unique-swot 0.1.0__tar.gz

unique_swot-0.1.0/LICENSE

@@ -0,0 +1 @@
+`unique_swot` is covered by the [`Unique License v1`](https://github.com/Unique-AG/license/releases/tag/unique-license.v1), unless the/a header or a nested LICENSE specifies another license.

unique_swot-0.1.0/PKG-INFO

@@ -0,0 +1,215 @@
+Metadata-Version: 2.1
+Name: unique-swot
+Version: 0.1.0
+Summary: Unique Swot package is a tool that for conducting a swot analysis based on internal and external documents
+License: Proprietary
+Author: Rami Azouz
+Author-email: rami.ext@unique.ch
+Requires-Python: >=3.12,<4.0
+Classifier: License :: Other/Proprietary License
+Classifier: Programming Language :: Python :: 3
+Classifier: Programming Language :: Python :: 3.12
+Requires-Dist: beautifulsoup4 (>=4.14.2,<5.0.0)
+Requires-Dist: jinja2 (>=3.1.0,<4.0.0)
+Requires-Dist: pydantic (>=2.8.2,<3.0.0)
+Requires-Dist: python-docx (>=1.2.0,<2.0.0)
+Requires-Dist: ruff (>=0.11.7,<0.12.0)
+Requires-Dist: unique-toolkit (>=1.22.1,<2.0.0)
+Description-Content-Type: text/markdown
+
+# SWOT Analysis Tool
+
+A sophisticated AI-powered tool for generating comprehensive SWOT (Strengths, Weaknesses, Opportunities, Threats) analysis reports based on internal documents, earnings calls, and external web resources.
+
+## Overview
+
+The SWOT Analysis Tool is an agentic tool built on the Unique Toolkit framework that automatically analyzes multiple data sources to produce structured, well-cited SWOT analysis reports. It leverages large language models with structured output to extract insights and generate professional-grade analysis documents.
+
+## Architecture
+
+![Overall Architecture Flow](docs/images/Overall%20Architecture%20Flow.svg)
+
+## Project Structure
+```
+unique_swot/
+├── service.py                 # Main SwotAnalysisTool
+├── services/
+│   ├── citations.py          # Citation management
+│   ├── executor.py           # Execution orchestration
+│   ├── notifier.py           # Progress notifications
+│   ├── collection/           # Source collection
+│   ├── generation/           # LLM-based generation
+│   ├── memory/              # State management
+│   └── report/              # Report rendering
+├── tests/                   # Comprehensive test suite
+└── docs/                    # Architecture documentation
+```
+
+## How It Works
+
+### 1. **Plan Reception & Validation**
+The tool receives a SWOT plan specifying which components to analyze (Strengths, Weaknesses, Opportunities, Threats) and the analysis objective. Each component can be set to:
+- `GENERATE` - Create new analysis from sources
+- `MODIFY` - Update existing analysis with new information
+- `NOT_REQUESTED` - Skip this component
+
+### 2. **Multi-Source Data Collection**
+The tool collects relevant information from multiple sources:
+- **Knowledge Base**: Internal documents filtered by metadata
+- **Earnings Calls**: Financial earnings call transcripts
+- **Web Sources**: External web research and articles
+
+All collected content is registered in a central registry with unique identifiers for citation tracking.
+
+### 3. **Two-Phase Generation Process**
+
+For each requested SWOT component:
+
+#### Phase 1: Extraction
+- Sources are split into manageable batches based on token limits
+- Each batch is processed by the language model with structured output enforcement
+- The LLM extracts specific SWOT items (e.g., individual strengths) with reasoning
+- Extracted items include citation placeholders linking back to source chunks
+- Results from all batches are aggregated into a structured model
+
+#### Phase 2: Summarization
+- The aggregated extraction results are synthesized
+- The LLM generates a coherent narrative summary
+- The summary maintains all citation references
+- Output is formatted with proper structure and citations
+
+### 4. **Citation Management**
+The tool implements a sophisticated dual-citation system:
+
+- **Inline Citations**: `[bullet_chunk_X]` → Converted to `[1]`, `[2]`, etc.
+- **Consolidated References**: `[chunk_X]` → Expanded to `[1] [Document Title: page 5]`
+
+This ensures every claim in the report is traceable back to its source material.
+
+### 5. **Report Delivery**
+
