- Removing the Convenient Functions to unify the Class instaniation approach

- Warnings Suppression fix
- Lazy imports for faster loading time of PDFstract CLI
- Adding Logs into Embedding factory
- Docs Update

Author: AKSarav

.gitignore

@@ -15,4 +15,6 @@ ultralytics-cache/*
 results/
 *chunks.json
 .env.example
-uv.lock
+uv.lock
+test.ipynb
+notebooks/

.project_doc_record/meta-info.json

@@ -0,0 +1,7 @@
+{
+  "doc_version": "",
+  "in_generation_process": true,
+  "fake_file_reflection": {},
+  "jump_files": [],
+  "deleted_items_from_older_meta": []
+}

.project_doc_record/project_hierarchy.json

@@ -29,10 +29,10 @@ pip install pdfstract[all]
 ### Convert a PDF
 
 ```python
-from pdfstract import convert_pdf
+from pdfstract import PDFStract
 
-# One-liner conversion
-text = convert_pdf('document.pdf', library='marker')
+pdfstract = PDFStract()
+text = pdfstract.convert('document.pdf', library='marker')
 print(text)
 ```
 
@@ -141,7 +141,7 @@ pdfstract compare sample.pdf -l marker -l docling -l pymupdf4llm
 
 ## Embeddings
 
-PDFStract can generate embeddings using multiple providers (OpenAI, Azure OpenAI, Google Generative, Ollama, Sentence-Transformers, Model2Vec). Use the Python API `embed_text` / `embed_texts` or the CLI `pdfstract embeddings-list` and `pdfstract embed-text` commands. Hosted providers require API keys; local providers require installed models or running services.
+PDFStract can generate embeddings using multiple providers (OpenAI, Azure OpenAI, Google Generative, Ollama, Sentence-Transformers, Model2Vec). Use a `PDFStract()` instance and call `embed_text` / `embed_texts`, or use the CLI `pdfstract embeddings-list` and `pdfstract embed-text` commands. Hosted providers require API keys; local providers require installed models or running services.
 
 ## Web UI & Docker
 

CLI_README.md

@@ -62,7 +62,7 @@ chunks=ps.convert_chunk("path/to/pdf", library="auto", chunker="auto")
 
 ## Embeddings
 
-PDFStract can also generate vector embeddings from text using pluggable providers (OpenAI, Azure OpenAI, Google Generative, Ollama, Sentence-Transformers, Model2Vec). Use `PDFStract.embed_text(s)` or the convenience `embed_text(s)` functions. Hosted providers require API keys (see docs), while local providers like Sentence-Transformers and Ollama run locally.
+PDFStract can also generate vector embeddings from text using pluggable providers (OpenAI, Azure OpenAI, Google Generative, Ollama, Sentence-Transformers, Model2Vec). Use `pdfstract.embed_text(s)` / `pdfstract.embed_texts(...)` on a `PDFStract()` instance. Hosted providers require API keys (see docs), while local providers like Sentence-Transformers and Ollama run locally.
 
 # or do it in two steps
 # convert first with your library of choice
@@ -134,13 +134,13 @@ pdfstract download marker
 
 You don't need to use the CLI! PDFStract can be easily integrated into your Python applications as a library. 
 
-#### Convert a PDF (One-liner)
+#### Convert a PDF
 
 ```python
-from pdfstract import convert_pdf
+from pdfstract import PDFStract
 
-# Quick conversion with default settings
-result = convert_pdf('sample.pdf', library='marker')
+pdfstract = PDFStract()
+result = pdfstract.convert('sample.pdf', library='marker')
 print(result)  # Markdown content
 ```
 
@@ -150,8 +150,6 @@ print(result)  # Markdown content
 from pdfstract import PDFStract
 
