Model/scripts/hyde/elmhurst_fill.py

"""Fill Elmhurst RdSAP-10-Online pages from known field values (POC).

Re-keys UPRN 10092973954 (the flat in elmhurst_inputs.md) into Elmhurst's web
entry tool. Reuses the saved persistent session and the viewer display.

    # dry-run one page (fills, screenshots, does NOT navigate away => not saved):
    DISPLAY=:99 python scripts/hyde/elmhurst_fill.py --page dimensions
    # commit it (clicks Next > which saves AND advances):
    DISPLAY=:99 python scripts/hyde/elmhurst_fill.py --page dimensions --commit

Mechanics learned from the live DOM (see elmhurst_explore.py):
  * ASP.NET WebForms. Almost every control **AutoPostBacks** — the whole page
    reloads on change — so we set ONE field, wait for the postback, then
    re-locate the next field (old handles go stale). A short settle after each
    interaction stops the next change being missed.
  * Enter via the Guid URL; section nav is a full-form __doPostBack. Navigating
    to another section **saves** the current page (so Next > = save + advance).
  * NEVER re-click the section tab you're already on — it double-postbacks and
    misbehaves. goto_section() checks the current URL and only clicks if the
    target is a different page.
  * Dropdown option values carry a code prefix ("F Flat", "L 2012-2022").
"""

from __future__ import annotations

import argparse
import sys
from dataclasses import dataclass
from pathlib import Path
from typing import Callable, Literal

from playwright.sync_api import (
    Page,
    TimeoutError as PlaywrightTimeoutError,
    sync_playwright,
)

HERE = Path(__file__).parent
SESSION_DIR = HERE / ".elmhurst-session"
SHOT_DIR = HERE / "elmhurst_dom"

ASSESSMENT_GUID = "B44A0DB4-4C08-4241-B818-86F060172105"
ENTRY_URL = (
    "https://rdsap10online.elmhurstenergy.co.uk/Processing/WebFormAddress.aspx"
    f"?Guid={ASSESSMENT_GUID}"
)
FIELD_PREFIX = "ContentBody_ContentPlaceHolder1_"
NAV_PREFIX = "ContentBody_buttonAction"

# WebForms reattaches client JS after each postback; act too fast and the next
# change is missed. A short settle after every interaction fixes the flake.
SETTLE_MS = 600


@dataclass
class Field:
    """One value to set. ``suffix`` is appended to FIELD_PREFIX to form the id.
    ``kind`` selects how we set it; ``autopostback`` whether the page reloads."""

    suffix: str
    kind: Literal["text", "select", "checkbox"]
    value: str  # for checkbox: "on" to check, "off" to uncheck
    autopostback: bool
    note: str = ""


@dataclass
class PageSpec:
    """A section: how to reach it, how to know we're on it, and what to set."""

    nav: str  # the buttonAction<Section> id segment, e.g. "PropertyDescription"
    url_fragment: str  # appears in page.url when this section is showing
    fields: list[Field]


# Dimensions Main-tab controls sit under a deep WebUserControl path.
_DIM_MAIN = "TabContainer_TabPanelMain_WebUserControlDimensionsMain_"
# Walls main external-wall controls sit under an even deeper nested-tab path.
_WALL_MAIN = (
    "TabContainer_TabPanelMain_InnerTabContainerMain_"
    "TabPanelExternalWallMain_WebUserControlWallMain_"
)