+The final report can be delivered in two formats:
+
+**DOCX Mode** (Document):
+- Markdown converted to professional Word document
+- Full citations with document titles and page numbers
+- Uploaded as downloadable attachment
+
+**Chat Mode** (Markdown):
+- Rich markdown formatting displayed in chat
+- Inline superscript citations
+- Clickable references to source documents
+
+### 6. **Memory & Caching**
+- Extraction results are cached in memory for potential modifications
+- Content chunk registry persisted for citation lookup
+- Supports iterative refinement of analysis
+
+## Key Features
+
+### 🎯 Comprehensive Analysis
+- Analyzes all four SWOT dimensions
+- Processes multiple sources simultaneously
+- Maintains context across large document sets
+
+### 📊 Intelligent Batch Processing
+- Automatically splits sources based on token limits
+- Processes batches in parallel where possible (planned but not supported yet)
+- Handles sources of any size
+
+### 🔗 Advanced Citation System
+- Every point backed by source references
+- Dual-level citation (inline + consolidated)
+- Traceable to specific pages and documents
+
+### 📈 Real-Time Progress Tracking
+- Step-by-step progress updates
+- Percentage completion calculation
+- Detailed execution logs
+
+### 💾 State Management
+- Caches extraction results for modifications
+- Persistent storage in Knowledge Base
+- Quick lookups via Short-Term Memory
+
+### 🎨 Multiple Output Formats
+- Professional DOCX reports
+- Rich markdown for chat
+- Customizable templates
+
+### 🛡️ Robust Error Handling
+- Graceful degradation on failures
+- Automatic retry logic
+- Detailed error logging
+
+## Workflow Example
+
+```
+User Request → SWOT Plan
+    ↓
+Source Collection (KB + Earnings + Web)
+    ↓
+For Each Component (Strengths, Weaknesses, Opportunities, Threats):
+    ├── Split Sources into Batches
+    ├── Extract SWOT Items from Each Batch (LLM)
+    ├── Aggregate Extraction Results
+    ├── Generate Summary (LLM)
+    └── Save to Memory
+    ↓
+Format Citations
+    ↓
+Render Report (DOCX or Markdown)
+    ↓
+Deliver to User
+```
+
+## Core Services
+
+| Service | Responsibility |
+|---------|---------------|
+| **SwotAnalysisTool** | Main orchestrator, validates plans, manages workflow |
+| **SourceCollectionManager** | Collects data from KB, earnings calls, and web |
+| **SWOTExecutionManager** | Executes analysis for each SWOT component |
+| **BatchProcessor** | Manages batch processing with token limits |
+| **CitationManager** | Tracks and formats citations |
+| **MemoryService** | Persists state and caches results |
+| **ReportDeliveryService** | Renders and delivers final reports |
+| **ProgressNotifier** | Provides real-time progress updates |
+
+## Configuration
+
+The tool is configured via `SwotAnalysisToolConfig` which includes:
+- Report generation settings (batch size, token limits)
+- Language model configuration
+- Prompt templates for extraction and summarization
+- Report rendering preferences (DOCX vs Chat)
+- Cache scope for state management
+
+## Technology Stack
+
+- **Framework**: Unique Toolkit (Python)
+- **Validation**: Pydantic models with strict typing
+- **LLM Integration**: Structured output with schema enforcement
+- **Storage**: Knowledge Base for persistence, Short-Term Memory for caching
+- **Templating**: Jinja2 for report templates
+- **Document Processing**: DOCX generation, Markdown rendering
+- **Progress Tracking**: Real-time message execution updates
+
+## Use Cases
+
+- **Strategic Planning**: Generate comprehensive SWOT analyses for business strategy
+- **Market Analysis**: Analyze market position based on multiple data sources
+- **Competitive Intelligence**: Assess strengths and weaknesses vs competitors
+- **Investment Research**: Evaluate opportunities and threats for investments
+- **Due Diligence**: Comprehensive analysis for M&A or partnerships
+
+## Output Quality
+
+The tool produces high-quality analysis by:
+- Using structured output to enforce consistency
+- Extracting before summarizing for thorough coverage
+- Maintaining citation integrity throughout the pipeline
+- Validating all outputs against Pydantic schemas
+- Applying domain-specific prompt engineering
+
+

