Hypergraph is more powerful. Transform unstructured text into structured knowledge with LLMs. Graphs, hypergraphs, and spatio-temporal extractions — with one command.
chore: 迁移测试样本文件并添加新测试数据
yifanfeng97b9a54601a79e
diff --git a/examples/test_samples/en/finance/.gitkeep b/tests/fixtures/test_samples/en/finance/.gitkeepsimilarity index 100%rename from examples/test_samples/en/finance/.gitkeeprename to tests/fixtures/test_samples/en/finance/.gitkeepdiff --git a/tests/fixtures/test_samples/en/general/arbitrary_tech_news.md b/tests/fixtures/test_samples/en/general/arbitrary_tech_news.mdnew file mode 100644index 0000000..392c1db--- /dev/null+++ b/tests/fixtures/test_samples/en/general/arbitrary_tech_news.md@@ -0,0 +1,96 @@+# Galactic Interstellar Announces Successful Maiden Flight of Shenzhou-50, Mars Travel Prices Drop to 2 Million Yuan++**Date**: June 18, 2090+**Author**: Space Journalist, Star Wanderer++---++## Key Announcement++Today, **Galactic Interstellar Travel Corporation**, the world's leading commercial space travel company, held a grand press conference at the **Wenchang Space Launch Center**, officially announcing the successful maiden flight of its new-generation manned spacecraft—**Shenzhou-50**!++**Li Xinghe**, Founder and CEO of Galactic Interstellar, excitedly stated at the conference: "The successful maiden flight of Shenzhou-50 marks a new era in human space travel. Our goal is to let everyone experience the charm of space!"++---++## Shenzhou-50 Highlights++### 1. Revolutionary Propulsion System++Shenzhou-50 is equipped with Galactic Interstellar's independently developed **"Galactic Engine" nuclear-powered propulsion system**, a major breakthrough in the history of human spaceflight:+- Travel time from Earth to Mars reduced from 6 months to **30 days**+- Propulsion efficiency increased by 500%+- Maximum speed reaches 2% of the speed of light++### 2. Comfortable Space Capsule Experience++Shenzhou-50 can accommodate 20 passengers, providing a luxurious and comfortable space travel experience:+- Each cabin is equipped with a panoramic sunroof, offering magnificent views of the universe+- Zero-gravity fitness center and leisure entertainment area+- Artificial gravity chamber to reduce space discomfort+- Michelin 3-star space restaurant++### 3. Safety and Security System++For the special risks of space travel, Shenzhou-50 provides multiple layers of safety and security:+- Triple redundant life support system+- Micro black hole shield+- Emergency return capsule, capable of returning to Earth within 24 hours+- Compliant with ISO Space Safety Certification++---++## Space Team++The R&D of Shenzhou-50 was led by **Wang Yuzhou**, CTO of Galactic Interstellar, with core team members including:+- **Zhang Xingchen**: Chief Designer, previously involved in the construction of the International Space Station+- **Liu Xingyun**: Propulsion System Lead, nuclear power expert+- **Zhao Yinhe**: Mission Planning Director, with 20 years of space mission experience+- **Sun Xingji**: Security System Architect++The entire R&D team consists of more than 5,000 people and took 10 years to complete the development of Shenzhou-50.++---++## Strategic Partnerships++At the conference, Galactic Interstellar also announced strategic partnerships with several organizations:+- **International Astronautical Agency (IAA)**: Co-develop deep space exploration technology+- **Mars Base Construction Company**: Build the first commercial base on Mars+- **Space Hotel Group**: Co-operate orbital space station hotels+- **Interstellar Travel Alliance**: Establish global space travel standards++---++## Development History++Galactic Interstellar was founded in 2050, headquartered in Shanghai:+- October 2050: Received angel round financing of ¥1 billion+- May 2055: Completed Series A financing of ¥5 billion+- August 2060: Successful maiden flight of Shenzhou-30, achieving suborbital travel+- March 2070: Series B financing of ¥20 billion, valuation reached ¥100 billion+- December 2080: Successful maiden flight of Shenzhou-40, achieving lunar commercial travel+- June 2090: Successful maiden flight of Shenzhou-50, ushering in the era of Mars travel++---++## Pricing and Booking++Shenzhou-50 Mars travel will officially accept bookings from October 1, 2090:+- **Economy Class**: ¥2 million per person+- **Business Class**: ¥5 million per person+- **Luxury Class**: ¥20 million per person++Early booking users can enjoy a 50% discount!++---++## Market Outlook++Industry analysts believe that the release of Shenzhou-50 will completely transform human travel. It is expected that by 2100, the global space travel market will exceed ¥10 trillion.++Galactic Interstellar stated that it will continue to invest in R&D, planning to launch **Shenzhou-60** in 2095 to achieve Saturn travel, and achieve commercial travel to any planet in the solar system by 2100.++---++"The universe is so big, we all want to see it!" —— Li Xinghe, Founder of Galactic Interstellardiff --git a/tests/fixtures/test_samples/en/general/biography_scientist.md b/tests/fixtures/test_samples/en/general/biography_scientist.mdnew file mode 100644index 0000000..dbbfb82--- /dev/null+++ b/tests/fixtures/test_samples/en/general/biography_scientist.md@@ -0,0 +1,165 @@+# Zu Chongzhi: A Giant of Ancient Chinese Science++**Author**: Historical Biography, Chen Jingyi+**Publication Year**: 2026++---++## Preface++Zu Chongzhi was an outstanding mathematician, astronomer, and inventor during the Northern and Southern Dynasties period in China. He made remarkable achievements in multiple fields including mathematics, astronomical calendars, and mechanical engineering, and stands as a monument in the history of ancient Chinese science.++This book aims to record the extraordinary life of Zu Chongzhi, showcasing his persistent pursuit of science and his great contributions to Chinese civilization.++---++## Chapter 1: Early Years (429-450)++### 1.1 Birth and Family++In 429 (the 6th year of Yuanjia during the reign of Emperor Wen of Song), Zu Chongzhi was born into an official family in **Qiuxian County, Fanyang Commandery** (now Laishui County, Hebei Province).++His great-grandfather **Zu Taizhi** served as a Palace Attendant and Grand Master of Glorious Happiness during the Eastern Jin Dynasty; his grandfather **Zu Chang** served as the Director of Imperial Workshops (a position in charge of civil engineering) during the Liu Song Dynasty; his father **Zu Shuozhi** served as a Court Gentleman for Audience. In such a family environment, Zu Chongzhi received a good education from an early age and developed a strong interest in science and technology.++### 1.2 Diligent in Studies++From an early age, Zu Chongzhi was diligent in his studies and widely read, with a particular fondness for books on mathematics, astronomy, and mechanical engineering.++Historical records state that Zu Chongzhi "studied ancient texts from a young age and had an ingenious mind," meaning he studied ancient classics from childhood and possessed extraordinary wit and thinking ability.++### 1.3 Entering Officialdom++Around 450, Zu Chongzhi entered officialdom, serving as a Staff Officer in Southern Xuzhou (now Zhenjiang, Jiangsu), and later as a Military Advisor to the Duke. Although in official service, he never gave up his pursuit of scientific research.++---++## Chapter 2: Mathematical Achievements (450-470)++### 2.1 Calculation of Pi++Zu Chongzhi's most outstanding contribution in mathematics was his precise calculation of **pi (π)**.