更新markdown示例：添加RUG Rust单元测试生成演示内容

- 修改主题为黑色样式以提高可读性
- 更新markdown.md为RUG论文演示内容
- 添加相关图片资源到images目录
- 调整演示尺寸为1920x1080以适应现代显示器
- 移除原有的示例幻灯片，专注于学术演示内容

examples/markdown.html

@@ -7,7 +7,7 @@
 		<title>reveal.js - Markdown Example</title>
 
 		<link rel="stylesheet" href="../dist/reveal.css">
-		<link rel="stylesheet" href="../dist/theme/white.css" id="theme">
+		<link rel="stylesheet" href="../dist/theme/black.css" id="theme">
 
         <link rel="stylesheet" href="../plugin/highlight/monokai.css">
 	</head>
@@ -19,7 +19,7 @@
 			<div class="slides">
 
                 <!-- Use external markdown resource, separate slides by three newlines; vertical slides by two newlines -->
-                <section style="text-align: left;" data-markdown="markdown.md" data-separator="---" data-separator-vertical="^\n\n"></section>
+                <section data-markdown="markdown.md" data-separator="^\n\n\n" data-separator-vertical="^\n\n"></section>
 
             </div>
 		</div>

examples/markdown.md

@@ -1,153 +1,133 @@
-# JSONite: High-Performance Embedded Database for Semi-Structured Data
+# RUG: Turbo LLM for Rust Unit Test Generation
 
----
+Keywords: LLM, Rust, Unit Test
 
-## The JSON Performance Crisis
+Research date: 2022, published date: 2025
-
-**JSON is Everywhere:**
-- Web APIS, IoT, logs, configurations
-- Semi-structured, flexible, human-readable
-
-**But Current Solutions Fail:**
-- **Large Databases**: People use MongoDB or PostgreSQL's JSONB to store data
-- **Embeded Databases**: RocksDB and PoloDB lack of ACID and SQL support
-- **Serialization to String**: Or serialize JSON into strings and store in SQLite
 
 
-Serialized JSON with SQL example
 
-```sql
+#### Introduction
-insert into http_request_log (ip, headers)
-values ('127.0.0.1', '{
-    "Content-Type": "application/oct-stream",
-    "X-Forwarded-For": "100.64.0.1",
-}');
-```
 
----
+*   Unit testing is crucial but costly.
 
-## Introducing JSONite
+*   Rust's strict type system.
 
-**Best of Both Worlds:**
+*   Existing LLM approaches often fail.
-- SQLite's based
-- Native JSON optimization
 
-**Key Advantages:**
-- ✅ ACID compliance
-- ✅ SQL simplicity  
-- ✅ Serverless C library
-- ✅ Lightning-fast JSON access
 
----
+#### Rust Unit Test
 
-## Smart Key Optimization
+```rust
-
+/// Returns the sum of two numbers
-**Key Sorting by Length:**
+///
-```
+/// # Examples
-{
+///
-  "id": 1,
+/// ```
-  "address": {...}
+/// assert_eq!(add(2, 3), 5);
-  "name": "John",
+/// assert_eq!(add(-1, 1), 0);
-  "email": "john@example.com",
+/// ```
+fn add(a: i32, b: i32) -> i32 {
+    a + b
 }
 ```
 
-**Sorted as:**
+
-```
+#### Challenge
-{
+
-  "id", (2 chars)
+```rust
-  "name", (4 chars)  
+fn encode<E: Encoder>(&self: char, encoder: E) -> Result<EncodeError> // target function
-  "email", (5 chars)
+
-  "address", (7 chars)
+impl<W: Writer, C: Config> Encoder for EncoderImpl
-}
+
+pub struct EncoderImpl<W: Writer, C: Config>
+impl Writer for SliceWriter
+impl Writer for IoWriter
+
+impl<T> Config for T where T: R1 + R2 + R3
+pub struct Configuration<R1, R2, R3>
 ```
 
-**Binary search on length → Fast lookups**
+Simplified python version
 
----
+```python
+def encode(char_data, encoder):
+    result = encoder.process(char_data)
+    return result
 
-## Handling Massive Data: Smart TOAST
+class Encoder:
+    def __init__(self, writer, config):
+        self.config = config
     
