rasgapdf 0.2.3__tar.gz → 0.2.4__tar.gz

{rasgapdf-0.2.3 → rasgapdf-0.2.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rasgapdf
-Version: 0.2.3
+Version: 0.2.4
 Summary: Ultra-fast 2D spatial PDF parser powered by pure Rust — extract exact [x0, y0, x1, y1] bounding boxes, text blocks, tables, images & vector paths in milliseconds for RAG & LLMs
 Author: Messias Martins
 Project-URL: Homepage, https://martinsmessias.github.io/rasgapdf/
@@ -191,6 +191,33 @@ print(manifest["written"], "files written")
 
 ---
 
+---
+
+## 🦜️🔗 AI Framework Integrations (LangChain & LlamaIndex)
+
+### 🦜️ LangChain Document Loader
+`python
+from rasgapdf import RasgaPDFLoader
+
+# 1 Document per page with markdown tables:
+loader = RasgaPDFLoader('document.pdf', mode='pages')
+docs = loader.load()
+
+# Spatial RAG: 1 Document per block/table with exact [x0, y0, x1, y1] coordinates:
+spatial_loader = RasgaPDFLoader('document.pdf', mode='elements')
+chunks = spatial_loader.load()
+for chunk in chunks:
+    print(chunk.metadata['bbox'], chunk.page_content[:40])
+`
+
+### 🦙 LlamaIndex Reader
+`python
+from rasgapdf import RasgaPDFReader
+
+reader = RasgaPDFReader(mode='pages')
+documents = reader.load_data('financial_report.pdf')
+`
+
 ## 📄 License
 
 GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). Copyright (c) 2026 Messias Martins.
@@ -198,3 +225,19 @@ GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). Copyright (
 RasgaPDF embeds and links against [MuPDF](https://mupdf.com/) (Artifex Software),
 which is distributed under the AGPL-3.0. That is why this project is licensed
 under the same terms.
+
+---
+
+## 💼 Commercial & Enterprise Licensing (AGPL-3.0 Waiver)
+
+Need to use **RasgaPDF** in closed-source commercial software, proprietary SaaS platforms, private cloud environments, or corporate applications without releasing your source code under the AGPL-3.0?
+
+A **Commercial / Enterprise License** is available directly from the author:
+
+- **Complete AGPL-3.0 Exemption:** Keep your intellectual property, backend code, and SaaS infrastructure 100% closed-source and proprietary.
+- **Enterprise SLA & Support:** Direct architecture advisory, guaranteed response times, and priority bug resolution.
+- **On-Premise & Air-Gapped Appliances:** Ready-to-deploy private Docker containers for confidential banking, healthcare, and legal compliance (LGPD / GDPR).
+- **Custom Spatial Features:** Tailored table extractors for bank statements, invoices, and court proceedings.
+
+📬 **Inquiries & Licensing:**  
+[👉 **Click here to Request a Commercial License & Pricing**](mailto:messiasads2017@gmail.com?subject=RasgaPDF%20Commercial%20License%20Inquiry) with details about your organization, use case, and estimated volume.

{rasgapdf-0.2.3 → rasgapdf-0.2.4}/README.md

@@ -163,6 +163,33 @@ print(manifest["written"], "files written")
 
 ---
 
+---
+
+## 🦜️🔗 AI Framework Integrations (LangChain & LlamaIndex)
+
+### 🦜️ LangChain Document Loader
+`python
+from rasgapdf import RasgaPDFLoader
+
+# 1 Document per page with markdown tables:
+loader = RasgaPDFLoader('document.pdf', mode='pages')
+docs = loader.load()
+
+# Spatial RAG: 1 Document per block/table with exact [x0, y0, x1, y1] coordinates:
+spatial_loader = RasgaPDFLoader('document.pdf', mode='elements')
+chunks = spatial_loader.load()
+for chunk in chunks:
+    print(chunk.metadata['bbox'], chunk.page_content[:40])
+`
+
+### 🦙 LlamaIndex Reader
+`python
+from rasgapdf import RasgaPDFReader
+
+reader = RasgaPDFReader(mode='pages')
+documents = reader.load_data('financial_report.pdf')
+`
+
 ## 📄 License
 
 GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). Copyright (c) 2026 Messias Martins.
@@ -170,3 +197,19 @@ GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). Copyright (
 RasgaPDF embeds and links against [MuPDF](https://mupdf.com/) (Artifex Software),
 which is distributed under the AGPL-3.0. That is why this project is licensed
 under the same terms.
+
+---
+
+## 💼 Commercial & Enterprise Licensing (AGPL-3.0 Waiver)
+
+Need to use **RasgaPDF** in closed-source commercial software, proprietary SaaS platforms, private cloud environments, or corporate applications without releasing your source code under the AGPL-3.0?
+
+A **Commercial / Enterprise License** is available directly from the author:
+
+- **Complete AGPL-3.0 Exemption:** Keep your intellectual property, backend code, and SaaS infrastructure 100% closed-source and proprietary.
+- **Enterprise SLA & Support:** Direct architecture advisory, guaranteed response times, and priority bug resolution.
+- **On-Premise & Air-Gapped Appliances:** Ready-to-deploy private Docker containers for confidential banking, healthcare, and legal compliance (LGPD / GDPR).
+- **Custom Spatial Features:** Tailored table extractors for bank statements, invoices, and court proceedings.
+
+📬 **Inquiries & Licensing:**  
+[👉 **Click here to Request a Commercial License & Pricing**](mailto:messiasads2017@gmail.com?subject=RasgaPDF%20Commercial%20License%20Inquiry) with details about your organization, use case, and estimated volume.

{rasgapdf-0.2.3 → rasgapdf-0.2.4}/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "rasgapdf"
-version = "0.2.3"
+version = "0.2.4"
 authors = [
   { name="Messias Martins" },
 ]

{rasgapdf-0.2.3 → rasgapdf-0.2.4}/rasgapdf/__init__.py

@@ -2,7 +2,8 @@ import os
 import sys
 import json
 import ctypes
-from typing import Dict, Any, List, Optional
+from pathlib import Path
+from typing import Dict, Any, List, Optional, Union
 
 def _find_library() -> str:
     lib_name = "librasgapdf.so" if sys.platform != "win32" else "rasgapdf.dll"
@@ -251,7 +252,7 @@ def parse_raw_bytes(data: bytes) -> Dict[str, Any]:
         _lib.rasgapdf_free_string(res_ptr)
 
 
-def extract_text(data_or_path: bytes | str) -> str:
+def extract_text(data_or_path: Union[bytes, str]) -> str:
     """
     🚀 Extrai o texto completo do PDF em ordem natural de blocos de leitura
     diretamente de `bytes` ou caminho de arquivo.
@@ -339,4 +340,24 @@ def extract_images_bytes(data: bytes, output_dir: str) -> Dict[str, Any]:
     finally:
         _lib.rasgapdf_free_string(res_ptr)
 
-__all__ = ["parse_raw", "parse_raw_bytes", "parse_pdf", "extract_text", "extract_images", "extract_images_bytes", "version"]
+def __getattr__(name: str):
+    if name == "RasgaPDFLoader":
+        from .integrations.langchain import RasgaPDFLoader
+        return RasgaPDFLoader
+    if name == "RasgaPDFReader":
+        from .integrations.llama_index import RasgaPDFReader
+        return RasgaPDFReader
+    raise AttributeError(f"module '{__name__}' has no attribute '{name}'")
+
+
+__all__ = [
+    "parse_raw",
+    "parse_raw_bytes",
+    "parse_pdf",
+    "extract_text",
+    "extract_images",
+    "extract_images_bytes",
+    "version",
+    "RasgaPDFLoader",
+    "RasgaPDFReader",
+]

rasgapdf-0.2.4/rasgapdf/integrations/__init__.py

@@ -0,0 +1,6 @@
+"""Official AI & RAG Integrations for RasgaPDF (LangChain & LlamaIndex)."""
+
+from .langchain import RasgaPDFLoader
+from .llama_index import RasgaPDFReader
+
+__all__ = ["RasgaPDFLoader", "RasgaPDFReader"]

