scientific-writer 2.0.0__py3-none-any.whl

scientific_writer/cli.py

@@ -0,0 +1,295 @@
+#!/usr/bin/env python3
+"""
+Scientific Writer CLI Tool
+A command-line interface for scientific writing powered by Claude Code.
+"""
+
+import os
+import sys
+import time
+import asyncio
+from pathlib import Path
+from typing import Optional
+
+from claude_agent_sdk import query, ClaudeAgentOptions
+
+from .core import (
+    get_api_key,
+    load_system_instructions,
+    ensure_output_folder,
+    get_data_files,
+    process_data_files,
+    create_data_context_message,
+)
+from .utils import find_existing_papers, detect_paper_reference
+
+
+async def main():
+    """Main CLI loop for the scientific writer."""
+    # Get API key (verify it exists)
+    try:
+        get_api_key()
+    except ValueError as e:
+        print(f"Error: {e}")
+        sys.exit(1)
+    
+    # Get the current working directory (project root)
+    cwd = Path(__file__).parent.parent.absolute()
+    
+    # Ensure paper_outputs folder exists
+    output_folder = ensure_output_folder(cwd)
+    
+    # Load system instructions from CLAUDE.md
+    system_instructions = load_system_instructions(cwd)
+    
+    # Add conversation continuity instruction  
+    # Note: The Python CLI handles session tracking via current_paper_path
+    # These instructions only apply WITHIN a single CLI session, not across different chat sessions
+    system_instructions += "\n\n" + """
+IMPORTANT - CONVERSATION CONTINUITY:
+- The user will provide context in their prompt if they want to continue working on an existing paper
+- If the prompt includes [CONTEXT: You are currently working on a paper in: ...], continue editing that paper
+- If no such context is provided, this is a NEW paper request - create a new paper directory
+- Do NOT assume there's an existing paper unless explicitly told in the prompt context
+- Each new chat session should start with a new paper unless context says otherwise
+"""
+    
+    # Configure the Claude agent options
+    options = ClaudeAgentOptions(
+        system_prompt=system_instructions,
+        model="claude-sonnet-4-20250514",  # Always use Claude Sonnet 4.5
+        allowed_tools=["Read", "Write", "Edit", "Bash", "research-lookup"],  # Default Claude Code tools + research lookup
+        permission_mode="bypassPermissions",  # Execute immediately without approval prompts
+        setting_sources=["project"],  # Load skills from .claude/skills/
+        cwd=str(cwd),  # Set working directory to project root
+    )
+    
+    # Track conversation state
+    current_paper_path = None
+    conversation_history = []
+    
+    # Print welcome message
+    print("=" * 70)
+    print("Scientific Writer CLI - Powered by Claude Sonnet 4.5")
+    print("=" * 70)
+    print("\nWelcome! I'm your scientific writing assistant.")
+    print("\nI can help you with:")
+    print("  • Writing scientific papers (IMRaD structure)")
+    print("  • Literature reviews and citation management")
+    print("  • Peer review feedback")
+    print("  • Real-time research lookup using Perplexity Sonar Pro")
+    print("  • Document manipulation (docx, pdf, pptx, xlsx)")
+    print("\n📋 Workflow:")
+    print("  1. I'll present a brief plan and immediately start execution")
+    print("  2. I'll provide continuous updates during the process")
+    print("  3. All outputs saved to: paper_outputs/<timestamp_description>/")
+    print("  4. Progress tracked in real-time in progress.md")
+    print(f"\n📁 Output folder: {output_folder}")
+    print(f"\n📦 Data Files:")
+    print("  • Place files in the 'data/' folder to include them in your paper")
+    print("  • Data files → copied to paper's data/ folder")
+    print("  • Images → copied to paper's figures/ folder")
+    print("  • Original files are automatically deleted after copying")
+    print("\n🤖 Intelligent Paper Detection:")
+    print("  • I automatically detect when you're referring to a previous paper")
+    print("  • Continue: 'continue', 'update', 'edit', 'the paper', etc.")
+    print("  • Search: 'look for', 'find', 'show me', 'where is', etc.")
+    print("  • Or reference the paper topic (e.g., 'find the acoustics paper')")
+    print("  • Say 'new paper' to explicitly start a fresh paper")
+    print("\nType 'exit' or 'quit' to end the session.")
+    print("Type 'help' for usage tips.")
