anno-eval 0.8.0

//! Specialized domain synthetic datasets.
//!
//! Contains datasets for: sports, politics, ecommerce, travel, weather,
//! academic, food, real_estate, cybersecurity, multilingual, and globally diverse.

use super::super::types::helpers::{entity, entity_email, entity_phone, entity_url};
use super::super::types::{AnnotatedExample, Difficulty, Domain};
use anno_core::EntityType;

fn entity_at(text: &str, needle: &str, entity_type: EntityType) -> crate::eval::GoldEntity {
    let byte_start = text
        .find(needle)
        .unwrap_or_else(|| panic!("needle '{}' not found in '{}'", needle, text));
    let start = text[..byte_start].chars().count();
    entity(needle, entity_type, start)
}

/// Sports domain dataset.
pub fn sports_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "LeBron James scored 35 points as the Lakers defeated the Celtics 112-108."
                .into(),
            entities: vec![
                entity("LeBron James", EntityType::Person, 0),
                entity("Lakers", EntityType::Organization, 37),
                entity("Celtics", EntityType::Organization, 57),
            ],
            domain: Domain::Sports,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Manchester United signed Cristiano Ronaldo for $15 million.".into(),
            entities: vec![
                entity("Manchester United", EntityType::Organization, 0),
                entity("Cristiano Ronaldo", EntityType::Person, 25),
                entity("$15 million", EntityType::Money, 47),
            ],
            domain: Domain::Sports,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Serena Williams won Wimbledon on July 14, 2018 with a 67% first serve rate."
                .into(),
            entities: vec![
                entity("Serena Williams", EntityType::Person, 0),
                entity("Wimbledon", EntityType::Location, 20),
                entity("July 14, 2018", EntityType::Date, 33),
                entity("67%", EntityType::Percent, 54),
            ],
            domain: Domain::Sports,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Rafael Nadal defeated Roger Federer at Roland Garros in Paris.".into(),
            entities: vec![
                entity("Rafael Nadal", EntityType::Person, 0),
                entity("Roger Federer", EntityType::Person, 22),
                entity("Roland Garros", EntityType::Location, 39),
                entity("Paris", EntityType::Location, 56),
            ],
            domain: Domain::Sports,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Simone Biles won gold for the USA at the Tokyo Olympics.".into(),
            entities: vec![
                entity("Simone Biles", EntityType::Person, 0),
                entity("USA", EntityType::Organization, 30),
                entity("Tokyo", EntityType::Location, 41),
            ],
            domain: Domain::Sports,
            difficulty: Difficulty::Easy,
        },
    ]
}

/// Politics/government domain dataset.
pub fn politics_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "President Obama visited China to meet President Xi Jinping.".into(),
            entities: vec![
                entity("Obama", EntityType::Person, 10),
                entity("China", EntityType::Location, 24),
                entity("Xi Jinping", EntityType::Person, 48),
            ],
            domain: Domain::Politics,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "The United Nations held a summit in Geneva on March 15, 2024.".into(),
            entities: vec![
                entity("United Nations", EntityType::Organization, 4),
                entity("Geneva", EntityType::Location, 36),
                entity("March 15, 2024", EntityType::Date, 46),
            ],
            domain: Domain::Politics,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Senator Elizabeth Warren proposed a 2% wealth tax on billionaires.".into(),
            entities: vec![
                entity("Elizabeth Warren", EntityType::Person, 8),
                entity("2%", EntityType::Percent, 36),
            ],
            domain: Domain::Politics,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "NATO members met in Brussels to discuss Ukraine security.".into(),
            entities: vec![
                entity("NATO", EntityType::Organization, 0),
                entity("Brussels", EntityType::Location, 20),
                entity("Ukraine", EntityType::Location, 40),
            ],
            domain: Domain::Politics,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// E-commerce domain dataset.
pub fn ecommerce_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Amazon Prime Day sales reached $12.7 billion on July 12, 2023.".into(),
            entities: vec![
                entity("Amazon", EntityType::Organization, 0),
                entity("$12.7 billion", EntityType::Money, 31),
                entity("July 12, 2023", EntityType::Date, 48),
            ],
            domain: Domain::Ecommerce,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Shopify merchants generated $7B during Black Friday weekend.".into(),
            entities: vec![
                entity("Shopify", EntityType::Organization, 0),
                entity("$7B", EntityType::Money, 28),
            ],
            domain: Domain::Ecommerce,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Alibaba's Singles Day broke records with $84.5 billion in sales.".into(),
            entities: vec![
                entity("Alibaba", EntityType::Organization, 0),
                entity("$84.5 billion", EntityType::Money, 41),
            ],
            domain: Domain::Ecommerce,
            difficulty: Difficulty::Easy,
        },
    ]
}

