marketmyna 0.2.0__py3-none-any.whl

marketmyna/core/tenant.py

@@ -0,0 +1,130 @@
+"""Per-user data scoping for accounts mode.
+
+When accounts mode is on (`enable()` is called at app start), every ORM
+SELECT / UPDATE / DELETE on an owned model gets `user_id = <current user>`
+added automatically, and every new owned row gets the current user's id on
+flush. The current user lives in a context variable that AuthMiddleware sets
+per request, so routers and services don't have to pass user ids around.
+
+Scopes:
+- a user id (int): only that user's rows.
+- SYSTEM: no filter. Only for code that must see every user's rows
+  (admin counts, account deletion, background jobs that loop over users).
+- unset: fail closed. Queries match nothing and inserts raise, so code that
+  forgot to pick a scope can't read or write another user's data.
+
+With accounts off (CLI runs, one local user) nothing is filtered.
+"""
+
+from __future__ import annotations
+
+import contextlib
+import logging
+from contextvars import ContextVar
+from typing import Iterator, Optional, Union
+
+from sqlalchemy import event
+from sqlalchemy.orm import ORMExecuteState, Session, with_loader_criteria
+
+logger = logging.getLogger(__name__)
+
+
+class _System:
+    def __repr__(self) -> str:
+        return "SYSTEM"
+
+
+SYSTEM = _System()
+Scope = Union[int, _System, None]
+
+_current: ContextVar[Scope] = ContextVar("marketmyna_user_scope", default=None)
+_enabled = False
+_owned: tuple[type, ...] = ()
+# Owned for inserts (user_id is filled in) but readable across users: the
+# admin reads all feedback and usage, the platform scheduler / alert config
+# rows belong to the platform, not one user.
+_insert_only: tuple[type, ...] = ()
+
+
+def configure(owned: tuple[type, ...], insert_only: tuple[type, ...] = ()) -> None:
+    global _owned, _insert_only
+    _owned, _insert_only = owned, insert_only
+
+
+def owned_models() -> tuple[type, ...]:
+    """Models scoped to one user (each has a user_id column)."""
+    return _owned
+
+
+def enable(on: bool = True) -> None:
+    global _enabled
+    _enabled = on
+
+
+def enabled() -> bool:
+    return _enabled
+
+
+def current() -> Scope:
+    return _current.get()
+
+
+def current_user_id() -> Optional[int]:
+    scope = _current.get()
+    return scope if isinstance(scope, int) else None
+
+
+def set_scope(scope: Scope):
+    """Sets the scope for the rest of this context; returns a reset token."""
+    return _current.set(scope)
+
+
+def reset_scope(token) -> None:
+    _current.reset(token)
+
+
+@contextlib.contextmanager
+def user_scope(scope: Scope) -> Iterator[None]:
+    token = _current.set(scope)
+    try:
+        yield
+    finally:
+        _current.reset(token)
+
+
+def system_scope():
+    return user_scope(SYSTEM)
+
+
+@event.listens_for(Session, "do_orm_execute")
+def _scope_queries(state: ORMExecuteState) -> None:
+    if not _enabled or not _owned:
+        return
+    if not (state.is_select or state.is_update or state.is_delete):
+        return
+    if state.is_column_load or state.is_relationship_load:
+        return  # the parent row was already scoped
+    scope = _current.get()
+    if scope is SYSTEM:
+        return
+    # No user id is ever -1, so an unset scope matches nothing.
+    uid = scope if isinstance(scope, int) else -1
+    state.statement = state.statement.options(
+        *(with_loader_criteria(model, model.user_id == uid, include_aliases=True) for model in _owned)
+    )
+
+
+@event.listens_for(Session, "before_flush")
+def _stamp_owner(session: Session, flush_context, instances) -> None:
+    if not _enabled:
+        return
+    scope = _current.get()
+    for obj in session.new:
+        if not isinstance(obj, _owned + _insert_only):
+            continue
+        if getattr(obj, "user_id", None) is not None:
+            continue
+        if isinstance(scope, int):
+            obj.user_id = scope
+        elif scope is None and isinstance(obj, _owned):
+            raise RuntimeError(f"Refusing to create {type(obj).__name__} with no user scope")

marketmyna/data/__init__.py

@@ -0,0 +1,8 @@
+"""Data fetching module for market data."""
