langroid 0.35.1__tar.gz → 0.36.1__tar.gz

{langroid-0.35.1 → langroid-0.36.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: langroid
-Version: 0.35.1
+Version: 0.36.1
 Summary: Harness LLMs with Multi-Agent Programming
 Author-email: Prasad Chalasani <pchalasani@gmail.com>
 License: MIT
@@ -75,6 +75,7 @@ Requires-Dist: sqlalchemy<3.0.0,>=2.0.19; extra == 'all'
 Requires-Dist: torch<3.0.0,>=2.0.0; extra == 'all'
 Requires-Dist: transformers<5.0.0,>=4.40.1; extra == 'all'
 Requires-Dist: unstructured[docx,pdf,pptx]<0.10.18,>=0.10.16; extra == 'all'
+Requires-Dist: weaviate-client>=4.9.6; extra == 'all'
 Provides-Extra: arango
 Requires-Dist: arango-datasets<2.0.0,>=1.2.2; extra == 'arango'
 Requires-Dist: python-arango<9.0.0,>=8.1.2; extra == 'arango'
@@ -148,6 +149,9 @@ Requires-Dist: chromadb<=0.4.23,>=0.4.21; extra == 'vecdbs'
 Requires-Dist: lancedb<0.9.0,>=0.8.2; extra == 'vecdbs'
 Requires-Dist: pyarrow<16.0.0,>=15.0.0; extra == 'vecdbs'
 Requires-Dist: tantivy<0.22.0,>=0.21.0; extra == 'vecdbs'
+Requires-Dist: weaviate-client>=4.9.6; extra == 'vecdbs'
+Provides-Extra: weaviate
+Requires-Dist: weaviate-client>=4.9.6; extra == 'weaviate'
 Description-Content-Type: text/markdown
 
 <div align="center">
@@ -288,20 +292,28 @@ teacher_task.run()
 <summary> <b>Click to expand</b></summary>
 
 - **Jan 2025:**
-  - [0.33.0](https://github.com/langroid/langroid/releases/tag/0.33.3) Move from Poetry to uv!
+  - [0.36.0](https://github.com/langroid/langroid/releases/tag/0.36.0): Weaviate vector-db support (thanks @abab-dev).
+  - [0.35.0](https://github.com/langroid/langroid/releases/tag/0.35.0): Capture/Stream reasoning content from 
+    Reasoning LLMs (e.g. DeepSeek, OpenAI o1) in addition to final answer.
+  - [0.34.0](https://github.com/langroid/langroid/releases/tag/0.34.0): DocChatAgent 
+    chunk enrichment to improve retrieval. (collaboration with @dfm88). 
+  - [0.33.0](https://github.com/langroid/langroid/releases/tag/0.33.3) Move from Poetry to uv! (thanks @abab-dev).
   - [0.32.0](https://github.com/langroid/langroid/releases/tag/0.32.0) DeepSeek v3 support.
 - **Dec 2024:**
   - [0.31.0](https://github.com/langroid/langroid/releases/tag/0.31.0) Azure OpenAI Embeddings
-  - [0.30.0](https://github.com/langroid/langroid/releases/tag/0.30.0) Llama-cpp embeddings.
-  - [0.29.0](https://github.com/langroid/langroid/releases/tag/0.29.0) Custom Azure OpenAI Client 
+  - [0.30.0](https://github.com/langroid/langroid/releases/tag/0.30.0) Llama-cpp embeddings (thanks @Kwigg).
+  - [0.29.0](https://github.com/langroid/langroid/releases/tag/0.29.0) Custom Azure OpenAI Client (thanks 
+    @johannestang).
   - [0.28.0](https://github.com/langroid/langroid/releases/tag/0.28.0) `ToolMessage`: `_handler` field to override 
-default handler method name in `request` field. 
+default handler method name in `request` field (thanks @alexagr).
   - [0.27.0](https://github.com/langroid/langroid/releases/tag/0.27.0) OpenRouter Support.
   - [0.26.0](https://github.com/langroid/langroid/releases/tag/0.26.0) Update to latest Chainlit.
-  - [0.25.0](https://github.com/langroid/langroid/releases/tag/0.25.0) True Async Methods for agent and user-response.
+  - [0.25.0](https://github.com/langroid/langroid/releases/tag/0.25.0) True Async Methods for agent and 
+    user-response (thanks @alexagr).
 - **Nov 2024:**
   - **[0.24.0](https://langroid.github.io/langroid/notes/structured-output/)**: 
      Enables support for `Agent`s with strict JSON schema output format on compatible LLMs and strict mode for the OpenAI tools API.
+    (thanks @nilspalumbo).
   - **[0.23.0](https://langroid.github.io/langroid/tutorials/local-llm-setup/#local-llms-hosted-on-glhfchat)**: 
       support for LLMs (e.g. `Qwen2.5-Coder-32b-Instruct`) hosted on glhf.chat 
   - **[0.22.0](https://langroid.github.io/langroid/notes/large-tool-results/)**: 

