HowdenParser 0.1.8__py3-none-any.whl

HowdenParser/__init__.py

@@ -0,0 +1,3 @@
+from .parser import ParserFactory
+
+__all__ = ["ParserFactory"]

HowdenParser/parameter/__init__.py

@@ -0,0 +1,15 @@
+import os
+import importlib
+
+# Get all .py files in this folder except __init__.py
+current_dir = os.path.dirname(__file__)
+for filename in os.listdir(current_dir):
+    if filename.endswith(".py") and filename != "__init__.py":
+        module_name = filename[:-3]  # remove .py
+        module = importlib.import_module(f".{module_name}", package=__name__)
+
+        # Add all classes from the module to the package namespace
+        for attr_name in dir(module):
+            attr = getattr(module, attr_name)
+            if isinstance(attr, type):  # only classes
+                globals()[attr_name] = attr

HowdenParser/parameter/huggingface.py

@@ -0,0 +1,6 @@
+from typing import Literal
+from pydantic import BaseModel
+
+class Parameter(BaseModel):
+    model: str = "huggingface:HURIDOCS/pdf-segmentation"
+

HowdenParser/parameter/llamaparser.py

@@ -0,0 +1,8 @@
+
+from typing import Literal
+from pydantic import BaseModel
+
+class Parameter(BaseModel):
+    model: str = "llamaparser:"
+    result_type: Literal["md"] = "md"
+    mode: bool = False

HowdenParser/parameter/mistralocr.py

@@ -0,0 +1,7 @@
+from typing import Literal
+from pydantic import BaseModel
+
+class Parameter(BaseModel):
+    model: str = "mistralocr:"
+    result_type: Literal["md"] = "md"
+    mode: bool = False

HowdenParser/parser.py

@@ -0,0 +1,137 @@
+from abc import ABC, abstractmethod
+from dotenv import load_dotenv
+import os
+from pathlib import Path
+
+load_dotenv()
+
+class BaseParser(ABC):
+    @abstractmethod
+    def parse(self, text: str) -> dict:
+        pass
+
+
+class MistralOCRParser(BaseParser):
+    def __init__(self, result_type: str, **kwargs) -> None:
+        from mistralai import Mistral
+        name = "MISTRAL-OCR-API-TOKEN"
+        api_key = os.getenv(name)
+        if not api_key:
+            raise EnvironmentError(f"Missing {name} in .env file.")
+
+        self.result_type = "md" if result_type.lower() in ("md", "markdown") else "text"
+        self.client = Mistral(api_key=api_key)
+
+    def parse(self, file_path: Path) -> str:
+        def upload_pdf(filename):
+            uploaded_pdf = self.client.files.upload(
+            file={
+            "file_name": filename,
+                "content": open(filename, "rb"),
+            },
+            purpose="ocr"
+            )
+            signed_url = self.client.files.get_signed_url(file_id=uploaded_pdf.id)
+            return signed_url.url
+
+        ocr_response = self.client.ocr.process(
+            model="mistral-ocr-latest",
+            document={
+            "type": "document_url",
+            "document_url": upload_pdf(file_path),
+            },
+            include_image_base64=True,
+            )
+
+        return "\n".join(doc.markdown for doc in ocr_response.pages)
+
+
+class LangChainParser(BaseParser):
+    def __init__(self, model_name: str):
+        from langchain.llms import OpenAI
+        self.model_name = model_name
+        self.model = OpenAI(model_name=model_name)
+
+    def parse(self, text: str) -> dict:
+        response = self.model(text)
+        return {"source": "LangChain", "output": response}
+
+
+class LlamaParser(BaseParser):
+    def __init__(self, result_type: str, mode: bool, **kwargs) -> None:
+        from llama_parse import LlamaParse, ResultType
+        if result_type.lower() == "md" or result_type.lower() == "markdown":
+            self.result_type = ResultType.MD
+        name = "LLAMA-PARSER-API-TOKEN"
+        self.api_key = os.getenv(name)
+        if not self.api_key:
+            raise EnvironmentError(f"Missing {name} in .env file.")
+
+        self.parser = LlamaParse(
+            api_key=self.api_key,
+            result_type=self.result_type,
+            premium_mode=mode
+        )
+
+    def parse(self, file_path: Path) -> str:
+        documents = self.parser.load_data(str(file_path))
+        return "\n".join(doc.text for doc in documents)
+
+class HuggingFaceParser(BaseParser):
+    def __init__(self, result_type: str, mode: bool, **kwargs) -> None:
+        from transformers import pipeline
+
+        if result_type.lower() in ("md", "markdown"):
+            self.result_type = "markdown"
+        else:
+            self.result_type = "text"
+
+        name = "HF-API-TOKEN"
+        self.api_key = os.getenv(name)
+        if not self.api_key:
+            raise EnvironmentError(f"Missing {name} in .env file.")
+
+        # OCR + text extraction pipeline
+        # Example model: microsoft/layoutlmv3-base-finetuned-docvqa
+        self.parser = pipeline(
+            task="document-question-answering",
+            model="impira/layoutlm-document-qa",
+            use_auth_token=self.api_key
+        )
+
+    def parse(self, file_path: Path) -> str:
+        import fitz  # PyMuPDF for PDF → images
+
+        pdf_doc = fitz.open(file_path)
+        output_parts = []
+
+        for page in pdf_doc:
+            pix = page.get_pixmap(dpi=200)
+            img_bytes = pix.tobytes("png")
+            response = self.parser(img_bytes, question="Extract all text")
+            if response and "answer" in response[0]:
+                output_parts.append(response[0]["answer"])
+
+        if self.result_type == "markdown":
+            # Here you could add rules to format into markdown
+            return "\n\n".join(output_parts)
+        else:
+            return " ".join(output_parts)
+
+# === Step 3: Dynamic factory using string input ===
+class ParserFactory:
+    @staticmethod
+    def get_parser(parser_type: str, **kwargs) -> BaseParser:
+        provider = parser_type.partition(":")[0].lower()
+        model = parser_type.partition(":")[2]
+        if provider == "langchain":
+            return LangChainParser(model_name=model)
+        elif provider == "llamaparser":
+            return LlamaParser(**kwargs)
+        elif provider == "huggingface":
+            return HuggingFaceParser(model_name=model, **kwargs)
+        elif provider == "mistralocr":
+            return MistralOCRParser(**kwargs)
+        else:
+            raise ValueError(f"Unknown parser type: {parser_type}")
+

