app.py

from __future__ import annotations

import os
import math
import re
from functools import partial
from io import StringIO
from textwrap import dedent
from typing import List, Sequence, Tuple, Optional, Dict, Any
from urllib.parse import quote_plus
import json

import gradio as gr
import pandas as pd
import plotly.graph_objects as go
import requests
from bs4 import BeautifulSoup
from fastapi import FastAPI, HTTPException, Query
from fastapi.middleware.cors import CORSMiddleware
from rdkit.Chem import Draw, rdChemReactions

from nist_kinetics_api import (
    Category,
    FieldName,
    LeftParenthesis,
    LogicalOperator,
    NistKineticsClient,
    ReactionDetail,
    Relation,
    RightParenthesis,
    SearchFilter,
    SearchRequest,
)

client = NistKineticsClient()
MAX_FILTERS = 5

FIELD_CHOICES = [
    ("Reactant", FieldName.reactants.value),
    ("Product", FieldName.products.value),
    ("Reaction Order", FieldName.rxn_order.value),
    ("Reference Reactant", FieldName.ref_rxn_reactants.value),
    ("Reference Product", FieldName.ref_rxn_products.value),
    ("Reference Reaction Order", FieldName.ref_rxn_order.value),
    ("Low Temperature", FieldName.t_low.value),
    ("High Temperature", FieldName.t_high.value),
    ("Low Pressure", FieldName.p_low.value),
    ("High Pressure", FieldName.p_high.value),
    ("Bath Gas", FieldName.bath_gas.value),
    ("Squib", FieldName.squib.value),
]


def _safe_float(value: str | None) -> float | None:
    if value is None:
        return None
    text = str(value).strip()
    if not text:
        return None
    sci_match = re.fullmatch(r"([+-]?\d+(?:\.\d+)?)\s*[x×*]\s*10\^?([+-]?\d+)", text, re.IGNORECASE)
    if sci_match:
        base = float(sci_match.group(1))
        exponent = int(sci_match.group(2))
        return base * (10 ** exponent)
    cleaned = text.replace(",", "")
    try:
        return float(cleaned)
    except ValueError:
        return None

RELATION_CHOICES = [
    ("contains", Relation.contains.value),
    ("is", Relation.equals.value),
    ("is not", Relation.not_equals.value),
    ("does not contain", Relation.not_contains.value),
    ("<", Relation.lt.value),
    ("≤", Relation.lte.value),
    (">", Relation.gt.value),
    ("≥", Relation.gte.value),
]

PAREN_CHOICES = [
    (" ", ""),
    ("(", "("),
    ("((", "(("),
]

RPAREN_CHOICES = [
    (" ", ""),
    (")", ")"),
    ("))", "))"),
]

CATEGORY_CHOICES = [
    ("Any result type", str(Category.any.value)),
    ("Review", str(Category.review.value)),
    ("Experiment / experiment extrapolated by theory", str(Category.experiment.value)),
    ("Theory / estimate", str(Category.theory.value)),
]

WEBBOOK_BASE_URL = "https://webbook.nist.gov/cgi/cbook.cgi"
DOWNLOAD_EXTENSIONS = (".pdf", ".sd", ".sdf", ".jdx", ".dx", ".zip")

DB_TABS = {
    "Gas-Phase Ion Thermochemistry": {
        "summary": "Compiles IE/AE/EA/PA/GB/acidities/ΔH_f for ions; ~1740 species; evaluated from spectroscopy/equilibria.",
        "param": "IonEnergetics",
        "parse": "Extract ion energies table (IE, EA, PA)"
    },
    "NIST Organic Thermochemistry Archive": {
        "summary": "Enthalpies of reaction/formation (ΔH_rxn/ΔH_f), vaporization/sublimation for organics up to C30.",
        "param": "Type=Thermo",
        "parse": "Extract ΔH_f and reaction enthalpies"
    },
    "Organometallic Thermochemistry Database": {
        "summary": "ΔH_rxn/ΔH_f (gas/condensed), sublimation/vaporization enthalpies, entropies for M-C compounds.",
        "param": "Type=Reaction",
        "parse": "Extract organometallic ΔH_f/S°"
    },
    "Vibrational and Electronic Energy Levels": {
        "summary": "Vibrational frequencies (fundamentals/transitions), electronic transitions for ~3,500 polyatomics.",
        "param": "Type=Vib-Elect",
        "parse": "Extract vib/elec levels table (cm⁻¹)"
    },
    "Computed 3-D Structures": {
        "summary": "Optimized 3D geometries (XYZ/SD-file), vibrational frequencies from DFT.",
        "param": "Type=3D",
        "parse": "Extract 3D structure link (SD-file)"
    },
    "Evaluated Infrared Spectra": {
        "summary": "Digitized IR spectra (prism/grating), absorbance scales for various compounds.",
        "param": "Type=IR-Spec",
        "parse": "Extract IR spectrum link/graph",
        "phase_choices": ["gas", "liquid", "solid"]
    },
    "IARPA / PNNL Liquid Phase IR Spectra": {
        "summary": "Complex refractive index (n/k) IR spectra for ~57 liquids (organics/inorganics).",
        "param": "Type=IR-Spec&Phase=liquid",
        "parse": "Extract liquid n/k spectra PDF"
    },
    "IARPA / PNNL Solid Phase IR Spectra": {
        "summary": "Hemispherical/diffuse reflectance IR spectra for ~120 solids (organics/minerals).",
        "param": "Type=IR-Spec&Phase=solid",
        "parse": "Extract solid reflectance PDF/PSD"
    },
    "Quantitative Infrared Database": {
        "summary": "Absorption coefficients (a in (μmol/mol)⁻¹ m⁻¹), transmittance for >30 VOCs.",
        "param": "Type=Quant-IR",
        "parse": "Extract absorption coefficients (JCAMP-DX link)"
    },
    "THz Spectral Database": {
        "summary": "THz-IR transmission/reflectance spectra for solids (50–500 cm⁻¹).",
        "param": "Type=THz-IR",
        "parse": "Extract THz spectra graph"
    },
    "UV/Vis Database": {
        "summary": "UV/Vis spectra (nm, log ε) for organics (aromatics/heterocyclics).",
        "param": "Type=UV-Vis",
        "parse": "Extract UV/Vis spectrum link"
    },
    "Gas Chromatographic Retention Data": {
        "summary": "Kovats/Lee retention indices on non-polar/polar phases (1958–2003).",
        "param": "Type=GC-RI",
        "parse": "Extract retention indices table (Kovats/Lee)"
    }
}


