additory 0.1.0a1__py3-none-any.whl

additory/augment/builtin_lists.py

@@ -0,0 +1,430 @@
+"""
+Built-in Lists Registry for Data Augmentation
+
+Provides globally available lists for choice-based augmentation strategies.
+These lists are loaded in memory for instant access (O(1) lookup, zero I/O).
+
+Users can override any list using add.register_list("list_name", [...])
+
+List Resolution Order:
+    1. User-registered lists (highest priority)
+    2. Built-in lists (this module)
+    3. Error if not found
+"""
+
+# =============================================================================
+# NAMES - Globally diverse names representing major world regions
+# =============================================================================
+
+FIRST_NAMES = [
+    # India & South Asia
+    "Arjun", "Vikram", "Samuel", "Harpreet", "Kabir", "Aarav", "Aadhya",
+    "Fatima", "Anita", "Zoya", "Meera", "Nisha", "Ayesha", "Ritika", "Ira",
+    "Dev", "Karan", "Rhea", "Tanvi", "Varun", "Neha", "Ishaan", "Tara",
+    "Aditya", "Kavya", "Suresh", "Priya", "Manish", "Shreya", "Naveen",
+    "Diya", "Rohit", "Ananya", "Kunal", "Pooja", "Raj", "Sneha", "Suraj",
+    "Lavanya", "Gaurav", "Shalini", "Mohan", "Vandana", "Karthik", "Bhavana",
+    "Raghav", "Simran", "Tejas", "Anjali", "Nitin", "Radhika", "Abhishek",
+    "Shweta", "Parth", "Ishita", "Siddharth", "Malini", "Vivek", "Yamini",
+    "Pranav", "Aarti", "Rajesh", "Komal", "Sanjay", "Madhuri", "Deepak",
+    "Sonali", "Hemant", "Kirti", "Ashok", "Rupa", "Jatin", "Leela", "Vinay",
+    "Charu", "Harish", "Nandini", "Shankar", "Pallavi", "Arvind", "Kusum",
+    "Manoj", "Gayatri", "Sagar", "Ritu", "Bharat", "Chitra", "Keshav", "Lata",
+    "Om", "Padma", "Yash", "Seema", "Kartik", "Madhav", "Veena", "Rohan",
+    "Amar", "Abdul", "Antony",
+    # Americas & Europe
+    "James", "Mary", "Robert", "Oliver", "Olivia", "George",
+    # East Asia
+    "Wei", "Mei", "Jun", "Haruto", "Yui", "Ren", "Min-jun", "Seo-yeon", "Ji-ho",
+    # Europe
+    "Gabriel", "Emma", "Louis", "Noah", "Emilia", "Leon", "Hugo", "Lucia",
+    "Mateo", "Leonardo", "Sofia", "Francesco", "Miguel", "Helena", "Arthur",
+    "Santiago", "Maria", "Jose", "Liam", "Charlotte", "William", "Alexander",
+    "Dmitry", "Mia", "Ethan",
+    # Africa
+    "Chinedu", "Aisha", "Oluwaseun",
+    # Middle East
+    "Mohamed", "Fatima", "Ahmed", "Yusuf", "Elif", "Mustafa", "Abdullah",
+    "Omar", "Mariam",
+    # Southeast Asia
+    "Putra", "Siti", "Agus", "Juan", "Angel", "Jose", "Rahim", "Ayesha",
+    "Karim", "Anh", "Linh", "Minh", "Niran", "Mali", "Arun",
+    # Additional
+    "Chloe", "Jayden", "Saif", "Khalid", "Daan", "Sem", "Lucas", "Alice",
+    "William", "Jakob", "Nora", "Emil", "Oscar", "Alma", "Jack", "Aoife",
+    "Conor", "Leo", "Isla"
+]
+
+LAST_NAMES = [
+    # India & South Asia
+    "Rao", "Sharma", "Gupta", "Kumar", "Reddy", "Iyer", "Nair", "Pillai",
+    "Shetty", "Gowda", "Verma", "Yadav", "Chauhan", "Thakur", "Jha", "Tiwari",
+    "Mishra", "Pandey", "Shukla", "Saxena", "Bhat", "Naidu", "Kulkarni",
+    "Deshmukh", "Patil", "Joshi", "Shinde", "Gaikwad", "Salvi", "Khatri",