+    print("=" * 70)
+    print()
+    
+    # Main loop
+    while True:
+        try:
+            # Get user input
+            user_input = input("\n> ").strip()
+            
+            # Handle special commands
+            if user_input.lower() in ["exit", "quit"]:
+                print("\nThank you for using Scientific Writer CLI. Goodbye!")
+                break
+            
+            if user_input.lower() == "help":
+                _print_help()
+                continue
+            
+            if not user_input:
+                continue
+            
+            # Get all existing papers
+            existing_papers = find_existing_papers(output_folder)
+            
+            # Check if user wants to start a new paper
+            new_paper_keywords = ["new paper", "start fresh", "start afresh", "create new", "different paper", "another paper"]
+            is_new_paper_request = any(keyword in user_input.lower() for keyword in new_paper_keywords)
+            
+            # Try to detect reference to existing paper
+            detected_paper_path = None
+            if not is_new_paper_request:
+                detected_paper_path = detect_paper_reference(user_input, existing_papers)
+                
+                # If we detected a paper reference and it's different from current, update it
+                if detected_paper_path and str(detected_paper_path) != current_paper_path:
+                    current_paper_path = str(detected_paper_path)
+                    print(f"\n🔍 Detected reference to existing paper: {detected_paper_path.name}")
+                    print(f"📂 Working on: {current_paper_path}\n")
+                elif detected_paper_path and str(detected_paper_path) == current_paper_path:
+                    # Already working on the right paper, just confirm
+                    print(f"📂 Continuing with: {Path(current_paper_path).name}\n")
+            
+            # Check for data files and process them if we have a current paper
+            data_context = ""
+            data_files = get_data_files(cwd)
+            
+            if data_files and current_paper_path and not is_new_paper_request:
+                print(f"📦 Found {len(data_files)} file(s) in data folder. Processing...")
+                processed_info = process_data_files(cwd, data_files, current_paper_path)
+                if processed_info:
+                    data_context = create_data_context_message(processed_info)
+                    data_count = len(processed_info['data_files'])
+                    image_count = len(processed_info['image_files'])
+                    if data_count > 0:
+                        print(f"   ✓ Copied {data_count} data file(s) to data/")
+                    if image_count > 0:
+                        print(f"   ✓ Copied {image_count} image(s) to figures/")
+                    print("   ✓ Deleted original files from data folder\n")
+            elif data_files and not current_paper_path:
+                # Store data files info for later processing once paper is created
+                print(f"\n📦 Found {len(data_files)} file(s) in data folder.")
+                print("   They will be processed once the paper directory is created.\n")
+            
+            # Build contextual prompt
+            contextual_prompt = user_input
+            
+            # Add context about current paper if one exists and not starting new
+            if current_paper_path and not is_new_paper_request:
+                contextual_prompt = f"""[CONTEXT: You are currently working on a paper in: {current_paper_path}]
+[INSTRUCTION: Continue editing this existing paper. Do NOT create a new paper directory.]
+{data_context}
+User request: {user_input}"""
+            elif is_new_paper_request:
+                # Reset paper tracking when explicitly starting new
+                current_paper_path = None
+                print("📝 Starting a new paper...\n")
+            
+            # Send query to Claude
+            print()  # Add blank line before response
+            async for message in query(prompt=contextual_prompt, options=options):
+                # Handle AssistantMessage with content blocks
+                if hasattr(message, "content") and message.content:
+                    for block in message.content:
+                        if hasattr(block, "text"):
+                            print(block.text, end="", flush=True)
+            
+            print()  # Add blank line after response
+            
+            # Try to detect if a new paper directory was created
+            if not current_paper_path or is_new_paper_request:
+                # Look for the most recently modified directory in paper_outputs
+                # Only update if it was modified in the last 10 seconds (indicating it was just created)
+                try:
+                    paper_dirs = [d for d in output_folder.iterdir() if d.is_dir()]
+                    if paper_dirs:
+                        most_recent = max(paper_dirs, key=lambda d: d.stat().st_mtime)
+                        time_since_modification = time.time() - most_recent.stat().st_mtime
+                        
+                        # Only set as current paper if it was modified very recently (within last 10 seconds)
+                        if time_since_modification < 10:
+                            current_paper_path = str(most_recent)
+                            print(f"\n📂 Working on: {most_recent.name}")
+                            
+                            # Process any remaining data files now that we have a paper path
+                            remaining_data_files = get_data_files(cwd)
+                            if remaining_data_files:
+                                print(f"\n📦 Processing {len(remaining_data_files)} data file(s)...")
