app.py

import os
import uuid
import shutil
import tempfile
import datetime
import pandas as pd
import gradio as gr

from PIL import Image

# =========================
# PDF (reportlab)
# =========================
from reportlab.platypus import SimpleDocTemplate, Paragraph, Spacer, PageBreak
from reportlab.platypus import Image as RLImage
from reportlab.lib.styles import getSampleStyleSheet
from reportlab.lib.pagesizes import A4
from reportlab.lib.units import cm

# =========================
# CONFIG
# =========================
XLSX_PATH = "Soild Anaylsis EN(1).xlsx"
IND_SHEET = "Indicators"

SEASON_COLS = {
    "Fall": "Available in Fall",
    "Winter": "Available in Winter",
    "Spring": "Available in Spring",
    "Summer": "Available in Summer",
}

SCORE_MAP = {"Low": 1, "Medium": 2, "High": 3}

# =========================
# LOAD DATA
# =========================
if not os.path.exists(XLSX_PATH):
    raise FileNotFoundError("Missing Soil Analysis Excel file")

ind_df = pd.read_excel(XLSX_PATH, sheet_name=IND_SHEET)

# =========================
# HELPERS
# =========================
def normalize_text(x):
    if pd.isna(x):
        return ""
    return str(x).strip()

def is_available(row, season):
    return normalize_text(row.get(SEASON_COLS[season], "")).lower() == "x"

def build_indicators(season):
    return [r for _, r in ind_df.iterrows() if is_available(r, season)]

def requires_self_assessment(data_entry):
    data_entry = data_entry.lower()
    return any(
        k in data_entry
        for k in ["smell", "odor", "temperature", "moisture", "value", "on site"]
    )

# =========================
# SAFE IMAGE HANDLING (FIX)
# =========================
def save_uploaded_images(uploaded):
    """
    Normalize Gradio uploads into safe local file paths.
    Accepts PIL, str, tuple, list of them.
    """
    if not uploaded:
        return []

    if not isinstance(uploaded, list):
        uploaded = [uploaded]

    saved = []

    for item in uploaded:
        if isinstance(item, tuple):
            item = item[0]

        if isinstance(item, str) and os.path.exists(item):
            dst = f"/tmp/{uuid.uuid4().hex}{os.path.splitext(item)[1]}"
            shutil.copy(item, dst)
            saved.append(dst)

        elif isinstance(item, Image.Image):
            dst = f"/tmp/{uuid.uuid4().hex}.png"
            item.save(dst)
            saved.append(dst)

    return saved

# =========================
# PLACEHOLDER "MODEL"
# =========================
def visual_score_placeholder(image_paths, indicator_name):
    """
    Lightweight heuristic placeholder.
    Replace later with CLIP / ViT safely.
    """
    if not image_paths:
        return "Medium", "No relevant photo detected; defaulting to Medium."

    name = indicator_name.lower()

    if "earthworm" in name or "fauna" in name:
        return "High", "Visible earthworms indicate active soil biology."

    if "erosion" in name:
        return "Low", "Visible rills and sediment indicate erosion."

    if "structure" in name:
        return "Medium", "Soil aggregates visible but not clearly crumbly."

    if "cover" in name:
        return "High", "Dense vegetation cover visible."

    return "Medium", "Visual indicators are inconclusive."

# =========================
# PDF REPORT
# =========================
def soil_quality_label(total):
    if 15 <= total <= 25:
        return "Low soil quality"
    if 26 <= total <= 37:
        return "Medium soil quality"
    if 38 <= total <= 48:
        return "High soil quality"
    return "Outside expected range"

def generate_pdf_report(state):
    tmp = tempfile.NamedTemporaryFile(delete=False, suffix=".pdf")
    doc = SimpleDocTemplate(tmp.name, pagesize=A4)
    styles = getSampleStyleSheet()
    story = []

    story.append(Paragraph("<b>Soil Quality Assessment Report</b>", styles["Title"]))
    story.append(Spacer(1, 12))
    story.append(Paragraph(f"Season: {state['season']}", styles["Normal"]))
    story.append(Paragraph(f"Location: {state['location']}", styles["Normal"]))
    story.append(Paragraph(f"Date: {datetime.date.today().isoformat()}", styles["Normal"]))
    story.append(Spacer(1, 20))

    total = 0

    for r in state["results"]:
        total += r["score"]

        story.append(Paragraph(
            f"<b>{r['id']}. {r['name']}</b>",
            styles["Heading2"]
        ))

        if r.get("image"):
            story.append(RLImage(r["image"], width=8*cm, height=6*cm))
            story.append(Spacer(1, 6))

        story.append(Paragraph(
            f"<b>Assessment:</b> {r['assessment']} ({r['score']} points)",
            styles["Normal"]
        ))

        story.append(Paragraph(
            f"<b>Explanation:</b><br/>{r['explanation']}",
            styles["Normal"]
        ))

        story.append(Paragraph(
            f"<b>Motivation:</b><br/>{r['motivation']}",
            styles["Normal"]
        ))

        story.append(Spacer(1, 12))

    story.append(PageBreak())
    story.append(Paragraph("<b>Overall Result</b>", styles["Heading1"]))
    story.append(Paragraph(f"<b>Total score:</b> {total} / 48", styles["Normal"]))
    story.append(Paragraph(
        f"<b>Soil quality classification:</b> {soil_quality_label(total)}",
        styles["Normal"]
    ))

    doc.build(story)
    tmp.close()
    return tmp.name

# =========================
# CORE LOGIC
# =========================
def run_assessment(season, location, uploaded_images, self_scores):
    if self_scores is None:
        self_scores = {}

    image_paths = save_uploaded_images(uploaded_images)
    indicators = build_indicators(season)

    results = []

    for _, row in enumerate(indicators):
        ind_id = int(row["Indicator ID"])
        name = normalize_text(row["Indicator Name"])
        data_entry = normalize_text(row["Data entry"])
        motivation = normalize_text(row["Motivation"])

        if requires_self_assessment(data_entry):
            label = self_scores.get(str(ind_id), "Medium")
            explanation = (
                "This indicator cannot be reliably assessed from photos alone. "
                "The score is based on the user's self-assessment."
            )
            img = None
        else:
            label, explanation = visual_score_placeholder(image_paths, name)
            img = image_paths[0] if image_paths else None

        results.append({
            "id": ind_id,
            "name": name,
            "assessment": label,
            "score": SCORE_MAP[label],
            "motivation": motivation,
            "explanation": explanation,
            "image": img
        })

    state = {
        "season": season,
        "location": location,
        "results": results
    }

    pdf = generate_pdf_report(state)
    return pdf

# =========================
# UI
# =========================
with gr.Blocks(title="Soil Quality Analyzer") as demo:
    gr.Markdown("# 🌱 Soil Quality Analyzer (Photo-Based)")

    season = gr.Dropdown(["Fall", "Winter", "Spring", "Summer"], value="Spring", label="Season")
    location = gr.Textbox(label="Location")

    uploads = gr.File(
        file_types=["image"],
        file_count="multiple",
        label="Upload all field photos"
    )

    gr.Markdown("### Self-assessment (only required where photos cannot assess)")
    self_scores = gr.JSON(label="Self scores (e.g. {'3':'High','6':'Low'})")

    run_btn = gr.Button("Run assessment")

    pdf_out = gr.File(label="📄 Download PDF report")

    run_btn.click(
        run_assessment,
        inputs=[season, location, uploads, self_scores],
        outputs=[pdf_out]
    )

demo.launch()