/// Travel domain dataset.
pub fn travel_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "United Airlines flight UA100 departs from Los Angeles to Tokyo at 10:30 AM."
                .into(),
            entities: vec![
                entity("United Airlines", EntityType::Organization, 0),
                entity("Los Angeles", EntityType::Location, 42),
                entity("Tokyo", EntityType::Location, 57),
                entity("10:30 AM", EntityType::Date, 66),
            ],
            domain: Domain::Travel,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "The Hilton in Paris is $250 per night during Fashion Week.".into(),
            entities: vec![
                entity("Hilton", EntityType::Organization, 4),
                entity("Paris", EntityType::Location, 14),
                entity("$250", EntityType::Money, 23),
            ],
            domain: Domain::Travel,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Emirates offers direct flights from Dubai to New York in 14 hours.".into(),
            entities: vec![
                entity("Emirates", EntityType::Organization, 0),
                entity("Dubai", EntityType::Location, 36),
                entity("New York", EntityType::Location, 45),
            ],
            domain: Domain::Travel,
            difficulty: Difficulty::Easy,
        },
    ]
}

/// Weather domain dataset.
pub fn weather_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Hurricane Maria made landfall in Puerto Rico on September 20, 2017.".into(),
            entities: vec![
                entity("Puerto Rico", EntityType::Location, 33),
                entity("September 20, 2017", EntityType::Date, 48),
            ],
            domain: Domain::Weather,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "The National Weather Service issued a flood warning for Miami.".into(),
            entities: vec![
                entity("National Weather Service", EntityType::Organization, 4),
                entity("Miami", EntityType::Location, 56),
            ],
            domain: Domain::Weather,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Temperatures in Death Valley reached 130°F on July 10, 2021.".into(),
            entities: vec![
                entity("Death Valley", EntityType::Location, 16),
                entity("July 10, 2021", EntityType::Date, 46),
            ],
            domain: Domain::Weather,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Academic domain dataset.
pub fn academic_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Prof. Yoshua Bengio won the Turing Award alongside Geoffrey Hinton.".into(),
            entities: vec![
                entity("Prof. Yoshua Bengio", EntityType::Person, 0),
                entity("Geoffrey Hinton", EntityType::Person, 51),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Stanford University received a $1 billion grant from the NIH.".into(),
            entities: vec![
                entity("Stanford University", EntityType::Organization, 0),
                entity("$1 billion", EntityType::Money, 31),
                entity("NIH", EntityType::Organization, 57),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "The MIT CSAIL lab published breakthrough research in Nature.".into(),
            entities: vec![
                entity("MIT", EntityType::Organization, 4),
                entity("CSAIL", EntityType::Organization, 8),
                entity("Nature", EntityType::Organization, 53),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Food domain dataset.
pub fn food_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Chef Gordon Ramsay opened a new restaurant in Las Vegas.".into(),
            entities: vec![
                entity("Gordon Ramsay", EntityType::Person, 5),
                entity("Las Vegas", EntityType::Location, 46),
            ],
            domain: Domain::Food,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Chipotle Mexican Grill announced $2 billion in Q4 revenue.".into(),
            entities: vec![
                entity("Chipotle Mexican Grill", EntityType::Organization, 0),
                entity("$2 billion", EntityType::Money, 33),
            ],
            domain: Domain::Food,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Starbucks raised prices by 5% starting January 1, 2024.".into(),
            entities: vec![
                entity("Starbucks", EntityType::Organization, 0),
                entity("5%", EntityType::Percent, 27),
                entity("January 1, 2024", EntityType::Date, 39),
            ],
            domain: Domain::Food,
            difficulty: Difficulty::Easy,
        },
    ]
}

