docqwise 0.2.1__tar.gz → 0.3.1__tar.gz

{docqwise-0.2.1 → docqwise-0.3.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: docqwise
-Version: 0.2.1
+Version: 0.3.1
 Summary: Read, Extract, Retrieve:  Document intelligence that adapts, accelerates, and scales.
 Author: Venkatkumar Rajan
 License: Apache-2.0
@@ -95,7 +95,7 @@ Dynamic: license-file
 </p>
 
 <p align="center">
-  <a href="https://pypi.org/project/docqwise/"><img src="https://img.shields.io/badge/PyPI-docqwise-blue" alt="PyPI"></a>
+  <a href="https://pypi.org/project/docqwise/"><img src="https://img.shields.io/badge/PyPI-v0.3.1-green" alt="PyPI"></a>
   <a href="https://github.com/VK-Ant/docqwise/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache%202.0-blue.svg" alt="License"></a>
   <a href="https://github.com/VK-Ant/docqwise/blob/main/notebooks/docqwise_getting_started.ipynb">
         <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a>
@@ -241,6 +241,38 @@ Return JSON matching the schema exactly.
 """)
 ```
 
+## System Prompt
+
+Set the LLM's role before extraction. Works with all backends and all RAG modes.
+
+```python
+# Set the role — LLM behaves as this specialist
+dq.extract_fields("invoice.pdf",
+    system_prompt="You are a maritime document specialist with 20 years experience.",
+    prompt="Extract vessel name, port, and total cost.
+
+Document:
+{context}
+
+JSON:",
+)
+
+# System prompt + schema + template — full control
+dq.extract_fields("contract.pdf",
+    system_prompt="You are a legal analyst specializing in shipping law.",
+    schema={"parties": {"type": "array"}, "liability": {"type": "number"}},
+    prompt_template="Extract {schema} from:
+{context}
+JSON:",
+)
+
+# System prompt with GraphRAG
+dq.ask_rag("What is the liability cap?",
+    mode="graphrag",
+    system_prompt="You are a risk analyst. Be precise with numbers.",
+)
+```
+
 ## Self-Improving Corrections
 
 ```python
@@ -258,6 +290,64 @@ dq.ask("Which vendor has highest sales?")    # GROUP BY + MAX
 dq.ask("How many invoices are overdue?")     # COUNT + WHERE
 ```
 
+## Source Attribution
+
+Every answer tells you which document it came from.
+
+```python
+# Simple ask — shows source in output
+answer = dq.ask("What is the total amount?")
+print(answer)
+# 2,18,300.00
+#   Source: sample_invoice.pdf
+
+# Detailed answer with metadata
+result = dq.ask_with_source("What is the governing law?")
+print(result.answer)        # "laws of India"
+print(result.source_name)   # "sample_contract.pdf"
+print(result.confidence)    # 0.85
+print(result.method)        # "llm"
+```
+
+## RAG Modes
+
+Choose your retrieval strategy:
+
+```python
+dq = Docqwise()
+dq.ingest("documents/")
+
+# General RAG (default) — chunk → embed → retrieve → LLM
+result = dq.ask_rag("total amount?", mode="general")
+
+# GraphRAG — entity graph → graph traversal → cross-document answers
+result = dq.ask_rag("Who is the vendor for PO-2012?", mode="graphrag")
+print(result["answer"])         # "SuperStore"
+print(result["source"])         # "invoice_001.pdf"
+print(result["evidence"])       # [{entity, relation, target}, ...]
+
+# Multimodal RAG — text + images → vision LLM
+result = dq.ask_rag("What is in this scan?", mode="multimodal", source="scan.jpg")
+```
+
+## Graph Visualization
+
+Build and visualize document knowledge graphs:
+
+```python
+dq = Docqwise()
+dq.ingest("documents/")
+
+# Build knowledge graph from all documents
+graphrag = dq.build_document_graph()
+graph = graphrag.get_graph()
+print(f"Nodes: {graph.node_count}, Edges: {graph.edge_count}")
+
+# Generate interactive HTML visualization
+dq.visualize_graph(output="my_graph.html")
+# Open my_graph.html in browser — interactive, color-coded, draggable nodes
+```
+
 ## All Features
 