++After long and arduous research, Zu Chongzhi concluded:+- **Excess value (upper limit)**: 3.1415927+- **Deficit value (lower limit)**: 3.1415926++In other words, the value of pi is **between 3.1415926 and 3.1415927**, accurate to the 7th decimal place. This achievement was **over a thousand years ahead of Western mathematicians**!++For calculation convenience, Zu Chongzhi also provided two fractional approximations:+- **Approximate ratio**: 22/7 (≈3.1429)+- **Precise ratio**: 355/113 (≈3.1415929)++Among these, the precise ratio 355/113 is the best approximation with both numerator and denominator within 1000, another achievement that led the world by over a thousand years.++### 2.2 Zhui Shu++Zu Chongzhi compiled his mathematical research results into a work titled **Zhui Shu** (Method of Interpolation).++Zhui Shu had profound content and was called a "mathematical classic," later becoming one of the textbooks in the Imperial Academy's School of Mathematics during the Tang Dynasty. Unfortunately, this precious mathematical work was lost during the Northern Song Dynasty.++---++## Chapter 3: Astronomy and Calendars (450-464)++### 3.1 Discovering Calendar Problems++While studying astronomical calendars, Zu Chongzhi discovered that the then-popular **Yuanjia Calendar** had numerous errors.++The Yuanjia Calendar was formulated by the astronomer He Chengtian. Although more precise than previous calendars, it still had shortcomings. After years of observation and research, Zu Chongzhi pointed out multiple defects in the Yuanjia Calendar.++### 3.2 Creating the Daming Calendar++In 462 (the 6th year of Daming during the reign of Emperor Xiaowu of Song), Zu Chongzhi compiled a new calendar—the **Daming Calendar**.++The main innovations of the Daming Calendar included:+1. **Introduction of precession**: A major breakthrough in the history of Chinese calendars, distinguishing between the tropical year and sidereal year+2. **Improved intercalation cycle**: Changed the old calendar's 7 intercalary months in 19 years to 144 intercalary months in 391 years, more precise+3. **Determination of nodical month**: Accurately calculated the number of days in a nodical month as 27.21223 days, extremely close to the modern measured value of 27.21222 days+4. **Precise determination of planetary cycles**: Accurately measured the orbital periods of the five major planets: Venus, Jupiter, Mercury, Mars, and Saturn++### 3.3 Calendar Debate++After the Daming Calendar was compiled, Zu Chongzhi submitted a memorial to Emperor Xiaowu of Song, requesting its promulgation and implementation. However, this proposal was opposed by the powerful official **Dai Faxing**.++Dai Faxing believed that Zu Chongzhi was "defying heaven and betraying the classics," and that changing ancient methods was a great treason. Undaunted, Zu Chongzhi wrote the famous **"Refutation"**, arguing forcefully and refuting Dai Faxing's erroneous viewpoints point by point.++This debate is a famous event in the history of ancient Chinese science, demonstrating Zu Chongzhi's scientific spirit of adhering to truth and defying powerful officials.++Unfortunately, due to Dai Faxing's obstruction, the Daming Calendar was not promulgated and implemented during Zu Chongzhi's lifetime. It was not until 510 (the 9th year of Tianjian during the reign of Emperor Wu of Liang) that, through the efforts of his son **Zu Geng**, the Daming Calendar was officially adopted, 10 years after Zu Chongzhi's death.++---++## Chapter 4: Mechanical Inventions (470-500)++### 4.1 South-Pointing Chariot++Zu Chongzhi was not only an outstanding mathematician and astronomer, but also an excellent mechanical inventor.++He successfully created the **South-Pointing Chariot**. The South-Pointing Chariot was a vehicle that could indicate direction, said to have existed since the time of the Yellow Emperor, but its manufacturing method had long been lost. Based on ancient records, after repeated research and experimentation, Zu Chongzhi finally successfully restored the South-Pointing Chariot.++### 4.2 Thousand-Li Boat++Zu Chongzhi also invented a type of boat called the "**Thousand-Li Boat**."++According to the "Book of Southern Qi · Biography of Zu Chongzhi": "Testing it on the Xinting River, it traveled over a hundred li in a day." In other words, when the Thousand-Li Boat was tested on the Xinting River (now near Nanjing, Jiangsu), it could travel over a hundred li in one day, a remarkable achievement for that time.++### 4.3 Water-Powered Trip Hammer and Mill++Zu Chongzhi also created a **water-powered trip hammer and mill**, a grain-processing tool driven by water power.++The water-powered trip hammer and mill combined a water-powered trip hammer and a water mill, capable of hulling rice and grinding wheat simultaneously, greatly improving production efficiency. This invention played an important role in the development of agricultural production.++### 4.4 Timekeeping Devices++In addition, Zu Chongzhi also created a **qi vessel** (a type of timekeeping device) and other ingenious mechanical devices, demonstrating his exceptional talent in mechanical engineering.++---++## Chapter 5: Historical Influence and Evaluation++### 5.1 International Recognition++Zu Chongzhi's scientific achievements have been widely recognized by the international scientific community.++To commemorate Zu Chongzhi's outstanding contributions:+- In 1964, an asteroid discovered by the Purple Mountain Observatory was named **"Zuchongzhi"**+- A crater on the Moon was named **"Zu Chongzhi Crater"**++### 5.2 Historical Evaluation++The "Book of Southern Qi · Biography of Zu Chongzhi" evaluates him: "Zu Chongzhi understood music, chess, and various games; he was unparalleled in his time, and no one could match him."++This means Zu Chongzhi was proficient in music, chess, and various arts; he was unparalleled in his time, and no one could match him.++**Joseph Needham** (a famous British historian of science) highly praised Zu Chongzhi in "Science and Civilization in China," calling him "one of the greatest mathematicians of ancient China."++### 5.3 Scientific Spirit++In Zu Chongzhi, we see the excellent qualities of ancient Chinese scientists:+- **Seeking truth from facts**: Insisting on starting from actual observations+- **Daring to innovate**: Daring to break through old theories and propose new ideas+- **Rigorous and meticulous**: Striving for perfection in scientific research+- **Defying powerful officials**: Daring to adhere to truth and argue forcefully++---++## Conclusion++Zu Chongzhi was a versatile talent in the history of ancient Chinese science, making outstanding achievements in multiple fields including mathematics, astronomy, and mechanics.++His calculation of pi led the world by over a thousand years, his Daming Calendar was a milestone in the history of Chinese calendars, and his mechanical inventions demonstrated the wisdom of the ancient Chinese people.++Zu Chongzhi's scientific spirit and achievements are not only the pride of the Chinese nation, but also a precious heritage in the treasure house of world scientific culture. His name will forever be inscribed in the long river of human scientific development.