-**The Oversized-Attribute Storage Technique**
+    def process(self, data):
-- Standard approach: arbitrary chunking
+        output = self.writer.write(data, self.config)
-- JSONite's innovation: **Data-Type Aware TOAST**
+        return output
 
-**Intelligent Chunking:**
+class Config:
-- Arrays split between elements
+    def __init__(self):
-- Objects split between key-value pairs  
+        self.settings = {}
-- Text falls back to fixed chunks
 
-**Enables "Slice Detoasting":**
+config = Config()
-- `$.logs[1000000:1000010]` fetches only 10 elements
+encoder = Encoder(stdout, config)
-- Not the entire multi-gigabyte array
 
-
+# Test code
-Smart Chunking Example
+result = encode('A', encoder)
-
-```json
-{
-    "id": 1,
-    "title": "some text",
-    "html": <pointer to TOAST of 200k text>,
-    "photos": [<pointer to TOAST of binary data>],
-    "crawl_logs": [<pointer to TOAST of array of texts>]
-}
 ```
 
----
+LLM generated code are hard to pass the compiler.
 
-## Query Power
 
-**Full SQL + JSON Support:**
-- PostgreSQL-compatible JSONB path operators
-- GIN indexes for instant search
 
-```sql
+#### RUG design
-SELECT *
-FROM accounts
-WHERE data @> '{"status": "active"}'
-```
 
----
+<img src="./images/Screenshot_20251125_010053.jpeg"
+  width="75%">
 
-## Performance Validation: Benchmark Datasets
 
-**Three Specialized Workloads:**
 
-1. **YCSB-Style Read Benchmark**
+<img src="./images/Screenshot_20251125_011029.jpeg" width="80%">
-   - Yahoo! Cloud Serving Benchmark
-   - 1M JSON documents (1KB-100KB each)
 
-2. **TPC-C Inspired Update Benchmark**  
+<img src="./images/Screenshot_20251125_011348.jpeg" width="80%">
-   - Transaction Processing Performance Council
-   - 100K transactional JSON records
-   - Frequent small field updates
 
-3. **Large-Array Slice Benchmark**
-   - Multi-gigabyte JSON documents
-   - Massive arrays (10M+ elements)
 
-**Comparison Targets:** SQLite JSONB vs MongoDB vs PostgreSQL vs JSONite
 
----
+#### Implementation
 
-## JSONite: The Future of Embedded Data Storage
+- gpt-3.5-turbo-16k-0613
+- gpt-4-1106
+- presence penalty set to -1
+- frequency_penalty set to 0.5
+- temperature set to 1 (by default)
 
-**Why It Matters Today:**
-- **Edge Computing**: Lightweight, handles sensor data efficiently
-- **Modern Apps**: SQL power + JSON flexibility, no schema migrations
 
-**The Vision:**
-- Open source implementation
-- Community-driven development  
-- Becoming the default choice for embedded JSON storage
-- Bridging SQL reliability with NoSQL flexibility
 
----
+#### Eval: Comparison with Traditional Tools
 
-## Thank You
+<img src="./images/Screenshot_20251125_014355.jpeg" width="75%">
 
-**Questions?**
 
-*CHEN Yongyuan*  
+#### Token Consumption
-*2025-11-01*
+
+- GPT-4 cost 1000$ in baseline method (send the whole context)
+- RUG saved 51.3% tokens (process unique dependency only once)
+
+
+
+#### Real-World Usability
+
+> We directly leverage RUG's generated tests, without changing test bodies and send them as PRs to the open source projects.
+> To our surprise, the developers are happy to merge these machine generated tests.
+> RUG generated a total of 248 unit tests, of which we submitted 113 to the corresponding crates based on their quality and priority.
+> So far, 53 of these unit tests have been merged with positive feedback.
+
+> Developers chose not to merge 17 tests for two main reasons:
+> first, the target functions are imported from external libraries(16),
+> and the developers do not intend to include tests
+
+
+
+#### 2025 Situation
+
+<img src="./images/Screenshot_20251125_015416.jpeg" width="60%">
+
+
+<img src="./images/Screenshot_20251125_015705.jpeg" width="60%">