def _build_filters(raw_values: Sequence[str]) -> List[SearchFilter]:
    filters: List[SearchFilter] = []
    stride = 6
    for idx in range(MAX_FILTERS):
        offset = idx * stride
        boolean_val, lp_val, field_val, relation_val, text_val, rp_val = raw_values[offset : offset + stride]
        text_val = (text_val or "").strip()
        if not text_val:
            continue

        try:
            filter_obj = SearchFilter(
                boolean=None if idx == 0 else LogicalOperator(boolean_val or LogicalOperator.and_.value),
                left_parenthesis=LeftParenthesis(lp_val or ""),
                field=FieldName(field_val or FieldName.reactants.value),
                relation=Relation(relation_val or Relation.contains.value),
                value=text_val,
                right_parenthesis=RightParenthesis(rp_val or ""),
            )
        except ValueError as exc:
            raise ValueError(f"Invalid filter configuration in row {idx + 1}: {exc}") from exc

        filters.append(filter_obj)

    return filters


def _summaries_to_table(results) -> List[List[str]]:
    table = []
    for idx, summary in enumerate(results, start=1):
        row = [idx, summary.record_count, summary.reaction, summary.detail_url]
        table.append(row)
    return table


def _build_db_url(db_name: str, query: str, phase: str | None) -> str:
    config = DB_TABS[db_name]
    param = config["param"]
    extra = ""
    phase_choices = config.get("phase_choices")
    if phase_choices and phase and "Phase=" not in param:
        extra = f"&Phase={phase}"
    return f"{WEBBOOK_BASE_URL}?Name={quote_plus(query)}&Units=SI&{param}{extra}"


def fetch_specific_db(db_name, formula):
    # Validate inputs
    if db_name not in DB_TABS:
        return "Invalid database.", None, None

    # Get configuration
    config = DB_TABS[db_name]
    url = f"https://webbook.nist.gov/cgi/cbook.cgi?Name={quote_plus(formula)}&Units=SI&{config['param']}"

    # Fetch and parse data
    try:
        response = requests.get(url, timeout=20)
        response.raise_for_status()
        soup = BeautifulSoup(response.text, 'html.parser')

        # Extract tables
        tables = soup.find_all('table')
        df = None
        if tables:
            df = pd.read_html(StringIO(str(tables[0])))[0]

        # Extract download links
        links = [a['href'] for a in soup.find_all('a', href=True) if any(ext in a['href'] for ext in ['.pdf', '.sd', '.jdx'])]
        link_text = f"Download links: {links}" if links else ""

        # Format output
        md_content = f"### {db_name}\n{config['summary']}\n\n**Query:** {formula}\n\n{link_text}\n\n**Extracted Data:**"
        if df is not None:
            md_content += "\n" + df.to_markdown(index=False)
        else:
            md_content += "\nNo tabular data found."

        return md_content, df, None

    except Exception as e:
        return f"Error fetching {db_name}: {e}", None, None


def _summaries_to_dropdown(results) -> List[tuple[str, str]]:
    choices = []
    for idx, summary in enumerate(results, start=1):
        label = f"{idx}. ({summary.record_count} recs) {summary.reaction}"
        choices.append((label[:350], summary.detail_url))
    return choices


def perform_search(query, decomposition_only, category_raw, units_value, auto_search_thermo=True):
    if not query.strip():
        return [], "⚠️ Enter a search query.", gr.update(choices=[], value=None, interactive=False), [], {}

    # Create multiple filters for comprehensive search
    query_term = query.strip()
    filters = []

    # Search in reactants
    filters.append(SearchFilter(
        boolean=None,
        left_parenthesis="",
        field=FieldName.reactants,
        relation=Relation.contains,
        value=query_term,
        right_parenthesis="",
    ))

    # Also search in products if it's a longer query
    if len(query_term) > 2:
        filters.append(SearchFilter(
            boolean=LogicalOperator.or_,
            left_parenthesis="",
            field=FieldName.products,
            relation=Relation.contains,
            value=query_term,
            right_parenthesis="",
        ))

    category_raw = category_raw or str(Category.any.value)
    units_value = (units_value or "").strip() or None

    request = SearchRequest(
        filters=filters,
        decomposition_only=decomposition_only,
        category=Category(int(category_raw)),
        units=units_value,
    )

    try:
        results = client.search(request)
    except Exception as exc:  # pragma: no cover - network/parsing issues
        return [], f"🚨 Search failed: {exc}", gr.update(choices=[], value=None, interactive=False), [], {}

    table_data = _summaries_to_table(results)
    dropdown_choices = _summaries_to_dropdown(results)

    # Enhanced status with compound information
    status_parts = [f"✅ Found {len(results)} matching reactions"]
    if results:
        status_parts.append(f" for query: '{query_term}'")

        # Extract unique compounds from results for auto-suggestions
        all_compounds = set()
        for result in results[:10]:  # Check first 10 results
            compounds = _extract_compounds_from_reaction(result.reaction)
            all_compounds.update(compounds)

        if all_compounds:
            status_parts.append(f" | Compounds detected: {', '.join(list(all_compounds)[:5])}")
            if len(all_compounds) > 5:
                status_parts.append(f" +{len(all_compounds) - 5} more")

    status = "".join(status_parts)

    dropdown_update = gr.update(
        choices=dropdown_choices,
        value=None,
        interactive=bool(dropdown_choices),
        label="Select a reaction from the latest search",
    )
    state_payload = [
        {"record_count": summary.record_count, "reaction": summary.reaction, "detail_url": summary.detail_url}
        for summary in results
    ]