+                                processed_info = process_data_files(cwd, remaining_data_files, current_paper_path)
+                                if processed_info:
+                                    data_count = len(processed_info['data_files'])
+                                    image_count = len(processed_info['image_files'])
+                                    if data_count > 0:
+                                        print(f"   ✓ Copied {data_count} data file(s) to data/")
+                                    if image_count > 0:
+                                        print(f"   ✓ Copied {image_count} image(s) to figures/")
+                                    print("   ✓ Deleted original files from data folder")
+                except Exception:
+                    pass  # Silently fail if we can't detect the directory
+            
+        except KeyboardInterrupt:
+            print("\n\nInterrupted. Type 'exit' to quit or continue with a new prompt.")
+            continue
+        except Exception as e:
+            print(f"\nError: {str(e)}")
+            print("Please try again or type 'exit' to quit.")
+
+
+def _print_help():
+    """Print help information."""
+    print("\n" + "=" * 70)
+    print("HELP - Scientific Writer CLI")
+    print("=" * 70)
+    print("\n📝 What I Can Do:")
+    print("  • Create complete scientific papers (LaTeX, Word, Markdown)")
+    print("  • Literature reviews with citation management")
+    print("  • Peer review feedback on drafts")
+    print("  • Real-time research lookup using Perplexity Sonar Pro")
+    print("  • Format citations in any style (APA, IEEE, Nature, etc.)")
+    print("  • Document manipulation (docx, pdf, pptx, xlsx)")
+    print("\n🔄 How I Work:")
+    print("  1. You describe what you need")
+    print("  2. I present a brief plan and start execution immediately")
+    print("  3. I provide continuous progress updates")
+    print("  4. All files organized in paper_outputs/ folder")
+    print("\n💡 Example Requests:")
+    print("  'Create a NeurIPS paper on transformer attention mechanisms'")
+    print("  'Write a literature review on CRISPR gene editing'")
+    print("  'Review my methods section in draft.docx'")
+    print("  'Research recent advances in quantum computing 2024'")
+    print("  'Create a Nature paper on climate change impacts'")
+    print("  'Format 20 citations in IEEE style'")
+    print("\n📁 File Organization:")
+    print("  All work saved to: paper_outputs/<timestamp>_<description>/")
+    print("  - drafts/ - Working versions")
+    print("  - final/ - Completed documents")
+    print("  - references/ - Bibliography files")
+    print("  - figures/ - Images and charts")
+    print("  - data/ - Data files for the paper")
+    print("  - progress.md - Real-time progress log")
+    print("  - SUMMARY.md - Project summary and instructions")
+    print("\n📦 Data Files:")
+    print("  Place files in the 'data/' folder at project root:")
+    print("  • Data files (csv, txt, json, etc.) → copied to paper's data/")
+    print("  • Images (png, jpg, svg, etc.) → copied to paper's figures/")
+    print("  • Files are used as context for the paper")
+    print("  • Original files automatically deleted after copying")
+    print("\n🎯 Pro Tips:")
+    print("  • Be specific about journal/conference (e.g., 'Nature', 'NeurIPS')")
+    print("  • Mention citation style if you have a preference")
+    print("  • I'll make smart defaults if you don't specify details")
+    print("  • Check progress.md for detailed execution logs")
+    print("\n🔄 Intelligent Paper Detection:")
+    print("  • I automatically detect when you're referring to a previous paper")
+    print("  • Continue working: 'continue the paper', 'update my paper', 'edit the poster'")
+    print("  • Search/find: 'look for the X paper', 'find the paper about Y'")
+    print("  • Or mention the paper topic: 'show me the acoustics paper'")
+    print("  • Keywords like 'continue', 'update', 'edit', 'look for', 'find' trigger detection")
+    print("  • I'll find the most relevant paper based on topic matching")
+    print("  • Say 'new paper' or 'start fresh' to explicitly begin a new one")
+    print("  • Current working paper is tracked throughout the session")
+    print("=" * 70)
+
+
+def cli_main():
+    """Entry point for the CLI script."""
+    try:
+        asyncio.run(main())
+    except KeyboardInterrupt:
+        print("\n\nExiting...")
+        sys.exit(0)
+
+
+if __name__ == "__main__":
+    cli_main()
+

scientific_writer/core.py

@@ -0,0 +1,219 @@
+"""Core utilities for scientific writer."""