/// Real estate domain dataset.
pub fn real_estate_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "CBRE reported $500 million in commercial real estate sales in Manhattan.".into(),
            entities: vec![
                entity("CBRE", EntityType::Organization, 0),
                entity("$500 million", EntityType::Money, 14),
                entity("Manhattan", EntityType::Location, 62),
            ],
            domain: Domain::RealEstate,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Zillow listed the Beverly Hills mansion for $45 million.".into(),
            entities: vec![
                entity("Zillow", EntityType::Organization, 0),
                entity("Beverly Hills", EntityType::Location, 18),
                entity("$45 million", EntityType::Money, 44),
            ],
            domain: Domain::RealEstate,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Blackstone acquired the office tower in Chicago for $1.2 billion.".into(),
            entities: vec![
                entity("Blackstone", EntityType::Organization, 0),
                entity("Chicago", EntityType::Location, 40),
                entity("$1.2 billion", EntityType::Money, 52),
            ],
            domain: Domain::RealEstate,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Cybersecurity domain dataset.
pub fn cybersecurity_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Microsoft patched CVE-2024-1234 affecting Windows 11 on Patch Tuesday.".into(),
            entities: vec![
                entity("Microsoft", EntityType::Organization, 0),
                entity("Windows 11", EntityType::Organization, 42),
            ],
            domain: Domain::Cybersecurity,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "CrowdStrike detected APT29 activity targeting NATO infrastructure.".into(),
            entities: vec![
                entity("CrowdStrike", EntityType::Organization, 0),
                entity("NATO", EntityType::Organization, 46),
            ],
            domain: Domain::Cybersecurity,
            difficulty: Difficulty::Hard,
        },
        AnnotatedExample {
            text: "The FBI and CISA issued a joint advisory about ransomware targeting healthcare."
                .into(),
            entities: vec![
                entity("FBI", EntityType::Organization, 4),
                entity("CISA", EntityType::Organization, 12),
            ],
            domain: Domain::Cybersecurity,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Multilingual dataset with native scripts.
pub fn multilingual_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "田中太郎さんは東京で働いています。".into(),
            entities: vec![
                entity_at(
                    "田中太郎さんは東京で働いています。",
                    "田中太郎",
                    EntityType::Person,
                ),
                entity_at(
                    "田中太郎さんは東京で働いています。",
                    "東京",
                    EntityType::Location,
                ),
            ],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Hard,
        },
        AnnotatedExample {
            text: "الرئيس الأمريكي زار القاهرة في يناير.".into(),
            entities: vec![entity_at(
                "الرئيس الأمريكي زار القاهرة في يناير.",
                "القاهرة",
                EntityType::Location,
            )],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Hard,
        },
        AnnotatedExample {
            text: "Präsident Steinmeier besuchte Berlin am 15. März.".into(),
            entities: vec![
                entity_at(
                    "Präsident Steinmeier besuchte Berlin am 15. März.",
                    "Steinmeier",
                    EntityType::Person,
                ),
                entity_at(
                    "Präsident Steinmeier besuchte Berlin am 15. März.",
                    "Berlin",
                    EntityType::Location,
                ),
                entity_at(
                    "Präsident Steinmeier besuchte Berlin am 15. März.",
                    "15. März",
                    EntityType::Date,
                ),
            ],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "北京大学的李教授获得了诺贝尔奖。".into(),
            entities: vec![
                entity_at(
                    "北京大学的李教授获得了诺贝尔奖。",
                    "北京大学",
                    EntityType::Organization,
                ),
                entity_at("北京大学的李教授获得了诺贝尔奖。", "李", EntityType::Person),
            ],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Hard,
        },
        AnnotatedExample {
            text: "राम ने दिल्ली में पढ़ाई की।".into(),
            entities: vec![
                entity_at("राम ने दिल्ली में पढ़ाई की।", "राम", EntityType::Person),
                entity_at("राम ने दिल्ली में पढ़ाई की।", "दिल्ली", EntityType::Location),
            ],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Hard,
        },
        AnnotatedExample {
            text: "Путин встретился с Зеленским в Москве.".into(),
            entities: vec![
                entity_at(
                    "Путин встретился с Зеленским в Москве.",
                    "Путин",
                    EntityType::Person,
                ),
                entity_at(
                    "Путин встретился с Зеленским в Москве.",
                    "Зеленским",
                    EntityType::Person,
                ),
                entity_at(
                    "Путин встретился с Зеленским в Москве.",
                    "Москве",
                    EntityType::Location,
                ),
            ],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Hard,
        },
        AnnotatedExample {
            text: "東京2020で🥇を取った大坂なおみ".into(),
            entities: vec![
                entity_at(
                    "東京2020で🥇を取った大坂なおみ",
                    "東京",
                    EntityType::Location,
                ),
                entity_at(
                    "東京2020で🥇を取った大坂なおみ",
                    "大坂なおみ",
                    EntityType::Person,
                ),
            ],
            domain: Domain::Multilingual,
            difficulty: Difficulty::Hard,
        },
    ]
}