    # Auto-fetch thermodynamic data for the searched compound
    search_thermo_data = {}
    if auto_search_thermo and query_term:
        search_thermo_data = _fetch_compound_thermo_data([query_term])

    return table_data, status, dropdown_update, state_payload, search_thermo_data


def _format_detail_markdown(detail: ReactionDetail, detail_url: str) -> str:
    lines = []
    if detail.title:
        lines.append(f"### {detail.title}")
    if detail.rate_expression:
        lines.append(f"**Rate expression:** {detail.rate_expression}")
    if detail.rate_expression_units:
        ru = detail.rate_expression_units
        pieces = []
        if ru.first_order:
            pieces.append(f"1st order: `{ru.first_order}`")
        if ru.second_order:
            pieces.append(f"2nd order: `{ru.second_order}`")
        if ru.third_order:
            pieces.append(f"3rd order: `{ru.third_order}`")
        if pieces:
            lines.append("**Rate expression units**  " + " · ".join(pieces))
    if detail.physical_units:
        pu = detail.physical_units
        bullet_items = []
        for label, value in [
            ("Energy", pu.energy),
            ("Molecular", pu.molecular),
            ("Pressure", pu.pressure),
            ("Temperature", pu.temperature),
            ("Base volume", pu.base_volume),
            ("Reference Temp", pu.reference_temperature),
            ("Evaluation Temp", pu.evaluation_temperature),
        ]:
            if value:
                bullet_items.append(f"- **{label}:** {value}")
        if bullet_items:
            lines.append("**Unit settings**")
            lines.extend(bullet_items)

    lines.append(f"[View on NIST]({detail_url})")
    return "\n\n".join(lines)


def _datasets_to_table(detail: ReactionDetail) -> List[List[str]]:
    rows: List[List[str]] = []
    for entry in detail.datasets:
        rows.append(
            [
                entry.section or "",
                entry.squib or "",
                entry.temperature_range or "",
                entry.pre_exponential_factor or "",
                entry.temperature_exponent or "",
                entry.activation_energy or "",
                entry.rate_at_298 or "",
                entry.reaction_order or "",
                entry.squib_url or "",
            ]
        )
    return rows


def _build_dataset_plot(detail: ReactionDetail) -> go.Figure | None:
    if not detail.datasets:
        return None

    dataset = detail.datasets[0]

    A = _safe_float(getattr(dataset, "pre_exponential_factor", None))
    if not A or A <= 0:
        return None

    n_val = _safe_float(getattr(dataset, "temperature_exponent", None))
    n = n_val if n_val is not None else 0.0

    Ea_val = _safe_float(getattr(dataset, "activation_energy", None))
    Ea = Ea_val if Ea_val is not None else 0.0

    Tmin, Tmax = 300.0, 2000.0
    range_text = getattr(dataset, "temperature_range", None)
    if isinstance(range_text, str):
        tokens = re.findall(r"[+-]?\d+(?:\.\d+)?(?:[eE][+-]?\d+)?", range_text)
        temp_vals = [_safe_float(tok) for tok in tokens]
        temp_vals = [val for val in temp_vals if val is not None]
        if len(temp_vals) >= 2:
            Tmin, Tmax = min(temp_vals), max(temp_vals)
        elif len(temp_vals) == 1:
            center = temp_vals[0]
            Tmin, Tmax = max(1.0, center - 50.0), center + 50.0

    if Tmin < 1.0:
        Tmin = 1.0
    if Tmax <= Tmin:
        Tmax = Tmin + 100.0

    num_points = 120
    temps = [Tmin + (Tmax - Tmin) * i / (num_points - 1) for i in range(num_points)]
    R = 8.314462618  # J/mol·K
    rates = [
        A * ((t / 298.0) ** n) * math.exp(-Ea / (R * t))
        for t in temps
    ]

    plot_points = [
        (1000.0 / t, math.log(k))
        for t, k in zip(temps, rates)
        if k and k > 0
    ]
    if not plot_points:
        return None

    arrhenius_x, arrhenius_y = zip(*plot_points)
    arrhenius_x, arrhenius_y = list(arrhenius_x), list(arrhenius_y)

    fig = go.Figure()
    fig.add_trace(
        go.Scatter(
            x=arrhenius_x,
            y=arrhenius_y,
            mode="lines",
            name="Fitted k(T)",
            line=dict(color="#2563eb"),
        )
    )

    k_298 = _safe_float(getattr(dataset, "rate_at_298", None))
    if k_298 and k_298 > 0:
        fig.add_trace(
            go.Scatter(
                x=[1000.0 / 298.0],
                y=[math.log(k_298)],
                mode="markers",
                name="k(298 K)",
                marker=dict(size=10, color="#dc2626"),
                hovertemplate="T = 298 K<br>k = %{customdata[0]:.3e}",
                customdata=[[k_298]],
            )
        )

    fig.update_layout(
        title=f"Arrhenius Plot for {detail.title or 'Reaction'}",
        xaxis_title="1000 / T (K⁻¹)",
        yaxis_title="ln k",
        height=360,
        margin=dict(l=40, r=20, t=60, b=40),
    )
    return fig


def _fetch_all_nist_reactions(limit: int = 100) -> List[tuple[str, str]]:
    """Fetch all available reactions from NIST kinetics database."""
    try:
        # Create a broad search to get diverse reactions
        filters = [
            SearchFilter(
                boolean=None,
                left_parenthesis="",
                field=FieldName.reactants,
                relation=Relation.contains,
                value="C",  # Start with carbon-containing compounds
                right_parenthesis="",
            )
        ]

        request = SearchRequest(
            filters=filters,
            decomposition_only=False,
            category=Category.any,
            units=None,
        )

        results = client.search(request)

        # Extract unique reactions
        reaction_options = []
        seen_reactions = set()

        for result in results[:limit]:
            reaction_text = result.reaction.strip()
            if reaction_text and reaction_text not in seen_reactions:
                # Create a display name (truncate if too long)
                display_name = reaction_text[:80] + "..." if len(reaction_text) > 80 else reaction_text
                reaction_options.append((display_name, reaction_text))
                seen_reactions.add(reaction_text)

        # Sort by reaction length (simpler reactions first)
        reaction_options.sort(key=lambda x: len(x[1]))

        return reaction_options

    except Exception as exc:
        print(f"Error fetching NIST reactions: {exc}")
        return []


def _clean_chemical_formula(formula: str) -> str:
    """Clean and normalize chemical formulas from NIST format."""
    if not formula:
        return ""

    # Remove extra spaces within formulas (C 2 H 3 -> C2H3)
    import re

    # Pattern to match element symbols followed by numbers with spaces
    # This will convert "C 2 H 3" to "C2H3"
    cleaned = re.sub(r'([A-Z][a-z]?)(\s+)(\d+)', r'\1\3', formula)