rasgapdf-0.2.4/rasgapdf/integrations/langchain.py

@@ -0,0 +1,185 @@
+"""Official LangChain Document Loader for RasgaPDF.
+
+Ultra-fast 2D spatial PDF loading with exact bounding boxes, structured tables,
+and zero-loss layout preservation for RAG pipelines.
+"""
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any, Dict, Iterator, List, Optional, Union
+
+try:
+    from langchain_core.document_loaders import BaseLoader
+    from langchain_core.documents import Document
+except ImportError:
+    # Standalone fallback when langchain_core is not installed
+    class BaseLoader:
+        def load(self) -> List["Document"]:
+            return list(self.lazy_load())
+
+    @dataclass
+    class Document:
+        page_content: str
+        metadata: Dict[str, Any] = field(default_factory=dict)
+
+
+class RasgaPDFLoader(BaseLoader):
+    """LangChain Document Loader powered by RasgaPDF.
+
+    Args:
+        file_path_or_bytes: Path to PDF file or raw bytes of a PDF.
+        mode: Extraction granularity:
+            - "pages" (default): One Document per page with layout-preserved text & markdown tables.
+            - "elements": One Document per spatial element (blocks and tables) with exact 2D bounding boxes.
+            - "text": A single Document containing the entire normalized text stream.
+        detect_tables: Whether to detect and reconstruct physical 2D tables. Default True.
+        markdown_tables: Format tables as Markdown matrices inside page content. Default True.
+    """
+
+    def __init__(
+        self,
+        file_path_or_bytes: Union[str, Path, bytes],
+        mode: str = "pages",
+        detect_tables: bool = True,
+        markdown_tables: bool = True,
+    ):
+        self.file_input = file_path_or_bytes
+        self.mode = mode
+        self.detect_tables = detect_tables
+        self.markdown_tables = markdown_tables
+        self.source = str(file_path_or_bytes) if not isinstance(file_path_or_bytes, (bytes, bytearray)) else "bytes_stream"
+
+    def lazy_load(self) -> Iterator[Document]:
+        import rasgapdf
+
+        if self.mode == "text":
+            text = rasgapdf.extract_text(self.file_input)
+            yield Document(
+                page_content=text,
+                metadata={"source": self.source, "mode": "text"},
+            )
+            return
+
+        doc = rasgapdf.parse_pdf(self.file_input, detect_tables=self.detect_tables)
+        pages = doc.get("pages", [])
+
+        if self.mode == "elements":
+            for page in pages:
+                page_num = page.get("page_number", 1)
+                width = page.get("width", 0.0)
+                height = page.get("height", 0.0)
+
+                # 1. Text blocks
+                for block in page.get("text_blocks", []):
+                    lines_text = [line.get("text", "") for line in block.get("lines", []) if line.get("text")]
+                    content = "\n".join(lines_text).strip()
+                    if content:
+                        yield Document(
+                            page_content=content,
+                            metadata={
+                                "source": self.source,
+                                "page": page_num,
+                                "type": "text_block",
+                                "bbox": block.get("bbox", {}),
+                                "reading_order": block.get("reading_order", 0),
+                                "page_width": width,
+                                "page_height": height,
+                            },
+                        )
+
+                # 2. Tables
+                for t_idx, table in enumerate(page.get("tables", [])):
+                    table_md = self._format_table_markdown(table) if self.markdown_tables else "\n".join(
+                        ["\t".join([str(c) for c in row]) if isinstance(row, list) else str(row) for row in table.get("rows", [])]
+                    )
+                    if table_md.strip():
+                        yield Document(
+                            page_content=table_md,
+                            metadata={
+                                "source": self.source,
+                                "page": page_num,
+                                "type": "table",
+                                "table_id": table.get("table_id", t_idx),
+                                "bbox": table.get("bbox", {}),
+                                "row_count": table.get("row_count", len(table.get("rows", []))),
+                                "col_count": table.get("col_count", 0),
+                                "page_width": width,
+                                "page_height": height,
+                            },
+                        )
+            return
+
+        # Default mode: "pages"
+        for page in pages:
+            page_num = page.get("page_number", 1)
+            parts = []
+
+            # Add text blocks
+            for block in page.get("text_blocks", []):
+                lines_text = [line.get("text", "") for line in block.get("lines", []) if line.get("text")]
+                text = "\n".join(lines_text).strip()
+                if text:
+                    parts.append(text)
+
+            # Add formatted tables
+            for table in page.get("tables", []):
+                if self.markdown_tables:
+                    table_str = self._format_table_markdown(table)
+                else:
+                    table_str = "\n".join(
+                        ["\t".join([str(c) for c in row]) if isinstance(row, list) else str(row) for row in table.get("rows", [])]
+                    )
+                if table_str:
+                    parts.append(table_str)
+
+            page_content = "\n\n".join(parts).strip()
+            yield Document(
+                page_content=page_content,
+                metadata={
+                    "source": self.source,
+                    "page": page_num,
+                    "total_pages": len(pages),
+                    "width": page.get("width", 0.0),
+                    "height": page.get("height", 0.0),
+                    "block_count": len(page.get("text_blocks", [])),
+                    "table_count": len(page.get("tables", [])),
+                },
+            )
+
+    def load(self) -> List[Document]:
+        return list(self.lazy_load())
+
+    @staticmethod
+    def _format_table_markdown(table: Dict[str, Any]) -> str:
+        rows = table.get("rows", [])
+        headers = table.get("headers", [])
+        if not rows and not headers:
+            return ""
+
+        matrix = []
+        if headers and isinstance(headers, list):
+            matrix.append([str(h).replace("\n", " ").strip() for h in headers])
+
+        for row in rows:
+            if isinstance(row, list):
+                row_cells = [str(c).replace("\n", " ").strip() for c in row]
+            elif isinstance(row, dict):
+                row_cells = [str(c.get("text", "")).replace("\n", " ").strip() for c in row.get("cells", [])]
+            else:
+                row_cells = [str(row)]
+            matrix.append(row_cells)
+
+        if not matrix:
+            return ""
+
+        max_cols = max(len(r) for r in matrix)
+        for r in matrix:
+            while len(r) < max_cols:
+                r.append("")
+
+        md_lines = []
+        header = matrix[0]
+        md_lines.append("| " + " | ".join(header) + " |")
+        md_lines.append("| " + " | ".join(["---"] * max_cols) + " |")
+        for r in matrix[1:]:
+            md_lines.append("| " + " | ".join(r) + " |")
+        return "\n".join(md_lines)