# UPRN 10092973954 (flat) — values from elmhurst_inputs.md.
# NOTE age band: 2020 build => Elmhurst "L 2012-2022" (by year), though our
# engine labels construction_age_band "M".
PAGES: dict[str, PageSpec] = {
    "property_description": PageSpec(
        nav="PropertyDescription",
        url_fragment="WebFormPropertyDescription.aspx",
        fields=[
            Field("DropDownListPropertyType1", "select", "F Flat", True, "property_type 2"),
            Field("DropDownListPropertyType2", "select", "D Detached", True, "built_form 1"),
            Field("TextBoxStoreys", "text", "1", True, "single storey"),
            Field("TextBoxHabitableRooms", "text", "3", False, "habitable_rooms 3"),
            Field("TextBoxHeatedHabitableRooms", "text", "3", True, "all heated"),
            Field("DropDownListDateBuiltMain", "select", "L 2012-2022", True, "2020 build"),
            Field("DropDownListDateBuiltFirst", "select", "", True, "no 1st extension — clear"),
            Field("DropDownListRoomInRoofMain", "select", "", True, "no room in roof — clear"),
        ],
    ),
    "dimensions": PageSpec(
        nav="Dimensions",
        url_fragment="WebFormDimensions.aspx",
        fields=[
            # Single-storey ground-floor flat => everything on the Lowest Floor
            # row of the Main tab. Heat-loss perimeter 38.14 m, no party wall.
            # (Type is already "Internal" and locks once dimensions exist — skip.)
            Field(f"{_DIM_MAIN}TextBoxFloorAreaLowestFloor", "text", "67.84", True, "engine TFA"),
            Field(f"{_DIM_MAIN}TextBoxRoomHeightLowestFloor", "text", "2.40", True, "vol/area"),
            Field(f"{_DIM_MAIN}TextBoxWallPerimeterLowestFloor", "text", "38.14", True, "heat-loss perimeter"),
            Field(f"{_DIM_MAIN}TextBoxPartyWallLengthLowestFloor", "text", "0", True, "no party wall"),
            # NB: Room-in-Roof area field is auto-DISABLED once Property
            # Description declares no room-in-roof, so we don't (can't) set it.
        ],
    ),
    "flats": PageSpec(
        nav="Flats",
        url_fragment="WebFormFlats.aspx",
        fields=[
            Field("DropDownListPositionOfFlat", "select", "Ground Floor", True, "ground-floor flat"),
            Field("TextBoxFloor", "text", "0", True, "ground = storey 0"),
            # Corridor None => all-exposed build (matches engine: stair wall is
            # exposed, not sheltered). Leaving "Unheated corridor" forces a
            # sheltered alternative wall on the Walls page (validation ✖).
            Field("RadioButtonListFlatCoridor", "select", "None", True, "no unheated corridor — all-exposed"),
            # 🔸 StoriesInBlock is assessor-only (EPC can't supply).
        ],
    ),
    "space_heating": PageSpec(
        nav="SpaceHeating",
        url_fragment="WebFormSpaceHeating.aspx",
        fields=[
            # NB: PCDF boiler reference is READ-ONLY (set via SelectBoilerDialog,
            # not typeable) — boiler PCDB 17929 must be driven through the dialog.
            # Room-sealed boiler => Balanced flue (cert is_flue_fan_present true).
            Field(
                "TabPanelMainHeating1_MainHeating1_RadioButtonListFlueType",
                "select", "Balanced", True, "room-sealed; main_heating_flue_type 2",
            ),
            # secondary_heating_category 1 = None — clears the placeholder 'RWM'.
            Field("DropDownListSecondaryHeatingPresent", "select", "No", True, "no secondary heating"),
            # 🔸 boiler PCDB 17929 + control 2110 go via search dialogs (manual).
        ],
    ),
    "water_heating": PageSpec(
        nav="WaterHeating",
        url_fragment="WebFormWaterHeating.aspx",
        fields=[
            # Combi / instantaneous — no cylinder (clears CylinderSize/Insulated/…).
            Field("TabPanelWaterHeating_CheckBoxHotWaterCylinder", "checkbox", "off", True, "no cylinder"),
            # 🔸 water-heating code (from main, 901) via its dialog (manual).
        ],
    ),
    "walls": PageSpec(
        nav="Walls",
        url_fragment="WebFormWalls.aspx",
        fields=[
            # Main wall is already correct (TI Timber Frame / As Built). Just
            # clear the leftover Alternative Wall 1 (Solid Brick placeholder) by
            # zeroing its area — our wall is a single timber-frame element.
            Field(f"{_WALL_MAIN}TextBoxAddWallArea", "text", "0", True, "clear placeholder alt wall"),
        ],
    ),
    "roofs": PageSpec(
        nav="Roofs",
        url_fragment="WebFormRoofs.aspx",
        fields=[
            # roof = "(other premises above)" => no roof heat loss. Selecting
            # this disables the insulation/thickness fields (nothing to clear).
            Field(
                "TabContainer_TabPanelMain_WebUserControlRoofMain_DropDownListType",
                "select", "A Another dwelling above", True, "roof_w_per_k = 0",
            ),
        ],
    ),
}


def _commit(page: Page, action: Callable[[], object]) -> None:
    """Run a control change that triggers an ASP.NET postback and wait for the
    page to come back: full-form AutoPostBack => navigation; fall back to
    networkidle for a partial/no postback."""
    try:
        with page.expect_navigation(wait_until="load", timeout=8_000):
            action()
    except PlaywrightTimeoutError:
        page.wait_for_load_state("networkidle", timeout=8_000)


def _set_field(page: Page, f: Field) -> None:
    loc = page.locator(f"#{FIELD_PREFIX}{f.suffix}")  # re-located each call
    if f.kind == "select":
        if f.autopostback:
            _commit(page, lambda: loc.select_option(f.value))
        else:
            loc.select_option(f.value)
    elif f.kind == "checkbox":
        check = loc.check if f.value == "on" else loc.uncheck
        if f.autopostback:
            _commit(page, check)
        else:
            check()
    else:  # text
        loc.fill(f.value)
        if f.autopostback:
            _commit(page, lambda: loc.press("Tab"))  # blur fires onchange
    page.wait_for_timeout(SETTLE_MS)  # half-beat so WebForms reattaches
    print(f"  set {f.suffix.split('_')[-1]:32s} = {f.value!r}   ({f.note})", flush=True)