+    "Kapoor", "Khanna", "Mehta", "Shah", "Soni", "Modi", "Jain", "Agarwal",
+    "Bansal", "Mahajan", "Chawla", "Sethi", "Malhotra", "Anand", "Chopra",
+    "Gill", "Sandhu", "Sidhu", "Ahluwalia", "Bajwa", "Rawat", "Negi", "Bisht",
+    "Rana", "Sinha", "Bose", "Roy", "Dutta", "Mukherjee", "Sengupta",
+    "Chakraborty", "Basu", "Deb", "Paul", "Sen", "Konar", "Mondal", "Pal",
+    "Saha", "Halder", "Barik", "Behera", "Pradhan", "Swain", "Mohanty", "Raut",
+    "Jadhav", "Pawar", "Kamble", "More", "Wagh", "Lokhande", "Bendre", "Naik",
+    "Hegde", "Babu", "Varma", "Menon", "Cherian", "Kurian", "Thomas", "Mathew",
+    "George", "Abraham", "Philip", "Sunny", "Chacko", "Joseph", "Antony",
+    # Americas & Europe
+    "Smith", "Johnson", "Williams", "Jones", "Taylor", "Wang", "Chen", "Zhao",
+    "Sato", "Suzuki", "Takahashi", "Kim", "Lee", "Park", "Martin", "Bernard",
+    "Dubois", "Muller", "Schmidt", "Schneider", "Garcia", "Fernandez", "Lopez",
+    "Rossi", "Russo", "Ferrari", "Silva", "Santos", "Oliveira", "Hernandez",
+    "Martinez", "Brown", "Tremblay", "Ivanov", "Smirnov", "Kuznetsov",
+    # Africa
+    "Nkosi", "Botha", "Dlamini", "Okafor", "Ibrahim", "Adeyemi",
+    # Middle East
+    "Hassan", "Ali", "Youssef", "Yilmaz", "Kaya", "Demir", "Al-Saud",
+    "Al-Harbi", "Al-Qahtani",
+    # Southeast Asia
+    "Pratama", "Wijaya", "Saputra", "Dela Cruz", "Reyes", "Khan", "Malik",
+    "Chowdhury", "Islam", "Rahman", "Ahmed", "Nguyen", "Tran", "Le", "Saetang",
+    "Boonmee", "Chaiyaporn", "Tan", "Lim", "Al-Nahyan", "Al-Mansoori",
+    "Al-Falasi", "de Jong", "Jansen", "de Vries", "Andersson", "Johansson",
+    "Karlsson", "Hansen", "Johansen", "Olsen", "Nielsen", "Jensen", "Murphy",
+    "Kelly", "O'Brien", "Patel", "Das", "Chatterjee", "Banerjee", "Ghosh"
+]
+
+# =============================================================================
+# FINANCIAL - Major global banks and financial institutions
+# =============================================================================
+
+BANKS = [
+    # India
+    "State Bank of India (SBI)", "HDFC Bank", "ICICI Bank", "Axis Bank",
+    "Kotak Mahindra Bank", "Bank of Baroda", "Punjab National Bank",
+    "Canara Bank", "Union Bank of India", "IDFC First Bank",
+    # China
+    "Industrial and Commercial Bank of China", "China Construction Bank",
+    "Agricultural Bank of China", "Bank of China", "Bank of Communications",
+    "China Merchants Bank", "Postal Savings Bank of China", "Industrial Bank Co",
+    "Shanghai Pudong Development Bank",
+    # Japan
+    "Mitsubishi UFJ Financial Group", "Sumitomo Mitsui Financial Group",
+    "Mizuho Financial Group", "Resona Holdings", "Japan Post Bank",
+    # United States
+    "JPMorgan Chase", "Bank of America", "Citibank", "Wells Fargo",
+    "Goldman Sachs", "Morgan Stanley", "US Bank", "PNC Financial",
+    "Truist Financial", "Capital One",
+    # United Kingdom
+    "HSBC", "Barclays", "Lloyds Banking Group", "NatWest Group",
+    "Standard Chartered",
+    # Canada
+    "Royal Bank of Canada", "Toronto Dominion Bank", "Scotiabank",
+    "Bank of Montreal", "CIBC",
+    # France
+    "BNP Paribas", "Credit Agricole", "Societe Generale", "BPCE Group",
