fix agent module issues

Agent_module/carbon_data/__init__.py

@@ -19,7 +19,6 @@
 from .embedders import BaseEmbedder, Embedder, LambdaEmbedder, NullEmbedder
 from .models import (
     Chunk,
-    Embedding,
     Entity,
     MemoryHit,
     MemoryItem,
@@ -55,7 +54,6 @@ def create(path: PathLike, *, exist_ok: bool = False) -> CarbonStore:
     "CarbonStore",
     "Entity",
     "Chunk",
-    "Embedding",
     "Relation",
     "MemoryItem",
     "MemoryHit",

Agent_module/carbon_data/chunkers/__init__.py

@@ -56,12 +56,8 @@ def _chunk(text: str) -> list[str]:
     return _chunk
 
 
-def by_paragraph(*, min_chars: int = 0) -> Chunker:
-    """
-    Split on runs of blank lines. Drops any chunk shorter than
-    ``min_chars`` after stripping; set to 0 to keep every paragraph.
-    """
-    splitter = re.compile(r"\n\s*\n")
+def _regex_chunker(splitter: "re.Pattern[str]", min_chars: int) -> Chunker:
+    """Split on ``splitter``; keep stripped pieces of at least ``min_chars``."""
 
     def _chunk(text: str) -> list[str]:
         return [
@@ -73,22 +69,21 @@ def _chunk(text: str) -> list[str]:
     return _chunk
 
 
+def by_paragraph(*, min_chars: int = 0) -> Chunker:
+    """
+    Split on runs of blank lines. Drops any chunk shorter than
+    ``min_chars`` after stripping; set to 0 to keep every paragraph.
+    """
+    return _regex_chunker(re.compile(r"\n\s*\n"), min_chars)
+
+
 def by_sentence(*, min_chars: int = 0) -> Chunker:
     """
     Naive sentence splitter: end-of-sentence punctuation followed by
     whitespace and a capital letter. Good enough for prose; for legal/
     technical text use ``by_tokens`` or a real tokenizer.
     """
-    splitter = re.compile(r"(?<=[.!?])\s+(?=[A-Z])")
-
-    def _chunk(text: str) -> list[str]:
-        return [
-            s.strip()
-            for s in splitter.split(text)
-            if s.strip() and len(s.strip()) >= min_chars
-        ]
-
-    return _chunk
+    return _regex_chunker(re.compile(r"(?<=[.!?])\s+(?=[A-Z])"), min_chars)
 
 
 __all__ = ["Chunker", "by_tokens", "by_paragraph", "by_sentence"]

Agent_module/carbon_data/models.py

@@ -4,16 +4,8 @@
 from __future__ import annotations
 
 import time
-import uuid
 from dataclasses import dataclass, field
-from typing import TYPE_CHECKING, Any, Optional
-
-if TYPE_CHECKING:
-    import numpy as np
-
-
-def _uuid() -> str:
-    return str(uuid.uuid4())
+from typing import Any, Optional
 
 
 @dataclass
@@ -39,14 +31,6 @@ class Chunk:
     token_count: Optional[int] = None
 
 
-@dataclass
-class Embedding:
-    chunk_id: str
-    model: str
-    dim: int
-    vector: "np.ndarray"
-
-
 @dataclass
 class Relation:
     src: str
@@ -121,7 +105,6 @@ class ValidationReport:
 __all__ = [
     "Entity",
     "Chunk",
-    "Embedding",
     "Relation",
     "MemoryItem",
     "SearchHit",