+
+import os
+import shutil
+from pathlib import Path
+from typing import Optional, List, Dict, Any
+from dotenv import load_dotenv
+
+# Load environment variables from .env file if it exists
+load_dotenv()
+
+
+def get_api_key(api_key: Optional[str] = None) -> str:
+    """
+    Get the Anthropic API key.
+    
+    Args:
+        api_key: Optional API key to use. If not provided, reads from environment.
+        
+    Returns:
+        The API key.
+        
+    Raises:
+        ValueError: If API key is not found.
+    """
+    if api_key:
+        return api_key
+    
+    env_key = os.getenv("ANTHROPIC_API_KEY")
+    if not env_key:
+        raise ValueError(
+            "ANTHROPIC_API_KEY not found. Either pass api_key parameter or set "
+            "ANTHROPIC_API_KEY environment variable."
+        )
+    return env_key
+
+
+def load_system_instructions(cwd: Path) -> str:
+    """
+    Load system instructions from CLAUDE.md file.
+    
+    Args:
+        cwd: Current working directory (project root).
+        
+    Returns:
+        System instructions string.
+    """
+    instructions_file = cwd / "CLAUDE.md"
+    
+    if instructions_file.exists():
+        with open(instructions_file, 'r', encoding='utf-8') as f:
+            return f.read()
+    else:
+        # Fallback if CLAUDE.md doesn't exist
+        return (
+            "You are a scientific writing assistant. Follow best practices for "
+            "scientific communication and always present a plan before execution."
+        )
+
+
+def ensure_output_folder(cwd: Path, custom_dir: Optional[str] = None) -> Path:
+    """
+    Ensure the paper_outputs folder exists.
+    
+    Args:
+        cwd: Current working directory (project root).
+        custom_dir: Optional custom output directory path.
+        
+    Returns:
+        Path to the output folder.
+    """
+    if custom_dir:
+        output_folder = Path(custom_dir).resolve()
+    else:
+        output_folder = cwd / "paper_outputs"
+    
+    output_folder.mkdir(exist_ok=True, parents=True)
+    return output_folder
+
+
+def get_image_extensions() -> set:
+    """Return a set of common image file extensions."""
+    return {'.png', '.jpg', '.jpeg', '.gif', '.bmp', '.tiff', '.tif', '.svg', '.webp', '.ico'}
+
+
+def get_data_files(cwd: Path, data_files: Optional[List[str]] = None) -> List[Path]:
+    """
+    Get data files either from provided list or from data folder.
+    
+    Args:
+        cwd: Current working directory (project root).
+        data_files: Optional list of file paths. If not provided, reads from data/ folder.
+        
+    Returns:
+        List of Path objects for data files.
+    """
+    if data_files:
+        return [Path(f).resolve() for f in data_files]
+    
+    data_folder = cwd / "data"
+    if not data_folder.exists():
+        return []
+    
+    files = []
+    for file_path in data_folder.iterdir():
+        if file_path.is_file():
+            files.append(file_path)
+    
+    return files
+
+
+def process_data_files(
+    cwd: Path, 
+    data_files: List[Path], 
+    paper_output_path: str,
+    delete_originals: bool = True
+) -> Optional[Dict[str, Any]]:
+    """
+    Process data files by copying them to the paper output folder.
+    Images go to figures/, other files go to data/.
+    
+    Args:
+        cwd: Current working directory (project root).
+        data_files: List of file paths to process.
+        paper_output_path: Path to the paper output directory.
+        delete_originals: Whether to delete original files after copying.
+        
+    Returns:
+        Dictionary with information about processed files, or None if no files.
+    """
+    if not data_files:
+        return None
+    
+    paper_output = Path(paper_output_path)
+    data_output = paper_output / "data"
+    figures_output = paper_output / "figures"
+    
+    # Ensure output directories exist
+    data_output.mkdir(parents=True, exist_ok=True)
+    figures_output.mkdir(parents=True, exist_ok=True)
+    
+    image_extensions = get_image_extensions()
+    processed_info = {
+        'data_files': [],
+        'image_files': [],
+        'all_files': []
+    }
+    
+    for file_path in data_files:
+        file_ext = file_path.suffix.lower()
+        file_name = file_path.name
+        
+        # Determine destination based on file type
+        if file_ext in image_extensions:
+            destination = figures_output / file_name
+            file_type = 'image'
+            processed_info['image_files'].append({
+                'name': file_name,
+                'path': str(destination),
+                'original': str(file_path)
+            })
+        else:
+            destination = data_output / file_name
+            file_type = 'data'
+            processed_info['data_files'].append({
+                'name': file_name,
+                'path': str(destination),
+                'original': str(file_path)
+            })
+        
+        # Copy the file
+        try:
+            shutil.copy2(file_path, destination)
+            processed_info['all_files'].append({
+                'name': file_name,
+                'type': file_type,
+                'destination': str(destination)
+            })
+            
+            # Delete the original file after successful copy if requested
+            if delete_originals:
+                file_path.unlink()
+            
+        except Exception as e:
+            print(f"Warning: Could not process {file_name}: {str(e)}")
+    
+    return processed_info
+
+
+def create_data_context_message(processed_info: Optional[Dict[str, Any]]) -> str:
+    """
+    Create a context message about available data files.
+    
+    Args:
+        processed_info: Dictionary with processed file information.
+        
+    Returns:
+        Context message string.
+    """
+    if not processed_info or not processed_info['all_files']:
+        return ""
+    
+    context_parts = ["\n[DATA FILES AVAILABLE]"]
+    
+    if processed_info['data_files']:
+        context_parts.append("\nData files (in data/ folder):")
+        for file_info in processed_info['data_files']:
+            context_parts.append(f"  - {file_info['name']}: {file_info['path']}")
+    
+    if processed_info['image_files']:
+        context_parts.append("\nImage files (in figures/ folder):")
+        for file_info in processed_info['image_files']:
+            context_parts.append(f"  - {file_info['name']}: {file_info['path']}")
+        context_parts.append("\nNote: These images can be referenced as figures in the paper.")
+    
+    context_parts.append("[END DATA FILES]\n")
+    
+    return "\n".join(context_parts)
+

scientific_writer/models.py

@@ -0,0 +1,76 @@
+"""Data models for scientific writer API responses."""
+
+from dataclasses import dataclass, field, asdict
+from typing import Optional, List, Dict, Any
+from datetime import datetime
+
+
+@dataclass
+class ProgressUpdate:
+    """Progress update during paper generation."""
+    type: str = "progress"
+    timestamp: str = field(default_factory=lambda: datetime.utcnow().isoformat() + "Z")
+    message: str = ""
+    stage: str = "initialization"  # initialization|research|writing|compilation|complete
+    percentage: int = 0
+    
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for JSON serialization."""
+        return asdict(self)
+
+
+@dataclass
+class PaperMetadata:
+    """Metadata about the generated paper."""
+    title: Optional[str] = None
+    created_at: str = field(default_factory=lambda: datetime.utcnow().isoformat() + "Z")
+    topic: str = ""
+    word_count: Optional[int] = None
+    
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for JSON serialization."""
+        return asdict(self)
+
+
+@dataclass
+class PaperFiles:
+    """File paths for all generated paper artifacts."""
+    pdf_final: Optional[str] = None
+    tex_final: Optional[str] = None
+    pdf_drafts: List[str] = field(default_factory=list)
+    tex_drafts: List[str] = field(default_factory=list)
+    bibliography: Optional[str] = None
+    figures: List[str] = field(default_factory=list)
+    data: List[str] = field(default_factory=list)
+    progress_log: Optional[str] = None
+    summary: Optional[str] = None
+    
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for JSON serialization."""
+        return asdict(self)
+
+
+@dataclass
+class PaperResult:
+    """Final result containing all information about the generated paper."""
+    type: str = "result"
+    status: str = "success"  # success|partial|failed
+    paper_directory: str = ""
+    paper_name: str = ""
+    metadata: PaperMetadata = field(default_factory=PaperMetadata)
+    files: PaperFiles = field(default_factory=PaperFiles)
+    citations: Dict[str, Any] = field(default_factory=dict)
+    figures_count: int = 0
+    compilation_success: bool = False
+    errors: List[str] = field(default_factory=list)
+    
+    def to_dict(self) -> Dict[str, Any]:
+        """Convert to dictionary for JSON serialization."""
+        result = asdict(self)
+        # Ensure nested objects are also dictionaries
+        if isinstance(self.metadata, PaperMetadata):
+            result['metadata'] = self.metadata.to_dict()
+        if isinstance(self.files, PaperFiles):
+            result['files'] = self.files.to_dict()
+        return result
+