 ```python

{docqwise-0.2.1 → docqwise-0.3.1}/README.md

@@ -7,7 +7,7 @@
 </p>
 
 <p align="center">
-  <a href="https://pypi.org/project/docqwise/"><img src="https://img.shields.io/badge/PyPI-docqwise-blue" alt="PyPI"></a>
+  <a href="https://pypi.org/project/docqwise/"><img src="https://img.shields.io/badge/PyPI-v0.3.1-green" alt="PyPI"></a>
   <a href="https://github.com/VK-Ant/docqwise/blob/main/LICENSE"><img src="https://img.shields.io/badge/License-Apache%202.0-blue.svg" alt="License"></a>
   <a href="https://github.com/VK-Ant/docqwise/blob/main/notebooks/docqwise_getting_started.ipynb">
         <img src="https://colab.research.google.com/assets/colab-badge.svg" alt="Open In Colab"></a>
@@ -153,6 +153,38 @@ Return JSON matching the schema exactly.
 """)
 ```
 
+## System Prompt
+
+Set the LLM's role before extraction. Works with all backends and all RAG modes.
+
+```python
+# Set the role — LLM behaves as this specialist
+dq.extract_fields("invoice.pdf",
+    system_prompt="You are a maritime document specialist with 20 years experience.",
+    prompt="Extract vessel name, port, and total cost.
+
+Document:
+{context}
+
+JSON:",
+)
+
+# System prompt + schema + template — full control
+dq.extract_fields("contract.pdf",
+    system_prompt="You are a legal analyst specializing in shipping law.",
+    schema={"parties": {"type": "array"}, "liability": {"type": "number"}},
+    prompt_template="Extract {schema} from:
+{context}
+JSON:",
+)
+
+# System prompt with GraphRAG
+dq.ask_rag("What is the liability cap?",
+    mode="graphrag",
+    system_prompt="You are a risk analyst. Be precise with numbers.",
+)
+```
+
 ## Self-Improving Corrections
 
 ```python
@@ -170,6 +202,64 @@ dq.ask("Which vendor has highest sales?")    # GROUP BY + MAX
 dq.ask("How many invoices are overdue?")     # COUNT + WHERE
 ```
 
+## Source Attribution
+
+Every answer tells you which document it came from.
+
+```python
+# Simple ask — shows source in output
+answer = dq.ask("What is the total amount?")
+print(answer)
+# 2,18,300.00
+#   Source: sample_invoice.pdf
+
+# Detailed answer with metadata
+result = dq.ask_with_source("What is the governing law?")
+print(result.answer)        # "laws of India"
+print(result.source_name)   # "sample_contract.pdf"
+print(result.confidence)    # 0.85
+print(result.method)        # "llm"
+```
+
+## RAG Modes
+
+Choose your retrieval strategy:
+
+```python
+dq = Docqwise()
+dq.ingest("documents/")
+
+# General RAG (default) — chunk → embed → retrieve → LLM
+result = dq.ask_rag("total amount?", mode="general")
+
+# GraphRAG — entity graph → graph traversal → cross-document answers
+result = dq.ask_rag("Who is the vendor for PO-2012?", mode="graphrag")
+print(result["answer"])         # "SuperStore"
+print(result["source"])         # "invoice_001.pdf"
+print(result["evidence"])       # [{entity, relation, target}, ...]
+
+# Multimodal RAG — text + images → vision LLM
+result = dq.ask_rag("What is in this scan?", mode="multimodal", source="scan.jpg")
+```
+
+## Graph Visualization
+
+Build and visualize document knowledge graphs:
+
+```python
+dq = Docqwise()
+dq.ingest("documents/")
+
+# Build knowledge graph from all documents
+graphrag = dq.build_document_graph()
+graph = graphrag.get_graph()
+print(f"Nodes: {graph.node_count}, Edges: {graph.edge_count}")
+
+# Generate interactive HTML visualization
+dq.visualize_graph(output="my_graph.html")
+# Open my_graph.html in browser — interactive, color-coded, draggable nodes
+```
+
 ## All Features
 
 ```python

docqwise-0.3.1/docqwise/_version.py

@@ -0,0 +1 @@
+__version__ = "0.3.1"

{docqwise-0.2.1 → docqwise-0.3.1}/docqwise/engine.py

@@ -226,7 +226,8 @@ class Docqwise:
     def extract_fields(self, source: str, schema: dict = None,
                        template: str = None, method: str = "auto",
                        llm=None, prompt: str = None,
-                       prompt_template: str = None, **kwargs) -> ExtractionResult:
+                       prompt_template: str = None,
+                       system_prompt: str = None, **kwargs) -> ExtractionResult:
         """Extract fields from a document.
 
         Args:
@@ -235,8 +236,9 @@ class Docqwise:
             template: pre-built template (invoice, contract, resume, receipt)
             method: auto | rag | llm | vision | regex
             llm: your own LLM instance
-            prompt: your own extraction prompt (full prompt text)
-            prompt_template: your prompt template with {text} and {schema} placeholders
+            prompt: your extraction prompt — use {context} for document text
+            prompt_template: your template — use {context} and {schema} placeholders
+            system_prompt: LLM system role (e.g. "You are a maritime document specialist")
             model: Ollama model name
         """
         self._ensure_initialized()
@@ -269,6 +271,7 @@ class Docqwise:
             result = extractor.extract_fields(
                 doc, schema=target_schema,
                 prompt=prompt, prompt_template=prompt_template,
+                system_prompt=system_prompt,
             )
 
         result._engine = self
@@ -365,10 +368,98 @@ class Docqwise:
                                               filters=kwargs.get("filters"))
         return []
 
+    def _ensure_qa_embedder(self):
+        if self._qa_engine._embedder is None:
+            from docqwise.factory import EmbedderFactory
+            embedder = EmbedderFactory.get(self._user_embedder)
+            if embedder:
+                self._qa_engine._embedder = embedder
+
     def ask(self, question: str, source: str = None, **kwargs) -> str:
         self._ensure_initialized()
+        self._ensure_qa_embedder()
         return self._qa_engine.ask(question, source=source)
 
