fields_config_plants.py

CONFIG = {
    "REPO_ID": "mengtaoguo/my-data-archive",
    "DATA_DIR": "project_plants/raw_html",
    "TARGET_TABLE": "bs4_plants",
    "TARGET_ID_COL": "plant_id",
    "FIELD_RULES": {
        # --- 基本信息 ---
        "latin": {"col": "latin_name", "html_id": "ContentPlaceHolder1_lbldisplatinname", "use_bleach": False},
        "common": {"col": "common_name", "html_id": "ContentPlaceHolder1_lblCommanName", "use_bleach": False},
        "fam": {"col": "family", "html_id": "ContentPlaceHolder1_lblFamily", "use_bleach": False},
        "author_info": {"col": "author", "html_id": "ContentPlaceHolder1_txtAuthor", "use_bleach": False},
        "title": {"col": "title", "html_id": "ContentPlaceHolder1_lblTitle", "use_bleach": False},
        
        # --- 评级与环境 ---
        "hardiness": {"col": "usda_hardiness", "html_id": "ContentPlaceHolder1_lblUSDAhardiness", "use_bleach": False},
        "hazards": {"col": "known_hazards", "html_id": "ContentPlaceHolder1_lblKnownHazards", "use_bleach": False},
        "weed": {"col": "weed_potential", "html_id": "ContentPlaceHolder1_lblWeedPotential", "use_bleach": False},
        "conservation": {"col": "conservation_status", "html_id": "ContentPlaceHolder1_lblConservationStatus", "use_bleach": False},
        
        # --- 详细描述 ---
        "physical": {"col": "physical_characteristics", "html_id": "ContentPlaceHolder1_lblPhystatment", "use_bleach": False},
        "habitats": {"col": "habitats", "html_id": "ContentPlaceHolder1_lblhabitats", "use_bleach": False},
        "range_dist": {"col": "range", "html_id": "ContentPlaceHolder1_lblRange", "use_bleach": False},
        "found": {"col": "found_in", "html_id": "ContentPlaceHolder1_lblFoundInText", "use_bleach": False},
        
        # --- 用途与栽培 ---
        "edible": {"col": "edible_uses", "html_id": "ContentPlaceHolder1_txtEdibleUses", "use_bleach": False},
        "medicinal": {"col": "medicinal_uses", "html_id": "ContentPlaceHolder1_txtMediUses", "use_bleach": False},
        "other": {"col": "other_uses", "html_id": "ContentPlaceHolder1_txtOtherUses", "use_bleach": False},
        "special": {"col": "special_uses", "html_id": "ContentPlaceHolder1_txtSpecialUses", "use_bleach": False},
        "cultivation": {"col": "cultivation_details", "html_id": "ContentPlaceHolder1_txtCultivationDetails", "use_bleach": True},
        "prop": {"col": "propagation", "html_id": "ContentPlaceHolder1_txtPropagation", "use_bleach": False},
    }
}