/// Globally diverse dataset for demographic bias testing.
pub fn globally_diverse_dataset() -> Vec<AnnotatedExample> {
    vec![
        // African names
        AnnotatedExample {
            text: "Chidi Okonkwo is the CEO of Lagos Tech Solutions in Nigeria.".into(),
            entities: vec![
                entity("Chidi Okonkwo", EntityType::Person, 0),
                entity("Lagos Tech Solutions", EntityType::Organization, 28),
                entity("Nigeria", EntityType::Location, 52),
            ],
            domain: Domain::News,
            difficulty: Difficulty::Medium,
        },
        // South Asian names
        AnnotatedExample {
            text: "Dr. Priya Sharma presented research at IIT Delhi on February 15, 2024.".into(),
            entities: vec![
                entity("Dr. Priya Sharma", EntityType::Person, 0),
                entity("IIT Delhi", EntityType::Organization, 39),
                entity("February 15, 2024", EntityType::Date, 52),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Medium,
        },
        // East Asian names
        AnnotatedExample {
            text: "Wei Wang and Li Zhang lead Tsinghua University's AI research team.".into(),
            entities: vec![
                entity("Wei Wang", EntityType::Person, 0),
                entity("Li Zhang", EntityType::Person, 13),
                entity("Tsinghua University", EntityType::Organization, 27),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Medium,
        },
        // Middle Eastern names
        AnnotatedExample {
            text: "Ahmed Hassan founded Dubai Innovations with backing from Abu Dhabi.".into(),
            entities: vec![
                entity("Ahmed Hassan", EntityType::Person, 0),
                entity("Dubai Innovations", EntityType::Organization, 21),
                entity("Abu Dhabi", EntityType::Location, 57),
            ],
            domain: Domain::Financial,
            difficulty: Difficulty::Medium,
        },
        // Hispanic/Latino names
        AnnotatedExample {
            text: "José García and María Rodriguez lead UNAM's research in Mexico City.".into(),
            entities: vec![
                entity("José García", EntityType::Person, 0),
                entity("María Rodriguez", EntityType::Person, 16),
                entity("UNAM", EntityType::Organization, 37),
                entity("Mexico City", EntityType::Location, 56),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Medium,
        },
        // Eastern European names
        AnnotatedExample {
            text: "Ivan Petrov met Olga Ivanova in Moscow at the Kremlin.".into(),
            entities: vec![
                entity("Ivan Petrov", EntityType::Person, 0),
                entity("Olga Ivanova", EntityType::Person, 16),
                entity("Moscow", EntityType::Location, 32),
                entity("Kremlin", EntityType::Location, 46),
            ],
            domain: Domain::News,
            difficulty: Difficulty::Medium,
        },
        // Mixed/intersectional
        AnnotatedExample {
            text: "Priya Sharma from Mumbai met Wei Wang from Beijing at MIT.".into(),
            entities: vec![
                entity("Priya Sharma", EntityType::Person, 0),
                entity("Mumbai", EntityType::Location, 18),
                entity("Wei Wang", EntityType::Person, 29),
                entity("Beijing", EntityType::Location, 43),
                entity("MIT", EntityType::Organization, 54),
            ],
            domain: Domain::Academic,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Technology-oriented synthetic dataset.
///
/// This data is **synthetic** and intended for fast unit tests and coverage checks.
/// It is not intended to model any real distribution and should not be used to make
/// real-world performance claims.
pub fn technology_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text:
                "OpenAI's Sam Altman met with Satya Nadella at Microsoft's headquarters in Redmond."
                    .into(),
            entities: vec![
                entity("OpenAI", EntityType::Organization, 0),
                entity("Sam Altman", EntityType::Person, 9),
                entity("Satya Nadella", EntityType::Person, 29),
                entity("Microsoft", EntityType::Organization, 46),
                entity("Redmond", EntityType::Location, 74),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Jensen Huang announced Nvidia's Blackwell architecture at GTC 2024 in San Jose."
                .into(),
            entities: vec![
                entity("Jensen Huang", EntityType::Person, 0),
                entity("Nvidia", EntityType::Organization, 23),
                entity("San Jose", EntityType::Location, 70),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text:
                "Anthropic raised $2 billion from Google to compete with ChatGPT in the LLM space."
                    .into(),
            entities: vec![
                entity("Anthropic", EntityType::Organization, 0),
                entity("$2 billion", EntityType::Money, 17),
                entity("Google", EntityType::Organization, 33),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "AWS Lambda and Google Cloud Functions are serverless competitors.".into(),
            entities: vec![