{langroid-0.35.1 → langroid-0.36.1}/README.md

@@ -136,20 +136,28 @@ teacher_task.run()
 <summary> <b>Click to expand</b></summary>
 
 - **Jan 2025:**
-  - [0.33.0](https://github.com/langroid/langroid/releases/tag/0.33.3) Move from Poetry to uv!
+  - [0.36.0](https://github.com/langroid/langroid/releases/tag/0.36.0): Weaviate vector-db support (thanks @abab-dev).
+  - [0.35.0](https://github.com/langroid/langroid/releases/tag/0.35.0): Capture/Stream reasoning content from 
+    Reasoning LLMs (e.g. DeepSeek, OpenAI o1) in addition to final answer.
+  - [0.34.0](https://github.com/langroid/langroid/releases/tag/0.34.0): DocChatAgent 
+    chunk enrichment to improve retrieval. (collaboration with @dfm88). 
+  - [0.33.0](https://github.com/langroid/langroid/releases/tag/0.33.3) Move from Poetry to uv! (thanks @abab-dev).
   - [0.32.0](https://github.com/langroid/langroid/releases/tag/0.32.0) DeepSeek v3 support.
 - **Dec 2024:**
   - [0.31.0](https://github.com/langroid/langroid/releases/tag/0.31.0) Azure OpenAI Embeddings
-  - [0.30.0](https://github.com/langroid/langroid/releases/tag/0.30.0) Llama-cpp embeddings.
-  - [0.29.0](https://github.com/langroid/langroid/releases/tag/0.29.0) Custom Azure OpenAI Client 
+  - [0.30.0](https://github.com/langroid/langroid/releases/tag/0.30.0) Llama-cpp embeddings (thanks @Kwigg).
+  - [0.29.0](https://github.com/langroid/langroid/releases/tag/0.29.0) Custom Azure OpenAI Client (thanks 
+    @johannestang).
   - [0.28.0](https://github.com/langroid/langroid/releases/tag/0.28.0) `ToolMessage`: `_handler` field to override 
-default handler method name in `request` field. 
+default handler method name in `request` field (thanks @alexagr).
   - [0.27.0](https://github.com/langroid/langroid/releases/tag/0.27.0) OpenRouter Support.
   - [0.26.0](https://github.com/langroid/langroid/releases/tag/0.26.0) Update to latest Chainlit.
-  - [0.25.0](https://github.com/langroid/langroid/releases/tag/0.25.0) True Async Methods for agent and user-response.
+  - [0.25.0](https://github.com/langroid/langroid/releases/tag/0.25.0) True Async Methods for agent and 
+    user-response (thanks @alexagr).
 - **Nov 2024:**
   - **[0.24.0](https://langroid.github.io/langroid/notes/structured-output/)**: 
      Enables support for `Agent`s with strict JSON schema output format on compatible LLMs and strict mode for the OpenAI tools API.
+    (thanks @nilspalumbo).
   - **[0.23.0](https://langroid.github.io/langroid/tutorials/local-llm-setup/#local-llms-hosted-on-glhfchat)**: 
       support for LLMs (e.g. `Qwen2.5-Coder-32b-Instruct`) hosted on glhf.chat 
   - **[0.22.0](https://langroid.github.io/langroid/notes/large-tool-results/)**: 