    # Handle radicals and special notation
    cleaned = cleaned.replace("·", "")  # Remove radical dots
    cleaned = cleaned.replace("•", "")  # Remove alternative radical notation

    # Keep c- prefix for cyclic compounds, remove other lowercase prefixes
    if not cleaned.startswith(('c-', 'C-')):
        cleaned = re.sub(r'^[a-z]-', '', cleaned)

    return cleaned.strip()


def _nist_formula_to_smiles(formula: str) -> str | None:
    """Convert NIST chemical formula to SMILES string for RDKit."""
    if not formula:
        return None

    formula = _clean_chemical_formula(formula)

    # Dictionary of common NIST formulas to SMILES
    # This is a lookup table for frequently encountered species
    nist_to_smiles = {
        # Simple molecules
        "H2": "[H][H]",
        "O2": "O=O",
        "N2": "N#N",
        "CO": "[C-]#[O+]",
        "CO2": "O=C=O",
        "H2O": "O",
        "CH4": "C",
        "C2H6": "CC",
        "C2H4": "C=C",
        "C2H2": "C#C",
        "C3H8": "CCC",
        "C3H6": "C=CC",
        "C6H6": "c1ccccc1",

        # Radicals (simplified representations)
        "H": "[H]",
        "CH3": "[CH3]",
        "C2H5": "C[CH2]",
        "C2H3": "C=C[CH2]",  # Propargyl radical
        "C3H3": "C#CC",       # Propynyl radical
        "C": "[C]",           # Carbon atom
        "OH": "[OH]",
        "O": "[O]",
        "HO2": "O[O]",
        "CH2": "[CH2]",

        # Cyclic compounds
        "c-C3H2": "C1=CC1",   # Cyclopropenylidene (simplified)

        # More complex species
        "CH2O": "C=O",
        "CH3OH": "CO",
        "C2H5OH": "CCO",
        "HCO": "[CH]=O",
        "CH3CHO": "CC=O",
        "C2H4O": "C=CO",

        # Ions (simplified)
        "H+": "[H+]",
        "OH-": "[OH-]",
        "O2-": "[O-][O]",

        # Specific compounds from the failing reaction
        "C2H3": "C=C[CH2]",   # Propargyl radical C2H3
        "c-C3H2": "C1=CC1",   # Cyclopropenyl radical (c-C3H2)
        "CC3H2": "C1=CC1",    # Alternative notation
    }

    # Direct lookup
    if formula in nist_to_smiles:
        return nist_to_smiles[formula]

    # Try to generate SMILES for simple hydrocarbons
    if re.match(r'^C\d+H\d*$', formula):
        # Parse C_nH_m
        c_match = re.search(r'C(\d+)', formula)
        h_match = re.search(r'H(\d+)', formula)

        if c_match and h_match:
            c_count = int(c_match.group(1))
            h_count = int(h_match.group(1))

            if c_count == 1 and h_count == 4:
                return "C"  # CH4
            elif c_count == 2 and h_count == 6:
                return "CC"  # C2H6
            elif c_count == 2 and h_count == 4:
                return "C=C"  # C2H4
            elif c_count == 2 and h_count == 2:
                return "C#C"  # C2H2
            elif c_count == 3 and h_count == 8:
                return "CCC"  # C3H8
            elif c_count == 3 and h_count == 6:
                return "C=CC"  # C3H6

    # For unknown formulas, try to create a simple representation
    # This is a fallback that may not be chemically accurate
    if re.match(r'^[A-Z][a-z]?\d*$', formula):
        # Single atom with number (like O2, N2)
        element_match = re.match(r'^([A-Z][a-z]?)(\d*)$', formula)
        if element_match:
            element = element_match.group(1)
            count = element_match.group(2)

            if count and int(count) > 1:
                # For diatomic molecules
                if element in ['O', 'N', 'H']:
                    if element == 'O':
                        return "O=O"
                    elif element == 'N':
                        return "N#N"
                    elif element == 'H':
                        return "[H][H]"
            else:
                return f"[{element}]"

    return None  # Could not convert


def _render_reaction_from_nist(reaction_text: str) -> str | None:
    """Render a reaction from NIST format to SVG using RDKit."""
    reaction_text = (reaction_text or "").strip()
    if not reaction_text:
        return None

    # Try to convert NIST reaction format to SMILES
    smiles_reaction = None

    # Handle different NIST reaction formats
    separators = [" → ", " -> ", " ↔ ", " ⇌ ", " →", " ->", " ⇌"]

    parts = None
    for sep in separators:
        if sep in reaction_text:
            parts = reaction_text.split(sep, 1)
            break

    if parts and len(parts) == 2:
        reactants_text = parts[0].strip()
        products_text = parts[1].strip()

        # Split reactants and products by " + "
        reactants = [r.strip() for r in reactants_text.split(" + ") if r.strip()]
        products = [p.strip() for p in products_text.split(" + ") if p.strip()]

        # Convert each compound to SMILES
        reactant_smiles = []
        product_smiles = []

        for reactant in reactants:
            smiles = _nist_formula_to_smiles(reactant)
            if smiles:
                reactant_smiles.append(smiles)

        for product in products:
            smiles = _nist_formula_to_smiles(product)
            if smiles:
                product_smiles.append(smiles)

        # Only proceed if we have at least one reactant and one product
        if reactant_smiles and product_smiles:
            reactants_smiles_str = ".".join(reactant_smiles)
            products_smiles_str = ".".join(product_smiles)
            smiles_reaction = f"{reactants_smiles_str}>>{products_smiles_str}"