howdenparser-0.1.8.dist-info/METADATA

@@ -0,0 +1,83 @@
+Metadata-Version: 2.3
+Name: HowdenParser
+Version: 0.1.8
+Summary: A simple configuration manager with Pydantic and JSON export.
+License: MIT
+Keywords: config,configuration,pydantic,json
+Author: JesperThoftIllemannJ
+Author-email: jesper.jaeger@howdendanmark.dk
+Classifier: License :: OSI Approved :: MIT License
+Classifier: Programming Language :: Python :: 2
+Classifier: Programming Language :: Python :: 2.7
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.4
+Classifier: Programming Language :: Python :: 3.5
+Classifier: Programming Language :: Python :: 3.6
+Classifier: Programming Language :: Python :: 3.7
+Classifier: Programming Language :: Python :: 3.8
+Classifier: Programming Language :: Python :: 3.9
+Classifier: Programming Language :: Python :: 3.10
+Classifier: Programming Language :: Python :: 3.11
+Classifier: Programming Language :: Python :: 3.12
+Classifier: Programming Language :: Python :: 3.13
+Project-URL: Documentation, https://github.com/yourusername/config
+Project-URL: Homepage, https://github.com/yourusername/config
+Project-URL: Repository, https://github.com/yourusername/config
+Description-Content-Type: text/markdown
+
+# OCR & LLM Parser
+
+A powerful Python package for parsing and processing documents using multiple providers:
+- **Mistral OCR** — Extracts text from PDFs and images with high accuracy.
+- **LangChain** — Processes or summarizes text using LLMs.
+- **Llama Parser** — Advanced parsing with Markdown or text output.
+- **HuggingFace** — OCR and document question answering with transformer models.
+
+The package provides a **unified interface** so you can switch between providers easily using a **factory pattern**.
+
+---
+
+## 🚀 Features
+- Extract text from PDFs or images
+- Summarize or process text using LLMs
+- Support for **Markdown** or **plain text** output
+- Plug-and-play factory to switch providers without changing much code
+- Handles environment variable loading for API keys automatically
+
+---
+
+# 🔑 Tokens
+
+Create a .env file in your project root and add the API keys for the services you want to use.
+
+### Mistral OCR
+MISTRAL-OCR-API-TOKEN=your_mistral_api_key
+
+### Llama Parser
+LLAMA-PARSER-API-TOKEN=your_llama_parser_api_key
+
+### HuggingFace
+HF-API-TOKEN=your_huggingface_api_key
+
+Only include the keys for the providers you plan to use.
+
+---
+
+# 🛠️ Usage
+
+from HowdenParser import ParserFactory
+
+from pathlib import Path
+
+parser = ParserFactory.get_parser("mistralocr:", result_type="md")
+text = parser.parse(Path("document.pdf"))
+print(text)
+
+if HowdenConfig package being used
+
+
+parser = ParserFactory.get_parser("mistralocr:", **config.parameter.dump_model()) 
+
+text = parser.parse(Path("document.pdf"))
+
+

howdenparser-0.1.8.dist-info/RECORD

@@ -0,0 +1,9 @@
+HowdenParser/__init__.py,sha256=y8Nntk25CoqKQXzlujosuOdEJtujaC2Ui-QCEn8doik,64
+HowdenParser/parameter/__init__.py,sha256=KMPnWvqWoTjibBUo9Pxg9mgyhUDWrlkyKZ0ZQG86Rqo,618
+HowdenParser/parameter/huggingface.py,sha256=aYT3fvnegkgYiayPbjC1fDFz86T2OeyHMZisKwZS5gg,151
+HowdenParser/parameter/llamaparser.py,sha256=RTtNmTiBK2i5Y3ikGOR3J6RSoELtxlcWNx9FXijm7pM,187
+HowdenParser/parameter/mistralocr.py,sha256=e7lgZrcvfloS6Df5_3DnWNU8MVOyl_VEHTaeox8WOfM,184
+HowdenParser/parser.py,sha256=rwlqT0Ve27GSUPYFB65bkkRKMkNk2aNbMqfjYp1gbX8,4778
+howdenparser-0.1.8.dist-info/METADATA,sha256=duOw1Br-viPqAbkDzapNl-2brcELXziKwr2mYvYP8xQ,2730
+howdenparser-0.1.8.dist-info/WHEEL,sha256=9jVypgI8LevED4l0j3YDHs5IZdbkcE7aWw9FT9n2UzI,92
+howdenparser-0.1.8.dist-info/RECORD,,

howdenparser-0.1.8.dist-info/WHEEL

@@ -0,0 +1,4 @@
+Wheel-Version: 1.0
+Generator: poetry-core 2.1.2
+Root-Is-Purelib: true
+Tag: py2.py3-none-any