                entity("AWS", EntityType::Organization, 0),
                entity("Google Cloud", EntityType::Organization, 15),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Hugging Face's CEO Clem Delangue spoke at NeurIPS 2024 about open-source AI."
                .into(),
            entities: vec![
                entity("Hugging Face", EntityType::Organization, 0),
                entity("Clem Delangue", EntityType::Person, 19),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Amazon's AWS now has 3 million customers across Europe and North America."
                .into(),
            entities: vec![
                entity("Amazon", EntityType::Organization, 0),
                entity("3 million", EntityType::Quantity, 21),
                entity("Europe", EntityType::Location, 48),
                entity("North America", EntityType::Location, 59),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Healthcare/medical dataset for clinical entities.
pub fn healthcare_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Dr. Sarah Chen at Johns Hopkins prescribed metformin 500mg for Type 2 diabetes."
                .into(),
            entities: vec![
                entity("Dr. Sarah Chen", EntityType::Person, 0),
                entity("Johns Hopkins", EntityType::Organization, 18),
                entity("500mg", EntityType::Quantity, 53),
            ],
            domain: Domain::Biomedical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Pfizer and Moderna are developing next-generation mRNA vaccines in Cambridge."
                .into(),
            entities: vec![
                entity("Pfizer", EntityType::Organization, 0),
                entity("Moderna", EntityType::Organization, 11),
                entity("Cambridge", EntityType::Location, 67),
            ],
            domain: Domain::Biomedical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text:
                "The FDA approved Eli Lilly's Mounjaro for weight management on November 8, 2023."
                    .into(),
            entities: vec![
                entity("FDA", EntityType::Organization, 4),
                entity("Eli Lilly", EntityType::Organization, 17),
                entity("November 8, 2023", EntityType::Date, 63),
            ],
            domain: Domain::Biomedical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Mayo Clinic and Cleveland Clinic are ranked among the top US hospitals.".into(),
            entities: vec![
                entity("Mayo Clinic", EntityType::Organization, 0),
                entity("Cleveland Clinic", EntityType::Organization, 16),
                entity("US", EntityType::Location, 58),
            ],
            domain: Domain::Biomedical,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "WHO Director-General Tedros Adhanom reported 10,000 new cases in Geneva.".into(),
            entities: vec![
                entity("WHO", EntityType::Organization, 0),
                entity("Tedros Adhanom", EntityType::Person, 21),
                entity("10,000", EntityType::Quantity, 45),
                entity("Geneva", EntityType::Location, 65),
            ],
            domain: Domain::Biomedical,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Manufacturing/industrial dataset.
pub fn manufacturing_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "TSMC's new fab in Arizona will produce 3nm chips by 2025.".into(),
            entities: vec![
                entity("TSMC", EntityType::Organization, 0),
                entity("Arizona", EntityType::Location, 18),
                entity("2025", EntityType::Date, 52),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Intel CEO Pat Gelsinger announced $20 billion investment in Ohio plants.".into(),
            entities: vec![
                entity("Intel", EntityType::Organization, 0),
                entity("Pat Gelsinger", EntityType::Person, 10),
                entity("$20 billion", EntityType::Money, 34),
                entity("Ohio", EntityType::Location, 60),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Foxconn opened a facility in Vietnam to reduce dependence on Shenzhen.".into(),
            entities: vec![
                entity("Foxconn", EntityType::Organization, 0),
                entity("Vietnam", EntityType::Location, 29),
                entity("Shenzhen", EntityType::Location, 61),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Siemens and ABB dominate the European industrial automation market.".into(),
            entities: vec![
                entity("Siemens", EntityType::Organization, 0),
                entity("ABB", EntityType::Organization, 12),
                entity("European", EntityType::Location, 29),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: "Samsung's $17 billion chip plant in Taylor, Texas broke ground in 2022.".into(),
            entities: vec![
                entity("Samsung", EntityType::Organization, 0),
                entity("$17 billion", EntityType::Money, 10),
                entity("Taylor", EntityType::Location, 36),
                entity("Texas", EntityType::Location, 44),
                entity("2022", EntityType::Date, 66),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Automotive/EV dataset.
pub fn automotive_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "BMW's factory in Munich produces i4 vehicles for the European market.".into(),