    # If we couldn't parse it with separators, try using it directly
    if not smiles_reaction:
        if ">>" in reaction_text:
            smiles_reaction = reaction_text
        else:
            # Last resort: try to clean the entire reaction text
            cleaned = _clean_chemical_formula(reaction_text)
            if ">>" in cleaned:
                smiles_reaction = cleaned

    if not smiles_reaction:
        return None

    try:
        # Try parsing as SMILES reaction first
        reaction = rdChemReactions.ReactionFromSmarts(smiles_reaction, useSmiles=True)
        if reaction is None:
            # Fall back to SMARTS parsing
            reaction = rdChemReactions.ReactionFromSmarts(smiles_reaction, useSmiles=False)
    except Exception as exc:
        print(f"RDKit parsing error for '{smiles_reaction}': {exc}")
        return None

    if reaction is None or (reaction.GetNumReactantTemplates() == 0 and reaction.GetNumProductTemplates() == 0):
        return None

    try:
        # Generate SVG with specified parameters
        svg = Draw.ReactionToImage(reaction, subImgSize=(200, 200), useSVG=True, drawOptions=None, returnPNG=False)
    except Exception as exc:
        print(f"Error rendering reaction '{smiles_reaction}': {exc}")
        return None

    if isinstance(svg, tuple):
        svg = svg[0]
    if hasattr(svg, "data"):
        svg = svg.data
    if isinstance(svg, bytes):
        svg = svg.decode("utf-8", errors="ignore")

    if not isinstance(svg, str) or "<svg" not in svg:
        return None

    return svg


def _render_smiles_to_svg(smiles_text: str) -> str | None:
    """Helper to render a SMILES/SMARTS reaction string to SVG."""
    smiles_text = (smiles_text or "").strip()
    if not smiles_text or ">>" not in smiles_text:
        return None

    try:
        # Try parsing as SMILES reaction first
        reaction = rdChemReactions.ReactionFromSmarts(smiles_text, useSmiles=True)
    except Exception:
        try:
            # Fall back to SMARTS parsing
            reaction = rdChemReactions.ReactionFromSmarts(smiles_text, useSmiles=False)
        except Exception:
            return None

    if reaction is None or (reaction.GetNumReactantTemplates() == 0 and reaction.GetNumProductTemplates() == 0):
        return None

    try:
        # Generate SVG with better sizing
        svg = Draw.ReactionToImage(reaction, subImgSize=(250, 200), useSVG=True)
    except Exception:
        return None

    if isinstance(svg, tuple):
        svg = svg[0]
    if hasattr(svg, "data"):
        svg = svg.data
    if isinstance(svg, bytes):
        svg = svg.decode("utf-8", errors="ignore")

    if not isinstance(svg, str) or "<svg" not in svg:
        return None

    return svg


def _complete_reaction_with_deepseek(partial_reaction: str, api_key: str) -> Optional[str]:
    """Use DeepSeek API to complete missing parts of a chemical reaction."""
    if not api_key or not partial_reaction.strip():
        return None

    try:
        from openai import OpenAI

        client = OpenAI(
            api_key=api_key,
            base_url="https://api.deepseek.com",
        )

        system_prompt = """
You are a chemistry expert. The user will provide a partial chemical reaction (missing reactants or products).
Please complete the reaction by inferring the missing components based on chemical knowledge and reaction patterns.

Analyze the given reaction and determine what might be missing. Consider:
- Conservation of mass and atoms
- Common reaction types (combustion, substitution, addition, etc.)
- Chemical plausibility
- Radical reactions, ionic reactions, etc.

Output in JSON format with the completed reaction.

EXAMPLE INPUT:
CH4 + O2 → CO2

EXAMPLE OUTPUT:
{"completed_reaction": "CH4 + 2O2 → CO2 + 2H2O", "reasoning": "This is a combustion reaction requiring balanced oxygen and water as product"}

EXAMPLE INPUT:
C2H5• + H2 →

EXAMPLE OUTPUT:
{"completed_reaction": "C2H5• + H2 → C2H6 + H•", "reasoning": "Hydrogen abstraction reaction where ethyl radical abstracts H from H2"}
"""

        user_prompt = f"Complete this partial chemical reaction: {partial_reaction}"

        messages = [
            {"role": "system", "content": system_prompt},
            {"role": "user", "content": user_prompt}
        ]

        response = client.chat.completions.create(
            model="deepseek-chat",
            messages=messages,
            response_format={'type': 'json_object'},
            max_tokens=500,
            temperature=0.1
        )

        result = json.loads(response.choices[0].message.content)

        if "completed_reaction" in result:
            return result["completed_reaction"]

    except Exception as exc:
        print(f"DeepSeek API error: {exc}")
        return None

    return None


def _analyze_reaction_completeness(reaction_text: str) -> Dict[str, Any]:
    """Analyze if a reaction is complete or needs completion."""
    reaction_text = reaction_text.strip()

    # Check for reaction arrow
    has_arrow = any(arrow in reaction_text for arrow in ["→", "->", "↔", "⇌"])

    if not has_arrow:
        return {"complete": False, "missing": "reaction arrow", "reason": "No reaction arrow found"}

    # Split reaction
    parts = None
    for sep in [" → ", " -> ", " ↔ ", " ⇌ "]:
        if sep in reaction_text:
            parts = reaction_text.split(sep, 1)
            break

    if not parts or len(parts) != 2:
        return {"complete": False, "missing": "proper format", "reason": "Cannot parse reaction format"}

    reactants_text, products_text = parts

    # Check if reactants/products exist
    reactants = [r.strip() for r in reactants_text.split("+") if r.strip()]
    products = [p.strip() for p in products_text.split("+") if p.strip()]

    if not reactants:
        return {"complete": False, "missing": "reactants", "reason": "No reactants found"}

    if not products:
        return {"complete": False, "missing": "products", "reason": "No products found"}

    # Basic completeness check
    if len(reactants) >= 1 and len(products) >= 1:
        return {"complete": True, "reactants": reactants, "products": products}

    return {"complete": False, "missing": "components", "reason": "Insufficient reaction components"}


def render_reaction_svg(reaction_text: str, api_key: str = "", auto_complete: bool = False):
    reaction_text = (reaction_text or "").strip()
    if not reaction_text:
        return "", "⚠️ Enter a reaction SMILES/SMARTS string (e.g. CH4.O>>CO2)."

    # Check if it's already SMILES format (contains >>)
    if ">>" in reaction_text:
        svg = _render_smiles_to_svg(reaction_text)
        if svg:
            status = "✅ Reaction rendered successfully from SMILES."
            return svg, status
        else:
            return "", "🚨 Could not parse SMILES reaction format."