rasgapdf-0.2.4/rasgapdf/integrations/llama_index.py

@@ -0,0 +1,73 @@
+"""Official LlamaIndex Reader for RasgaPDF.
+
+High-throughput 2D spatial PDF reader for LlamaIndex indexers and agents.
+"""
+from dataclasses import dataclass, field
+from pathlib import Path
+from typing import Any, Dict, List, Optional, Union
+
+try:
+    from llama_index.core.readers.base import BaseReader
+    from llama_index.core.schema import Document
+except ImportError:
+    class BaseReader:
+        pass
+
+    @dataclass
+    class Document:
+        text: str
+        extra_info: Dict[str, Any] = field(default_factory=dict)
+
+        @property
+        def metadata(self) -> Dict[str, Any]:
+            return self.extra_info
+
+
+class RasgaPDFReader(BaseReader):
+    """LlamaIndex Reader powered by RasgaPDF.
+
+    Args:
+        mode: Granularity mode:
+            - "pages" (default): Returns one Document per PDF page.
+            - "elements": Returns one Document per text block or table with exact bounding boxes.
+        detect_tables: Detect and extract physical tables. Default True.
+        markdown_tables: Format tables as Markdown. Default True.
+    """
+
+    def __init__(
+        self,
+        mode: str = "pages",
+        detect_tables: bool = True,
+        markdown_tables: bool = True,
+    ):
+        self.mode = mode
+        self.detect_tables = detect_tables
+        self.markdown_tables = markdown_tables
+
+    def load_data(
+        self,
+        file: Union[str, Path, bytes],
+        extra_info: Optional[Dict[str, Any]] = None,
+    ) -> List[Document]:
+        from .langchain import RasgaPDFLoader
+
+        loader = RasgaPDFLoader(
+            file_path_or_bytes=file,
+            mode=self.mode,
+            detect_tables=self.detect_tables,
+            markdown_tables=self.markdown_tables,
+        )
+
+        base_meta = extra_info or {}
+        docs = []
+
+        for doc in loader.lazy_load():
+            meta = {**base_meta, **doc.metadata}
+            try:
+                # Use official llama_index Document if available
+                from llama_index.core.schema import Document as LlamaDoc
+                docs.append(LlamaDoc(text=doc.page_content, extra_info=meta))
+            except ImportError:
+                docs.append(Document(text=doc.page_content, extra_info=meta))
+
+        return docs