++---++## Appendix: Chronology of Zu Chongzhi++- **429**: Born in Qiuxian County, Fanyang Commandery (now Laishui, Hebei)+- **Around 450**: Served as Staff Officer in Southern Xuzhou, Military Advisor to the Duke+- **462**: Compiled the Daming Calendar, submitted memorial to the court+- **462-464**: Debated calendars with Dai Faxing+- **470s**: Created South-Pointing Chariot, Thousand-Li Boat, Water-Powered Trip Hammer and Mill+- **494-498**: Served as Colonel of Changshui+- **500**: Passed away, aged 72+- **510**: Daming Calendar officially adopteddiff --git a/tests/fixtures/test_samples/en/general/encyclopedia_machine_learning.md b/tests/fixtures/test_samples/en/general/encyclopedia_machine_learning.mdnew file mode 100644index 0000000..4c8e201--- /dev/null+++ b/tests/fixtures/test_samples/en/general/encyclopedia_machine_learning.md@@ -0,0 +1,268 @@+# Machine Learning++**Category**: Artificial Intelligence+**Created**: February 22, 2026+**Last Updated**: February 22, 2026++---++## Overview++**Machine Learning (ML)** is an important branch of **Artificial Intelligence (AI)**, dedicated to studying how computer systems can automatically improve through experience.++The core idea of machine learning is: instead of relying on explicit hard-coded rules, learn patterns and regularities from data through algorithms.++---++## Definition++Machine learning is a multidisciplinary field involving **probability theory**, **statistics**, **approximation theory**, **convex analysis**, **computational complexity theory**, and many other disciplines.++It is the core of artificial intelligence and the fundamental way to make computers intelligent. Machine learning algorithms automatically discover patterns by analyzing large amounts of data, and use these patterns to make predictions or decisions on unknown data.++---++## Classification System++### 1. Supervised Learning++Supervised learning uses **labeled training data** for learning. The algorithm learns mapping relationships from input features and corresponding output labels.++**Common Algorithms**:+- Linear Regression+- Logistic Regression+- Decision Tree+- Random Forest+- Support Vector Machine (SVM)+- Neural Networks++**Application Scenarios**:+- House price prediction+- Spam classification+- Image recognition+- Medical diagnosis++### 2. Unsupervised Learning++Unsupervised learning uses **unlabeled data**, and the algorithm needs to discover structures and patterns in the data on its own.++**Common Algorithms**:+- Clustering: K-Means, Hierarchical Clustering, DBSCAN+- Dimensionality Reduction: PCA, t-SNE, Autoencoders+- Association Rule Mining: Apriori, FP-Growth++**Application Scenarios**:+- Customer segmentation+- Anomaly detection+- Recommendation systems+- Topic modeling++### 3. Reinforcement Learning (RL)++Reinforcement learning learns through **interaction between an agent and the environment**. The agent obtains rewards (Reward) through trial and error, with the goal of maximizing long-term cumulative rewards.++**Core Concepts**:+- State+- Action+- Policy+- Reward+- Value Function++**Common Algorithms**:+- Q-Learning+- Policy Gradient+- Actor-Critic+- DQN (Deep Q-Network)+- PPO (Proximal Policy Optimization)++**Application Scenarios**:+- Game AI (e.g., AlphaGo)+- Robot control+- Autonomous driving+- Resource scheduling++### 4. Semi-Supervised Learning++Combines a small amount of labeled data with a large amount of unlabeled data for learning.++### 5. Self-Supervised Learning++Constructs supervision signals from the data itself, without manual annotation.++---++## Key Concepts++### 1. Feature++Features are measurable attributes or variables that describe data. In machine learning, features are the representation of the raw data input to the model.++**Examples**:+- In house price prediction, features may include: area, number of rooms, location, construction year+- In image recognition, features may be pixel values or extracted visual features++### 2. Model++A model is a mathematical representation or function learned by a machine learning algorithm, used to predict outputs from input features.++**Common Model Types**:+- Linear models+- Tree models+- Neural network models+- Probabilistic graphical models++### 3. Training++Training is the process of adjusting model parameters using training data, so that the model can learn patterns in the data.++### 4. Inference/Prediction++Inference is the process of using a trained model to make predictions on new data.++### 5. Overfitting++Overfitting refers to the phenomenon where a model performs well on training data but has poor generalization ability on new data.++### 6. Underfitting++Underfitting refers to the phenomenon where a model fails to fully learn data patterns, performing poorly on both training and test sets.++---++## Related Fields++### 1. Deep Learning++Deep learning is a subfield of machine learning that uses multi-layer **neural networks** to learn representations of data.++**Common Network Architectures**:+- Convolutional Neural Networks (CNN): Suitable for images+- Recurrent Neural Networks (RNN)/Transformer: Suitable for sequential data (text, speech)+- Generative Adversarial Networks (GAN): Used to generate new data++### 2. Data Mining++Data mining is the process of extracting useful information and knowledge from large amounts of data, with many overlapping techniques with machine learning.++### 3. Natural Language Processing (NLP)++NLP focuses on enabling computers to understand, interpret, and generate human language.++**Applications**:+- Machine translation+- Sentiment analysis+- Question answering systems+- Text summarization++### 4. Computer Vision++Computer vision aims to enable computers to obtain meaningful information from images or videos.++**Applications**:+- Image recognition+- Object detection+- Image segmentation+- Face recognition++### 5. Statistics++Statistics provides the theoretical foundation and methodological support for machine learning.++---++## Application Scenarios++1. **Healthcare**: Disease diagnosis, drug discovery, genomic analysis+2. **Finance**: Credit scoring, fraud detection, algorithmic trading+3. **Retail**: Recommendation systems, demand forecasting, customer segmentation+4. **Transportation**: Autonomous driving, traffic prediction, route optimization+5. **Manufacturing**: Quality control, predictive maintenance, supply chain optimization+6. **Education**: Personalized learning, intelligent tutoring, learning analytics+7. **Entertainment**: Content recommendation, game AI, special effects generation++---++## Historical Development++### 1940s-1950s: Origins++- 1943: Walter Pitts and Warren McCulloch proposed the mathematical model of neurons+- 1950: Alan Turing proposed the "Turing Test"+- 1952: Arthur Samuel developed the first checkers learning program+- 1959: Samuel first used the term "machine learning"++### 1960s-1970s: Early Development++- 1967: Nearest Neighbor algorithm proposed+- 1969: Minsky and Papert published "Perceptrons", pointing out the limitations of single-layer perceptrons++### 1980s-1990s: Renaissance++- 1986: Backpropagation algorithm became popular again+- 1989: Convolutional neural network concept proposed+- 1995: Support Vector Machine (SVM) proposed+- 1997: LSTM (Long Short-Term Memory) proposed++### 2000s-2010s: Big Data Era++- 2006: Geoffrey Hinton proposed "Deep Belief Networks", deep learning began to revive+- 2012: AlexNet led significantly in ImageNet competition, marking the arrival of the deep learning era+- 2014: GAN (Generative Adversarial Network) proposed+- 2017: Transformer architecture proposed++### 2020s-Present: Large Model Era++- 2020: GPT-3 released+- 2022: ChatGPT released, causing global attention+- 2023: Multimodal large models emerged++---++## Important People and Institutions++### People++- **Geoffrey