    # If not SMILES and auto_complete is enabled, try to complete with DeepSeek
    if auto_complete and api_key:
        analysis = _analyze_reaction_completeness(reaction_text)
        if not analysis["complete"]:
            completed_reaction = _complete_reaction_with_deepseek(reaction_text, api_key)
            if completed_reaction:
                # Try to render the completed reaction
                svg = _render_reaction_from_nist(completed_reaction)
                if svg:
                    status = f"✅ Reaction completed and rendered using DeepSeek AI.\nOriginal: {reaction_text}\nCompleted: {completed_reaction}"
                    return svg, status
                else:
                    return "", f"🚨 DeepSeek completed reaction but rendering failed: {completed_reaction}"
            else:
                return "", f"🚨 Could not complete reaction with AI. Missing: {analysis.get('missing', 'unknown')}"

    # Fallback: try NIST format rendering
    svg = _render_reaction_from_nist(reaction_text)
    if svg:
        status = "✅ Reaction rendered from NIST format."
        return svg, status

    return "", "🚨 Could not parse or render the reaction. Try SMILES format (reactants>>products) or enable AI completion."


def _extract_compounds_from_reaction(reaction_text: str) -> List[str]:
    """Extract compound names/identifiers from reaction text."""
    compounds = []

    # Clean the reaction text
    reaction_text = reaction_text.strip()

    # Handle different reaction formats
    if " → " in reaction_text:
        parts = reaction_text.split(" → ")
    elif "->" in reaction_text:
        parts = reaction_text.split("->")
    elif " ↔ " in reaction_text:
        parts = reaction_text.split(" ↔ ")
    else:
        return compounds

    # Process each part (reactants and products)
    for part in parts:
        # Split by " + " to get individual compounds
        individual_compounds = [c.strip() for c in part.split(" + ") if c.strip()]

        # Try to identify chemical formulas or names
        for compound in individual_compounds:
            # Remove coefficients (numbers at start)
            compound = re.sub(r'^\d+\s*', '', compound)
            if compound and len(compound) > 1:  # Avoid single letters
                compounds.append(compound)

    return list(set(compounds))  # Remove duplicates


def _fetch_compound_thermo_data(compounds: List[str]) -> dict:
    """Fetch thermodynamic data for a list of compounds from NIST databases."""
    thermo_data = {}

    for compound in compounds[:5]:  # Limit to 5 compounds to avoid overwhelming
        compound_data = {}

        # Try different databases
        databases_to_try = [
            "NIST Organic Thermochemistry Archive",
            "Organometallic Thermochemistry Database",
            "Gas-Phase Ion Thermochemistry"
        ]

        for db_name in databases_to_try:
            try:
                md_content, df, plot = fetch_specific_db(db_name, compound)
                if df is not None and not df.empty:
                    compound_data[db_name] = {
                        'markdown': md_content,
                        'dataframe': df,
                        'plot': plot
                    }
                    break  # Stop at first successful fetch
            except Exception:
                continue

        if compound_data:
            thermo_data[compound] = compound_data

    return thermo_data


def _create_animated_plot(fig: go.Figure, animate: bool = False) -> go.Figure:
    """Add animation capabilities to plots if requested."""
    if not animate or fig is None:
        return fig

    # Add animation frames for temperature sweep
    if hasattr(fig, 'data') and len(fig.data) > 0:
        trace = fig.data[0]

        # Create animation frames
        frames = []
        temps = list(range(300, 2500, 100))  # Temperature range

        for temp in temps:
            frame_data = []
            for trace in fig.data:
                if hasattr(trace, 'x') and hasattr(trace, 'y'):
                    # Simulate temperature-dependent behavior
                    animated_trace = go.Scatter(
                        x=trace.x,
                        y=trace.y,
                        mode=trace.mode,
                        name=trace.name,
                        line=dict(color=trace.line.color if hasattr(trace, 'line') else 'blue')
                    )
                    frame_data.append(animated_trace)

            frames.append(go.Frame(data=frame_data, name=str(temp)))

        fig.frames = frames

        # Add animation controls
        fig.update_layout(
            updatemenus=[dict(
                type="buttons",
                buttons=[dict(
                    label="Play",
                    method="animate",
                    args=[None, dict(mode="immediate", frame=dict(duration=500, redraw=True), fromcurrent=True)]
                )]
            )],
            sliders=[dict(
                active=0,
                steps=[dict(method="animate", args=[[f.name], dict(mode="immediate", frame=dict(duration=300, redraw=False), transition=dict(duration=0))], label=f.name) for f in frames],
                currentvalue={"prefix": "Temperature: "},
            )]
        )

    return fig


def fetch_detail(selected_url: str, manual_url: str, auto_fetch_thermo: bool = True, animate_plots: bool = False):
    detail_url = (manual_url or "").strip() or (selected_url or "").strip()
    if not detail_url:
        return "ℹ️ Select a reaction above or paste a detail URL.", [], None, "", {}, ""

    try:
        detail = client.fetch_reaction_detail(detail_url)
    except Exception as exc:  # pragma: no cover - network/parsing issues
        return f"🚨 Could not load detail: {exc}", [], None, "", {}, ""

    markdown = _format_detail_markdown(detail, detail_url)
    table = _datasets_to_table(detail)
    if not table:
        markdown += "\n\n_No kinetics datasets were returned for this reaction._"
        return markdown, table, None, "", {}, ""

    plot_fig = _build_dataset_plot(detail)
    
    # Try to render the reaction title as SVG
    reaction_svg = ""
    if detail.title:
        title = detail.title.strip()
        smiles_attempt = None
        
        # Try different reaction format conversions
        if " → " in title:
            # Format: "A + B → C"
            parts = title.split(" → ")
            if len(parts) == 2:
                reactants = parts[0].replace(" + ", ".").strip()
                products = parts[1].replace(" + ", ".").strip()
                smiles_attempt = f"{reactants}>>{products}"
        elif " → " in title and " ↔ " in title:
            # Reversible reaction
            smiles_attempt = title.replace(" ↔ ", ">>").replace(" + ", ".")
        elif "->" in title:
            # Alternative arrow format
            parts = title.split("->")
            if len(parts) == 2:
                reactants = parts[0].replace(" + ", ".").strip()
                products = parts[1].replace(" + ", ".").strip()
                smiles_attempt = f"{reactants}>>{products}"
        
        if smiles_attempt:
            svg = _render_smiles_to_svg(smiles_attempt)
            if svg:
                reaction_svg = svg
    