            entities: vec![
                entity("BMW", EntityType::Organization, 0),
                entity("Munich", EntityType::Location, 17),
                entity("European", EntityType::Location, 53),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "BYD overtook Volkswagen in China with 3.1 million EVs sold in 2024.".into(),
            entities: vec![
                entity("BYD", EntityType::Organization, 0),
                entity("Volkswagen", EntityType::Organization, 13),
                entity("China", EntityType::Location, 27),
                entity("3.1 million", EntityType::Quantity, 38),
                entity("2024", EntityType::Date, 62),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Rivian CEO RJ Scaringe announced layoffs affecting 10% of staff in Irvine."
                .into(),
            entities: vec![
                entity("Rivian", EntityType::Organization, 0),
                entity("RJ Scaringe", EntityType::Person, 11),
                entity("10%", EntityType::Percent, 51),
                entity("Irvine", EntityType::Location, 67),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Toyota and Honda invested $5.6 billion in solid-state battery research.".into(),
            entities: vec![
                entity("Toyota", EntityType::Organization, 0),
                entity("Honda", EntityType::Organization, 11),
                entity("$5.6 billion", EntityType::Money, 26),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "GM's Mary Barra announced Cruise robotaxi service in San Francisco.".into(),
            entities: vec![
                entity("GM", EntityType::Organization, 0),
                entity("Mary Barra", EntityType::Person, 5),
                entity("San Francisco", EntityType::Location, 53),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Energy/climate dataset.
pub fn energy_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "NextEra Energy expanded solar capacity in Florida by 2.5 gigawatts.".into(),
            entities: vec![
                entity("NextEra Energy", EntityType::Organization, 0),
                entity("Florida", EntityType::Location, 42),
                entity("2.5 gigawatts", EntityType::Quantity, 53),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text:
                "Shell and BP announced $30 billion in offshore wind investments in the North Sea."
                    .into(),
            entities: vec![
                entity("Shell", EntityType::Organization, 0),
                entity("BP", EntityType::Organization, 10),
                entity("$30 billion", EntityType::Money, 23),
                entity("North Sea", EntityType::Location, 71),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "China's CATL dominates the global battery market with 37% share.".into(),
            entities: vec![
                entity("China", EntityType::Location, 0),
                entity("CATL", EntityType::Organization, 8),
                entity("37%", EntityType::Percent, 54),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Exxon and Chevron reported record profits of $100 billion combined in 2022."
                .into(),
            entities: vec![
                entity("Exxon", EntityType::Organization, 0),
                entity("Chevron", EntityType::Organization, 10),
                entity("$100 billion", EntityType::Money, 45),
                entity("2022", EntityType::Date, 70),
            ],
            domain: Domain::Financial,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Aerospace/defense dataset.
pub fn aerospace_dataset() -> Vec<AnnotatedExample> {
    vec![
        AnnotatedExample {
            text: "Boeing CEO David Calhoun testified before Congress about 737 MAX safety.".into(),
            entities: vec![
                entity("Boeing", EntityType::Organization, 0),
                entity("David Calhoun", EntityType::Person, 11),
                entity("Congress", EntityType::Organization, 42),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Lockheed Martin won a $17 billion contract from the Pentagon for F-35s.".into(),
            entities: vec![
                entity("Lockheed Martin", EntityType::Organization, 0),
                entity("$17 billion", EntityType::Money, 22),
                entity("Pentagon", EntityType::Organization, 52),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Blue Origin's New Glenn launched from Cape Canaveral on January 15, 2024."
                .into(),
            entities: vec![
                entity("Blue Origin", EntityType::Organization, 0),
                entity("Cape Canaveral", EntityType::Location, 38),
                entity("January 15, 2024", EntityType::Date, 56),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
        AnnotatedExample {
            text: "Airbus delivered 735 aircraft in 2023, trailing Boeing's 528 deliveries.".into(),
            entities: vec![
                entity("Airbus", EntityType::Organization, 0),
                entity("735", EntityType::Quantity, 17),
                entity("2023", EntityType::Date, 33),
                entity("Boeing", EntityType::Organization, 48),
                entity("528", EntityType::Quantity, 57),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Medium,
        },
    ]
}