{rasgapdf-0.2.3 → rasgapdf-0.2.4}/rasgapdf.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: rasgapdf
-Version: 0.2.3
+Version: 0.2.4
 Summary: Ultra-fast 2D spatial PDF parser powered by pure Rust — extract exact [x0, y0, x1, y1] bounding boxes, text blocks, tables, images & vector paths in milliseconds for RAG & LLMs
 Author: Messias Martins
 Project-URL: Homepage, https://martinsmessias.github.io/rasgapdf/
@@ -191,6 +191,33 @@ print(manifest["written"], "files written")
 
 ---
 
+---
+
+## 🦜️🔗 AI Framework Integrations (LangChain & LlamaIndex)
+
+### 🦜️ LangChain Document Loader
+`python
+from rasgapdf import RasgaPDFLoader
+
+# 1 Document per page with markdown tables:
+loader = RasgaPDFLoader('document.pdf', mode='pages')
+docs = loader.load()
+
+# Spatial RAG: 1 Document per block/table with exact [x0, y0, x1, y1] coordinates:
+spatial_loader = RasgaPDFLoader('document.pdf', mode='elements')
+chunks = spatial_loader.load()
+for chunk in chunks:
+    print(chunk.metadata['bbox'], chunk.page_content[:40])
+`
+
+### 🦙 LlamaIndex Reader
+`python
+from rasgapdf import RasgaPDFReader
+
+reader = RasgaPDFReader(mode='pages')
+documents = reader.load_data('financial_report.pdf')
+`
+
 ## 📄 License
 
 GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). Copyright (c) 2026 Messias Martins.
@@ -198,3 +225,19 @@ GNU Affero General Public License v3.0 or later (AGPL-3.0-or-later). Copyright (
 RasgaPDF embeds and links against [MuPDF](https://mupdf.com/) (Artifex Software),
 which is distributed under the AGPL-3.0. That is why this project is licensed
 under the same terms.
+
+---
+
+## 💼 Commercial & Enterprise Licensing (AGPL-3.0 Waiver)
+
+Need to use **RasgaPDF** in closed-source commercial software, proprietary SaaS platforms, private cloud environments, or corporate applications without releasing your source code under the AGPL-3.0?
+
+A **Commercial / Enterprise License** is available directly from the author:
+
+- **Complete AGPL-3.0 Exemption:** Keep your intellectual property, backend code, and SaaS infrastructure 100% closed-source and proprietary.
+- **Enterprise SLA & Support:** Direct architecture advisory, guaranteed response times, and priority bug resolution.
+- **On-Premise & Air-Gapped Appliances:** Ready-to-deploy private Docker containers for confidential banking, healthcare, and legal compliance (LGPD / GDPR).
+- **Custom Spatial Features:** Tailored table extractors for bank statements, invoices, and court proceedings.
+
+📬 **Inquiries & Licensing:**  
+[👉 **Click here to Request a Commercial License & Pricing**](mailto:messiasads2017@gmail.com?subject=RasgaPDF%20Commercial%20License%20Inquiry) with details about your organization, use case, and estimated volume.

{rasgapdf-0.2.3 → rasgapdf-0.2.4}/rasgapdf.egg-info/SOURCES.txt

@@ -6,4 +6,7 @@ rasgapdf/rasgapdf.dll
 rasgapdf.egg-info/PKG-INFO
 rasgapdf.egg-info/SOURCES.txt
 rasgapdf.egg-info/dependency_links.txt
-rasgapdf.egg-info/top_level.txt
+rasgapdf.egg-info/top_level.txt
+rasgapdf/integrations/__init__.py
+rasgapdf/integrations/langchain.py
+rasgapdf/integrations/llama_index.py