def select_boiler_combi(page: Page) -> None:
    """Drive the Main-Heating cascade dialog to a mains-gas condensing combi
    (EES code BGW = SAP 104). The exact-model PCDB path is disabled
    (aspNetDisabled), so the cascade's generic combi is the automatable choice;
    its efficiency barely moves SAP for fabric-dominated dwellings. Selecting a
    combi also drops the HW cylinder on the Water Heating page."""
    mh1 = (
        "ContentBody_ContentPlaceHolder1_TabContainer_TabPanelMainHeating1_"
        "WebUserControlMainHeating1_"
    )
    dd6 = "ContentBody_OutsideUpdatePanel_SelectHeatingDialog_DropDownList6"
    page.click(f"#{mh1}ButtonMainHeatingCode", timeout=10_000)  # open cascade
    page.wait_for_timeout(SETTLE_MS)
    # Levels 1-5 (Gas/Mains gas/Boilers/Post 1998/Condensing) are already the
    # default path; only the leaf differs (regular BGB -> combi BGW).
    page.select_option(f"#{dd6}", label="BGW Post 98 Combi condens. with auto ign.")
    page.wait_for_timeout(SETTLE_MS)
    page.get_by_text("OK", exact=True).first.click(timeout=10_000)  # apply
    page.wait_for_timeout(SETTLE_MS * 2)
    print("  boiler -> BGW (SAP 104, condensing combi)", flush=True)


def goto_section(page: Page, spec: PageSpec) -> None:
    """Navigate to a section — but never re-click the tab we're already on
    (that double-postbacks and misbehaves)."""
    if spec.url_fragment in page.url:
        print(f"already on {spec.nav}", flush=True)
        return
    print(f"navigating to {spec.nav} ...", flush=True)
    with page.expect_navigation(wait_until="networkidle", timeout=60_000):
        page.click(f"#{NAV_PREFIX}{spec.nav}_Link", timeout=15_000)
    page.wait_for_timeout(SETTLE_MS)


def fill_page(page_key: str, commit: bool, commit_mode: str = "saveclose") -> None:
    spec = PAGES[page_key]
    SHOT_DIR.mkdir(parents=True, exist_ok=True)
    with sync_playwright() as p:
        context = p.chromium.launch_persistent_context(
            user_data_dir=str(SESSION_DIR),
            headless=False,
            accept_downloads=True,
            viewport={"width": 1400, "height": 1000},
        )
        page = context.pages[0] if context.pages else context.new_page()
        page.on("dialog", lambda d: d.accept())  # auto-dismiss confirm dialogs

        print(f"entering assessment {ASSESSMENT_GUID} ...", flush=True)
        page.goto(ENTRY_URL, wait_until="networkidle", timeout=60_000)
        goto_section(page, spec)

        if page_key == "space_heating":
            # Boiler is a cascade-dialog field — set it before the row fields
            # (it postbacks and can reset flue, so flue is set after).
            select_boiler_combi(page)

        print(f"filling {page_key}:", flush=True)
        for f in spec.fields:
            _set_field(page, f)

        page.screenshot(path=str(SHOT_DIR / f"fill_{page_key}.png"))
        print(f"screenshot: {SHOT_DIR / f'fill_{page_key}.png'}", flush=True)

        if commit:
            # Save & Close commits to the DATABASE (visible to other sessions);
            # Next > only saves into the in-progress server session. Default to
            # Save & Close so saved results show up elsewhere.
            # The body button's container path varies per page (Address nests
            # it, content pages don't), so match by id suffix. The top-bar copy
            # is a *Link*ButtonSaveClose, so "_ButtonSaveClose" hits only body.
            suffix = "_ButtonSaveClose" if commit_mode == "saveclose" else "_ButtonNext"
            print(f"committing ({commit_mode}) ...", flush=True)
            with page.expect_navigation(wait_until="networkidle", timeout=60_000):
                page.click(f"[id$='{suffix}']", timeout=15_000)
            page.wait_for_timeout(SETTLE_MS)
            print(f"saved. now on: {page.url}", flush=True)
            page.screenshot(path=str(SHOT_DIR / f"after_{page_key}.png"))
        else:
            print("DRY-RUN: not saved. Re-run with --commit to persist.", flush=True)

        context.close()


def main(argv: list[str] | None = None) -> int:
    ap = argparse.ArgumentParser(description="Fill an Elmhurst RdSAP page (POC).")
    ap.add_argument("--page", default="property_description", choices=sorted(PAGES))
    ap.add_argument("--commit", action="store_true", help="persist (else dry-run)")
    ap.add_argument(
        "--commit-mode",
        default="saveclose",
        choices=("saveclose", "next"),
        help="saveclose = DB commit (visible elsewhere); next = save + advance",
    )
    args = ap.parse_args(argv)
    fill_page(args.page, commit=args.commit, commit_mode=args.commit_mode)
    return 0


if __name__ == "__main__":
    sys.exit(main())