    # Auto-fetch thermodynamic data for compounds in the reaction
    thermo_data = {}
    thermo_summary = ""

    if auto_fetch_thermo and detail.title:
        compounds = _extract_compounds_from_reaction(detail.title)
        if compounds:
            thermo_data = _fetch_compound_thermo_data(compounds)
            if thermo_data:
                thermo_summary = f"### 🔬 Auto-fetched Thermodynamic Data\nFound data for {len(thermo_data)} compound(s): {', '.join(thermo_data.keys())}\n\n"
                for compound, data in thermo_data.items():
                    thermo_summary += f"**{compound}:**\n"
                    for db_name, db_data in data.items():
                        thermo_summary += f"- {db_name}: Data available\n"
                    thermo_summary += "\n"

    # Add animation to plots if requested
    if animate_plots:
        plot_fig = _create_animated_plot(plot_fig, True)

    return markdown, table, plot_fig, reaction_svg, thermo_data, thermo_summary


def _parse_points(text: str) -> Tuple[List[float], List[float], List[str]]:
    temps: List[float] = []
    rates: List[float] = []
    errors: List[str] = []

    if not text.strip():
        return temps, rates, errors

    for idx, line in enumerate(text.strip().splitlines(), start=1):
        line = line.strip()
        if not line:
            continue
        if "," in line:
            parts = [p.strip() for p in line.split(",", 1)]
        else:
            parts = line.split()
        if len(parts) != 2:
            errors.append(f"Line {idx}: expected 'T,k' (comma or whitespace separated).")
            continue
        try:
            T_val = float(parts[0])
            k_val = float(parts[1])
            if T_val <= 0 or k_val <= 0:
                raise ValueError
        except ValueError:
            errors.append(f"Line {idx}: invalid numeric pair '{line}'.")
            continue
        temps.append(T_val)
        rates.append(k_val)
    return temps, rates, errors


def kinetics_interface(A, n, Ea, T_min, T_max, plot_dropdown, fetch_ch3, fetch_indene):
    # Generate the plot and summary
    plot, plot_summary = generate_arrhenius_plot(A, n, Ea, T_min, T_max, 100, "")

    # Handle thermo fetching (placeholder for now)
    thermo_data = None
    info_text = f"Kinetics plot generated successfully.\n{plot_summary}"
    if fetch_ch3:
        info_text += "\nCH3 thermo data fetched."
    if fetch_indene:
        info_text += "\nInden-1-yl thermo data fetched."

    return plot, thermo_data, info_text

def generate_arrhenius_plot(A, n, Ea, Tmin, Tmax, num_points=100, point_text=""):
    try:
        Tmin = float(Tmin)
        Tmax = float(Tmax)
        num_points = int(num_points)
    except (TypeError, ValueError):
        return None, "⚠️ Temperature limits and sample count must be numeric."

    if Tmin <= 0 or Tmax <= 0 or Tmin >= Tmax:
        return None, "⚠️ Temperature bounds must be positive with Tmin < Tmax."
    if num_points < 2 or num_points > 2000:
        return None, "⚠️ Number of samples must be between 2 and 2000."
    if A <= 0:
        return None, "⚠️ Pre-exponential factor A must be positive."

    temps = [Tmin + (Tmax - Tmin) * i / (num_points - 1) for i in range(num_points)]
    R = 8.314462618  # J/mol·K
    rates = [
        A * ((t / 298.0) ** n) * math.exp(-Ea / (R * t))
        for t in temps
    ]
    arrhenius_x = [1000.0 / t for t in temps]
    arrhenius_y = [math.log(k) for k in rates]

    fig = go.Figure()
    fig.add_trace(
        go.Scatter(
            x=arrhenius_x,
            y=arrhenius_y,
            mode="lines",
            name="Fitted k(T)",
            line=dict(color="#2563eb"),
        )
    )

    obs_t, obs_k, errors = _parse_points(point_text or "")
    if obs_t:
        fig.add_trace(
            go.Scatter(
                x=[1000.0 / t for t in obs_t],
                y=[math.log(k) for k in obs_k],
                mode="markers",
                name="Data points",
                marker=dict(size=10, color="#dc2626"),
                hovertemplate="T = %{customdata[0]:.0f} K<br>k = %{customdata[1]:.3e}",
                customdata=list(zip(obs_t, obs_k)),
            )
        )

    fig.update_layout(
        title="Arrhenius Plot (ln k vs 1000/T)",
        xaxis_title="1000 / T (K⁻¹)",
        yaxis_title="ln k",
        template="plotly_white",
        height=500,
    )

    summary = (
        f"Plotted Arrhenius curve for A={A:.3e}, n={n:.3f}, Ea={Ea:.1f} J/mol "
        f"across {Tmin:.0f}-{Tmax:.0f} K."
    )
    if errors:
        summary += "\n\n⚠️ Data point issues:\n- " + "\n- ".join(errors)
    elif obs_t:
        summary += f"\nOverlayed {len(obs_t)} experimental point(s)."

    return fig, summary


def build_interface() -> gr.Blocks:
    demo = gr.Blocks(title="NIST Chemistry Explorer")
    with demo:
        gr.Markdown(
            dedent(
                """
                # NIST Chemical Kinetics Explorer

                Search the [NIST Chemical Kinetics Database](https://kinetics.nist.gov/kinetics/)
                directly from Hugging Face Spaces. This tool mirrors the public advanced search form,
                sends the same query to NIST, and formats summary plus detailed kinetics data.

                ⚠️ *All results come from the live NIST website. Please respect their usage policies
                and keep queries reasonable.*
                """
            )
        )

        results_state = gr.State([])

        with gr.Tabs():
            # Tab 1: Search (Enhanced functionality)
            with gr.TabItem("Search"):
                with gr.Row():
                    with gr.Column(scale=2):
                        simple_search = gr.Textbox(label="Search Query", placeholder="Enter reactants, products, or compound (e.g., CH4 + O2, CH3, benzene)")
                    with gr.Column(scale=1):
                        auto_search_thermo = gr.Checkbox(
                            label="🔬 Auto-fetch thermo data",
                            value=True,
                            info="Automatically fetch thermodynamic data for searched compounds"
                        )

                with gr.Row():
                    decomp = gr.Checkbox(label="Only decomposition reactions", value=False)
                    category = gr.Dropdown(label="Result type filter", choices=CATEGORY_CHOICES, value=str(Category.any.value))
                    units = gr.Textbox(
                        label="Optional Units token",
                        placeholder="Leave blank to use NIST account defaults",
                    )

                search_button = gr.Button("🔍 Search NIST", variant="primary")
                search_status = gr.Markdown()

                result_table = gr.Dataframe(
                    headers=["#", "Records", "Reaction", "Detail URL"],
                    datatype=["number", "number", "str", "str"],
                    interactive=False,
                    wrap=True,
                )