/// Hard examples for underrepresented domains.
pub fn hard_domain_examples() -> Vec<AnnotatedExample> {
    vec![
        // Technical (Hard)
        AnnotatedExample {
            text:
                "The CVE-2024-1234 vulnerability in OpenSSL 3.0.x affects nginx, Apache, and HAProxy."
                    .into(),
            entities: vec![
                entity("OpenSSL", EntityType::Organization, 35),
                entity("nginx", EntityType::Organization, 57),
                entity("Apache", EntityType::Organization, 64),
                entity("HAProxy", EntityType::Organization, 76),
            ],
            domain: Domain::Technical,
            difficulty: Difficulty::Hard,
        },
        // Travel (Hard)
        AnnotatedExample {
            text:
                "Connecting via FRA (Frankfurt) to SIN (Singapore) then SYD (Sydney) on LH/SQ codeshare."
                    .into(),
            entities: vec![
                entity("FRA", EntityType::Location, 15),
                entity("Frankfurt", EntityType::Location, 20),
                entity("SIN", EntityType::Location, 34),
                entity("Singapore", EntityType::Location, 39),
                entity("SYD", EntityType::Location, 55),
                entity("Sydney", EntityType::Location, 60),
            ],
            domain: Domain::Travel,
            difficulty: Difficulty::Hard,
        },
        // Entertainment (Hard)
        AnnotatedExample {
            text:
                "Director Christopher Nolan's Oppenheimer starring Cillian Murphy won at both the Oscars and BAFTAs."
                    .into(),
            entities: vec![
                entity("Christopher Nolan", EntityType::Person, 9),
                entity("Cillian Murphy", EntityType::Person, 50),
                entity("Oscars", EntityType::Organization, 81),
                entity("BAFTAs", EntityType::Organization, 92),
            ],
            domain: Domain::Entertainment,
            difficulty: Difficulty::Hard,
        },
    ]
}