Hinton**: "Father of Deep Learning", important contributor to the backpropagation algorithm+- **Yann LeCun**: Pioneer of convolutional neural networks, first director of Facebook AI Research+- **Yoshua Bengio**: Important scholar in the field of deep learning+- **Andrew Ng**: Co-founder of Google Brain, co-founder of Coursera+- **Demis Hassabis**: Founder of DeepMind, leader of the AlphaGo project++### Institutions++- **Google DeepMind**: Research institution for AlphaGo, AlphaFold+- **OpenAI**: Research institution for GPT series models+- **Meta AI** (formerly Facebook AI): PyTorch, FAIR+- **Microsoft Research Asia (MSRA)**+- **Stanford Artificial Intelligence Laboratory (SAIL)**+- **MIT Computer Science and Artificial Intelligence Laboratory (CSAIL)**++---++## Challenges and Future Directions++### Current Challenges++1. Data requirements: Deep learning requires large amounts of labeled data+2. Interpretability: Black-box models make it difficult to understand decision-making processes+3. Robustness: Adversarial examples can easily cause model failures+4. Fairness: Algorithms may have biases+5. Computational resources: Large models require enormous computing power+6. Security: Security safeguards for AI systems++### Future Directions++1. Few-shot Learning+2. Explainable AI+3. Federated Learning+4. Causal Inference+5. Multimodal Learning+6. Autonomous Learning++---++## References++- Goodfellow, I., Bengio, Y., & Courville, A. (2016). *Deep Learning*. MIT Press.+- Murphy, K. P. (2012). *Machine Learning: A Probabilistic Perspective*. MIT Press.+- Hastie, T., Tibshirani, R., & Friedman, J. (2009). *The Elements of Statistical Learning*. Springer.diff --git a/tests/fixtures/test_samples/en/general/regulation_company_policy.md b/tests/fixtures/test_samples/en/general/regulation_company_policy.mdnew file mode 100644index 0000000..ff3064b--- /dev/null+++ b/tests/fixtures/test_samples/en/general/regulation_company_policy.md@@ -0,0 +1,361 @@+# No.1 Slacker Company in the Universe Employee Attendance and Performance Management System++**Document Number**: MY-HR-2090-001+**Version**: V3.0+**Effective Date**: January 1, 2090+**Department**: Human Resources+**Approved by**: Slacker Emperor · Wang Dachui++---++## Chapter 1: General Provisions++### 1.1 Purpose++To standardize the company's employee attendance management, establish a scientific and reasonable performance evaluation system, protect the legitimate rights and interests of the company and employees, while carefully slacking off and efficiently completing work, in accordance with the "Galactic Labor Law," "Cosmic Labor Contract Law," and related interstellar laws and regulations, combined with the actual situation of the company, this system is formulated.++### 1.2 Scope of Application++This system applies to all regular employees, probationary employees, and alien interns of No.1 Slacker Company in the Universe (hereinafter referred to as the "Company").++### 1.3 Basic Principles++1. Principle of balance between slacking off and working+2. Principle of combining incentives and constraints+3. Principle of linking performance with slacking time+4. Principle of continuous slacking and continuous improvement++---++## Chapter 2: Attendance Management Regulations++### 2.1 Working Hours++#### 2.1.1 Standard Working Hours++- **Workdays**: Monday to Friday (can choose any two days per week to work from home and slack off)+- **Working Hours**: 10:00 AM - 12:00 PM (slacking time), 2:00 PM - 5:00 PM (working time)+- **Lunch Break**: 12:00 PM - 2:00 PM (mandatory slacking time)+- **Weekly Working Hours**: No more than 25 hours (the rest of the time can be freely arranged for slacking off)++#### 2.1.2 Flexible Working Hours++The Company implements a **super flexible work system**, where employees can flexibly arrange work and slacking off within the following ranges:+- Earliest arrival time: 9:00 AM+- Latest arrival time: 11:00 AM+- Earliest departure time: 4:00 PM+- Latest departure time: 6:00 PM++As long as 5 hours of work are completed each day, slacking time is unlimited.++#### 2.1.3 Special Positions++For positions requiring 24-hour on-duty coverage such as space exploration and alien communication, a **shift system** is implemented, with specific schedules arranged by the department supervisor, but every 4 consecutive hours of work must be followed by 1 hour of slacking off.++### 2.2 Check-In Management++#### 2.2.1 Check-In Methods++Employees must check in through the Company's interstellar attendance system:+- **Brainwave Check-In**: Direct check-in using a brain-computer interface+- **Holographic Projection Check-In**: Check-in via holographic image+- **Quantum Entanglement Check-In**: Can check in no matter which corner of the universe you are in++#### 2.2.2 Check-In Requirements++- Must check in twice daily: arrival check-in and departure check-in+- Arrival check-in should be completed within 30 minutes after arriving at work+- Departure check-in should be completed within 30 minutes before or after leaving work+- When unable to check in due to official business (including going to other planets), an "Outing Application" should be submitted through the OA system in advance++#### 2.2.3 Forgetting to Check In++- Allowed to forget to check in 5 times per month+- Forgetting to check in must be submitted through the OA system within 48 hours with a "Make-Up Check-In Application," effective after approval by the department supervisor+- Forgetting to check in more than 5 times, starting from the 6th time, each time will be treated as 0.5 days of absence (but slacking time can be offset)++### 2.3 Tardiness, Early Departure, and Absence++#### 2.3.1 Tardiness++- **Tardiness within 30 minutes**: Verbal warning (doesn't count)+- **Tardiness 30-60 minutes**: Written warning, deduct 1 slacking voucher+- **Tardiness 60-120 minutes**: Treated as 0.5 days of absence+- **Tardiness more than 120 minutes**: Treated as 1 day of absence+- **More than 5 cumulative tardiness instances per month**: Cancel the full attendance slacking award for the month++#### 2.3.2 Early Departure++- **Early departure within 30 minutes**: Verbal warning+- **Early departure 30-60 minutes**: Written warning, deduct 1 slacking voucher+- **Early departure more than 60 minutes**: Treated as 0.5 days of absence++#### 2.3.3 Absence++- **0.5 days of absence**: Deduct 1.5 times daily salary, deduct 2 slacking vouchers+- **1 day of absence**: Deduct 2 times daily salary, cancel full attendance slacking award for the month+- **3 consecutive days of absence or 5 cumulative days of absence per year**: Deemed a serious violation of company rules and regulations, and the Company has the right to terminate the labor contract (but can negotiate to offset with slacking vouchers)++### 2.4 Leave Management++#### 2.4.1 Leave Types++| Leave Type | Description | Salary Treatment | Slacking Impact |+|-----------|-------------|------------------|-----------------|+| Personal Leave | Leave for personal matters | Unpaid | No impact on slacking vouchers |+| Sick Leave | Leave for medical treatment or recuperation | Paid at 80% of local minimum wage | Extra slacking vouchers given |+| Marriage Leave | Leave for getting married | Paid (10 days) | Paid slacking |+| Maternity Leave | Leave for female employees giving birth | Paid (in accordance with interstellar regulations) | Full paid slacking |+| Paternity Leave | Leave for male employees whose spouse is giving birth | Paid (30 days) | Full paid slacking |+| Bereavement