unique_swot-0.1.0/README.md

@@ -0,0 +1,195 @@
+# SWOT Analysis Tool
+
+A sophisticated AI-powered tool for generating comprehensive SWOT (Strengths, Weaknesses, Opportunities, Threats) analysis reports based on internal documents, earnings calls, and external web resources.
+
+## Overview
+
+The SWOT Analysis Tool is an agentic tool built on the Unique Toolkit framework that automatically analyzes multiple data sources to produce structured, well-cited SWOT analysis reports. It leverages large language models with structured output to extract insights and generate professional-grade analysis documents.
+
+## Architecture
+
+![Overall Architecture Flow](docs/images/Overall%20Architecture%20Flow.svg)
+
+## Project Structure
+```
+unique_swot/
+├── service.py                 # Main SwotAnalysisTool
+├── services/
+│   ├── citations.py          # Citation management
+│   ├── executor.py           # Execution orchestration
+│   ├── notifier.py           # Progress notifications
+│   ├── collection/           # Source collection
+│   ├── generation/           # LLM-based generation
+│   ├── memory/              # State management
+│   └── report/              # Report rendering
+├── tests/                   # Comprehensive test suite
+└── docs/                    # Architecture documentation
+```
+
+## How It Works
+
+### 1. **Plan Reception & Validation**
+The tool receives a SWOT plan specifying which components to analyze (Strengths, Weaknesses, Opportunities, Threats) and the analysis objective. Each component can be set to:
+- `GENERATE` - Create new analysis from sources
+- `MODIFY` - Update existing analysis with new information
+- `NOT_REQUESTED` - Skip this component
+
+### 2. **Multi-Source Data Collection**
+The tool collects relevant information from multiple sources:
+- **Knowledge Base**: Internal documents filtered by metadata
+- **Earnings Calls**: Financial earnings call transcripts
+- **Web Sources**: External web research and articles
+
+All collected content is registered in a central registry with unique identifiers for citation tracking.
+
+### 3. **Two-Phase Generation Process**
+
+For each requested SWOT component:
+
+#### Phase 1: Extraction
+- Sources are split into manageable batches based on token limits
+- Each batch is processed by the language model with structured output enforcement
+- The LLM extracts specific SWOT items (e.g., individual strengths) with reasoning
+- Extracted items include citation placeholders linking back to source chunks
+- Results from all batches are aggregated into a structured model
+
+#### Phase 2: Summarization
+- The aggregated extraction results are synthesized
+- The LLM generates a coherent narrative summary
+- The summary maintains all citation references
+- Output is formatted with proper structure and citations
+
+### 4. **Citation Management**
+The tool implements a sophisticated dual-citation system:
+
+- **Inline Citations**: `[bullet_chunk_X]` → Converted to `[1]`, `[2]`, etc.
+- **Consolidated References**: `[chunk_X]` → Expanded to `[1] [Document Title: page 5]`
+
+This ensures every claim in the report is traceable back to its source material.
+
+### 5. **Report Delivery**
+
+The final report can be delivered in two formats:
+
+**DOCX Mode** (Document):
+- Markdown converted to professional Word document
+- Full citations with document titles and page numbers
+- Uploaded as downloadable attachment
+
+**Chat Mode** (Markdown):
+- Rich markdown formatting displayed in chat
+- Inline superscript citations
+- Clickable references to source documents
+
+### 6. **Memory & Caching**
+- Extraction results are cached in memory for potential modifications
+- Content chunk registry persisted for citation lookup
