scraper2-hj3415 2.4.1__py3-none-any.whl → 2.6.0__py3-none-any.whl

scraper2/adapters/out/sinks/memory/c108_memory_sink.py

@@ -1,26 +0,0 @@
-#scraper2/adapters/out/sinks/memory/c108_memory_sink.py
-from __future__ import annotations
-
-from datetime import datetime
-from typing import Iterable, Optional
-
-from contracts.nfs.c108 import C108DTO
-from scraper2.adapters.out.sinks.memory.store import InMemoryStore
-from scraper2.app.ports.sinks.c108_sink_port import C108SinkPort
-
-_ENDPOINT = "c108"
-
-class MemoryC108Sink(C108SinkPort):
-    def __init__(self, store: InMemoryStore[C108DTO]):
-        self._store = store
-
-    async def write(self, dto: C108DTO, *, asof: Optional[datetime] = None) -> None:
-        await self._store.put(_ENDPOINT, dto.코드, dto)
-
-    async def write_many(
-            self,
-            dtos: Iterable[C108DTO],
-            *,
-            asof: Optional[datetime] = None,
-    ) -> None:
-        await self._store.put_many(_ENDPOINT, ((d.코드, d) for d in dtos))

scraper2/adapters/out/sinks/mongo/__init__.py

@@ -1,14 +0,0 @@
-# scraper2/adapters/out/sinks/mongo/__init__.py
-from .c101_mongo_sink import MongoC101Sink
-from .c103_mongo_sink import MongoC103Sink
-from .c104_mongo_sink import MongoC104Sink
-from .c106_mongo_sink import MongoC106Sink
-from .c108_mongo_sink import MongoC108Sink
-
-__all__ = [
-    "MongoC101Sink",
-    "MongoC103Sink",
-    "MongoC104Sink",
-    "MongoC106Sink",
-    "MongoC108Sink",
-]

scraper2/adapters/out/sinks/mongo/c101_mongo_sink.py

@@ -1,43 +0,0 @@
-# scraper2/adapters/out/sinks/mongo/c101_mongo_sink.py
-from __future__ import annotations
-
-from datetime import datetime
-from typing import Iterable, Optional
-
-from pymongo.asynchronous.database import AsyncDatabase
-
-from contracts.nfs.c101 import C101DTO
-from scraper2.app.ports.sinks.c101_sink_port import C101SinkPort
-
-from db2.nfs import (
-    upsert_latest,
-    upsert_latest_many,
-    insert_snapshot,
-    insert_snapshots_many,
-)
-
-_ENDPOINT = "c101"
-
-
-class MongoC101Sink(C101SinkPort):
-    def __init__(self, db: AsyncDatabase):
-        self._db = db
-
-    async def write(self, dto: C101DTO, *, asof: Optional[datetime] = None) -> None:
-        # 최신 상태는 upsert
-        await upsert_latest(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-        # 히스토리는 insert
-        await insert_snapshot(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-
-    async def write_many(
-        self,
-        dtos: Iterable[C101DTO],
-        *,
-        asof: Optional[datetime] = None,
-    ) -> None:
-        dtos_list = list(dtos)
-        if not dtos_list:
-            return
-
-        await upsert_latest_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)
-        await insert_snapshots_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)

scraper2/adapters/out/sinks/mongo/c103_mongo_sink.py

@@ -1,41 +0,0 @@
-# scraper2/adapters/out/sinks/mongo/c103_mongo_sink.py
-from __future__ import annotations
-
-from datetime import datetime
-from typing import Iterable, Optional
-
-from pymongo.asynchronous.database import AsyncDatabase
-
-from contracts.nfs.c103 import C103DTO
-from scraper2.app.ports.sinks.c103_sink_port import C103SinkPort
-
-from db2.nfs import (
-    upsert_latest,
-    upsert_latest_many,
-    insert_snapshot,
-    insert_snapshots_many,
-)
-
-_ENDPOINT = "c103"
-
-
-class MongoC103Sink(C103SinkPort):
-    def __init__(self, db: AsyncDatabase):
-        self._db = db
-
-    async def write(self, dto: C103DTO, *, asof: Optional[datetime] = None) -> None:
-        await upsert_latest(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-        await insert_snapshot(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-
-    async def write_many(
-        self,
-        dtos: Iterable[C103DTO],
-        *,
-        asof: Optional[datetime] = None,
-    ) -> None:
-        dtos_list = list(dtos)
-        if not dtos_list:
-            return
-
-        await upsert_latest_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)
-        await insert_snapshots_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)