Leave | Leave for death of immediate family member | Paid (3-7 days) | Paid slacking |+| Annual Leave | Paid annual leave | Paid | Paid slacking |+| Slacking Leave | Leave specifically for slacking off | Paid | Special slacking |++#### 2.4.2 Leave Process++1. **Apply in advance**: In principle, a "Leave Application" should be submitted through the OA system 3 days in advance+2. **Fill in information**: Clearly state leave type, leave time, and reason for leave (slacking leave requires a slacking plan)+3. **Upload proof**: Sick leave, marriage leave, etc., must upload relevant proof materials (slacking leave can upload slacking photos)+4. **Approval process**:+ - Leave within 3 days: Approved by department supervisor+ - Leave 3-7 days: Approved by department supervisor + HR+ - Leave more than 7 days: Approved by department supervisor + HR + Slacker Emperor+5. **Work handover**: Work handover should be completed before taking leave+6. **Emergency leave**: In case of emergency where advance application is not possible, notify the department supervisor via quantum communication within 1 hour before work, and complete the leave application procedure within 48 hours after the event++#### 2.4.3 Annual Leave Regulations++| Company Seniority | Annual Leave Days | Slacking Days |+|------------------|------------------|--------------|+| 1-5 years | 10 days | 5 days |+| 5-10 years | 15 days | 10 days |+| More than 10 years | 20 days | 15 days |++- Annual leave can be used concentratedly or in segments during the year, cannot be carried over to the next year+- Slacking days can be used in conjunction with annual leave+- If the Company cannot arrange annual leave for employees due to work needs, with the consent of the employee, unpaid annual leave salary can be paid at 5 times the daily salary++---++## Chapter 3: Performance Evaluation Process++### 3.1 Evaluation Cycle++- **Monthly Evaluation**: Conduct evaluation for the previous month at the beginning of each month+- **Quarterly Evaluation**: Conduct evaluation for the previous quarter in the first month of each quarter+- **Annual Evaluation**: Conduct evaluation for the previous year in January each year+- **Slacking Evaluation**: Conduct slacking quality assessment every weekend++### 3.2 Evaluation Content++#### 3.2.1 Performance Evaluation Indicators (KPI + MFI)++| Indicator Category | Weight | Specific Content |+|------------------|--------|-----------------|+| Work Performance | 40% | Target completion, work quality, work efficiency |+| Slacking Quality | 30% | Slacking duration, diversity of slacking methods, creativity in slacking |+| Work Ability | 15% | Professional ability, communication ability, teamwork, learning ability |+| Work Attitude | 10% | Sense of responsibility, execution, discipline, enthusiasm for slacking |+| Attendance | 5% | Attendance rate, tardiness and early departure |++#### 3.2.2 Evaluation Focus for Different Positions++- **Space Exploration Positions**: Focus on technological innovation, task completion quality, creativity in slacking+- **Alien Communication Positions**: Focus on communication ability, cultural understanding, slacking time control+- **Slacking Guidance Positions**: Focus on slacking skills, slacking training effectiveness, slacking innovation+- **Functional Positions**: Focus on service quality, process optimization, slacking efficiency++### 3.3 Evaluation Process++#### 3.3.1 Goal Setting (P1)++- **Time**: Beginning of evaluation period (first month of each month/quarter/year)+- **Process**:+ 1. Employees set personal work goals and slacking goals based on department goals and job responsibilities+ 2. Communicate with department supervisor to determine final goals+ 3. Enter evaluation goals into the performance system++#### 3.3.2 Process Management (P2)++- **Time**: During evaluation period+- **Process**:+ 1. Employees regularly record work progress and slacking situation+ 2. Department supervisor has 1-2 mid-term communications with employees+ 3. Adjust goals in a timely manner or provide necessary slacking guidance++#### 3.3.3 Performance Evaluation (P3)++- **Time**: End of evaluation period (last month of each month/quarter/year)+- **Process**:+ 1. Employees conduct self-evaluation, fill out "Performance Self-Evaluation Form" and "Slacking Self-Evaluation Form"+ 2. Department supervisor conducts evaluation based on employee performance, target completion, and slacking quality+ 3. HR department reviews evaluation results++#### 3.3.4 Performance Feedback (P4)++- **Time**: Within 5 working days after evaluation results are determined+- **Process**:+ 1. Department supervisor has a 1-on-1 performance interview with the employee+ 2. Acknowledge achievements, point out shortcomings, jointly formulate improvement plans (including both work and slacking aspects)+ 3. Employee signs and confirms on the "Performance Interview Record Form"++### 3.4 Evaluation Grades++| Grade | Definition | Score Range | Proportion Limit | Slacking Reward |+|-------|------------|-------------|-----------------|----------------|+| S | Outstanding, far exceeding expectations | 95-100 points | ≤5% | 10 slacking vouchers + 5 days paid slacking leave |+| A | Excellent, exceeding expectations | 85-94 points | ≤20% | 5 slacking vouchers + 2 days paid slacking leave |+| B | Good, meeting expectations | 70-84 points | Around 50% | 2 slacking vouchers |+| C | Qualified, basically meeting expectations | 60-69 points | ≥20% | 1 slacking voucher |+| D | Unqualified, not meeting expectations | Below 60 points | ≤5% | No slacking reward, need to participate in slacking training |++---++## Chapter 4: Reward and Punishment Rules++### 4.1 Rewards++#### 4.1.1 Reward Types++| Reward Type | Description | Reward Method |+|------------|-------------|---------------|+| Monthly Slacker Star | Employee with highest slacking quality selected each month | ¥5,000 bonus + 10 slacking vouchers + certificate |+| Quarterly Slacker Star | Excellent slacking employee selected each quarter | ¥15,000 bonus + 20 slacking vouchers + certificate + 3 days paid slacking leave |+| Annual Slacker Master | Slacker master selected each year | ¥100,000 bonus + 50 slacking vouchers + certificate + 10 days paid slacking leave + priority for promotion |+| Innovation Award | Outstanding contribution to innovation in slacking methods or work methods | ¥10,000-100,000 bonus + certificate |+| Team Award | Reward for excellent slacking teams | ¥10,000-50,000 team slacking activity fund |++#### 4.1.2 Selection Criteria++- Monthly Slacker Star: Slacking evaluation Grade A or above for 2 consecutive months, work performance not below Grade B, no disciplinary records+- Quarterly Slacker Star: Quarterly slacking evaluation Grade S or A, work performance not below Grade B, no disciplinary records, excellent team evaluation+- Annual Slacker Master: Annual slacking evaluation Grade S, work performance not below Grade A, no disciplinary records throughout the year, significant contribution to slacking culture+- Innovation Award: Proposed innovation scheme adopted and achieved significant results+- Team Award: Team goals exceeded, good slacking atmosphere++#### 4.1.3 Selection Process++1. Department recommendation or employee self-nomination+2. Human Resources Department review+3. Slacking Committee evaluation+4. Public notice for 3 working days+5. Official recognition++### 4.2 Punishments++#### 4.2.1 Punishment Types++| Punishment Type | Applicable Scenarios | Punishment Method |+|----------------|---------------------|------------------|+| Verbal Warning | Minor violation of rules and regulations | Verbal criticism and education, deduct 1 slacking voucher |+| Written Warning | Multiple minor violations or more serious violations | Written warning, deduct 3-5 slacking vouchers |+| Record of Demerit | Serious violation of rules and regulations | Demerit on record, deduct 5-10 slacking vouchers, cancel slacking evaluation qualification for the year |+| Salary Reduction/Demotion | Incompetent for work or excessive slacking seriously affecting work | Salary reduction of 10%-30% or demotion, reduce slacking time |+| Termination of Labor Contract | Serious violation of rules and regulations, serious dereliction of duty, etc. | Termination of labor contract, no need to pay economic compensation |++#### 4.2.2 Comparison Table of Specific Violations and Punishments++| Violation | Punishment Grade |+|----------|-----------------|+| Tardiness or early departure within 5 times in 1 month | Verbal warning |+| Tardiness or early departure 5-10 times in 1 month | Written warning |+| 1 day of absence | Written warning |+| 2-3 consecutive days of absence | Record of demerit |+| More than 3 consecutive days of absence or 5 cumulative days of absence per year | Termination of labor contract |+| Excessive slacking during work hours, seriously affecting work | Written warning, record of demerit for recurrence |+| Leaking company interstellar secrets | Record of demerit, termination of labor contract if causing serious losses |+| Forging attendance or slacking records | Record of demerit, termination of labor contract for recurrence |+| Deceiving the company, fraud | Record of demerit, termination of labor contract in serious cases |+| Other serious violations of company rules and regulations | Corresponding punishment according to severity |++#### 4.2.3 Punishment Process++1. **Investigation and verification**: After discovering a violation, the HR department or relevant department conducts an investigation+2. **Hear statement**: Listen to the party's statement and defense+3. **Make decision**: Make a punishment decision based on investigation results and system regulations+4. **Deliver notice**: Deliver the "Punishment Notice" to the party+5. **Right to appeal**: If the party disagrees with the punishment decision, they may submit a written appeal to the HR department within 3 working days++---++## Chapter 5: Appeal Procedure++### 5.1 Appeal Scope++Employees may file an appeal if they disagree with the following matters:+- Performance evaluation results+- Slacking evaluation results+- Attendance records+- Reward and punishment decisions+- Other matters where rights and interests are believed to have been infringed++### 5.2 Appeal Process++#### 5.2.1 First-Level Appeal++1. **Time**: Within 3 working days after receiving the relevant decision or record+2. **Method**: Submit a written appeal to the direct supervisor or department supervisor+3. **Processing**: The direct supervisor or department supervisor should give a response within 3 working days++#### 5.2.2 Second-Level Appeal++1. **Prerequisite**: Not satisfied with the first-level appeal result+2. **Time**: Within 3 working days after receiving the first-level response+3. **Method**: Submit a written appeal to the HR department, along with relevant supporting materials+4. **Processing**: The HR department should conduct an investigation and give a response within 5 working days++#### 5.2.3 Third-Level Appeal++1. **Prerequisite**: Still not satisfied with the second-level appeal result+2. **Time**: Within 5 working days after receiving the second-level response+3. **Method**: Submit a written appeal to the Company Slacking Dispute Mediation Committee+4. **Processing**: The Mediation Committee should hold a meeting within 10 working days to conduct mediation and make a final decision++### 5.3 Appeal Materials++Appeal materials should include:+- Appeal letter (stating the appeal matter, reasons, and requirements)+- Relevant supporting materials+- Slacking records or work records++---++## Chapter 6: Supplementary Provisions++### 6.1 System Revision++Revisions to this system are proposed by the Human Resources Department and published for implementation after approval by the Slacker Emperor.++### 6.2 System Interpretation++The right to interpret this system belongs to the Company's Human Resources Department and Slacking Committee.++### 6.3 Effective Date++This system shall take effect on January 1, 2090, and the original "No.1 Slacker Company in the Universe Employee Attendance and Performance Management System (V2.0)" shall be repealed simultaneously.++### 6.4 Publicity and Training++The Human Resources Department and Slacking Committee are responsible for the publicity and training of this system, ensuring that every employee understands and complies with this system, while also being able to fully enjoy the fun of slacking off.++---++**Attachments**:+1. "Leave Application Form"+2. "Performance Evaluation Form"+3. "Slacking Evaluation Form"+4. "Performance Interview Record Form"+5. "Reward/Punishment Approval Form"+6. "Slacking Voucher Usage Guide"+7. "Appeal Letter Template"++---++**System Change Record**:+- V1.0 (January 1, 2080): First release, establishing slacking culture+- V2.0 (June 1, 2085): Revised performance evaluation section, added slacking evaluation indicators+- V3.0 (January 1, 2090): Improved reward and punishment rules, optimized slacking system++---++"Carefully slack off, work happily!" —— Slacker Emperor · Wang Dachuidiff --git a/examples/test_samples/en/industry/.gitkeep b/tests/fixtures/test_samples/en/industry/.gitkeepsimilarity index 100%rename from examples/test_samples/en/industry/.gitkeeprename to tests/fixtures/test_samples/en/industry/.gitkeepdiff --git a/examples/test_samples/zh/finance/.gitkeep b/tests/fixtures/test_samples/zh/finance/.gitkeepsimilarity index 100%rename from examples/test_samples/zh/finance/.gitkeeprename to tests/fixtures/test_samples/zh/finance/.gitkeepdiff --git a/tests/fixtures/test_samples/zh/general/arbitrary_tech_news.md b/tests/fixtures/test_samples/zh/general/arbitrary_tech_news.mdnew file mode 100644index 0000000..c18b323--- /dev/null+++ b/tests/fixtures/test_samples/zh/general/arbitrary_tech_news.md@@ -0,0 +1,96 @@+# 银河星际宣布神舟-50首飞成功,火星旅行票价降至200万元++**发布日期**:2090年6月18日+**作者**:太空记者 星空漫游者++---++## 核心发布++今日,全球领先的商业太空旅行企业**银河星际旅行公司**在**文昌航天发射中心**举行了盛大的新闻发布会,正式宣布其新一代载人飞船——**神舟-50**首飞成功!++银河星际创始人兼 CEO **李星河**在发布会上激动地表示:"神舟-50 的首飞成功,标志着人类太空旅行进入了一个新纪元。我们的目标是让每个普通人都能体验太空的魅力!"++---++## 神舟-50 亮点++### 1. 革命性的推进系统++神舟-50 搭载了银河星际自主研发的**"银河引擎"核动力推进系统**,这是人类航天史上的重大突破:+- 地球到火星的旅行时间从 6 个月缩短至 **30 天**+- 推进效率提升 500%+- 最大速度可达光速的 2%++### 2. 舒适的太空舱体验++神舟-50 可容纳 20 名乘客,提供豪华舒适的太空旅行体验:+- 每间客舱都配备全景天窗,可欣赏壮丽的宇宙景色+- 零重力健身中心和休闲娱乐区+- 人工重力舱,减少太空不适+- 米其林三星级太空餐厅++### 3. 安全保障系统++针对太空旅行的特殊风险,神舟-50 提供了多重安全保障:+- 三重冗余生命维持系统+- 微型黑洞防护罩+- 紧急返回舱,可在 24 小时内返回地球+- 符合 ISO 太空安全认证++---++## 航天团队++神舟-50 的研发由银河星际 CTO **王宇宙**领衔,核心团队成员包括:+- **张星辰**:总设计师,曾参与国际空间站建设+- **刘星云**:推进系统负责人,核动力专家+- **赵银河**:任务规划总监,拥有 20 年航天任务经验+- **孙星际**:安全系统架构师++整个研发团队共 5000 余人,历时 10 年完成神舟-50 的研发。++---++## 战略合作++发布会上,银河星际同时宣布与多家机构达成战略合作:+- **国际宇航局(IAA)**:共同开发深空探索技术+- **火星基地建设公司**:建设火星第一座商业基地+- **太空酒店集团**:共同运营轨道空间站酒店+- **星际旅游联盟**:建立全球太空旅行标准++---++## 发展历程++银河星际成立于 2050 年,总部位于上海:+- 2050 年 10 月:获得天使轮融资 10 亿元+- 2055 年 5 月:完成 A 轮融资 50 亿元+- 2060 年 8 月:神舟-30 首飞成功,实现亚轨道旅行+- 2070 年 3 月:B 轮融资 200 亿元,估值达 1000 亿元+- 2080 年 12 月:神舟-40 首飞成功,实现月球商业旅行+- 2090 年 6 月:神舟-50 首飞成功,开启火星旅行时代++---++## 价格与预订++神舟-50 火星旅行将于 2090 年 10 月 1 日正式接受预订:+- **经济舱**:200 万元/人+- **商务舱**:500 万元/人+- **豪华舱**:2000 万元/人++早期预订用户可享受 5 折优惠!