-Dist: transformers==4.53.3; extra == 'model'
+Description-Content-Type: text/markdown
+
+# verbatim-core
+
+Lightweight verbatim span extraction -- the RAG-agnostic core of [verbatim-rag](https://github.com/KRLabsOrg/verbatim-rag).
+
+Extract exact, verbatim text spans from documents that answer a question. No vector databases, no embeddings, no heavy ML dependencies -- just `openai` and `pydantic`.
+
+## Installation
+
+```bash
+pip install verbatim-core
+```
+
+## Quick Start
+
+```python
+from verbatim_core import VerbatimTransform
+
+vt = VerbatimTransform()
+response = vt.transform(
+    question="What is the main finding?",
+    context=[
+        {"content": "The study found that X leads to Y.", "title": "Paper A"},
+        {"content": "Results show Z is statistically significant.", "title": "Paper B"},
+    ],
+)
+
+print(response.answer)
+
+# Access individual highlights and citations
+for doc in response.documents:
+    for highlight in doc.highlights:
+        print(f"  [{highlight.start}:{highlight.end}] {highlight.text}")
+```
+
+## What This Package Includes
+
+- **VerbatimTransform** -- question + context -> cited, grounded answer
+- **LLMSpanExtractor** -- extract verbatim spans using an LLM
+- **LLMClient** -- unified OpenAI API wrapper (sync + async)
+- **TemplateManager** -- response formatting with multiple template strategies
+- **@verbatim_enhance** -- decorator to enhance existing RAG functions
+- **CLI** (`verbatim-enhance`) -- batch processing from the command line
+
+## Model-Based Extraction
+
+For ModernBERT or Zilliz semantic highlight extractors (adds torch, transformers):
+
+```bash
+pip install verbatim-core[model]
+```
+
+## Environment
+
+```bash
+export OPENAI_API_KEY=your_api_key_here
+```
+
+## Full RAG System
+
+For the complete RAG pipeline with vector indexing, embeddings, and document processing, install the full package:
+
+```bash
+pip install verbatim-rag
+```

verbatim_core-0.1.9/README.md

@@ -0,0 +1,64 @@
+# verbatim-core
+
+Lightweight verbatim span extraction -- the RAG-agnostic core of [verbatim-rag](https://github.com/KRLabsOrg/verbatim-rag).
+
+Extract exact, verbatim text spans from documents that answer a question. No vector databases, no embeddings, no heavy ML dependencies -- just `openai` and `pydantic`.
+
+## Installation
+
+```bash
+pip install verbatim-core
+```
+
+## Quick Start
+
+```python
+from verbatim_core import VerbatimTransform
+
+vt = VerbatimTransform()
+response = vt.transform(
+    question="What is the main finding?",
+    context=[
+        {"content": "The study found that X leads to Y.", "title": "Paper A"},
+        {"content": "Results show Z is statistically significant.", "title": "Paper B"},
+    ],
+)
+
+print(response.answer)
+
+# Access individual highlights and citations
+for doc in response.documents:
+    for highlight in doc.highlights:
+        print(f"  [{highlight.start}:{highlight.end}] {highlight.text}")
+```
+
+## What This Package Includes
+
+- **VerbatimTransform** -- question + context -> cited, grounded answer
+- **LLMSpanExtractor** -- extract verbatim spans using an LLM
+- **LLMClient** -- unified OpenAI API wrapper (sync + async)
+- **TemplateManager** -- response formatting with multiple template strategies
+- **@verbatim_enhance** -- decorator to enhance existing RAG functions
+- **CLI** (`verbatim-enhance`) -- batch processing from the command line
+
+## Model-Based Extraction
+
+For ModernBERT or Zilliz semantic highlight extractors (adds torch, transformers):
+
+```bash
+pip install verbatim-core[model]
+```
+
+## Environment
+
+```bash
+export OPENAI_API_KEY=your_api_key_here
+```
+
+## Full RAG System
+
+For the complete RAG pipeline with vector indexing, embeddings, and document processing, install the full package:
+
+```bash
+pip install verbatim-rag
+```

verbatim_core-0.1.9/pyproject.toml

@@ -0,0 +1,40 @@
+[build-system]
+requires = ["hatchling"]
+build-backend = "hatchling.build"
+
+[project]
+name = "verbatim-core"
+version = "0.1.9"
+description = "Lightweight verbatim span extraction -- the RAG-agnostic core of verbatim-rag"
+readme = "README.md"
+requires-python = ">=3.10"
+license = {text = "MIT"}
+authors = [
+    {name = "Adam Kovacs", email = "kovacs@krlabs.eu"},
+]
+classifiers = [
+    "Programming Language :: Python :: 3",
+    "License :: OSI Approved :: MIT License",
+    "Operating System :: OS Independent",
+]
+dependencies = [
+    "openai>=1.3.0",
+    "pydantic>=2.0.0",
+]
+
+[project.optional-dependencies]
+model = [
+    "torch>=2.6.0",
+    "transformers==4.53.3",
+    "scikit-learn==1.6.1",
+]
+
+[project.urls]
+"Homepage" = "https://github.com/krlabsorg/verbatim-rag"
+"Bug Tracker" = "https://github.com/krlabsorg/verbatim-rag/issues"
+
+[project.scripts]
+verbatim-enhance = "verbatim_core.cli:main"
+
+[tool.hatch.build.targets.wheel]
+packages = ["verbatim_core"]