+    "La Banque Postale",
+    # Germany
+    "Deutsche Bank", "Commerzbank", "KfW Bankengruppe", "DZ Bank",
+    "Landesbank Baden Wuerttemberg",
+    # Switzerland
+    "UBS", "Julius Baer", "Raiffeisen Switzerland",
+    # Italy
+    "UniCredit", "Intesa Sanpaolo", "Banco BPM", "BPER Banca",
+    # Spain
+    "Santander", "BBVA", "CaixaBank", "Bankinter",
+    # Netherlands
+    "ING Group", "Rabobank", "ABN AMRO",
+    # Sweden
+    "Nordea", "SEB", "Swedbank", "Handelsbanken",
+    # Norway & Denmark
+    "DNB", "Danske Bank",
+    # Singapore
+    "DBS Bank", "OCBC Bank", "UOB",
+    # Malaysia
+    "Maybank", "CIMB Group", "RHB Bank",
+    # Indonesia
+    "Bank Mandiri", "Bank Rakyat Indonesia", "Bank Central Asia",
+    # South Korea
+    "KB Financial Group", "Shinhan Financial Group", "Hana Financial Group",
+    "Woori Bank",
+    # Australia
+    "Commonwealth Bank of Australia", "Westpac", "ANZ", "NAB",
+    # Brazil
+    "Itau Unibanco", "Banco Bradesco", "Banco do Brasil",
+    "Caixa Economica Federal",
+    # Mexico
+    "BBVA Mexico", "Banorte"
+]
+
+CREDIT_CARD_NETWORKS = [
+    # Global majors
+    "Visa", "Mastercard", "American Express", "Discover",
+    "Diners Club International", "UnionPay",
+    # Japan
+    "JCB",
+    # India
+    "RuPay", "NCMC National Common Mobility Card",
+    # United States (additional networks)
+    "Pulse", "Star", "NYCE", "Accel", "Shazam", "Interlink", "Maestro",
+    "Cirrus", "Plus",
+    # Europe
+    "Girocard", "Cartes Bancaires", "Bancontact", "V Pay",
+    "EAPS European Alliance Payment Scheme", "Link UK ATM Network",
+    "PostePay Italy", "Multibanco Portugal",
+    # United Kingdom
+    "Switch", "Solo",
+    # Canada
+    "Interac", "Debit Mastercard Canada", "Visa Debit Canada",
+    # Latin America
+    "Elo Brazil", "Hipercard Brazil", "Cabal Argentina", "Naranja X Argentina",
+    "Red Link Argentina", "Banelco Argentina",
+    # Middle East
+    "Meeza Egypt", "Benefit Bahrain", "KNET Kuwait", "Mada Saudi Arabia",
+    "QatarPay", "OmanNet", "UAE Switch",
+    # Africa
+    "Verve Nigeria", "Interswitch", "GhLink Ghana", "Zimswitch Zimbabwe",
+    "KenSwitch Kenya", "Namclear Namibia",
+    # Southeast Asia
+    "GPN Indonesia", "PromptCard Thailand", "NETS Singapore",
+    "Touch n Go Malaysia", "MyDebit Malaysia", "BancNet Philippines",
+    "Vietcombank Card Vietnam",
+    # East Asia (non China/Japan)
+    "T Money Korea", "Cashbee Korea",
+    # China (additional)
+    "QuickPass", "China T Union Transit Card",
+    # Australia / New Zealand
+    "eftpos Australia", "BPAY", "Paymark New Zealand",
+    # Digital only / fintech card networks
+    "Revolut Card Network", "Nubank Card Network", "Monzo Card Network",
+    "Starling Card Network", "Chime Card Network", "Cash App Card Network",
+    # Private label / store card networks
+    "Target RedCard Network", "Walmart Card Network", "Costco Card Network",
+    "Amazon Store Card Network", "Best Buy Card Network"
+]
+
+# =============================================================================
+# RETAIL - Major global retailers and e-commerce platforms