+    def ask_with_source(self, question: str, source: str = None, **kwargs):
+        """Ask a question and get answer with source document attribution.
+
+        Returns:
+            QAAnswer with .answer, .source, .source_name, .confidence, .method
+        """
+        self._ensure_initialized()
+        self._ensure_qa_embedder()
+        return self._qa_engine.ask_with_source(question, source=source)
+
+    def ask_rag(self, question: str, mode: str = "general",
+                source: str = None, prompt: str = None,
+                system_prompt: str = None, **kwargs) -> dict:
+        """Ask using specific RAG strategy.
+
+        Modes:
+            general     — chunk → embed → retrieve → LLM (default)
+            graphrag    — extract entities → build graph → graph retrieval → LLM
+            multimodal  — text + images → vision LLM → combined
+
+        Args:
+            question: your question
+            mode: general | graphrag | multimodal
+            source: specific document path (optional)
+            prompt: custom LLM prompt — use {context} for document text, {question} for the question
+        """
+        self._ensure_initialized()
+        self._ensure_qa_embedder()
+        from docqwise.retrieval.rag_strategy import RAGStrategy
+        strategy = RAGStrategy(engine=self, mode=mode)
+        return strategy.ask(question, source=source, prompt=prompt, **kwargs)
+
+    def build_document_graph(self, sources: list[str] = None) -> Any:
+        """Build knowledge graph from documents for GraphRAG."""
+        self._ensure_initialized()
+        from docqwise.retrieval.graphrag import GraphRAGEngine
+        graphrag = GraphRAGEngine(self)
+        graphrag.build_from_ingested(sources=sources)
+        self._doc_graph = graphrag
+        return graphrag
+
+    def visualize_graph(self, output: str = "docqwise_graph.png",
+                        sources: list[str] = None) -> str:
+        """Generate graph visualization as image or HTML.
+
+        Args:
+            output: file path — .png, .svg, .pdf, .jpg, or .html
+            sources: list of document paths (builds from these, or uses all ingested)
+
+        Returns:
+            path to saved file
+        """
+        self._ensure_initialized()
+        if not hasattr(self, '_doc_graph') or self._doc_graph is None:
+            self.build_document_graph(sources=sources)
+
+        from docqwise.retrieval.graphrag import GraphVisualizer
+        graph = self._doc_graph.get_graph()
+        return GraphVisualizer.save(graph, output=output)
+
+    def visualize_query(self, question: str, output: str = "docqwise_query_graph.png",
+                        sources: list[str] = None) -> dict:
+        """Ask a question with GraphRAG and visualize the answer path.
+
+        Returns:
+            dict with "answer", "source", "evidence", "graph_file"
+        """
+        self._ensure_initialized()
+        if not hasattr(self, '_doc_graph') or self._doc_graph is None:
+            self.build_document_graph(sources=sources)
+
+        result = self._doc_graph.query(question)
+
+        from docqwise.retrieval.graphrag import GraphVisualizer
+        graph = self._doc_graph.get_graph()
+        GraphVisualizer.from_query(graph, result, output=output)
+
+        result["graph_file"] = output
+        return result
+
     def sql(self, query: str) -> list[dict]:
         from docqwise.databases.sqlite_db import SQLiteDB
         db = SQLiteDB(os.path.join(self._store_path, "structured.db"))

{docqwise-0.2.1 → docqwise-0.3.1}/docqwise/extractors/field_extractor.py

@@ -20,7 +20,7 @@ class RegexFieldExtractor(BaseFieldExtractor):
     }
 
     def extract_fields(self, document: DocqwiseDocument, schema: dict = None,
-                       prompt: str = None, prompt_template: str = None) -> ExtractionResult:
+                       prompt: str = None, prompt_template: str = None, system_prompt: str = None) -> ExtractionResult:
         text = document.text
         fields = {}
 

{docqwise-0.2.1 → docqwise-0.3.1}/docqwise/extractors/llm_extractor.py

@@ -30,8 +30,8 @@ class LLMFieldExtractor(BaseFieldExtractor):
 
         # Priority 1: Ollama
         try:
-            import requests
-            requests.get("http://localhost:11434/api/tags", timeout=2)
+            from urllib.request import urlopen
+            urlopen("http://localhost:11434/api/tags", timeout=2)
             from docqwise.llm.ollama import OllamaLLM
             model = self._model or "nemotron-mini"
             self._llm = OllamaLLM(model=model)
@@ -64,7 +64,7 @@ class LLMFieldExtractor(BaseFieldExtractor):
 
     def extract_fields(self, document: DocqwiseDocument,
                        schema: dict = None, prompt: str = None,
-                       prompt_template: str = None) -> ExtractionResult:
+                       prompt_template: str = None, system_prompt: str = None) -> ExtractionResult:
         llm = self._get_llm()
 
         if llm is None: