adding pdf-tex summarizer example

Author: HadiYoussef2005

workflows/pdf_summarizer_tex_workflow/.env.example

@@ -0,0 +1,6 @@
+MODEL=gpt-4o-mini
+ENVIRONMENT=dev
+OPENAI_API_KEY=
+OPENROUTER_MODEL=gryphe/mythomax-l2-13b
+SWARMZERO_LOG_LEVEL=INFO
+LANGTRACE_API_KEY=03ef836c358dc2f6edf599393d671ea180f23df1a2e3a99a16748b5adcf530fd

workflows/pdf_summarizer_tex_workflow/.gitignore

@@ -0,0 +1,3 @@
+.venv
+swarmzero-data
+.env

workflows/pdf_summarizer_tex_workflow/README.md

@@ -0,0 +1,39 @@
+# PDF Summarizer to Tex Workflow
+
+A sequential workflow built using SwarmZero framework that enables concise summarization of large pdf's to be outputted in a tex file 
+
+## Description
+
+This workflow utilizes PyMuPDF for extracting text out of a pdf and is built on top fo the SwarmZero framework, providing enhanced summarization cpabilities wiht AI-powered processing. 
+
+## Prerequisites
+
+- Python 3.11 or higher
+- Poetry package manager
+- OpenAI API Key
+- Langtrace API Key
+
+## Installation
+
+1. Clone the repository:
+```bash
+git clone https://github.com/swarmzero/examples.git
+cd examples/workflows/pdf_summarizer_tex_workflow
+```
+
+2. Install dependencies using Poetry:
+```bash
+poetry install --no-root
+```
+
+3. Set up environment variables:
+Create a `.env` file in the root directory and add your API keys based on the .env.example file
+## Usage
+
+Run the workflow on a pdf like so:
+``` bash
+python main.py path/to/pdf/text.pdf
+```
+
+## Learn more
+Visit [SwarmZero](https://swarmzero.ai) to learn more about the SwarmZero framework.

workflows/pdf_summarizer_tex_workflow/main.py

@@ -0,0 +1,144 @@
+import asyncio
+import os
+import sys
+import uuid
+from typing import List
+
+import fitz  # PyMuPDF
+
+from swarmzero.agent import Agent
+from swarmzero.sdk_context import SDKContext
+from swarmzero.workflow import Workflow, WorkflowStep, StepMode
+
+
+def extract_text_from_pdf(pdf_path: str) -> str:
+    doc = fitz.open(pdf_path)
+    text = "\n".join(page.get_text() for page in doc)
+    doc.close()
+    return text
+
+
+def chunk_text(text: str, max_chars: int = 2000) -> List[str]:
+    paragraphs = text.split("\n")
+    chunks, current = [], ""
+    for para in paragraphs:
+        if len(current) + len(para) < max_chars:
+            current += para + "\n"
+        else:
+            chunks.append(current.strip())
+            current = para + "\n"
+    if current:
+        chunks.append(current.strip())
+    return chunks
+
+
+def summarize_bullets(prompt: str, **kwargs):
+    lines = [line.strip() for line in prompt.split(".") if line.strip()]
+    bullets = "\n".join(f"- {line}" for line in lines)
+    return f"[Bullet Point Summary]:\n{bullets}"
+
+def format_latex(prompt: str, **kwargs) -> str:
+    bullet_lines = prompt.strip().split("\n")
+    latex_items = "\n".join(f"\\item {line.lstrip('- ').strip()}" for line in bullet_lines if line.startswith("-"))
+
+    latex_doc = (
+        "\\documentclass{article}\n"
+        "\\usepackage[utf8]{inputenc}\n"
+        "\\usepackage{enumitem}\n"
+        "\\begin{document}\n"
+        "\\section*{Summary Notes}\n"
+        "\\begin{itemize}[leftmargin=*, label=--]\n"
+        f"{latex_items}\n"
+        "\\end{itemize}\n"
+        "\\end{document}"
+    )
+    return latex_doc
+
+
+
+CONFIG_PATH = os.path.join(os.path.dirname(__file__), "swarmzero_config.toml")
+
+sdk_context = SDKContext(CONFIG_PATH)
+
+summarizer_agent = Agent(
+    name="BulletSummarizerAgent",
+    functions=[summarize_bullets],
+    instruction="Summarize input in bullet points.",
+    agent_id=str(uuid.uuid4()),
+    config_path=CONFIG_PATH,
+    sdk_context=sdk_context,
+    chat_only_mode=True
+)
+
+
+async def run_agent(prompt, **kwargs):
+    return await summarizer_agent.chat(prompt)
+
+latex_agent = Agent(
+    name="LatexFormatterAgent",
+    functions=[format_latex],
+    instruction="Format input bullet points into a LaTeX document.",
+    agent_id=str(uuid.uuid4()),
+    config_path=CONFIG_PATH,
+    sdk_context=sdk_context,
+    chat_only_mode=True
+)
+
+async def run_latex_agent(prompt, **kwargs):
+    return await latex_agent.chat(prompt)
+
+workflow = Workflow(
+    name="pdf_latex_summary",
+    instruction="Summarize PDF and convert to LaTeX notes.",
+    description="A workflow that summarizes and formats PDF text as LaTeX.",
+    steps=[
+        WorkflowStep(name="SummarizeBullets", runner=run_agent, mode=StepMode.SEQUENTIAL),
+        WorkflowStep(name="FormatLatex", runner=run_latex_agent, mode=StepMode.SEQUENTIAL),
+    ],
+    sdk_context=sdk_context,
+)
+
+async def main():
+    pdf_path = sys.argv[1] if len(sys.argv) > 1 else "sample.pdf"
+    text = extract_text_from_pdf(pdf_path)
+    chunks = chunk_text(text)
+
+    # Step 1: Summarize each chunk into bullets
+    bullet_summaries = []
+    for chunk in chunks:
+        summary = await run_agent(chunk)
+        bullet_summaries.append(summary)
+
+    # Step 2: Merge all bullet summaries into one bullet string
+    merged_bullets = "\n".join(bullet_summaries)
+
+    # Step 3: Extract just the bullet lines
+    bullet_lines = [line.strip() for line in merged_bullets.split("\n") if line.strip().startswith("-")]
+    latex_items = "\n".join(f"\\item {line.lstrip('- ').strip()}" for line in bullet_lines)
+
+    # Step 4: Hardcoded LaTeX wrapping
+    final_latex = (
+        "\\documentclass{article}\n"
+        "\\usepackage[utf8]{inputenc}\n"
+        "\\usepackage{enumitem}\n"
+        "\\begin{document}\n"
+        "\\section*{Summary Notes}\n"
+        "\\begin{itemize}[leftmargin=*, label=--]\n"
+        f"{latex_items}\n"
+        "\\end{itemize}\n"
+        "\\end{document}"
+    )
+
+    # Step 5: Write the final LaTeX doc to file
+    base_name = os.path.splitext(os.path.basename(pdf_path))[0]
+    output_path = f"{base_name}_summary.tex"
+    with open(output_path, "w", encoding="utf-8") as f:
+        f.write(final_latex)
+
+    print(f"Summary written to: {output_path}")
+
+
+
+
+if __name__ == "__main__":
+    asyncio.run(main())