{langroid-0.35.1 → langroid-0.36.1}/langroid/agent/special/doc_chat_agent.py

@@ -15,6 +15,7 @@ pip install "langroid[hf-embeddings]"
 """
 
 import logging
+import textwrap
 from collections import OrderedDict
 from functools import cache
 from typing import Any, Callable, Dict, List, Optional, Set, Tuple, no_type_check
@@ -81,7 +82,7 @@ You will be given various passages from these documents, and asked to answer que
 about them, or summarize them into coherent answers.
 """
 
-CHUNK_ENRICHMENT_DELIMITER = "<##-##-##>"
+CHUNK_ENRICHMENT_DELIMITER = "\n<##-##-##>"
 
 has_sentence_transformers = False
 try:
@@ -810,9 +811,11 @@ class DocChatAgent(ChatAgent):
         return "\n".join(
             [
                 f"""
-                [{i+1}]
+                -----[EXTRACT #{i+1}]----------
                 {content}
                 {source}
+                -----END OF EXTRACT------------
+                
                 """
                 for i, (content, source) in enumerate(zip(contents, sources))
             ]
@@ -949,12 +952,13 @@ class DocChatAgent(ChatAgent):
                     continue
 
                 # Combine original content with questions in a structured way
-                combined_content = f"""
-                {doc.content}
-                
+                combined_content = textwrap.dedent(
+                    f"""\
+                {doc.content}                
                 {enrichment_config.delimiter}
                 {enrichment}
-                """.strip()
+                """
+                )
 
                 new_doc = doc.copy(
                     update={
@@ -1440,7 +1444,7 @@ class DocChatAgent(ChatAgent):
         delimiter = self.config.chunk_enrichment_config.delimiter
         return [
             (
-                doc.copy(update={"content": doc.content.split(delimiter)[0].strip()})
+                doc.copy(update={"content": doc.content.split(delimiter)[0]})
                 if doc.content and getattr(doc.metadata, "has_enrichment", False)
                 else doc
             )

{langroid-0.35.1 → langroid-0.36.1}/langroid/parsing/parser.py

@@ -267,9 +267,11 @@ class Parser:
                 # Truncate the chunk text at the punctuation mark
                 chunk_text = chunk_text[: last_punctuation + 1]
 
-            # Remove any newline characters and strip any leading or
-            # trailing whitespace
-            chunk_text_to_append = re.sub(r"\n{2,}", "\n", chunk_text).strip()
+            # Replace redundant (3 or more) newlines with 2 newlines to preser
+            # paragraph separation!
+            # But do NOT strip leading/trailing whitespace, to preserve formatting
+            # (e.g. code blocks, or in case we want to stitch chunks back together)
+            chunk_text_to_append = re.sub(r"\n{3,}", "\n\n", chunk_text)
 
             if len(chunk_text_to_append) > self.config.discard_chunk_chars:
                 # Append the chunk text to the list of chunks

{langroid-0.35.1 → langroid-0.36.1}/langroid/parsing/utils.py

@@ -310,9 +310,9 @@ def extract_numbered_segments(s: str, specs: str) -> str:
         ]
 
         # If we extracted any segments from this paragraph,
-        # join them and append to results
+        # join them with ellipsis (...) and append to results.
         if extracted_segments:
-            extracted_paragraphs.append(" ".join(extracted_segments))
+            extracted_paragraphs.append("...".join(extracted_segments))
 
     return "\n\n".join(extracted_paragraphs)
 