+- Supports iterative refinement of analysis
+
+## Key Features
+
+### 🎯 Comprehensive Analysis
+- Analyzes all four SWOT dimensions
+- Processes multiple sources simultaneously
+- Maintains context across large document sets
+
+### 📊 Intelligent Batch Processing
+- Automatically splits sources based on token limits
+- Processes batches in parallel where possible (planned but not supported yet)
+- Handles sources of any size
+
+### 🔗 Advanced Citation System
+- Every point backed by source references
+- Dual-level citation (inline + consolidated)
+- Traceable to specific pages and documents
+
+### 📈 Real-Time Progress Tracking
+- Step-by-step progress updates
+- Percentage completion calculation
+- Detailed execution logs
+
+### 💾 State Management
+- Caches extraction results for modifications
+- Persistent storage in Knowledge Base
+- Quick lookups via Short-Term Memory
+
+### 🎨 Multiple Output Formats
+- Professional DOCX reports
+- Rich markdown for chat
+- Customizable templates
+
+### 🛡️ Robust Error Handling
+- Graceful degradation on failures
+- Automatic retry logic
+- Detailed error logging
+
+## Workflow Example
+
+```
+User Request → SWOT Plan
+    ↓
+Source Collection (KB + Earnings + Web)
+    ↓
+For Each Component (Strengths, Weaknesses, Opportunities, Threats):
+    ├── Split Sources into Batches
+    ├── Extract SWOT Items from Each Batch (LLM)
+    ├── Aggregate Extraction Results
+    ├── Generate Summary (LLM)
+    └── Save to Memory
+    ↓
+Format Citations
+    ↓
+Render Report (DOCX or Markdown)
+    ↓
+Deliver to User
+```
+
+## Core Services
+
+| Service | Responsibility |
+|---------|---------------|
+| **SwotAnalysisTool** | Main orchestrator, validates plans, manages workflow |
+| **SourceCollectionManager** | Collects data from KB, earnings calls, and web |
+| **SWOTExecutionManager** | Executes analysis for each SWOT component |
+| **BatchProcessor** | Manages batch processing with token limits |
+| **CitationManager** | Tracks and formats citations |
+| **MemoryService** | Persists state and caches results |
+| **ReportDeliveryService** | Renders and delivers final reports |
+| **ProgressNotifier** | Provides real-time progress updates |
+
+## Configuration
+
+The tool is configured via `SwotAnalysisToolConfig` which includes:
+- Report generation settings (batch size, token limits)
+- Language model configuration
+- Prompt templates for extraction and summarization
+- Report rendering preferences (DOCX vs Chat)
+- Cache scope for state management
+
+## Technology Stack
+
+- **Framework**: Unique Toolkit (Python)
+- **Validation**: Pydantic models with strict typing
+- **LLM Integration**: Structured output with schema enforcement
+- **Storage**: Knowledge Base for persistence, Short-Term Memory for caching
+- **Templating**: Jinja2 for report templates
+- **Document Processing**: DOCX generation, Markdown rendering
+- **Progress Tracking**: Real-time message execution updates
+
+## Use Cases
+
+- **Strategic Planning**: Generate comprehensive SWOT analyses for business strategy
+- **Market Analysis**: Analyze market position based on multiple data sources
+- **Competitive Intelligence**: Assess strengths and weaknesses vs competitors
+- **Investment Research**: Evaluate opportunities and threats for investments
+- **Due Diligence**: Comprehensive analysis for M&A or partnerships
+
+## Output Quality
+
+The tool produces high-quality analysis by:
+- Using structured output to enforce consistency
+- Extracting before summarizing for thorough coverage
+- Maintaining citation integrity throughout the pipeline
+- Validating all outputs against Pydantic schemas
+- Applying domain-specific prompt engineering
+