verbatim_core-0.1.9/verbatim_core/__init__.py

@@ -0,0 +1,24 @@
+"""
+Lightweight, RAG-agnostic core for verbatim transformation.
+
+This subpackage provides:
+- VerbatimTransform: question + context -> cited, grounded answer (sync/async)
+- RAGProvider (interface): minimal retrieval contract (no index/Milvus deps)
+- UniversalDocument: simple context container
+- @verbatim_enhance decorator: drop-in enhancement for existing RAG fns
+- CLI: `verbatim-enhance` for batch processing JSON(L)
+"""
+
+from .enhance import verbatim_enhance
+from .providers import RAGProvider
+from .transform import VerbatimTransform, verbatim_query, verbatim_query_async
+from .universal_document import UniversalDocument
+
+__all__ = [
+    "VerbatimTransform",
+    "verbatim_query",
+    "verbatim_query_async",
+    "RAGProvider",
+    "UniversalDocument",
+    "verbatim_enhance",
+]

verbatim_core-0.1.9/verbatim_core/cli.py

@@ -0,0 +1,71 @@
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from typing import Any, Dict, Iterable, List
+
+from .transform import VerbatimTransform
+
+
+def _iter_records(fp) -> Iterable[Dict[str, Any]]:
+    """Yield JSON objects from a file path or stdin. Supports JSONL or JSON array."""
+    try:
+        data = fp.read()
+    except Exception as e:
+        print(f"Error reading input: {e}", file=sys.stderr)
+        return
+    data = (data or "").strip()
+    if not data:
+        return
+    # Try JSONL first
+    if "\n" in data and not data.startswith("["):
+        for line in data.splitlines():
+            line = line.strip()
+            if not line:
+                continue
+            try:
+                yield json.loads(line)
+            except Exception as e:
+                print(f"Skipping malformed JSONL line: {e}", file=sys.stderr)
+    else:
+        try:
+            obj = json.loads(data)
+        except Exception as e:
+            print(f"Malformed JSON input: {e}", file=sys.stderr)
+            return
+        if isinstance(obj, list):
+            for item in obj:
+                yield item
+        else:
+            yield obj
+
+
+def main(argv: List[str] | None = None) -> int:
+    p = argparse.ArgumentParser(description="Verbatim enhance JSON(L) records")
+    p.add_argument("--input", "-i", help="Input file (JSON or JSONL). Defaults to stdin.")
+    p.add_argument("--output", "-o", help="Output file (JSONL). Defaults to stdout.")
+    p.add_argument("--max-spans", type=int, default=5, help="Max display spans")
+    args = p.parse_args(argv)
+
+    fin = open(args.input, "r", encoding="utf-8") if args.input else sys.stdin
+    fout = open(args.output, "w", encoding="utf-8") if args.output else sys.stdout
+
+    vt = VerbatimTransform(max_display_spans=args.max_spans)
+
+    for rec in _iter_records(fin) or []:
+        question = rec.get("question") or ""
+        context = rec.get("context") or rec.get("sources") or []
+        answer = rec.get("answer")
+        resp = vt.transform(question=question, context=context, answer=answer)
+        fout.write(json.dumps(resp.model_dump()) + "\n")
+
+    if fin is not sys.stdin:
+        fin.close()
+    if fout is not sys.stdout:
+        fout.close()
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

verbatim_core-0.1.9/verbatim_core/enhance.py

@@ -0,0 +1,76 @@
+from __future__ import annotations
+
+from functools import wraps
+from typing import Any, Callable, Dict, Iterable, Mapping
+
+from .transform import VerbatimTransform
+
+
+def _to_context_dicts(obj: Any) -> list[dict]:
+    """Best-effort conversion of sources/context to context dicts."""
+    ctx: list[dict] = []
+    if obj is None:
+        return ctx
+    if isinstance(obj, Mapping):
+        # Single dict
+        data = dict(obj)
+        if "content" in data or "text" in data:
+            ctx.append(
+                {
+                    "content": data.get("content") or data.get("text"),
+                    "title": data.get("title", ""),
+                    "source": data.get("source", ""),
+                    "metadata": data.get("metadata") or {},
+                }
+            )
+        return ctx
+    if isinstance(obj, (list, tuple)):
+        for item in obj:
+            ctx.extend(_to_context_dicts(item))
+        return ctx
+    # Fallback: treat as raw text
+    if isinstance(obj, str) and obj.strip():
+        ctx.append({"content": obj})
+    return ctx
+
+
+def verbatim_enhance(
+    max_display_spans: int = 5,
+) -> Callable[[Callable[..., Any]], Callable[..., Any]]:
+    """Decorator to enhance existing RAG functions with verbatim transformation.
+
+    The wrapped function can return:
+    - dict with keys: 'question'?, 'answer'?, 'context' or 'sources'
+    - tuple: (answer, sources)
+    - or just 'sources' (context list/dict) if no answer
+    Currently, provided 'answer' is ignored; verbatim answer is derived from context.
+    """
+
+    def decorator(fn: Callable[..., Any]) -> Callable[..., Any]:
+        @wraps(fn)
+        def wrapper(*args, **kwargs):
+            result = fn(*args, **kwargs)
+            question = kwargs.get("question") or (args[0] if args else "")
+            answer = None
+            context: Iterable[Dict[str, Any]] = []
+
+            if isinstance(result, dict):
+                answer = result.get("answer")
+                context = result.get("context") or result.get("sources") or []
+            elif isinstance(result, (list, tuple)):
+                if len(result) == 2:
+                    answer, context = result
+                else:
+                    context = result
+            else:
+                context = result
+
+            context_dicts = _to_context_dicts(context)
+
+            vt = VerbatimTransform(max_display_spans=max_display_spans)
+            resp = vt.transform(question=question or "", context=context_dicts, answer=answer)
+            return resp
+
+        return wrapper
+
+    return decorator