langroid-0.36.1/langroid/utils/output/citations.py

@@ -0,0 +1,61 @@
+def extract_markdown_references(md_string: str) -> list[int]:
+    """
+    Extracts markdown references (e.g., [^1], [^2]) from a string and returns
+    them as a sorted list of integers.
+
+    Args:
+        md_string (str): The markdown string containing references.
+
+    Returns:
+        list[int]: A sorted list of unique integers from the markdown references.
+    """
+    import re
+
+    # Regex to find all occurrences of [^<number>]
+    matches = re.findall(r"\[\^(\d+)\]", md_string)
+    # Convert matches to integers, remove duplicates with set, and sort
+    return sorted(set(int(match) for match in matches))
+
+
+def format_footnote_text(content: str, width: int = 0) -> str:
+    """
+    Formats the content so that each original line is individually processed.
+    - If width=0, no wrapping is done (lines remain as is).
+    - If width>0, lines are wrapped to that width.
+    - Blank lines remain blank (with indentation).
+    - Everything is indented by 4 spaces (for markdown footnotes).
+
+    Args:
+        content (str): The text of the footnote to be formatted.
+        width (int): Maximum width of the text lines. If 0, lines are not wrapped.
+
+    Returns:
+        str: Properly formatted markdown footnote text.
+    """
+    import textwrap
+
+    indent = "    "  # 4 spaces for markdown footnotes
+    lines = content.split("\n")  # keep original line structure
+
+    output_lines = []
+    for line in lines:
+        # If the line is empty (or just spaces), keep it blank (but indented)
+        if not line.strip():
+            output_lines.append(indent)
+            continue
+
+        if width > 0:
+            # Wrap each non-empty line to the specified width
+            wrapped = textwrap.wrap(line, width=width)
+            if not wrapped:
+                # If textwrap gives nothing, add a blank (indented) line
+                output_lines.append(indent)
+            else:
+                for subline in wrapped:
+                    output_lines.append(indent + subline)
+        else:
+            # No wrapping: just indent the original line
+            output_lines.append(indent + line)
+
+    # Join them with newline so we preserve the paragraph/blank line structure
+    return "\n".join(output_lines)

{langroid-0.35.1 → langroid-0.36.1}/langroid/vector_store/__init__.py

@@ -48,3 +48,14 @@ try:
     __all__.extend(["chromadb", "ChromaDBConfig", "ChromaDB"])
 except ImportError:
     pass
+
+try:
+    from . import weaviatedb
+    from .weaviatedb import WeaviateDBConfig, WeaviateDB
+
+    weaviatedb
+    WeaviateDB
+    WeaviateDBConfig
+    __all__.extend(["weaviatedb", "WeaviateDB", "WeaviateDBConfig"])
+except ImportError:
+    pass

{langroid-0.35.1 → langroid-0.36.1}/langroid/vector_store/base.py

@@ -59,6 +59,7 @@ class VectorStore(ABC):
         from langroid.vector_store.meilisearch import MeiliSearch, MeiliSearchConfig
         from langroid.vector_store.momento import MomentoVI, MomentoVIConfig
         from langroid.vector_store.qdrantdb import QdrantDB, QdrantDBConfig
+        from langroid.vector_store.weaviatedb import WeaviateDB, WeaviateDBConfig
 
         if isinstance(config, QdrantDBConfig):
             return QdrantDB(config)
@@ -70,6 +71,8 @@ class VectorStore(ABC):
             return LanceDB(config)
         elif isinstance(config, MeiliSearchConfig):
             return MeiliSearch(config)
+        elif isinstance(config, WeaviateDBConfig):
+            return WeaviateDB(config)
 
         else:
             logger.warning(
@@ -261,7 +264,7 @@ class VectorStore(ABC):
             metadata = copy.deepcopy(id2metadata[w[0]])
             metadata.window_ids = w
             document = Document(
-                content=" ".join([d.content for d in self.get_documents_by_ids(w)]),
+                content="".join([d.content for d in self.get_documents_by_ids(w)]),
                 metadata=metadata,
             )
             # make a fresh id since content is in general different

langroid-0.36.1/langroid/vector_store/weaviatedb.py

@@ -0,0 +1,271 @@
+import logging
+import os
+import re
+from typing import Any, List, Optional, Sequence, Tuple
+
+from dotenv import load_dotenv
+
+from langroid.embedding_models.base import (
+    EmbeddingModelsConfig,
+)
+from langroid.embedding_models.models import OpenAIEmbeddingsConfig
+from langroid.exceptions import LangroidImportError
+from langroid.mytypes import DocMetaData, Document, EmbeddingFunction
+from langroid.utils.configuration import settings
+from langroid.vector_store.base import VectorStore, VectorStoreConfig
+
+logger = logging.getLogger(__name__)
+try:
+    import weaviate
+    from weaviate.classes.config import (
+        Configure,
+        VectorDistances,
+    )
+    from weaviate.classes.init import Auth
+    from weaviate.classes.query import Filter, MetadataQuery
+    from weaviate.util import generate_uuid5, get_valid_uuid
+except ImportError:
+    raise LangroidImportError("weaviate", "weaviate")
+
+
+class WeaviateDBConfig(VectorStoreConfig):
+    collection_name: str | None = "temp"
+    embedding: EmbeddingModelsConfig = OpenAIEmbeddingsConfig()
+    distance: str = VectorDistances.COSINE
+
+
+class WeaviateDB(VectorStore):
+    def __init__(self, config: WeaviateDBConfig = WeaviateDBConfig()):
+        super().__init__(config)
+        self.config: WeaviateDBConfig = config
+        self.embedding_fn: EmbeddingFunction = self.embedding_model.embedding_fn()
+        self.embedding_dim = self.embedding_model.embedding_dims
+        load_dotenv()
+        key = os.getenv("WEAVIATE_API_KEY")
+        url = os.getenv("WEAVIATE_API_URL")
+        if None in [key, url]:
+            logger.warning(
+                """WEAVIATE_API_KEY, WEAVIATE_API_URL env variable must be set to use
+                WeaviateDB in cloud mode. Please set these values
+                in your .env file.
+                """
+            )
+        self.client = weaviate.connect_to_weaviate_cloud(
+            cluster_url=url,
+            auth_credentials=Auth.api_key(key),
+        )
+        if config.collection_name is not None:
+            WeaviateDB.validate_and_format_collection_name(config.collection_name)
+
+    def clear_empty_collections(self) -> int:
+        colls = self.client.collections.list_all()
+        n_deletes = 0
+        for coll_name, _ in colls.items():
+            val = self.client.collections.get(coll_name)
+            if len(val) == 0:
+                n_deletes += 1
+                self.client.collections.delete(coll_name)
+        return n_deletes
+
+    def list_collections(self, empty: bool = False) -> List[str]:
+        colls = self.client.collections.list_all()
+        if empty:
+            return list(colls.keys())
+        non_empty_colls = [
+            coll_name
+            for coll_name in colls.keys()
+            if len(self.client.collections.get(coll_name)) > 0
+        ]
+
+        return non_empty_colls
+
+    def clear_all_collections(self, really: bool = False, prefix: str = "") -> int:
+        if not really:
+            logger.warning(
+                "Not really deleting all collections ,set really=True to confirm"
+            )
+            return 0
+        coll_names = [
+            c for c in self.list_collections(empty=True) if c.startswith(prefix)
+        ]
+        if len(coll_names) == 0:
+            logger.warning(f"No collections found with prefix {prefix}")
+            return 0
+        n_empty_deletes = 0
+        n_non_empty_deletes = 0
+        for name in coll_names:
+            info = self.client.collections.get(name)
+            points_count = len(info)
+
+            n_empty_deletes += points_count == 0
+            n_non_empty_deletes += points_count > 0
+            self.client.collections.delete(name)
+        logger.warning(
+            f"""
+            Deleted {n_empty_deletes} empty collections and
+            {n_non_empty_deletes} non-empty collections.
+            """
+        )
+        return n_empty_deletes + n_non_empty_deletes
+
+    def delete_collection(self, collection_name: str) -> None:
+        self.client.collections.delete(name=collection_name)
+
+    def create_collection(self, collection_name: str, replace: bool = False) -> None:
+        collection_name = WeaviateDB.validate_and_format_collection_name(
+            collection_name
+        )
+        self.config.collection_name = collection_name
+        if self.client.collections.exists(name=collection_name):
+            coll = self.client.collections.get(name=collection_name)
+            if len(coll) > 0:
+                logger.warning(f"Non-empty Collection {collection_name} already exists")
+                if not replace:
+                    logger.warning("Not replacing collection")
+                    return
+                else:
+                    logger.warning("Recreating fresh collection")
+            self.client.collections.delete(name=collection_name)
+
+        vector_index_config = Configure.VectorIndex.hnsw(
+            distance_metric=VectorDistances.COSINE,
+        )
+        if self.config.embedding == OpenAIEmbeddingsConfig:
+            vectorizer_config = Configure.Vectorizer.text2vec_openai(
+                model=self.embedding_model
+            )
+        else:
+            vectorizer_config = None
+
+        collection_info = self.client.collections.create(
+            name=collection_name,