scraper2/adapters/out/sinks/mongo/c104_mongo_sink.py

@@ -1,41 +0,0 @@
-# scraper2/adapters/out/sinks/mongo/c104_mongo_sink.py
-from __future__ import annotations
-
-from datetime import datetime
-from typing import Iterable, Optional
-
-from pymongo.asynchronous.database import AsyncDatabase
-
-from contracts.nfs.c104 import C104DTO
-from scraper2.app.ports.sinks.c104_sink_port import C104SinkPort
-
-from db2.nfs import (
-    upsert_latest,
-    upsert_latest_many,
-    insert_snapshot,
-    insert_snapshots_many,
-)
-
-_ENDPOINT = "c104"
-
-
-class MongoC104Sink(C104SinkPort):
-    def __init__(self, db: AsyncDatabase):
-        self._db = db
-
-    async def write(self, dto: C104DTO, *, asof: Optional[datetime] = None) -> None:
-        await upsert_latest(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-        await insert_snapshot(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-
-    async def write_many(
-        self,
-        dtos: Iterable[C104DTO],
-        *,
-        asof: Optional[datetime] = None,
-    ) -> None:
-        dtos_list = list(dtos)
-        if not dtos_list:
-            return
-
-        await upsert_latest_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)
-        await insert_snapshots_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)

scraper2/adapters/out/sinks/mongo/c106_mongo_sink.py

@@ -1,41 +0,0 @@
-# scraper2/adapters/out/sinks/mongo/c106_mongo_sink.py
-from __future__ import annotations
-
-from datetime import datetime
-from typing import Iterable, Optional
-
-from pymongo.asynchronous.database import AsyncDatabase
-
-from contracts.nfs.c106 import C106DTO
-from scraper2.app.ports.sinks.c106_sink_port import C106SinkPort
-
-from db2.nfs import (
-    upsert_latest,
-    upsert_latest_many,
-    insert_snapshot,
-    insert_snapshots_many,
-)
-
-_ENDPOINT = "c106"
-
-
-class MongoC106Sink(C106SinkPort):
-    def __init__(self, db: AsyncDatabase):
-        self._db = db
-
-    async def write(self, dto: C106DTO, *, asof: Optional[datetime] = None) -> None:
-        await upsert_latest(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-        await insert_snapshot(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-
-    async def write_many(
-        self,
-        dtos: Iterable[C106DTO],
-        *,
-        asof: Optional[datetime] = None,
-    ) -> None:
-        dtos_list = list(dtos)
-        if not dtos_list:
-            return
-
-        await upsert_latest_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)
-        await insert_snapshots_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)

scraper2/adapters/out/sinks/mongo/c108_mongo_sink.py

@@ -1,41 +0,0 @@
-# scraper2/adapters/out/sinks/mongo/c108_mongo_sink.py
-from __future__ import annotations
-
-from datetime import datetime
-from typing import Iterable, Optional
-
-from pymongo.asynchronous.database import AsyncDatabase
-
-from contracts.nfs.c108 import C108DTO
-from scraper2.app.ports.sinks.c108_sink_port import C108SinkPort
-
-from db2.nfs import (
-    upsert_latest,
-    upsert_latest_many,
-    insert_snapshot,
-    insert_snapshots_many,
-)
-
-_ENDPOINT = "c108"
-
-
-class MongoC108Sink(C108SinkPort):
-    def __init__(self, db: AsyncDatabase):
-        self._db = db
-
-    async def write(self, dto: C108DTO, *, asof: Optional[datetime] = None) -> None:
-        await upsert_latest(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-        await insert_snapshot(self._db, endpoint=_ENDPOINT, dto=dto, asof=asof)
-
-    async def write_many(
-        self,
-        dtos: Iterable[C108DTO],
-        *,
-        asof: Optional[datetime] = None,
-    ) -> None:
-        dtos_list = list(dtos)
-        if not dtos_list:
-            return
-
-        await upsert_latest_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)
-        await insert_snapshots_many(self._db, endpoint=_ENDPOINT, dtos=dtos_list, asof=asof)