++---++## 市场展望++行业分析师认为,神舟-50 的发布将彻底改变人类的旅行方式。预计到 2100 年,全球太空旅行市场规模将突破 10 万亿元。++银河星际表示,未来将继续投入研发,计划在 2095 年推出**神舟-60**,实现土星旅行,并在 2100 年实现太阳系内任意行星的商业旅行。++---++"宇宙那么大,我们都想去看看!" —— 银河星际创始人 李星河diff --git a/tests/fixtures/test_samples/zh/general/biography_scientist.md b/tests/fixtures/test_samples/zh/general/biography_scientist.mdnew file mode 100644index 0000000..4192b17--- /dev/null+++ b/tests/fixtures/test_samples/zh/general/biography_scientist.md@@ -0,0 +1,165 @@+# 祖冲之:中国古代科学巨匠++**作者**:历史传记 陈静怡+**出版年份**:2026年++---++## 序言++祖冲之是中国南北朝时期杰出的数学家、天文学家和发明家。他在数学、天文历法、机械制造等多个领域都取得了卓越的成就,是中国古代科学史上的一座丰碑。++本书旨在记录祖冲之不平凡的一生,展现他对科学的执着追求和对中华文明的伟大贡献。++---++## 第一章:早年岁月(429-450)++### 1.1 出生与家庭++429年(南朝宋文帝元嘉六年),祖冲之出生于**范阳郡遒县**(今河北省涞水县)的一个官宦世家。++曾祖父**祖台之**,东晋时官至侍中、光禄大夫;祖父**祖昌**,南朝宋时任大匠卿(掌管土木工程的官职);父亲**祖朔之**,任奉朝请。在这样的家庭环境中,祖冲之从小就受到了良好的教育,对科学技术产生了浓厚兴趣。++### 1.2 少年好学++祖冲之自幼勤奋好学,博览群书,尤其对数学、天文和机械制造等方面的书籍情有独钟。++史书载,祖冲之"少稽古,有机思",意思是他从小就研究古代典籍,有非凡的机智和思考能力。++### 1.3 步入仕途++450年左右,祖冲之进入官场,任南徐州(今江苏镇江)从事史,后又任公府参军。虽然身在仕途,但他始终没有放弃对科学研究的追求。++---++## 第二章:数学成就(450-470)++### 2.1 圆周率计算++祖冲之在数学上最杰出的贡献,是对圆周率(π)的精确计算。++经过长期艰苦的研究,祖冲之得出:+- **盈数(上限)**:3.1415927+- **朒数(下限)**:3.1415926++也就是说,圆周率的值在 **3.1415926 和 3.1415927 之间**,精确到了小数点后第7位。这一成就比西方数学家早了**一千多年**!++为了计算方便,祖冲之还给出了两个分数形式的近似值:+- **约率**:22/7(≈3.1429)+- **密率**:355/113(≈3.1415929)++其中,密率 355/113 是分子分母都在1000以内的最佳近似分数,这一成就也领先世界一千多年。++### 2.2 《缀术》++祖冲之将自己的数学研究成果写成了一部名为**《缀术》**的著作。++《缀术》内容深奥,被称为"算学经书",后来成为唐代国子监算学馆的教科书之一。可惜的是,这部珍贵的数学著作在北宋时失传了。++---++## 第三章:天文历法(450-464)++### 3.1 发现历法问题++祖冲之在研究天文历法时,发现当时通行的**《元嘉历》**存在诸多误差。++《元嘉历》由天文学家何承天制定,虽比之前的历法精密,但仍有不足之处。祖冲之经过多年观测和研究,指出了《元嘉历》的多个缺陷。++### 3.2 创制《大明历》++462年(南朝宋孝武帝大明六年),祖冲之编成了新的历法——**《大明历》**。++《大明历》的主要创新包括:+1. **引入岁差**:这是中国历法史上的重大突破,使回归年和恒星年得到区分+2. **改进闰周**:将旧历法的19年7闰改为391年144闰,更加精确+3. **测定交点月**:准确计算出交点月的日数为27.21223日,与现代测量值27.21222日相差极微+4. **精密测定五星周期**:对金、木、水、火、土五大行星的运行周期进行了精确测定++### 3.3 历法辩论++《大明历》编成后,祖冲之上表给宋孝武帝,请求颁布施行。然而,这一建议遭到了权臣**戴法兴**的反对。++戴法兴认为祖冲之"诬天背经",改变古法是大逆不道。祖冲之毫不畏惧,写下了著名的**《驳议》**一文,据理力争,逐条驳斥戴法兴的错误观点。++这场辩论是中国古代科学史上的著名事件,展现了祖冲之坚持真理、不畏权贵的科学精神。++可惜的是,由于戴法兴的阻挠,《大明历》在祖冲之生前未能颁布施行。直到510年(南朝梁武帝天监九年),在他的儿子**祖暅**的努力下,《大明历》才被正式采用,此时距离祖冲之去世已有10年。++---++## 第四章:机械发明(470-500)++### 4.1 指南车++祖冲之不仅是一位杰出的数学家和天文学家,还是一位出色的机械发明家。++他成功制造了**指南车**。指南车是一种能够指示方向的车辆,相传黄帝时代就有,但制法早已失传。祖冲之根据古籍记载,经过反复研究和试验,终于成功复原了指南车。++### 4.2 千里船++祖冲之还发明了一种名为"**千里船**"的船只。++据《南齐书·祖冲之传》记载:"于新亭江试之,日行百余里。"也就是说,千里船在新亭江(今江苏南京附近)试航时,一天能航行一百多里,这在当时是非常了不起的成就。++### 4.3 水碓磨++祖冲之还制造了**水碓磨**,这是一种利用水力驱动的粮食加工工具。++水碓磨将水碓和水磨结合起来,能够同时舂米和磨面,大大提高了生产效率。这一发明对农业生产的发展起到了重要作用。++### 4.4 计时器++此外,祖冲之还制造了**欹器**(一种计时器)和其他精巧的机械装置,展现了他卓越的机械制造才能。++---++## 第五章:历史影响与评价++### 5.1 国际认可++祖冲之的科学成就得到了国际科学界的广泛认可。++为了纪念祖冲之的杰出贡献:+- 1964年,紫金山天文台发现的一颗小行星被命名为**"祖冲之星"**+- 月球上的一座环形山被命名为**"祖冲之环形山"**++### 5.2 历史评价++《南齐书·祖冲之传》评价他:"冲之解钟律博塞,当时独绝,莫能对者。"++意思是祖冲之精通音律、棋艺等各种技艺,当时无人能及。++**李约瑟**(英国著名科学史学家)在《中国科学技术史》中对祖冲之给予了高度评价,称他是"中国古代最伟大的数学家之一"。++### 5.3 科学精神++祖冲之的身上,体现了中国古代科学家的优秀品质:+- **实事求是**:坚持从实际观测出发+- **勇于创新**:敢于突破旧说,提出新见+- **严谨细致**:对科学研究精益求精+- **不畏权贵**:敢于坚持真理,据理力争++---++## 结语++祖冲之是中国古代科学史上的一位全才,他在数学、天文、机械等多个领域都取得了卓越成就。++他的圆周率计算领先世界一千多年,他的《大明历》是中国历法史上的里程碑,他的机械发明展现了中国古代人民的智慧。++祖冲之的科学精神和成就,不仅是中华民族的骄傲,也是世界科学文化宝库中的珍贵遗产。他的名字,永远铭刻在人类科学发展的历史长河中。++---++## 附录:祖冲之年谱++- **429年**:出生于范阳郡遒县(今河北涞水)+- **450年左右**:任南徐州从事史、公府参军+- **462年**:编成《大明历》,上表朝廷+- **462-464年**:与戴法兴辩论历法+- **470年代**:制造指南车、千里船、水碓磨+- **494-498年**:任长水校尉+- **500年**:逝世,享年72岁+- **510年**:《大明历》被正式采用diff --git a/tests/fixtures/test_samples/zh/general/encyclopedia_machine_learning.md b/tests/fixtures/test_samples/zh/general/encyclopedia_machine_learning.mdnew file mode 100644index 0000000..b221fbb--- /dev/null+++ b/tests/fixtures/test_samples/zh/general/encyclopedia_machine_learning.md@@ -0,0 +1,268 @@+# 机器学习++**学科分类**:人工智能+**创建日期**:2026年2月22日+**最后更新**:2026年2月22日++---++## 概述++**机器学习(Machine Learning,ML)**是**人工智能(Artificial Intelligence,AI)**的一个重要分支,致力于研究如何使计算机系统通过经验自动改进。++机器学习的核心思想是:不依赖明确的硬编码规则,而是通过算法从数据中学习模式和规律。++---++## 定义++机器学习是一门多领域交叉学科,涉及**概率论**、**统计学**、**逼近论**、**凸分析**、**算法复杂度理论**等多门学科。++它是人工智能的核心,是使计算机具有智能的根本途径。机器学习算法通过分析大量数据,自动发现规律,并利用这些规律对未知数据进行预测或决策。++---++## 分类体系++### 1. 监督学习(Supervised Learning)++监督学习使用**带标签的训练数据**进行学习。算法从输入特征和对应的输出标签中学习映射关系。++**常见算法**:+- 线性回归(Linear Regression)+- 逻辑回归(Logistic Regression)+- 决策树(Decision Tree)+- 随机森林(Random Forest)+- 支持向量机(SVM)+- 神经网络(Neural Networks)++**应用场景**:+- 房价预测+- 垃圾邮件分类+- 图像识别+- 医疗诊断++### 2. 无监督学习(Unsupervised Learning)++无监督学习使用**不带标签的数据**,算法需要自行发现数据中的结构和模式。++**常见算法**:+- 聚类(Clustering):K-Means、层次聚类、DBSCAN+- 降维(Dimensionality Reduction):PCA、t-SNE、自编码器+- 关联规则挖掘(Association Rule Mining):Apriori、FP-Growth++**应用场景**:+- 客户分群+- 异常检测+- 推荐系统+- 主题建模++### 3. 强化学习(Reinforcement Learning,RL)++强化学习通过**智能体(Agent)与环境的交互**进行学习。智能体通过试错获得奖励(Reward),目标是最大化长期累积奖励。++**核心概念**:+- 状态(State)+- 动作(Action)+- 策略(Policy)+- 奖励(Reward)+- 价值函数(Value Function)++**常见算法**:+- Q-Learning+- 策略梯度(Policy Gradient)+- Actor-Critic+- DQN(Deep Q-Network)+- PPO(Proximal Policy Optimization)++**应用场景**:+- 游戏AI(如AlphaGo)+- 机器人控制+- 自动驾驶+- 资源调度++### 4. 半监督学习(Semi-Supervised Learning)++结合少量标注数据和大量未标注数据进行学习。++### 5. 自监督学习(Self-Supervised Learning)++从数据本身构造监督信号,无需人工标注。++---++## 关键概念++### 1. 特征(Feature)++特征是描述数据的可测量属性或变量。在机器学习中,特征是输入模型的原始数据的表示。++**示例**:+- 在房价预测中,特征可能包括:面积、房间数、位置、建筑年代+- 在图像识别中,特征可能是像素值或经过提取的视觉特征++### 2. 模型(Model)++模型是机器学习算法学习到的数学表示或函数,用于从输入特征预测输出。++**常见模型类型**:+- 线性模型+- 树模型+- 神经网络模型+- 概率图模型++### 3. 训练(Training)++训练是使用训练数据调整模型参数的过程,使模型能够学习到数据中的规律。++### 4. 推理/预测(Inference/Prediction)++推理是使用训练好的模型对新数据进行预测的过程。++### 5. 过拟合(Overfitting)++过拟合是指模型在训练数据上表现良好,但在新数据上泛化能力差的现象。++### 6. 欠拟合(Underfitting)++欠拟合是指模型未能充分学习数据规律,在训练集和测试集上表现都不好的现象。++---++## 相关领域++### 1. 深度学习(Deep Learning)++深度学习是机器学习的一个子领域,使用多层**神经网络(Neural Networks)**来学习数据的表示。++**常见网络结构**:+- 卷积神经网络(CNN):适用于图像+- 循环神经网络(RNN)/Transformer:适用于序列数据(文本、语音)+- 生成对抗网络(GAN):用于生成新数据++### 2. 数据挖掘(Data Mining)++数据挖掘是从大量数据中提取有用信息和知识的过程,与机器学习有很多重叠的技术。++### 3. 自然语言处理(Natural Language Processing,NLP)++NLP 专注于让计算机理解、解释和生成人类语言。++**应用**:+- 机器翻译+- 情感分析+- 问答系统+- 文本摘要++### 4. 计算机视觉(Computer Vision)++计算机视觉致力于让计算机从图像或视频中获取有意义的信息。++**应用**:+- 图像识别+- 目标检测+- 图像分割+- 人脸识别++### 5. 统计学(Statistics)++统计学为机器学习提供了理论基础和方法论支持。++---++## 应用场景++1. **医疗健康**:疾病诊断、药物发现、基因组分析+2. **金融**:信用评分、欺诈检测、算法交易+3. **零售**:推荐系统、需求预测、客户分群+4. **交通**:自动驾驶、交通预测、路线优化+5. **制造**:质量控制、预测性维护、供应链优化+6. **教育**:个性化学习、智能辅导、学习分析+7. **娱乐**:内容推荐、游戏AI、特效生成++---++## 历史发展++### 1940s-1950s:起源++- 1943年:Walter Pitts 和 Warren McCulloch 提出神经元数学模型+- 1950年:Alan Turing 提出"图灵测试"+- 1952年:Arthur Samuel 开发了第一个跳棋学习程序+- 1959年:Samuel 首次使用"机器学习"一词++### 1960s-1970s:早期发展++- 1967年:最近邻算法(Nearest Neighbor)提出+- 1969年:Minsky 和 Papert 出版《感知机》,指出单层感知机的局限性++### 1980s-1990s:复兴++- 1986年:反向传播(Backpropagation)算法重新流行+- 1989年:卷积神经网络概念提出+- 1995年:支持向量机(SVM)提出+- 1997年:LSTM(长短期记忆网络)提出++### 2000s-2010s:大数据时代++- 2006年:Geoffrey Hinton 提出"深度信念网络",深度学习开始复兴+- 2012年:AlexNet 在 ImageNet 竞赛中大幅领先,标志着深度学习时代的到来+- 2014年:GAN(生成对抗网络)提出+- 2017年:Transformer 架构提出++### 2020s 至今:大模型时代++- 2020年:GPT-3 发布+- 2022年:ChatGPT 发布,引起全球关注+- 2023年:多模态大模型兴起++---++## 重要人物和机构++### 人物++- **Geoffrey Hinton**:"深度学习之父",反向传播算法的重要贡献者+- **Yann LeCun**:卷积神经网络的先驱,Facebook AI 研究院首任院长+- **Yoshua Bengio**:深度学习领域的重要学者+- **Andrew Ng**:Google Brain 联合创始人,Coursera 联合创始人+- **Demis Hassabis**:DeepMind 创始人,AlphaGo 项目领导者++### 机构++- **Google DeepMind**:AlphaGo、AlphaFold 的研发机构+- **OpenAI**:GPT 系列模型的研发机构+- **Meta AI**(原 Facebook AI):PyTorch、FAIR+- **微软亚洲研究院(MSRA)**+- **斯坦福人工智能实验室(SAIL)**+- **麻省理工学院计算机科学与人工智能实验室(CSAIL)**++---++## 挑战与未来方向++### 当前挑战++1. 数据需求:深度学习需要大量标注数据+2. 可解释性:黑盒模型难以理解决策过程+3. 