+            vector_index_config=vector_index_config,
+            vectorizer_config=vectorizer_config,
+        )
+        collection_info = self.client.collections.get(name=collection_name)
+        assert len(collection_info) in [0, None]
+        if settings.debug:
+            level = logger.getEffectiveLevel()
+            logger.setLevel(logging.INFO)
+            logger.info(collection_info)
+            logger.setLevel(level)
+
+    def add_documents(self, documents: Sequence[Document]) -> None:
+        super().maybe_add_ids(documents)
+        colls = self.list_collections(empty=True)
+        for doc in documents:
+            doc.metadata.id = str(self._create_valid_uuid_id(doc.metadata.id))
+        if len(documents) == 0:
+            return
+
+        document_dicts = [doc.dict() for doc in documents]
+        embedding_vecs = self.embedding_fn([doc.content for doc in documents])
+        if self.config.collection_name is None:
+            raise ValueError("No collection name set, cannot ingest docs")
+        if self.config.collection_name not in colls:
+            self.create_collection(self.config.collection_name, replace=True)
+        coll_name = self.client.collections.get(self.config.collection_name)
+        with coll_name.batch.dynamic() as batch:
+            for i, doc_dict in enumerate(document_dicts):
+                id = doc_dict["metadata"].pop("id", None)
+                batch.add_object(properties=doc_dict, uuid=id, vector=embedding_vecs[i])
+
+    def get_all_documents(self, where: str = "") -> List[Document]:
+        if self.config.collection_name is None:
+            raise ValueError("No collection name set, cannot retrieve docs")
+        # cannot use filter as client does not support json type queries
+        coll = self.client.collections.get(self.config.collection_name)
+        return [self.weaviate_obj_to_doc(item) for item in coll.iterator()]
+
+    def get_documents_by_ids(self, ids: List[str]) -> List[Document]:
+        if self.config.collection_name is None:
+            raise ValueError("No collection name set, cannot retrieve docs")
+
+        docs = []
+        coll_name = self.client.collections.get(self.config.collection_name)
+
+        result = coll_name.query.fetch_objects(
+            filters=Filter.by_property("_id").contains_any(ids), limit=len(coll_name)
+        )
+
+        id_to_doc = {}
+        for item in result.objects:
+            doc = self.weaviate_obj_to_doc(item)
+            id_to_doc[doc.metadata.id] = doc
+
+        # Reconstruct the list of documents in the original order of input ids
+        docs = [id_to_doc[id] for id in ids if id in id_to_doc]
+
+        return docs
+
+    def similar_texts_with_scores(
+        self, text: str, k: int = 1, where: Optional[str] = None
+    ) -> List[Tuple[Document, float]]:
+        embedding = self.embedding_fn([text])[0]
+        if self.config.collection_name is None:
+            raise ValueError("No collections name set,cannot search")
+        coll = self.client.collections.get(self.config.collection_name)
+        response = coll.query.near_vector(
+            near_vector=embedding,
+            limit=k,
+            return_properties=True,
+            return_metadata=MetadataQuery(distance=True),
+        )
+        return [
+            (self.weaviate_obj_to_doc(item), 1 - item.metadata.distance)
+            for item in response.objects
+        ]
+
+    def _create_valid_uuid_id(self, id: str) -> Any:
+        try:
+            id = get_valid_uuid(id)
+            return id
+        except Exception:
+            return generate_uuid5(id)
+
+    def weaviate_obj_to_doc(self, input_object: Any) -> Document:
+        content = input_object.properties.get("content", "")
+        metadata_dict = input_object.properties.get("metadata", {})
+
+        window_ids = metadata_dict.pop("window_ids", [])
+        window_ids = [str(uuid) for uuid in window_ids]
+
+        # Ensure the id is a valid UUID string
+        id_value = get_valid_uuid(input_object.uuid)
+
+        metadata = DocMetaData(id=id_value, window_ids=window_ids, **metadata_dict)
+
+        return Document(content=content, metadata=metadata)
+
+    @staticmethod
+    def validate_and_format_collection_name(name: str) -> str:
+        """
+        Formats the collection name to comply with Weaviate's naming rules:
+        - Name must start with a capital letter.
+        - Name can only contain letters, numbers, and underscores.
+        - Replaces invalid characters with underscores.
+        """
+        if not name:
+            raise ValueError("Collection name cannot be empty.")
+
+        formatted_name = re.sub(r"[^a-zA-Z0-9_]", "_", name)
+
+        # Ensure the first letter is capitalized
+        if not formatted_name[0].isupper():
+            formatted_name = formatted_name.capitalize()
+
+        # Check if the name now meets the criteria
+        if not re.match(r"^[A-Z][A-Za-z0-9_]*$", formatted_name):
+            raise ValueError(
+                f"Invalid collection name '{name}'."
+                " Names must start with a capital letter "
+                "and contain only letters, numbers, and underscores."
+            )
+
+        if formatted_name != name:
+            logger.warning(
+                f"Collection name '{name}' was reformatted to '{formatted_name}' "
+                "to comply with Weaviate's rules."
+            )
+
+        return formatted_name