unique_swot-0.1.0/pyproject.toml

@@ -0,0 +1,36 @@
+[tool.poetry]
+name = "unique-swot"
+version = "0.1.0"
+description = "Unique Swot package is a tool that for conducting a swot analysis based on internal and external documents"
+authors = ["Rami Azouz <rami.ext@unique.ch>", "Fabian Schläpfer <fabian.schlaepfer@unique.ai>"]
+license = "Proprietary"
+readme = "README.md"
+
+
+[tool.poetry.dependencies]
+python = "^3.12"
+pydantic = "^2.8.2"
+ruff = "^0.11.7"
+jinja2 = "^3.1.0"
+python-docx = "^1.2.0"
+beautifulsoup4 = "^4.14.2"
+unique-toolkit = "^1.22.1"
+
+[tool.poetry.group.dev.dependencies]
+python = "^3.12"
+pydantic = "^2.8.2"
+pytest = "^8.3.5"
+pytest-mock = "^3.14.0"
+pytest-asyncio = "^1.2.0"
+unique-toolkit = { path = "../../unique_toolkit",  develop = true}
+
+[build-system]
+requires = ["poetry-core>=2.0.0,<3.0.0"]
+build-backend = "poetry.core.masonry.api"
+
+
+[tool.ruff]
+target-version = "py311"
+
+[tool.ruff.lint]
+extend-select = ["I"]

unique_swot-0.1.0/unique_swot/__init__.py

@@ -0,0 +1,7 @@
+from unique_swot.config import SwotAnalysisToolConfig
+from unique_swot.service import SwotAnalysisTool
+
+__all__ = [
+    "SwotAnalysisTool",
+    "SwotAnalysisToolConfig",
+]

unique_swot-0.1.0/unique_swot/config.py

@@ -0,0 +1,53 @@
+from pydantic import Field
+from pydantic.json_schema import SkipJsonSchema
+from unique_toolkit.agentic.tools.schemas import BaseToolConfig
+
+from unique_swot.services.generation import ReportGenerationConfig
+from unique_swot.services.report import ReportRendererConfig
+
+TOOL_DESCRIPTION = """
+This tool is used to perfom a SWOT analysis of a company by analyzing its strengths, weaknesses, opportunities, and threats.
+The user can either generate a new SWOT analysis or modify an existing one.
+If the user simply says RUN, the tool will generate a new SWOT analysis.
+
+If the user simply says RUN, It means that he expects the tool to generate a new SWOT analysis.
+"""
+
+
+class SwotAnalysisToolConfig(BaseToolConfig):
+    cache_scope_id: str = Field(
+        default="",
+        description="The scope id for the SWOT analysis cache.",
+    )
+    report_generation_config: ReportGenerationConfig = Field(
+        default_factory=ReportGenerationConfig,
+        description="The configuration for the report generation.",
+    )
+    report_renderer_config: ReportRendererConfig = Field(
+        default_factory=ReportRendererConfig,
+        description="The configuration for the report renderer.",
+    )
+    tool_description: str = Field(
+        default=TOOL_DESCRIPTION,
+        description="The description of the SWOT analysis tool.",
+    )
+    tool_description_for_system_prompt: str = Field(
+        default=TOOL_DESCRIPTION,
+        description="The system prompt for the SWOT analysis tool.",
+    )
+    tool_format_information_for_system_prompt: SkipJsonSchema[str] = Field(
+        default="",
+        description="The format information for the SWOT analysis tool.",
+    )
+    tool_description_for_user_prompt: SkipJsonSchema[str] = Field(
+        default="",
+        description="The user prompt for the SWOT analysis tool.",
+    )
+    tool_format_information_for_user_prompt: SkipJsonSchema[str] = Field(
+        default="",
+        description="The format information for the SWOT analysis tool.",
+    )
+    tool_format_reminder_for_user_prompt: SkipJsonSchema[str] = Field(
+        default="",
+        description="The format reminder for the SWOT analysis tool.",
+    )