鲁棒性:对抗样本易导致模型失败+4. 公平性:算法可能存在偏见+5. 计算资源:大模型需要巨大的算力+6. 安全性:AI 系统的安全保障++### 未来方向++1. 小样本学习(Few-shot Learning)+2. 可解释 AI(Explainable AI)+3. 联邦学习(Federated Learning)+4. 因果推断(Causal Inference)+5. 多模态学习+6. 自主学习(Autonomous Learning)++---++## 参考资料++- Goodfellow, I., Bengio, Y., & Courville, A. (2016). *Deep Learning*. MIT Press.+- Murphy, K. P. (2012). *Machine Learning: A Probabilistic Perspective*. MIT Press.+- Hastie, T., Tibshirani, R., & Friedman, J. (2009). *The Elements of Statistical Learning*. Springer.diff --git a/tests/fixtures/test_samples/zh/general/regulation_company_policy.md b/tests/fixtures/test_samples/zh/general/regulation_company_policy.mdnew file mode 100644index 0000000..8a7a121--- /dev/null+++ b/tests/fixtures/test_samples/zh/general/regulation_company_policy.md@@ -0,0 +1,361 @@+# 宇宙第一摸鱼公司员工考勤与绩效管理制度++**文件编号**:MY-HR-2090-001+**版本**:V3.0+**生效日期**:2090年1月1日+**制定部门**:人力资源部+**审批人**:摸鱼大帝·王大锤++---++## 第一章 总则++### 1.1 目的++为规范公司员工的考勤管理,建立科学、合理的绩效考核体系,保障公司和员工的合法权益,在认真摸鱼的同时也能高效完成工作,根据《银河系劳动法》《宇宙劳动合同法》及相关星际法律法规,结合公司实际情况,制定本制度。++### 1.2 适用范围++本制度适用于宇宙第一摸鱼公司(以下简称"公司")所有正式员工、试用期员工及外星实习生。++### 1.3 基本原则++1. 摸鱼与工作平衡原则+2. 激励与约束相结合原则+3. 绩效与摸鱼时长挂钩原则+4. 持续摸鱼持续改进原则++---++## 第二章 考勤管理规定++### 2.1 工作时间++#### 2.1.1 标准工作时间++- **工作日**:周一至周五(每周任选两天可以申请在家摸鱼)+- **工作时间**:上午 10:00 - 12:00(摸鱼时间),下午 14:00 - 17:00(工作时间)+- **午休时间**:12:00 - 14:00(强制摸鱼时间)+- **每周工作时间**:不超过 25 小时(其余时间可以自由安排摸鱼)++#### 2.1.2 弹性工作时间++公司实行**超级弹性工作制度**,员工可在以下范围内灵活安排工作和摸鱼:+- 最早到岗时间:9:00+- 最晚到岗时间:11:00+- 最早离岗时间:16:00+- 最晚离岗时间:18:00++只要保证每日工作满 5 小时即可,摸鱼时间不限。++#### 2.1.3 特殊岗位++对于太空探索、外星沟通等需要 24 小时值守的岗位,实行**轮班制度**,具体排班由部门主管安排,但每连续工作 4 小时必须摸鱼 1 小时。++### 2.2 打卡管理++#### 2.2.1 打卡方式++员工须通过公司星际考勤系统进行打卡:+- **脑电波打卡**:使用脑机接口直接打卡+- **全息投影打卡**:通过全息影像打卡+- **量子纠缠打卡**:无论你在宇宙哪个角落都能打卡++#### 2.2.2 打卡要求++- 每日须打卡两次:上班打卡和下班打卡+- 上班打卡应在到岗后 30 分钟内完成+- 下班打卡应在离岗前后 30 分钟内完成+- 因公外出(包括去其他星球)无法打卡时,应提前通过OA系统提交"外出申请"++#### 2.2.3 忘记打卡++- 每月允许忘记打卡 5 次+- 忘记打卡须在 48 小时内通过OA系统提交"补卡申请",经部门主管审批后生效+- 忘记打卡超过 5 次的,第 6 次起每次按旷工 0.5 天处理(但摸鱼时间可以抵消)++### 2.3 迟到、早退与旷工++#### 2.3.1 迟到++- **迟到 30 分钟以内**:口头警告(不算事)+- **迟到 30-60 分钟**:书面警告,扣罚摸鱼券 1 张+- **迟到 60-120 分钟**:按旷工 0.5 天处理+- **迟到 120 分钟以上**:按旷工 1 天处理+- **每月累计迟到 5 次以上**:取消当月全勤摸鱼奖++#### 2.3.2 早退++- **早退 30 分钟以内**:口头警告+- **早退 30-60 分钟**:书面警告,扣罚摸鱼券 1 张+- **早退 60 分钟以上**:按旷工 0.5 天处理++#### 2.3.3 旷工++- **旷工 0.5 天**:扣罚 1.5 倍日薪,扣摸鱼券 2 张+- **旷工 1 天**:扣罚 2 倍日薪,取消当月全勤摸鱼奖+- **连续旷工 3 天或年度累计旷工 5 天**:视为严重违反公司规章制度,公司有权解除劳动合同(但可以协商用摸鱼券抵消)++### 2.4 请假管理++#### 2.4.1 请假类型++| 请假类型 | 说明 | 薪资待遇 | 摸鱼影响 |+|---------|------|---------|---------|+| 事假 | 因个人事务请假 | 无薪 | 不影响摸鱼券 |+| 病假 | 因病就医或休养 | 按当地最低工资标准的 80% 支付 | 额外赠送摸鱼券 |+| 婚假 | 结婚休假 | 带薪(10天) | 带薪摸鱼 |+| 产假 | 女性员工生育 | 带薪(按星际规定执行) | 全薪摸鱼 |+| 陪产假 | 男性员工配偶生育 | 带薪(30天) | 全薪摸鱼 |+| 丧假 | 直系亲属去世 | 带薪(3-7天) | 带薪摸鱼 |+| 年假 | 带薪年休假 | 带薪 | 带薪摸鱼 |+| 摸鱼假 | 专门用于摸鱼的假期 | 带薪 | 专门摸鱼 |++#### 2.4.2 请假流程++1. **提前申请**:原则上应提前 3 天通过OA系统提交"请假申请"+2. **填写信息**:明确请假类型、请假时间、请假事由(摸鱼假需说明摸鱼计划)+3. **上传证明**:病假、婚假等须上传相关证明材料(摸鱼假可上传摸鱼照片)+4. **审批流程**:+ - 请假 3 天以内:部门主管审批+ - 请假 3-7 天:部门主管 + HR 审批+ - 请假 7 天以上:部门主管 + HR + 摸鱼大帝审批+5. **工作交接**:请假前应做好工作交接+6. **紧急请假**:突发情况无法提前申请的,应在上班前 1 小时内通过量子通讯通知部门主管,并在事后 48 小时内补办请假手续++#### 2.4.3 年假规定++| 司龄 | 年假天数 | 摸鱼天数 |+|------|---------|---------|+| 1-5年 | 10天 | 5天 |+| 5-10年 | 15天 | 10天 |+| 10年以上 | 20天 | 15天 |++- 年假可在当年内集中或分段使用,不可累计到下一年+- 摸鱼天数可以与年假叠加使用+- 公司因工作需要不能安排员工休年假的,经员工同意,可按日薪的 5 倍支付未休年假工资++---++## 第三章 绩效考核流程++### 3.1 考核周期++- **月度考核**:每月初进行上月考核+- **季度考核**:每季度首月进行上季度考核+- **年度考核**:每年 1 月进行上年度考核+- **摸鱼考核**:每周末进行摸鱼质量评估++### 3.2 考核内容++#### 3.2.1 绩效考核指标(KPI + MFI)++| 指标类别 | 权重 | 具体内容 |+|---------|------|---------|+| 工作业绩 | 40% | 目标完成情况、工作质量、工作效率 |+| 摸鱼质量 | 30% | 摸鱼时长、摸鱼方式多样性、摸鱼创造性 |+| 工作能力 | 15% | 专业能力、沟通能力、团队协作、学习能力 |+| 工作态度 | 10% | 责任心、执行力、纪律性、摸鱼积极性 |+| 考勤情况 | 5% | 出勤率、迟到早退情况 |++#### 3.2.2 不同岗位的考核侧重点++- **太空探索岗位**:侧重技术创新、任务完成质量、摸鱼创造性+- **外星沟通岗位**:侧重沟通能力、文化理解、摸鱼时间控制+- **摸鱼指导岗位**:侧重摸鱼技巧、摸鱼培训效果、摸鱼创新+- **职能岗位**:侧重服务质量、流程优化、摸鱼效率++### 3.3 考核流程++#### 3.3.1 目标设定(P1)++- **时间**:考核期初(每月/季度/年度首月)+- **流程**:+ 1. 员工根据部门目标和岗位职责,设定个人工作目标和摸鱼目标+ 2. 与部门主管沟通,确定最终目标+ 3. 在绩效系统中录入考核目标++#### 3.3.2 过程管理(P2)++- **时间**:考核期间+- **流程**:+ 1. 员工定期记录工作进展和摸鱼情况+ 2. 部门主管与员工进行 1-2 次中期沟通+ 3. 及时调整目标或提供必要的摸鱼指导++#### 3.3.3 绩效评估(P3)++- **时间**:考核期末(每月/季度/年度末月)+- **流程**:+ 1. 员工进行自我评价,填写《绩效自评表》和《摸鱼自评表》+ 2. 部门主管根据员工表现、目标完成情况和摸鱼质量进行评价+ 3. HR 部门审核考核结果++#### 3.3.4 绩效反馈(P4)++- **时间**:考核结果确定后 5 个工作日内+- **流程**:+ 1. 部门主管与员工进行 1 对 1 的绩效面谈+ 2. 肯定成绩,指出不足,共同制定改进计划(包括工作和摸鱼两方面)+ 3. 员工在《绩效面谈记录表》上签字确认++### 3.4 考核等级++| 等级 | 定义 | 分数区间 | 比例限制 | 摸鱼奖励 |+|------|------|---------|---------|---------|+| S | 卓越,远超预期 | 95-100分 | ≤5% | 摸鱼券 10 张 + 带薪摸鱼假 5 天 |+| A | 优秀,超出预期 | 85-94分 | ≤20% | 摸鱼券 5 张 + 带薪摸鱼假 2 天 |+| B | 良好,符合预期 | 70-84分 | 50%左右 | 摸鱼券 2 张 |+| C | 合格,基本符合预期 | 60-69分 | ≥20% | 摸鱼券 1 张 |+| D | 不合格,未达到预期 | 60分以下 | ≤5% | 无摸鱼奖励,需参加摸鱼培训 |++---++## 第四章 奖惩规则++### 4.1 奖励++#### 4.1.1 奖励类型++| 奖励类型 | 说明 | 奖励方式 |+|---------|------|---------|+| 月度摸鱼之星 | 每月评选的摸鱼质量最高员工 | 奖金 5000 元 + 摸鱼券 10 张 + 证书 |+| 季度摸鱼之星 | 每季度评选的摸鱼优秀员工 | 奖金 15000 元 + 摸鱼券 20 张 + 证书 + 带薪摸鱼假 3 天 |+| 年度摸鱼大师 | 每年评选的摸鱼大师 | 奖金 100000 元 + 摸鱼券 50 张 + 证书 + 带薪摸鱼假 10 天 + 晋升优先 |+| 创新奖 | 对摸鱼方式或工作方式创新有突出贡献 | 奖金 10000-100000 元 + 证书 |+| 团队奖 | 对优秀摸鱼团队的奖励 | 团队摸鱼活动经费 10000-50000 元 |++#### 4.1.2 评选标准++- 月度摸鱼之星:摸鱼考核连续 2 个月为 A 以上,工作绩效不低于 B,无违纪记录+- 季度摸鱼之星:季度摸鱼考核为 S 或 A,工作绩效不低于 B,无违纪记录,团队评价优秀+- 年度摸鱼大师:年度摸鱼考核为 S,工作绩效不低于 A,全年无违纪记录,对摸鱼文化有重大贡献+- 创新奖:提出的创新方案被采纳并取得显著成效+- 团队奖:团队目标超额完成,摸鱼氛围良好++#### 4.1.3 评选流程++1. 部门推荐或员工自荐+2. 人力资源部审核+3. 摸鱼委员会评审+4. 公示 3 个工作日+5. 正式表彰++### 4.2 处罚++#### 4.2.1 处罚类型++| 处罚类型 | 适用情形 | 处罚方式 |+|---------|---------|---------|+| 口头警告 | 轻微违反规章制度 | 口头批评教育,扣摸鱼券 1 张 |+| 书面警告 | 多次轻微违规或较严重违规 | 书面警告,扣摸鱼券 3-5 张 |+| 记过 | 严重违反规章制度 | 记过处分,扣摸鱼券 5-10 张,取消当年摸鱼评优资格 |+| 降薪/降职 | 不能胜任工作或严重摸鱼过度影响工作 | 降薪 10%-30% 或降职,减少摸鱼时间 |+| 解除劳动合同 | 严重违反规章制度、严重失职等 | 解除劳动合同,无需支付经济补偿 |++#### 4.2.2 具体违规行为与处罚对照表++| 违规行为 | 处罚等级 |+|---------|---------|+| 1个月内迟到早退 5 次以内 | 口头警告 |+| 1个月内迟到早退 5-10 次 | 书面警告 |+| 旷工 1 天 | 书面警告 |+| 连续旷工 2-3 天 | 记过 |+| 连续旷工 3 天以上或年度累计旷工 5 天 | 解除劳动合同 |+| 工作时间摸鱼过度,严重影响工作 | 书面警告,再次发生记过 |+| 泄露公司星际秘密 | 记过,造成严重损失的解除劳动合同 |+| 伪造考勤或摸鱼记录 | 记过,再次发生解除劳动合同 |+| 欺骗公司、弄虚作假 | 记过,情节严重的解除劳动合同 |+| 其他严重违反公司规章制度的行为 | 视情节轻重给予相应处罚 |++#### 4.2.3 处罚流程++1. **调查核实**:发现违规行为后,由 HR 部门或相关部门进行调查+2. **听取陈述**:听取当事人的陈述和申辩+3. **做出决定**:根据调查结果和制度规定,做出处罚决定+4. **送达通知**:向当事人送达《处罚通知书》+5. **申诉权利**:当事人对处罚决定不服的,可在 3 个工作日内向 HR 部门提出书面申诉++---++## 第五章 申诉程序++### 5.1 申诉范围++员工对以下事项有异议的,可提出申诉:+- 绩效考核结果+- 摸鱼考核结果+- 考勤记录+- 奖惩决定+- 其他认为权益受到侵害的事项++### 5.2 申诉流程++#### 5.2.1 第一级申诉++1. **时间**:收到相关决定或记录后 3 个工作日内+2. **方式**:向直接上级或部门主管提出书面申诉+3. **处理**:直接上级或部门主管应在 3 个工作日内给予答复++#### 5.2.2 第二级申诉++1. **前提**:对第一级申诉结果不满意+2. **时间**:收到第一级答复后 3 个工作日内+3. **方式**:向 HR 部门提出书面申诉,并附相关证明材料+4. **处理**:HR 部门应在 5 个工作日内进行调查并给予答复++#### 5.2.3 第三级申诉++1. **前提**:对第二级申诉结果仍不满意+2. **时间**:收到第二级答复后 5 个工作日内+3. **方式**:向公司摸鱼争议调解委员会提出书面申诉+4. **处理**:调解委员会应在 10 个工作日内召开会议,进行调解并做出最终决定++### 5.3 申诉材料++申诉材料应包括:+- 申诉书(说明申诉事项、理由、要求)+- 相关证明材料+- 摸鱼记录或工作记录++---++## 第六章 附则++### 6.1 制度修订++本制度的修订由人力资源部提出,经摸鱼大帝审批后公布实施。++### 6.2 制度解释++本制度的解释权归公司人力资源部和摸鱼委员会所有。++### 6.3 生效日期++本制度自 2090 年 1 月 1 日起生效,原《宇宙第一摸鱼公司员工考勤与绩效管理制度(V2.0)》同时废止。++### 6.4 宣传与培训++人力资源部和摸鱼委员会负责本制度的宣传和培训,确保每位员工都了解和遵守本制度,同时也能充分享受摸鱼的乐趣。++---++**附件**:+1. 《请假申请表》+2. 《绩效考核表》+3. 《摸鱼考核表》+4. 《绩效面谈记录表》+5. 《奖励/处罚审批表》+6. 《摸鱼券使用指南》+7. 《申诉书模板》++---++**制度变更记录**:+- V1.0(2080年1月1日):首次发布,确立摸鱼文化+- V2.0(2085年6月1日):修订绩效考核部分,增加摸鱼考核指标+- V3.0(2090年1月1日):完善奖惩规则,优化摸鱼制度++---++"认真摸鱼,快乐工作!" —— 摸鱼大帝·丰大锤diff --git a/examples/test_samples/zh/industry/.gitkeep b/tests/fixtures/test_samples/zh/industry/.gitkeepsimilarity index 100%rename from examples/test_samples/zh/industry/.gitkeeprename to tests/fixtures/test_samples/zh/industry/.gitkeep