{langroid-0.35.1 → langroid-0.36.1}/pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "langroid"
-version = "0.35.1"
+version = "0.36.1"
 authors = [
     {name = "Prasad Chalasani", email = "pchalasani@gmail.com"},
 ]
@@ -40,7 +40,7 @@ dependencies = [
     "pygments<3.0.0,>=2.15.1",
     "pyparsing<4.0.0,>=3.0.9",
     "pytest-rerunfailures<16.0,>=15.0",
-    "python-dotenv<2.0.0,>=1.0.0",
+    "python-dotenv>=1.0.0,<2.0.0",
     "python-magic<1.0.0,>=0.4.27",
     "pyyaml<7.0.0,>=6.0.1",
     "qdrant-client<2.0.0,>=1.8.0",
@@ -79,6 +79,7 @@ vecdbs = [
     "tantivy<0.22.0,>=0.21.0",
     "pyarrow<16.0.0,>=15.0.0",
     "chromadb<=0.4.23,>=0.4.21",
+    "weaviate-client>=4.9.6",
 ]
 
 db = [
@@ -103,6 +104,7 @@ all = [
     "transformers<5.0.0,>=4.40.1",
     "huggingface-hub<0.22.0,>=0.21.2",
     "chromadb<=0.4.23,>=0.4.21",
+    "weaviate-client>=4.9.6",
     "metaphor-python<0.2.0,>=0.1.23",
     "neo4j<6.0.0,>=5.14.1",
     "python-arango<9.0.0,>=8.1.2",
@@ -190,6 +192,9 @@ chainlit = [
 chromadb = [
     "chromadb<=0.4.23,>=0.4.21",
 ]
+weaviate = [
+    "weaviate-client>=4.9.6",
+]
 
 meilisearch = [
     "meilisearch-python-sdk<3.0.0,>=2.2.3",

langroid-0.35.1/langroid/utils/output/citations.py

@@ -1,41 +0,0 @@
-def extract_markdown_references(md_string: str) -> list[int]:
-    """
-    Extracts markdown references (e.g., [^1], [^2]) from a string and returns
-    them as a sorted list of integers.
-
-    Args:
-        md_string (str): The markdown string containing references.
-
-    Returns:
-        list[int]: A sorted list of unique integers from the markdown references.
-    """
-    import re
-
-    # Regex to find all occurrences of [^<number>]
-    matches = re.findall(r"\[\^(\d+)\]", md_string)
-    # Convert matches to integers, remove duplicates with set, and sort
-    return sorted(set(int(match) for match in matches))
-
-
-def format_footnote_text(content: str, width: int = 80) -> str:
-    """
-    Formats the content part of a footnote (i.e. not the first line that
-    appears right after the reference [^4])
-    It wraps the text so that no line is longer than the specified width and indents
-    lines as necessary for markdown footnotes.
-
-    Args:
-        content (str): The text of the footnote to be formatted.
-        width (int): Maximum width of the text lines.
-
-    Returns:
-        str: Properly formatted markdown footnote text.
-    """
-    import textwrap
-
-    # Wrap the text to the specified width
-    wrapped_lines = textwrap.wrap(content, width)
-    if len(wrapped_lines) == 0:
-        return ""
-    indent = "    "  # Indentation for markdown footnotes
-    return indent + ("\n" + indent).join(wrapped_lines)