+# =============================================================================
+
+RETAILERS = [
+    # India
+    "Reliance Retail", "DMart", "Tata Croma", "Big Bazaar", "Spencers Retail",
+    "V Mart", "More Retail", "Trent Westside", "Future Retail", "Shoppers Stop",
+    "Lifestyle India", "Pantaloons", "Fabindia", "Nykaa", "Myntra", "Flipkart",
+    "Tanishq", "Titan Company", "Apollo Pharmacy", "MedPlus",
+    # United States
+    "Walmart", "Amazon", "Costco", "Target", "Kroger", "Home Depot", "Lowes",
+    "Best Buy", "Walgreens", "CVS Pharmacy", "Macys", "Kohl's", "TJ Maxx",
+    "Dollar General", "Dollar Tree", "Nordstrom", "Sears", "Staples",
+    "Office Depot", "Whole Foods Market",
+    # United Kingdom
+    "Tesco", "Sainsbury's", "Asda", "Marks and Spencer", "John Lewis", "Argos",
+    "Boots", "Morrisons", "Next", "Primark",
+    # Europe (France, Germany, Spain, etc.)
+    "Carrefour", "Auchan", "Aldi", "Lidl", "Metro AG", "Edeka", "Rewe",
+    "El Corte Ingles", "MediaMarkt", "Decathlon",
+    # Japan
+    "Aeon", "Seven and I Holdings", "FamilyMart", "Lawson", "Don Quijote",
+    # China
+    "JD Retail", "Alibaba Tmall", "Suning", "Gome Retail", "Yonghui Superstores",
+    # South Korea
+    "Lotte Mart", "Shinsegae", "E Mart", "Homeplus",
+    # Southeast Asia
+    "Robinsons Philippines", "SM Retail", "Central Group Thailand",
+    "Big C Thailand", "VinMart Vietnam",
+    # Middle East
+    "Lulu Hypermarket", "Carrefour UAE", "Danube Saudi Arabia",
+    "Panda Retail Saudi Arabia", "Choithrams",
+    # Australia
+    "Woolworths Australia", "Coles", "JB Hi Fi", "Harvey Norman", "Myer",
+    # Africa
+    "Shoprite", "Pick n Pay", "Game Stores", "Massmart", "Nakumatt",
+    # Latin America
+    "Mercado Libre", "Falabella", "Cencosud", "Grupo Pao de Acucar",
+    "Liverpool Mexico"
+]
+
+# =============================================================================
+# TECHNOLOGY - Major global technology companies
+# =============================================================================
+
+TECH_COMPANIES = [
+    # India
+    "Tata Consultancy Services", "Infosys", "Wipro", "HCL Technologies",
+    "Tech Mahindra", "Larsen and Toubro Infotech", "Mindtree",
+    "Persistent Systems", "Mphasis", "Zensar Technologies", "Birlasoft",
+    "Hexaware Technologies", "NIIT Technologies", "Coforge",
+    "Reliance Jio Platforms", "Bharti Airtel Digital", "Zoho", "Freshworks",
+    "InMobi", "Quick Heal",
+    # Taiwan
+    "TSMC", "MediaTek", "Foxconn", "UMC", "ASE Technology",
+    "Realtek Semiconductor", "Novatek Microelectronics", "AU Optronics",
+    "Delta Electronics", "Pegatron", "Quanta Computer", "Compal Electronics",
+    "Wistron", "Macronix", "Winbond Electronics",
+    # United States
+    "Apple", "Microsoft", "Google", "Amazon Web Services", "Meta Platforms",
+    "IBM", "Intel", "Nvidia", "AMD", "Oracle", "Salesforce", "Adobe", "Cisco",
+    "Qualcomm", "HP", "Dell Technologies", "Uber", "Airbnb", "Palantir",
+    "Snowflake",
+    # United Kingdom
+    "ARM", "Sage Group", "Micro Focus", "BT Group", "Vodafone",
+    # Europe
+    "SAP", "Siemens", "Bosch", "Capgemini", "Atos", "Nokia", "Ericsson",
+    "Dassault Systemes", "Infineon Technologies", "STMicroelectronics",
+    # Japan
+    "Sony", "Panasonic", "Toshiba", "Fujitsu", "NEC", "Hitachi", "SoftBank",
+    "Rakuten", "Nintendo", "Canon",
+    # China
+    "Huawei", "Tencent", "Alibaba", "Baidu", "Xiaomi", "Lenovo", "ZTE",
+    "ByteDance", "JD Technology", "Hikvision",
+    # South Korea
+    "Samsung Electronics", "LG Electronics", "SK Hynix", "Naver", "Kakao",
+    # Southeast Asia
+    "Grab", "Sea Group", "GoTo Group", "FPT Software", "VNG Corporation",
+    # Middle East
+    "STC Solutions", "eFinance Egypt", "Etisalat Digital", "Aramco Digital",
+    "Beeah Digital",
+    # Australia
+    "Atlassian", "Telstra", "WiseTech Global", "Afterpay", "Canva",
+    # Africa
+    "MTN Group", "Safaricom", "Liquid Intelligent Technologies", "Naspers",
+    "Cell C",
+    # Latin America
+    "Mercado Libre", "Nubank", "TOTVS", "PagSeguro", "Globant"
+]
+
+EMAIL_PROVIDERS = [
+    # Global giants
+    "gmail.com", "yahoo.com", "outlook.com", "hotmail.com", "live.com",
+    "icloud.com", "aol.com", "protonmail.com", "zoho.com", "mail.com",
+    # India
+    "rediffmail.com", "in.com", "sify.com", "airtelmail.in", "bsnl.in",
+    # USA
+    "msn.com", "comcast.net", "verizon.net", "att.net", "me.com", "mac.com",
+    # Europe
+    "gmx.com", "gmx.de", "web.de", "orange.fr", "free.fr", "laposte.net",
+    "btinternet.com", "virginmedia.com", "sky.com",
+    # UK
+    "hotmail.co.uk", "yahoo.co.uk", "live.co.uk",
+    # Germany
+    "t-online.de", "posteo.de", "mailbox.org",
+    # France
+    "sfr.fr", "wanadoo.fr",
+    # Russia
+    "mail.ru", "yandex.com", "yandex.ru", "rambler.ru",
+    # China
+    "qq.com", "163.com", "126.com", "sina.com", "sohu.com",
+    # Japan
+    "yahoo.co.jp", "docomo.ne.jp", "ezweb.ne.jp", "softbank.ne.jp",
+    # Korea
+    "naver.com", "daum.net", "hanmail.net",
+    # Middle East
+    "emirates.net.ae", "etisalat.ae", "saudi.net.sa",
+    # Africa
+    "webmail.co.za", "mtn.co.za",
+    # Latin America
+    "uol.com.br", "bol.com.br", "terra.com.br", "hotmail.com.mx", "live.com.mx",
+    # Southeast Asia
+    "singnet.com.sg", "starhub.net.sg", "tm.net.my", "pldt.com.ph",
+    "yahoo.com.ph",
+    # Australia / NZ
+    "bigpond.com", "optusnet.com.au", "xtra.co.nz",
+    # Privacy-focused / SaaS
+    "fastmail.com", "hey.com",
+    # Generic / testing domains
+    "example.com", "test.com", "demo.com", "sample.com", "mailtest.com",
+    "placeholder.com", "domain.com"
+]
+
+# =============================================================================
+# BUSINESS - Common business categories and classifications
+# =============================================================================
+
+DEPARTMENTS = [
+    # Core Business Functions
+    "Human Resources", "Finance", "Accounting", "Sales", "Marketing",
+    "Operations", "Customer Service", "Customer Support",
+    # Technology & Engineering
+    "Information Technology", "Engineering", "Research and Development",
+    "Product Development", "Data Science", "Quality Assurance",
+    # Supply Chain & Logistics
+    "Supply Chain", "Procurement", "Logistics", "Warehouse",
+    # Legal & Compliance
