created by gemma4:12b

README.md

@@ -1,28 +1,31 @@
-# QCMS (Quality Control Management System)
+# Data-Transport Studio
 
-QCMS is a robust system built to streamline and manage quality control processes.
+**Data-Transport Studio** is a front-end for the [data-transport](https://github.com/lnyemba/data-transport) framework and library. It lowers the barrier to entry to the capabilities of data-transport by providing an intuitive interface that makes it easy to connect to, query, and analyze data across diverse sources—without needing to write boilerplate code.
 
-## Powered by data-transport
+Powered by the data-transport framework and library, Data-Transport Studio adds **AI-powered querying** and an **integrated spreadsheet** to empower users to explore, transform, and better understand their data.
 
-QCMS leverages **data-transport**, a powerful abstraction layer for seamless data communication across diverse systems.
+## What data-transport Provides
 
-### What is data-transport?
-
-**data-transport** provides a consistent interface to interact with a wide range of data sources, including:
+At its core, Data-Transport Studio is powered by data-transport—a lightweight abstraction layer that enables seamless, unified access to a wide variety of data environments:
 
 - **SQL Databases:** PostgreSQL, MySQL, SQL Server, SQLite3+, DuckDB
 - **NoSQL Databases:** MongoDB, CouchDB
-- **Cloud Storage & Warehouse:** S3, Nextcloud, Apache Iceberg, Apache Drill
+- **Cloud & Warehouse:** S3, Nextcloud, Apache Iceberg, Apache Drill
 - **Other Integrations:** Files, RabbitMQ, HTTP
 
-### Why use data-transport in QCMS?
+## Key Features
+
+- **Simple Connection Management:** Easily register and manage connections to diverse data sources through a label-based interface, eliminating the need to handle raw connection strings.
+- **Intuitive Querying:** Query your data directly through the interface without writing Python scripts or complex boilerplate.
+- **AI-Powered Data Exploration:** Translate natural language prompts into precise SQL queries and generate multi-sheet analytical reports using integrated providers like **Ollama** and **Azure OpenAI**.
+- **Integrated Spreadsheet:** View query results directly in an integrated spreadsheet, enabling ad-hoc analysis, data wrangling, augmentation, and export with ease.
+- **Immediate Results:** Interact with your data in real-time and iterate on analysis quickly through a spreadsheet-driven workflow.
 
-Integrating data-transport into the QCMS architecture provides several key advantages:
+## Why Use Data-Transport Studio?
 
-- **Unified Data Access:** QCMS can interact with multiple database types (SQL, NoSQL, Cloud) using a consistent code base, simplifying development and maintenance.
-- **Enhanced Security:** It masks raw database connection details, preventing the leak of sensitive credentials within the application.
-- **Seamless Data Manipulation:** By leveraging **Pandas** and **SQLAlchemy**, data-transport allows QCMS to perform complex data transformations and analysis efficiently.
-- **Scalable ETL Pipelines:** It facilitates the creation of robust, reproducible data processing pipelines by treating different storage backends as a unified "transport" layer.
-- **Portability:** It allows for easier sharing of scripts and notebooks across different environments without exposing specific connection strings.
+- **Empowers Non-Technical Users:** Analysts and domain experts can explore and analyze data without needing to write code.
+- **Accelerates Insights:** Natural language querying, automated analysis, and spreadsheet-based exploration help users understand their data faster.
+- **Reduces Complexity:** A single interface replaces the need to juggle multiple tools, clients, or scripts to work with different data sources.
+- **Focus on Analysis, Not Infrastructure:** Spend less time on connectivity and boilerplate, and more time uncovering actionable insights.
 
-For more information on the underlying technology, visit the [data-transport repository](https://github.com/lnyemba/data-transport).
+For more information on the underlying framework and library, visit the [data-transport repository](https://github.com/lnyemba/data-transport).

www/html/_plugins/parser.py

@@ -0,0 +1,49 @@
+"""
+This file implements the parsing strategy for a response to an LLM. This has been tested on :
+    - ollama    gemma4
+    - azure     openai  gpt-6-astra
+"""
+import json
+import pandas as pd
+def format(_data) :
+    if type(_data) == str :
+        _data = json.loads(_data)
+    return pd.DataFrame(_data).fillna(0).to_dict(orient='records')
+def gemma4_12b(_stream) :
+    try:
+        _data =  json.loads(_stream.replace('```json','').replace('```','').strip())
+        
+        _data['data'] = format(_data['data']) 
+        if 'analysis' in _data:
+            for _item in _data['analysis'] :
+                if 'data' in _item :
+                    _item['data'] = format(_item['data'])
+        #
+        #
+        return _data
+    except Exception as e:
+        # print ([' Error ',e])
+        pass
+    return _stream
+
+def gpt6_astra(_stream):
+    try:
+        _data = _stream[0]['text'] if _stream[0]['type'] == 'text' else _stream[1]['text']
+        
+        if type(_data) == str :
+            _data = json.loads(_data)
+        _data['data'] = format(_data['data']) 
+        
+        if 'analysis' in _data:
+            for _item in _data['analysis'] :
+                if 'data' in _item :
+                    _item['data'] = format(_item['data'])
+        #
+        #
+        return _data
+ 
+    except Exception as e:
+        print ([' Error ',e])
+        pass
+
+   return _stream