scraper2/app/composition.py

@@ -1,204 +0,0 @@
-# scraper2/app/composition.py
-from __future__ import annotations
-
-import os
-from dataclasses import dataclass
-from typing import Literal, Optional
-
-from pymongo.asynchronous.database import AsyncDatabase
-
-from scraper2.app.ports.browser.browser_factory_port import BrowserFactoryPort
-from scraper2.adapters.out.playwright.browser_factory import PlaywrightBrowserFactory
-
-from scraper2.app.usecases.fetch.fetch_c101 import FetchC101
-from scraper2.app.usecases.fetch.fetch_c103 import FetchC103
-from scraper2.app.usecases.fetch.fetch_c104 import FetchC104
-from scraper2.app.usecases.fetch.fetch_c106 import FetchC106
-from scraper2.app.usecases.fetch.fetch_c108 import FetchC108
-
-from scraper2.app.ports.ingest_port import IngestPort
-from scraper2.app.usecases.ingest.ingest_c101 import IngestC101
-from scraper2.app.usecases.ingest.ingest_c103 import IngestC103
-from scraper2.app.usecases.ingest.ingest_c104 import IngestC104
-from scraper2.app.usecases.ingest.ingest_c106 import IngestC106
-from scraper2.app.usecases.ingest.ingest_c108 import IngestC108
-
-from scraper2.adapters.out.sinks.memory.store import InMemoryStore
-from scraper2.adapters.out.sinks.memory.c101_memory_sink import MemoryC101Sink
-from scraper2.adapters.out.sinks.memory.c103_memory_sink import MemoryC103Sink
-from scraper2.adapters.out.sinks.memory.c104_memory_sink import MemoryC104Sink
-from scraper2.adapters.out.sinks.memory.c106_memory_sink import MemoryC106Sink
-from scraper2.adapters.out.sinks.memory.c108_memory_sink import MemoryC108Sink
-
-from scraper2.adapters.out.sinks.mongo.c101_mongo_sink import MongoC101Sink
-from scraper2.adapters.out.sinks.mongo.c103_mongo_sink import MongoC103Sink
-from scraper2.adapters.out.sinks.mongo.c104_mongo_sink import MongoC104Sink
-from scraper2.adapters.out.sinks.mongo.c106_mongo_sink import MongoC106Sink
-from scraper2.adapters.out.sinks.mongo.c108_mongo_sink import MongoC108Sink
-
-from scraper2.app.ports.sinks.c101_sink_port import C101SinkPort
-from scraper2.app.ports.sinks.c103_sink_port import C103SinkPort
-from scraper2.app.ports.sinks.c104_sink_port import C104SinkPort
-from scraper2.app.ports.sinks.c106_sink_port import C106SinkPort
-from scraper2.app.ports.sinks.c108_sink_port import C108SinkPort
-
-from db2.mongo import Mongo
-
-SinkKind = Literal["memory", "mongo"]
-
-
-def _env_bool(key: str, default: bool) -> bool:
-    v = os.getenv(key)
-    return default if v is None else v.strip().lower() in {"1", "true", "yes", "y", "on"}
-
-
-def _env_int(key: str, default: int) -> int:
-    v = os.getenv(key)
-    if v is None:
-        return default
-    try:
-        return int(v)
-    except ValueError:
-        return default
-
-
-def build_browser_factory() -> BrowserFactoryPort:
-    return PlaywrightBrowserFactory(
-        headless=_env_bool("SCRAPER_HEADLESS", True),
-        timeout_ms=_env_int("SCRAPER_TIMEOUT_MS", 20_000),
-        max_concurrency=_env_int("SCRAPER_MAX_CONCURRENCY", 2),
-    )
-
-
-# -------------------------
-# Bundles
-# -------------------------
-
-@dataclass(frozen=True)
-class FetchUsecases:
-    c101: FetchC101
-    c103: FetchC103
-    c104: FetchC104
-    c106: FetchC106
-    c108: FetchC108
-
-
-@dataclass(frozen=True)
-class Sinks:
-    c101: C101SinkPort
-    c103: C103SinkPort
-    c104: C104SinkPort
-    c106: C106SinkPort
-    c108: C108SinkPort
-
-
-@dataclass(frozen=True)
-class IngestUsecases:
-    c101: IngestPort
-    c103: IngestPort
-    c104: IngestPort
-    c106: IngestPort
-    c108: IngestPort
-
-
-@dataclass(frozen=True)
-class Usecases:
-    fetch: FetchUsecases
-    ingest: IngestUsecases
-    sinks: Sinks
-    store: InMemoryStore | None = None      # ✅ memory일 때만
-    mongo: Mongo | None = None              # ✅ mongo일 때만
-    db: AsyncDatabase | None = None         # ✅ mongo일 때만
-    browser_factory: Optional[BrowserFactoryPort] = None
-
-    async def aclose(self) -> None:
-        if self.browser_factory is not None:
-            await self.browser_factory.aclose()
-
-        if self.mongo is not None:
-            await self.mongo.close()
-
-# -------------------------
-# builders
-# -------------------------
-
-def build_fetch_usecases(*, factory: BrowserFactoryPort) -> FetchUsecases:
-    return FetchUsecases(
-        c101=FetchC101(factory=factory),
-        c103=FetchC103(factory=factory),
-        c104=FetchC104(factory=factory),
-        c106=FetchC106(factory=factory),
-        c108=FetchC108(factory=factory),
-    )
-
-
-@dataclass(frozen=True)
-class MemoryBundle:
-    store: InMemoryStore
-    sinks: Sinks
-
-
-def build_memory_bundle() -> MemoryBundle:
-    store = InMemoryStore()
-    sinks = Sinks(
-        c101=MemoryC101Sink(store),
-        c103=MemoryC103Sink(store),
-        c104=MemoryC104Sink(store),
-        c106=MemoryC106Sink(store),
-        c108=MemoryC108Sink(store),
-    )
-    return MemoryBundle(store=store, sinks=sinks)
-
-# ---- mongo bundle ----
-
-@dataclass(frozen=True)
-class MongoBundle:
-    mongo: Mongo
-    db: AsyncDatabase
-    sinks: Sinks
-
-
-def build_mongo_bundle() -> MongoBundle:
-    mongo = Mongo()        # settings는 db2가 env로 읽음 (DB2_MONGO_URI 등)
-    db = mongo.get_db()
-    sinks = Sinks(
-        c101=MongoC101Sink(db),
-        c103=MongoC103Sink(db),
-        c104=MongoC104Sink(db),
-        c106=MongoC106Sink(db),
-        c108=MongoC108Sink(db),
-    )
-    return MongoBundle(mongo=mongo, db=db, sinks=sinks)
-
-
-def build_ingest_usecases(*, fetch: FetchUsecases, sinks: Sinks) -> IngestUsecases:
-    return IngestUsecases(
-        c101=IngestC101(fetch=fetch.c101, sink=sinks.c101),
-        c103=IngestC103(fetch=fetch.c103, sink=sinks.c103),
-        c104=IngestC104(fetch=fetch.c104, sink=sinks.c104),
-        c106=IngestC106(fetch=fetch.c106, sink=sinks.c106),
-        c108=IngestC108(fetch=fetch.c108, sink=sinks.c108),
-    )
-
-
-def build_usecases(
-    *,
-    factory: BrowserFactoryPort | None = None,
-    sink_kind: SinkKind = "memory",
-) -> Usecases:
-    factory = factory or build_browser_factory()
-    fetch = build_fetch_usecases(factory=factory)
-
-    if sink_kind == "memory":
-        bundle = build_memory_bundle()
-        ingest = build_ingest_usecases(fetch=fetch, sinks=bundle.sinks)
-        return Usecases(fetch=fetch, ingest=ingest, sinks=bundle.sinks, store=bundle.store,
-            browser_factory=factory)
-
-    if sink_kind == "mongo":
-        bundle = build_mongo_bundle()
-        ingest = build_ingest_usecases(fetch=fetch, sinks=bundle.sinks)
-        return Usecases(fetch=fetch, ingest=ingest, sinks=bundle.sinks, mongo=bundle.mongo, db=bundle.db,
-            browser_factory=factory)
-
-    raise ValueError(f"Unknown sink_kind: {sink_kind}")

scraper2/app/parsing/_converters.py

@@ -1,85 +0,0 @@
-from __future__ import annotations
-import re
-from typing import Iterable
-
-
-_EMPTY_VALUES = {"", "-", "N/A", "NA", "null", "None"}
-
-
-def normalize(s: str | None) -> str:
-    if s is None:
-        return ""
-    return s.strip()
-
-
-def _is_empty(s: str) -> bool:
-    return s in _EMPTY_VALUES
-
-
-def to_int(s: str | None) -> int | None:
-    """
-    C101 파서용 정수 변환기
-
-    처리 규칙:
-    - None / '' / '-' / 'N/A' → None
-    - ',', '주', '원', '주식' 제거
-    - '1,234' → 1234
-    """
-    s = normalize(s)
-    if _is_empty(s):
-        return None
-
-    for ch in (",", "원", "주", "주식"):
-        s = s.replace(ch, "")
-
-    try:
-        return int(s)
-    except ValueError:
-        return None
-
-
-def to_float(s: str | None) -> float | None:
-    """
-    C101 파서용 실수 변환기
-
-    처리 규칙:
-    - None / '' / '-' / 'N/A' → None
-    - ',', '%', '원' 제거
-    - '12.34%' → 12.34
-    """
-    s = normalize(s)
-    if _is_empty(s):
-        return None
-
-    for ch in (",", "%", "원"):
-        s = s.replace(ch, "")
-
-    try:
-        return float(s)
-    except ValueError:
-        return None
-
-
-def parse_won(text: str) -> int:
-    """
-    한국 화폐 표현 문자열을 숫자로 변환 (조원, 억원, 만원, 원, 억 등 처리)
-    """
-    units = {
-        "조원": 1_000_000_000_000,
-        "억원": 100_000_000,
-        "억":   100_000_000,
-        "만원": 10_000,
-        "원": 1,
-    }
-
-    text = text.replace(",", "").strip()
-    match = re.match(r"([-+]?[0-9]*\.?[0-9]+)([가-힣]+)", text)
-
-    if not match:
-        raise ValueError(f"형식이 잘못된 금액 문자열: {text}")
-
-    number, unit = match.groups()
-    if unit not in units:
-        raise ValueError(f"알 수 없는 단위: {unit}")
-
-    return int(float(number) * units[unit])

scraper2/app/parsing/_normalize.py

@@ -1,134 +0,0 @@
-# scraper2/app/parsing/_normalize.py
-from __future__ import annotations
-
-import re
-from collections import Counter
-from typing import Any
-
-import numpy as np
-import pandas as pd
-
-
-# -----------------------------
-# 1) 항목(행의 "항목" 값) 정규화
-# -----------------------------
-_IFRS_PATTERN = re.compile(r"\(IFRS[^)]*\)")
-_ETC_PAREN_PATTERN = re.compile(r"\((E|YoY|QoQ)[^)]*\)")
-_BRACKET_PATTERN = re.compile(r"\[[^\]]*\]")  # [구K-IFRS] 등
-_EXTRA_WORDS_PATTERN = re.compile(r"(펼치기|연간컨센서스보기|연간컨센서스닫기)")
-_ALL_PAREN_PATTERN = re.compile(r"\([^)]*\)")  # ★ 모든 괄호 제거
-
-def normalize_c1034_item(text: str | None) -> str:
-    """
-    C103 항목명(행 값) 정규화
-    - 펼치기/컨센서스 제거
-    - 모든 괄호 내용 제거 (발표기준, 연결, 개별 등 포함)
-    - [구K-IFRS] 제거
-    - '*' 제거
-    - 공백 정리
-    """
-    if not text:
-        return ""
-
-    s = str(text)
-
-    # 1) 불필요한 키워드
-    s = _EXTRA_WORDS_PATTERN.sub("", s)
-
-    # 2) 대괄호 제거
-    s = _BRACKET_PATTERN.sub("", s)
-
-    # 3) 모든 괄호 제거 (중요)
-    s = _ALL_PAREN_PATTERN.sub("", s)
-
-    # 4) 별표 제거
-    s = s.replace("*", "")
-
-    # 5) 공백 정리
-    s = re.sub(r"\s+", " ", s).strip()
-
-    return s
-
-
-# -----------------------------
-# 2) 컬럼명 정규화
-# -----------------------------
-_COL_PAREN_PATTERN = re.compile(r"\((IFRS[^)]*|E|YoY|QoQ)[^)]*\)")
-_COL_EXTRA_WORDS = re.compile(r"(연간컨센서스보기|연간컨센서스닫기)")
-_COL_DOTNUM = re.compile(r"\.\d+$")  # pandas 중복 컬럼 suffix 제거용 (.1, .2 ...)
-_COL_MULTI_SPACE = re.compile(r"\s+")
-
-def normalize_c1034_col(col: str | None) -> str:
-    """
-    C103 컬럼명 정규화
-    예)
-      "2024/12 (IFRS연결)  연간컨센서스보기" -> "2024/12"
-      "2025/12(E) (IFRS연결)  연간컨센서스닫기" -> "2025/12"
-      "전년대비 (YoY)" -> "전년대비"
-      "전년대비 (YoY).1" -> "전년대비"  (중복은 후처리에서 _2/_3로 자동 분리)
-    """
-    if col is None:
-        return ""
-
-    s = str(col)
-
-    # 1) pandas가 붙인 .1 같은 suffix 제거 (정규화 충돌은 후단에서 처리)
-    s = _COL_DOTNUM.sub("", s)
-
-    # 2) 컨센서스 문구 제거
-    s = _COL_EXTRA_WORDS.sub("", s)
-
-    # 3) 괄호 주석 제거: (IFRS...), (E), (YoY), (QoQ)
-    s = _COL_PAREN_PATTERN.sub("", s)
-
-    # 4) 공백 정리
-    s = _COL_MULTI_SPACE.sub(" ", s).strip()
-
-    return s
-
-
-def _dedupe_columns(cols: list[str]) -> list[str]:
-    """
-    정규화 후 중복 컬럼명이 생기면 자동으로 _2, _3 ... 붙여서 유니크하게 만든다.
-    예) ["전년대비", "전년대비"] -> ["전년대비", "전년대비_2"]
-    """
-    seen: Counter[str] = Counter()
-    out: list[str] = []
-    for c in cols:
-        c = c or ""
-        seen[c] += 1
-        if seen[c] == 1:
-            out.append(c)
-        else:
-            out.append(f"{c}_{seen[c]}")
-    return out
-
-
-# -----------------------------
-# 3) DataFrame 전체 정규화 + records 변환
-# -----------------------------
-def normalize_c1034_df(df: pd.DataFrame) -> pd.DataFrame:
-    """
-    - 컬럼명 전체 정규화
-    - '항목' 값 정규화
-    - NaN -> None
-    - 중복 컬럼명 자동 분리(_2/_3)
-    """
-    if df is None or df.empty:
-        return df
-
-    df = df.copy()
-
-    # 컬럼명 정규화 + 중복 방지
-    norm_cols = [normalize_c1034_col(c) for c in df.columns.astype(str).tolist()]
-    df.columns = _dedupe_columns(norm_cols)
-
-    # 항목 값 정규화
-    if "항목" in df.columns:
-        df["항목"] = df["항목"].map(normalize_c1034_item)
-
-    # NaN -> None
-    df = df.replace({np.nan: None})
-    return df
-
-