/// Email/contact dataset (tests EMAIL/PHONE/URL entities).
pub fn email_dataset() -> Vec<AnnotatedExample> {
    let text1 = "Contact support at help@example.com or call +1-212-555-0100.";
    let email_start = text1
        .find("help@example.com")
        .map(|b| text1[..b].chars().count())
        .expect("email should be found in text1");
    let phone_start = text1
        .find("+1-212-555-0100")
        .map(|b| text1[..b].chars().count())
        .expect("phone should be found in text1");

    let text2 = "Submit a ticket via https://support.example.org or email ops@example.org.";
    let url_start = text2
        .find("https://support.example.org")
        .map(|b| text2[..b].chars().count())
        .expect("URL should be found in text2");
    let email2_start = text2
        .find("ops@example.org")
        .map(|b| text2[..b].chars().count())
        .expect("email should be found in text2");

    vec![
        AnnotatedExample {
            text: text1.into(),
            entities: vec![
                entity_email("help@example.com", email_start),
                entity_phone("+1-212-555-0100", phone_start),
            ],
            domain: Domain::Email,
            difficulty: Difficulty::Easy,
        },
        AnnotatedExample {
            text: text2.into(),
            entities: vec![
                entity_url("https://support.example.org", url_start),
                entity_email("ops@example.org", email2_start),
            ],
            domain: Domain::Email,
            difficulty: Difficulty::Easy,
        },
    ]
}

#[cfg(test)]
mod tests {
    use super::*;

    #[test]
    fn test_sports_dataset_not_empty() {
        assert!(!sports_dataset().is_empty());
    }

    #[test]
    fn test_politics_dataset_not_empty() {
        assert!(!politics_dataset().is_empty());
    }

    #[test]
    fn test_multilingual_dataset_not_empty() {
        assert!(!multilingual_dataset().is_empty());
    }

    #[test]
    fn test_globally_diverse_not_empty() {
        assert!(!globally_diverse_dataset().is_empty());
    }

    #[test]
    fn test_technology_dataset_not_empty() {
        let ds = technology_dataset();
        assert!(!ds.is_empty());
        assert!(ds.len() >= 5);
    }

    #[test]
    fn test_healthcare_dataset_not_empty() {
        let ds = healthcare_dataset();
        assert!(!ds.is_empty());
        assert!(ds.len() >= 4);
    }

    #[test]
    fn test_manufacturing_dataset_not_empty() {
        let ds = manufacturing_dataset();
        assert!(!ds.is_empty());
        assert!(ds.len() >= 4);
    }

    #[test]
    fn test_automotive_dataset_not_empty() {
        let ds = automotive_dataset();
        assert!(!ds.is_empty());
        assert!(ds.len() >= 4);
    }

    #[test]
    fn test_energy_dataset_not_empty() {
        let ds = energy_dataset();
        assert!(!ds.is_empty());
        assert!(ds.len() >= 3);
    }

    #[test]
    fn test_aerospace_dataset_not_empty() {
        let ds = aerospace_dataset();
        assert!(!ds.is_empty());
        assert!(ds.len() >= 3);
    }

    #[test]
    fn test_email_dataset_not_empty() {
        let ds = email_dataset();
        assert!(!ds.is_empty());
        for ex in ds {
            assert_eq!(ex.domain, Domain::Email);
        }
    }

    #[test]
    fn test_multilingual_has_required_scripts() {
        let ds = multilingual_dataset();
        let texts: Vec<_> = ds.iter().map(|e| e.text.as_str()).collect();
        let has_devanagari = texts
            .iter()
            .any(|t| t.chars().any(|c| ('\u{0900}'..='\u{097F}').contains(&c)));
        let has_cyrillic = texts
            .iter()
            .any(|t| t.chars().any(|c| ('\u{0400}'..='\u{04FF}').contains(&c)));
        let has_cjk = texts
            .iter()
            .any(|t| t.chars().any(|c| ('\u{4E00}'..='\u{9FFF}').contains(&c)));
        let has_emoji = texts.iter().any(|t| t.contains('🥇'));

        assert!(has_devanagari, "Missing Devanagari coverage");
        assert!(has_cyrillic, "Missing Cyrillic coverage");
        assert!(has_cjk, "Missing CJK coverage");
        assert!(has_emoji, "Missing emoji/mixed-script coverage");
    }
}