 pdfstract = PDFStract()
-
-# Get list of available libraries
 available = pdfstract.list_available_libraries()
 print(available)  # ['pymupdf4llm', 'marker', 'docling', ...]
 ```

README.md

@@ -4,6 +4,17 @@
 Provides: single conversions, multi-library comparisons, batch processing
 """
 
+# Suppress noisy warnings from third-party libraries during CLI runs
+import warnings
+warnings.filterwarnings(
+    "ignore",
+    message=".*urllib3.*or chardet.*doesn't match a supported version.*",
+    module="requests",
+)
+warnings.filterwarnings("ignore", category=UserWarning)
+warnings.filterwarnings("ignore", category=DeprecationWarning)
+warnings.filterwarnings("ignore", category=SyntaxWarning)
+
 import click
 import json
 import os
@@ -12,7 +23,6 @@
 from typing import List, Dict, Optional
 from datetime import datetime
 from concurrent.futures import ThreadPoolExecutor
-import csv
 import sys
 
 from rich.console import Console
@@ -24,7 +34,7 @@
 from services.cli_factory import CLILazyFactory
 from services.base import OutputFormat
 from services.logger import logger
-from services.chunker_factory import get_chunker_factory
+# get_chunker_factory imported lazily inside chunk/convert-chunk/chunkers to avoid loading chonkie on every CLI run
 
 # Import version for --version option (reads from api module which reads from pyproject.toml)
 try:
@@ -789,6 +799,7 @@ def batch_compare(input_dir: Path, libraries: tuple, format: str, output: str, m
 @pdfstract.command()
 def chunkers():
     """List all available text chunkers and their parameters"""
+    from services.chunker_factory import get_chunker_factory
     cli_app.print_banner()
 
     factory = get_chunker_factory()
@@ -880,6 +891,7 @@ def chunk(input_file: Path, chunker: str, chunk_size: int, chunk_overlap: int, o
     cli_app.print_info(f"Chunker: {chunker} | Size: {chunk_size} | Overlap: {chunk_overlap}")
 
     try:
+        from services.chunker_factory import get_chunker_factory
         factory = get_chunker_factory()
 
         with Progress(
@@ -1042,6 +1054,7 @@ def convert_chunk(
             # Step 2: Chunk the converted text
             progress.add_task("Chunking text...", total=None)
 
+            from services.chunker_factory import get_chunker_factory
             factory = get_chunker_factory()
             result = asyncio.run(
                 factory.chunk_with_result(chunker, converted_text, **chunker_params)

cli.py

@@ -183,14 +183,14 @@ print(f"Semantic chunker: {info}")
 Generate vector embeddings for text using pluggable providers.
 
 ```python
-# Embed multiple texts (synchronous)
-from pdfstract import embed_texts
-vectors = embed_texts(["First sentence.", "Second sentence."], model='auto')
+from pdfstract import PDFStract
 
-# Using the PDFStract object (async + sync)
 pdf = PDFStract()
-vecs = pdf.embed_texts(["Hello world"], model='sentence-transformers')
+# Sync
+vecs = pdf.embed_texts(["First sentence.", "Second sentence."], model='auto')
 print(len(vecs[0]))  # embedding dimension
+# Or single text
+e = pdf.embed_text("Hello world", model='sentence-transformers')
 ```
 
 Parameters:
@@ -201,39 +201,7 @@ Notes:
 - Credentials are validated internally and a clear error is raised if required environment variables are missing.
 - For hosted providers set `OPENAI_API_KEY`, `AZURE_OPENAI_KEY`, `GOOGLE_API_KEY`, etc. For Ollama, ensure a local Ollama daemon is running.
 
-## Convenience Functions
-
-PDFStract also provides standalone functions for quick operations:
-
-### PDF Conversion Functions
-
-```python
-from pdfstract import (
-    convert_with_docling,
-    convert_with_marker, 
-    convert_with_pymupdf4llm,
-    convert_with_unstructured
-)
-
-# Use specific converters directly
-text = convert_with_marker('document.pdf')
-text = convert_with_docling('document.pdf', extract_images=True)
-```
-
-### Chunking Functions
-
-```python
-from pdfstract import (
-    chunk_by_tokens,
-    chunk_semantically,
-    chunk_recursively,
-    chunk_by_sentences
-)
-
-# Use specific chunkers directly
-chunks = chunk_semantically(text, chunk_size=512)
-chunks = chunk_by_tokens(text, chunk_size=1024, overlap=100)
-```
+Use a single `PDFStract()` instance for all operations: `pdfstract.convert(...)`, `pdfstract.chunk_text(...)`, `pdfstract.embed_texts(...)`, etc. Instantiate once and reuse.
 
 ## Error Handling
 

docs/docs/api/overview.md

@@ -87,14 +87,12 @@ Set `chunker='auto'` to automatically select the best available method (typicall
 PDFStract can produce vector embeddings using multiple providers. Embeddings are useful for indexing, semantic search, and RAG pipelines.
 
 ```python
-# Embed a list of texts using auto-selected provider
-from pdfstract import embed_texts
-vecs = embed_texts(["First sentence", "Second sentence"], model='auto')
-print(len(vecs[0]))
+from pdfstract import PDFStract
 
-# Embed a single text
-from pdfstract import embed_text
-e = embed_text("Hello world", model='sentence-transformers')
+pdf = PDFStract()
+vecs = pdf.embed_texts(["First sentence", "Second sentence"], model='auto')
+print(len(vecs[0]))
+e = pdf.embed_text("Hello world", model='sentence-transformers')
 ```
 
 Notes:

docs/docs/quick-start.md

@@ -28,7 +28,7 @@
     See services/ for implementation details
 """
 
-from .api import PDFStract, convert_pdf, list_available_libraries, chunk_text, list_available_chunkers, list_available_embeddings, embed_text, embed_texts
+from .api import PDFStract
 
 # Version is managed in pyproject.toml
 # Read it dynamically to keep a single source of truth
@@ -51,13 +51,6 @@
 
 __all__ = [
     "PDFStract",
-    "convert_pdf",
-    "list_available_libraries",
-    "chunk_text",
-    "list_available_chunkers",
-    "list_available_embeddings",
-    "embed_text",
-    "embed_texts",
     "__version__",
 ]