unique_swot-0.1.0/unique_swot/docs/swot_generator_architecture.md

@@ -0,0 +1,89 @@
+---
+config:
+  layout: elk
+---
+flowchart TB
+ subgraph subGraph0["Generator Entry Point"]
+        A["Generation Request"]
+        B["Component Type"]
+        C["Sources"]
+  end
+ subgraph subGraph8["Request Types"]
+        D["Generation"]
+        E["Modification"]
+  end
+ subgraph subGraph1["Prompt Management System"]
+        F["Prompt Router"]
+        G["Strengths Prompt"]
+        H["Weaknesses Prompt"]
+        I["Opportunities Prompt"]
+        J["Threats Prompt"]
+  end
+ subgraph subGraph3["Processing Engine"]
+        N["Batch Processing"]
+        O["LLM Integration"]
+        P["Result Assembly"]
+  end
+ subgraph subGraph7["Prompt Features"]
+        CC["Component-Specific Instructions"]
+        DD["Output Format Guidelines"]
+        EE["Reference Requirements"]
+        FF["Analysis Depth Specifications"]
+  end
+ subgraph subGraph6["Prompt Templates"]
+        Y["Strengths Template"]
+        Z["Weaknesses Template"]
+        AA["Opportunities Template"]
+        BB["Threats Template"]
+        subGraph7
+  end
+     F --> G & H & I & J
+     Y --> CC
+     Z --> CC
+     AA --> CC
+     BB --> CC
+     CC --> DD
+     DD --> EE
+     EE --> FF
+     A --> subGraph8
+     B --> subGraph1
+     C --> subGraph3
+     subGraph8 --> subGraph1
+     subGraph1 --> subGraph6
+     subGraph3 --> subGraph1
+     A:::entry
+     B:::entry
+     C:::entry
+     D:::requestType
+     E:::requestType
+     F:::promptRouter
+     G:::strengthsPrompt
+     H:::weaknessesPrompt
+     I:::opportunitiesPrompt
+     J:::threatsPrompt
+     N:::processingEngine
+     O:::processingEngine
+     P:::processingEngine
+     CC:::promptFeature
+     DD:::promptFeature
+     EE:::promptFeature
+     FF:::promptFeature
+     Y:::promptTemplate
+     Z:::promptTemplate
+     AA:::promptTemplate
+     BB:::promptTemplate
+    classDef entry fill:#e3f2fd,stroke:#1976d2,stroke-width:3px
+    classDef requestType fill:#e8f5e8,stroke:#4caf50,stroke-width:2px
+    classDef promptRouter fill:#f3e5f5,stroke:#7b1fa2,stroke-width:3px
+    classDef strengthsPrompt fill:#e8f5e8,stroke:#4caf50,stroke-width:2px
+    classDef weaknessesPrompt fill:#fff3e0,stroke:#ff9800,stroke-width:2px
+    classDef opportunitiesPrompt fill:#e1f5fe,stroke:#2196f3,stroke-width:2px
+    classDef threatsPrompt fill:#ffebee,stroke:#f44336,stroke-width:2px
+    classDef modelRouter fill:#f3e5f5,stroke:#7b1fa2,stroke-width:3px
+    classDef strengthsModel fill:#e8f5e8,stroke:#4caf50,stroke-width:2px
+    classDef weaknessesModel fill:#fff3e0,stroke:#ff9800,stroke-width:2px
+    classDef opportunitiesModel fill:#e1f5fe,stroke:#2196f3,stroke-width:2px
+    classDef threatsModel fill:#ffebee,stroke:#f44336,stroke-width:2px
+    classDef processingEngine fill:#fff8e1,stroke:#f57c00,stroke-width:2px
+    classDef promptTemplate fill:#fce4ec,stroke:#c2185b,stroke-width:2px
+    classDef promptFeature fill:#f3e5f5,stroke:#7b1fa2,stroke-width:1px