+
+from .base import DataFetcher
+from .crypto import CryptoDataFetcher
+from .india import IndiaDataFetcher
+from .stocks import StockDataFetcher
+
+__all__ = ["DataFetcher", "StockDataFetcher", "CryptoDataFetcher", "IndiaDataFetcher"]

marketmyna/data/base.py

@@ -0,0 +1,36 @@
+"""Base data fetcher interface."""
+
+from abc import ABC, abstractmethod
+from datetime import datetime
+from typing import Optional
+
+import pandas as pd
+
+from ..core.constants import TimeFrame
+
+
+class DataFetcher(ABC):
+    """Abstract base class for data fetchers."""
+
+    @abstractmethod
+    def get_quote(self, symbol: str) -> dict:
+        """Get current quote for a symbol."""
+
+    @abstractmethod
+    def get_history(
+        self,
+        symbol: str,
+        period: str = "1y",
+        interval: TimeFrame = TimeFrame.DAY_1,
+        start: Optional[datetime] = None,
+        end: Optional[datetime] = None,
+    ) -> pd.DataFrame:
+        """Get historical price data."""
+
+    @abstractmethod
+    def search(self, query: str) -> list[dict]:
+        """Search for symbols matching query."""
+
+    @abstractmethod
+    def get_info(self, symbol: str) -> dict:
+        """Get detailed information about a symbol."""

marketmyna/data/categories.py

@@ -0,0 +1,272 @@
+"""One broad category list shared by every market, so a single filter
+("Technology", "Agriculture", "Natural Resources", ...) works whether the
+row came from NSE, NASDAQ or anywhere else.
+
+Resolution order for a row:
+1. industry keywords (the most specific signal, e.g. US "Farming/Seeds/Milling"),
+2. the source's own sector label (NSE industry / US sector),
+3. keywords in the company name - the only signal for most small NSE/SME
+   stocks, which no free bulk source classifies. Those rows are flagged
+   ``category_inferred`` so the UI can say so.
+"""
+
+from __future__ import annotations
+
+import re
+from typing import Optional
+
+TECHNOLOGY = "Technology"
+FINANCIALS = "Financials"
+HEALTHCARE = "Healthcare"
+CONSUMER = "Consumer"
+AUTOMOBILE = "Automobile"
+INDUSTRIALS = "Industrials"
+ENERGY = "Energy"
+NATURAL_RESOURCES = "Natural Resources"
+AGRICULTURE = "Agriculture"
+MATERIALS = "Chemicals & Materials"
+REAL_ESTATE = "Real Estate & Construction"
+TRUSTS = "REITs & InvITs"
+TELECOM_MEDIA = "Telecom & Media"
+UTILITIES = "Utilities"
+TEXTILES = "Textiles"
+SERVICES = "Services"
+OTHER = "Other"
+
+# Extra groups used only for ETFs (and index funds).
+BROAD_MARKET = "Broad Market"
+BONDS = "Bonds"
+COMMODITIES = "Commodities"
+LEVERAGED = "Leveraged & Inverse"
+INTERNATIONAL = "International"
+
+CATEGORIES = [
+    TECHNOLOGY, FINANCIALS, HEALTHCARE, CONSUMER, AUTOMOBILE, INDUSTRIALS, ENERGY,
+    NATURAL_RESOURCES, AGRICULTURE, MATERIALS, REAL_ESTATE, TRUSTS, TELECOM_MEDIA,
+    UTILITIES, TEXTILES, SERVICES, BROAD_MARKET, BONDS, COMMODITIES, INTERNATIONAL, LEVERAGED, OTHER,
+]
+
+_ETF_RULES: list[tuple[str, re.Pattern]] = [
+    (LEVERAGED, re.compile(r"\b[23]x\b|ultra ?pro|ultrashort|inverse|\bbull\b|\bbear\b|daily .*(long|short)|leveraged", re.I)),
+    (COMMODITIES, re.compile(r"gold|silver|commodit|platinum|copper|oil fund|natural gas|crude", re.I)),
+    (BONDS, re.compile(
+        r"bond|treasury|gilt|t-bill|income|credit|debt|fixed|municipal|muni|aggregate|"
+        r"liquid|money market|floating rate|high yield|duration|mortgage|sdl|g-sec|bharat bond", re.I)),
+    (INTERNATIONAL, re.compile(
+        r"international|emerging|ex[- ]u\.?s|global|world|europe|japan|china|india|brazil|"
+        r"korea|taiwan|germany|canada|mexico|spain|kuwait|saudi|asia|pacific|latin|frontier|"
+        r"developed markets|hang seng|msci [a-z]+ (etf|index)", re.I)),
+    (BROAD_MARKET, re.compile(
+        r"s&p 500|s&p500|total (stock )?market|total world|nifty 50\b|nifty next|sensex|"
+        r"russell|nasdaq[- ]100|dow jones industrial|msci (world|acwi|eafe|emerging)|all[- ]world|"
+        r"large[- ]?cap|mid[- ]?cap|small[- ]?cap|micro[- ]?cap|nifty 100|nifty 200|nifty 500|"
+        r"broad|total market|equal weight|dividend|value|growth|momentum|quality|low vol", re.I)),
+]
+
+# Source sector labels (NSE industries + NASDAQ sectors), lower-cased.
+_SECTOR_MAP = {
+    # Listed trusts (NSE series RR / IV, see listings.TRUST_SERIES)
+    "reit": TRUSTS,
+    "invit": TRUSTS,
+    # NSE
+    "information technology": TECHNOLOGY,
+    "financial services": FINANCIALS,
+    "healthcare": HEALTHCARE,
+    "fast moving consumer goods": CONSUMER,
+    "consumer durables": CONSUMER,
+    "consumer services": SERVICES,
+    "automobile and auto components": AUTOMOBILE,
+    "capital goods": INDUSTRIALS,
+    "oil gas & consumable fuels": ENERGY,
+    "power": ENERGY,
+    "metals & mining": NATURAL_RESOURCES,
+    "forest materials": NATURAL_RESOURCES,
+    "chemicals": MATERIALS,
+    "construction materials": MATERIALS,
+    "construction": REAL_ESTATE,
+    "realty": REAL_ESTATE,
+    "telecommunication": TELECOM_MEDIA,
+    "media entertainment & publication": TELECOM_MEDIA,
+    "utilities": UTILITIES,
+    "textiles": TEXTILES,
+    "services": SERVICES,
+    "diversified": OTHER,
+    # NASDAQ
+    "technology": TECHNOLOGY,
+    "finance": FINANCIALS,
+    "health care": HEALTHCARE,
+    "consumer discretionary": CONSUMER,
+    "consumer staples": CONSUMER,
+    "industrials": INDUSTRIALS,
+    "energy": ENERGY,
+    "basic materials": MATERIALS,
+    "real estate": REAL_ESTATE,
+    "telecommunications": TELECOM_MEDIA,
+}
+
+# Checked against industry first, then the company name. Order matters:
+# the first matching rule wins, so specific themes sit above generic ones.
+_KEYWORD_RULES: list[tuple[str, re.Pattern]] = [
+    (AGRICULTURE, re.compile(
+        r"\b(agri\w*|agro\w*|farm\w*|seeds?|fertili[sz]\w*|crop\w*|sugars?|tea|teas|coffee|"
+        r"plantation\w*|dairy|poultry|rice|flour|milling|feeds?|fisher\w*|aqua\w*|"
+        r"edible oils?|solvent|cattle|kisan|krishi|grains?)\b", re.I)),
+    (NATURAL_RESOURCES, re.compile(
+        r"\b(mining|mines?|minerals?|metals?|steel|iron|ore|alloys?|copper|zinc|"
+        r"alumin\w*|gold|silver|precious|coal|granite|marble|forest\w*|timber|"
+        r"paper|pulp|quarry\w*|ferro\w*|sponge)\b", re.I)),
+    (ENERGY, re.compile(
+        r"\b(oil|gas|petro\w*|refin\w*|energ\w*|power\w*|solar|wind|renewable\w*|"
+        r"hydro\w*|fuels?|lng|drilling)\b", re.I)),
+    (HEALTHCARE, re.compile(
+        r"\b(pharma\w*|health\w*|hospitals?|medic\w*|drugs?|bio\w*|labs?|"
+        r"laborator\w*|diagnostic\w*|life ?sciences?|remedies|clinic\w*|surgical|"
+        r"therapeutic\w*|formulations?)\b", re.I)),
+    (TECHNOLOGY, re.compile(
+        r"\b(software|tech\w*|infotech|info ?systems?|computer\w*|digital|data|"
+        r"cyber\w*|semiconductor\w*|electronics?|it services|infoway|internet|"
+        r"cloud|ai|analytics|netsol)\b", re.I)),
+    (FINANCIALS, re.compile(
+        r"\b(bank\w*|financ\w*|fincorp|fincap|capital|invest\w*|credit|leasing|"
+        r"securities|insurance|assurance|asset management|holdings?|fintech|"
+        r"money|broking|stock ?brokers?|nbfc)\b", re.I)),
+    (TEXTILES, re.compile(
+        r"\b(tex\w*|textiles?|spinning|spinners|cotton|yarns?|fabrics?|garments?|apparel|"
+        r"weav\w*|knit\w*|denim|silk|jute|mills?)\b", re.I)),
+    (AUTOMOBILE, re.compile(r"\b(auto\w*|motors?|tyres?|tires?|vehicles?|forg\w*|mobility|ev|wheels?)\b", re.I)),
+    (MATERIALS, re.compile(
+        r"\b(chemicals?|chem\w*|paints?|plastics?|polymers?|pipes?|cement|glass|"
+        r"ceramics?|packaging|rubber|dyes?|pigments?|fertichem|laminates?)\b", re.I)),
+    (REAL_ESTATE, re.compile(
+        r"\b(realty|real estate|estates?|housing|homes|builders?|construct\w*|"
+        r"infra\w*|developers?|properties|projects)\b", re.I)),
+    (TELECOM_MEDIA, re.compile(
+        r"\b(telecom\w*|communications?|media|entertainment|broadcast\w*|films?|"
+        r"studios?|publish\w*|news\w*|cable|networks?)\b", re.I)),
+    (UTILITIES, re.compile(r"\b(water|sewer|utilit\w*|electric utilit\w*)\b", re.I)),
+    (CONSUMER, re.compile(
+        r"\b(foods?|beverages?|consumer|retail\w*|jewel\w*|fashion|footwear|"
+        r"cosmetics?|brands?|appliances?|furniture|breweries|distiller\w*|"
+        r"spirits|tobacco|restaurants?)\b", re.I)),
+    (SERVICES, re.compile(
+        r"\b(hotels?|resorts?|travels?|tours?|logistics|shipping|transport\w*|"
+        r"airlines?|aviation|education\w*|services|hospitality|courier|trading|traders|"
+        r"impex|exports?|overseas|ventures)\b", re.I)),
+    (INDUSTRIALS, re.compile(
+        r"\b(engineer\w*|machin\w*|equipments?|electricals?|cables?|industrial\w*|"
+        r"tools?|pumps?|valves?|bearings?|defen[cs]e|aerospace|castings?|"
+        r"fabricat\w*|compressors?|transformers?|turbines?)\b", re.I)),
+]
+
+
+# Eastmoney's (Chinese) industry names for China A / Hong Kong. Substring
+# match, first hit wins, so e.g. 电力设备 (Industrials) precedes 电力 (Energy).
+_ZH_RULES: list[tuple[str, tuple[str, ...]]] = [
+    # Financials first: 农业银行 / 农商行 are banks, not farms.
+    (FINANCIALS, ("银行", "证券", "保险", "金融", "信托", "期货", "农商行")),
+    (AGRICULTURE, ("农", "种植", "养殖", "饲料", "渔业", "林业", "动物保健", "畜")),
+    (NATURAL_RESOURCES, ("煤炭", "有色", "金属", "钢", "贵金属", "黄金", "冶", "焦炭", "矿", "造纸", "非金属材料")),
+    (INDUSTRIALS, ("电力设备", "电网设备", "光伏设备", "风电设备", "电源设备", "电池", "电机", "机械", "专用设备",
+                   "通用设备", "自动化设备", "工程机械", "轨交", "航空装备", "航天", "航海装备", "地面兵装",
+                   "军工", "环保设备", "照明设备", "工业")),
+    (ENERGY, ("石油", "油气", "油服", "炼化", "电力", "燃气", "能源", "光伏", "风电", "新能源")),
+    (TECHNOLOGY, ("半导体", "软件", "IT服务", "计算机", "电子", "元件", "光学", "通信设备", "互联网", "科技", "数字")),
+    (HEALTHCARE, ("医", "药", "生物", "医疗")),
+    (AUTOMOBILE, ("汽车", "乘用车", "商用车", "摩托车")),
+    (MATERIALS, ("化学", "化工", "塑料", "橡胶", "玻璃", "水泥", "建材", "化纤", "农化", "包装")),
+    (REAL_ESTATE, ("房地产", "地产", "建筑", "基础建设", "工程", "装修", "房屋")),
+    (TELECOM_MEDIA, ("通信服务", "电信", "传媒", "广告", "游戏", "影视", "出版", "电视", "媒体")),
+    (UTILITIES, ("水务", "公用", "环境治理")),
+    (TEXTILES, ("纺织", "服装", "家纺")),
+    (CONSUMER, ("食品", "饮料", "乳", "酒", "家电", "家居", "零售", "化妆品", "个护", "饰品", "文娱", "调味",
+                "休闲", "电商", "连锁", "美容")),
+    (SERVICES, ("物流", "航运", "港口", "机场", "铁路", "公路", "旅游", "酒店", "餐饮", "教育", "贸易", "服务", "体育")),
+]
+
+
+def _match_zh(text: str) -> Optional[str]:
+    for category, words in _ZH_RULES:
+        if any(w in text for w in words):
+            return category
+    return None
+
+
+def _match(text: str) -> Optional[str]:
+    for category, pattern in _KEYWORD_RULES:
+        if pattern.search(text):
+            return category
+    return None
+
+
+def classify(row: dict) -> tuple[str, bool]:
+    """(category, inferred) for a stock/ETF row."""
+    industry = row.get("industry") or ""
+    sector = (row.get("sector") or "").strip()
+
+    # US industries carry themes their broad sector hides (Farming sits
+    # under Consumer Staples, Steel under Industrials, ...).
+    if industry:
+        hit = _match(industry)
+        if hit in (AGRICULTURE, NATURAL_RESOURCES, ENERGY, UTILITIES, TECHNOLOGY):
+            return hit, False
+
+    mapped = _SECTOR_MAP.get(sector.lower())
+    if mapped:
+        return mapped, False
+    if sector and row.get("market") in ("china_a", "hong_kong"):
+        zh = _match_zh(sector)
+        if zh:
+            return zh, False
+
+    if row.get("type") == "etf":
+        text = f"{row.get('symbol', '')} {row.get('name', '')} {row.get('underlying') or ''}"
+        # Sector themes first ("Technology Select Sector"), then asset class.
+        themed = _match(text)
+        if themed and themed not in (FINANCIALS, SERVICES):
+            return themed, True
+        for category, pattern in _ETF_RULES:
+            if pattern.search(text):
+                return category, True
+        return themed or OTHER, True
+
+    name = row.get("name") or ""
+    return _match(name) or _match_zh(row.get("name_local") or name) or OTHER, True
+
+
+def fund_category(row: dict) -> str:
+    """Mutual funds: AMFI scheme group, cleaned into a short label."""
+    group = (row.get("fund_group") or "").lower()
+    sub = (row.get("sector") or "").lower()
+    if "index" in sub or "etf" in sub:
+        return "Index / ETF"
+    if "fund of funds" in sub or "fof" in sub:
+        return "Fund of Funds"
+    if "elss" in sub or "elss" in group:
+        return "ELSS (Tax Saving)"
+    for key, label in (
+        ("equity", "Equity"),
+        ("debt", "Debt"),
+        ("income", "Debt"),
+        ("hybrid", "Hybrid"),
+        ("solution", "Solution Oriented"),
+        ("children", "Solution Oriented"),
+        ("retirement", "Solution Oriented"),
+        ("growth", "Equity"),
+    ):
+        if key in group:
+            return label
+    return "Other"
+
+
+def enrich(rows: list[dict]) -> list[dict]:
+    """Stamp ``category`` (+ ``category_inferred``) on every row in place."""
+    for row in rows:
+        if row.get("type") == "mf":
+            row["category"] = fund_category(row)
+            row["category_inferred"] = False
+        elif row.get("type") == "crypto":
+            row["category"] = "Crypto"
+            row["category_inferred"] = False
+        else:
+            row["category"], row["category_inferred"] = classify(row)
+    return rows