+    "Legal", "Compliance", "Risk Management", "Internal Audit",
+    # Strategic & Executive
+    "Strategy", "Business Development", "Corporate Development", "Executive",
+    # Specialized
+    "Manufacturing", "Production", "Facilities", "Security", "Administration",
+    "Training", "Communications", "Public Relations"
+]
+
+STATUSES = [
+    "Active", "Inactive", "Pending", "Completed", "Cancelled",
+    "In Progress", "On Hold", "Approved", "Rejected", "Draft"
+]
+
+SEVERITIES = [
+    "Low", "Medium", "High", "Critical", "Urgent"
+]
+
+PRIORITIES = [
+    "Low", "Medium", "High", "Critical"
+]
+
+# =============================================================================
+# MEDICAL - Common medical and healthcare terms
+# =============================================================================
+
+ADVERSE_EVENTS = [
+    "Headache", "Nausea", "Vomiting", "Dizziness", "Fatigue",
+    "Fever", "Chills", "Rash", "Itching", "Swelling",
+    "Pain", "Muscle Ache", "Joint Pain", "Weakness", "Insomnia",
+    "Anxiety", "Depression", "Confusion", "Tremor", "Seizure",
+    "Shortness of Breath", "Cough", "Chest Pain", "Palpitations",
+    "Edema", "Hypertension", "Hypotension", "Tachycardia"
+]
+
+# =============================================================================
+# MASTER REGISTRY - Maps list names to actual lists
+# =============================================================================
+
+BUILTIN_LISTS = {
+    # Names
+    "first_names": FIRST_NAMES,
+    "last_names": LAST_NAMES,
+    # Financial
+    "banks": BANKS,
+    "credit_card_networks": CREDIT_CARD_NETWORKS,
+    # Retail
+    "retailers": RETAILERS,
+    # Technology
+    "tech_companies": TECH_COMPANIES,
+    "email_providers": EMAIL_PROVIDERS,
+    # Business
+    "departments": DEPARTMENTS,
+    "statuses": STATUSES,
+    "severities": SEVERITIES,
+    "priorities": PRIORITIES,
+    # Medical
+    "adverse_events": ADVERSE_EVENTS,
+}
+
+
+def get_builtin_list(name: str):
+    """
+    Get a built-in list by name.
+    
+    Args:
+        name: List name (e.g., "first_names", "banks")
+    
+    Returns:
+        List of values or None if not found
+    """
+    return BUILTIN_LISTS.get(name)
+
+
+def list_builtin_names():
+    """
+    Get all available built-in list names.
+    
+    Returns:
+        List of list names
+    """
+    return list(BUILTIN_LISTS.keys())

additory/augment/distributions.py

@@ -0,0 +1,22 @@
+"""
+Distribution Strategies for Data Augmentation
+
+DEPRECATED: This module has been moved to additory.common.distributions
+Please update your imports to use additory.common.distributions instead.
+
+This file is kept for backward compatibility and will be removed in a future version.
+"""
+
+import warnings
+
+# Issue deprecation warning
+warnings.warn(
+    "additory.augment.distributions is deprecated. "
+    "Please use additory.common.distributions instead. "
+    "This module will be removed in a future version.",
+    DeprecationWarning,
+    stacklevel=2
+)
+
+# Import everything from common.distributions for backward compatibility
+from additory.common.distributions import *  # noqa: F401, F403