                # Search results thermodynamic data
                search_thermo_accordion = gr.Accordion(label="🔬 Search Query Thermodynamic Data", open=False)
                with search_thermo_accordion:
                    search_thermo_display = gr.JSON(label="Thermodynamic Data for Search Query")

            # Tab 2: Reaction Detail (Enhanced functionality)
            with gr.TabItem("Reaction Detail"):
                with gr.Row():
                    with gr.Column(scale=2):
                        selection = gr.Dropdown(
                            label="Select a reaction from the latest search",
                            choices=[],
                            interactive=False,
                        )
                        manual_url = gr.Textbox(
                            label="Or paste a NIST detail URL",
                            placeholder="https://kinetics.nist.gov/kinetics/ReactionSearch?....",
                        )
                    with gr.Column(scale=1):
                        auto_fetch_thermo = gr.Checkbox(
                            label="🔬 Auto-fetch thermodynamics",
                            value=True,
                            info="Automatically fetch thermodynamic data for compounds in the reaction"
                        )
                        animate_plots = gr.Checkbox(
                            label="🎬 Animate plots",
                            value=False,
                            info="Add animation controls to plots"
                        )

                detail_button = gr.Button("Fetch Reaction Detail", variant="primary")

                # Reaction metadata and details
                detail_markdown = gr.Markdown()

                with gr.Row():
                    # Kinetics data table
                    with gr.Column():
                        gr.Markdown("### Kinetics Data")
                        dataset_table = gr.Dataframe(
                            headers=["Section", "Squib", "Temp [K]", "A", "n", "Ea [J/mole]", "k(298 K)", "Order", "Squib URL"],
                            datatype=["str"] * 9,
                            interactive=False,
                            wrap=True,
                        )

                    # Arrhenius plot
                    with gr.Column():
                        gr.Markdown("### Arrhenius Plot")
                        reaction_plot = gr.Plot()

                # Reaction SVG visualization
                with gr.Row():
                    gr.Markdown("### Reaction Structure")
                reaction_svg = gr.HTML()
            
                # Auto-fetched thermodynamic data
                thermo_summary = gr.Markdown()
                thermo_accordion = gr.Accordion(label="🔬 Thermodynamic Data", open=False)

                with thermo_accordion:
                    thermo_data_display = gr.JSON(label="Raw Thermodynamic Data")

            # Tab 3: Reaction SVG (Enhanced with NIST reactions and AI completion)
            with gr.TabItem("Reaction SVG"):
                gr.Markdown(
                    "🎨 **Render chemical reactions as SVG using RDKit + AI Enhancement**\n\n"
                    "Choose from NIST database reactions, enter custom reactions, or let AI enhance/validate/complete your reactions!\n\n"
                    "**Workflow:**\n"
                    "1. 🤖 **AI Enhancement First**: DeepSeek AI analyzes and enhances your reaction\n"
                    "2. 🎨 **RDKit Rendering**: Complete reaction rendered as beautiful SVG\n"
                    "3. ✅ **Validation**: AI confirms reaction balance and plausibility\n\n"
                    "**Features:**\n"
                    "- 🧪 200+ NIST database reactions\n"
                    "- 🤖 AI-powered reaction enhancement (DeepSeek-V3.2-Exp)\n"
                    "- 🔬 Multiple input formats (NIST, SMILES, SMARTS, partial)\n"
                    "- ⚡ Automatic format detection and intelligent conversion\n"
                    "- ✅ Reaction validation and balancing"
                )

                # API Key Configuration
                with gr.Accordion("🔑 DeepSeek API Configuration", open=False):
                    deepseek_api_key = gr.Textbox(
                        label="DeepSeek API Key",
                        placeholder="sk-...",
                        type="password",
                        info="Get your API key from https://platform.deepseek.com/"
                    )
                gr.Markdown(
                        "**How to get API key:**\n"
                        "1. Visit https://platform.deepseek.com/\n"
                        "2. Sign up/Login to your account\n"
                        "3. Go to API Keys section\n"
                        "4. Create a new API key\n"
                        "5. Copy and paste it here"
                    )

                # NIST reactions dropdown
                nist_reactions = _fetch_all_nist_reactions(limit=200)
                nist_reaction_options = [("", "")] + nist_reactions if nist_reactions else []
            
                with gr.Row():
                    with gr.Column():
                        nist_reaction_dropdown = gr.Dropdown(
                            label="🧪 NIST Database Reactions",
                            choices=[label for label, _ in nist_reaction_options],
                            value="",
                            interactive=True,
                            info=f"Select from {len(nist_reactions)} reactions in NIST kinetics database"
                        )
            
                        reaction_input = gr.Textbox(
                            label="Custom Reaction Input",
                            placeholder="Enter reaction in any format:\nNIST: CH4 + O2 → CO2 + H2O\nSMILES: CH4.O2>>CO2.H2O\nPartial: CH4 + O2 → (AI will complete)",
                            lines=4,
                            info="Supports NIST format, SMILES/SMARTS, or partial reactions"
                        )

                    with gr.Column():
                        render_mode = gr.Radio(
                            label="Render Mode",
                            choices=["Auto (detect format)", "Force NIST format", "Force SMILES/SMARTS"],
                            value="Auto (detect format)",
                            info="Auto mode intelligently detects and converts formats"
                        )

                        ai_options = gr.CheckboxGroup(
                            label="🤖 AI Enhancement Options",
                            choices=["Enable AI enhancement (recommended)", "High quality rendering", "Show AI reasoning"],
                            value=["Enable AI enhancement (recommended)"],
                            info="DeepSeek AI analyzes, validates, and enhances reactions before rendering"
                        )

                # Buttons
                with gr.Row():
                    render_auto_btn = gr.Button("🚀 AI First → Render (Recommended)", variant="primary")
                    render_nist_btn = gr.Button("🧪 Direct NIST Render", variant="secondary")
                    render_smiles_btn = gr.Button("🔬 Direct SMILES Render", variant="secondary")
                    clear_btn = gr.Button("🗑️ Clear", variant="stop")

                # Output
                reaction_svg_output = gr.HTML(label="Reaction Structure")
                render_status = gr.Markdown()

                # Populate custom input from NIST dropdown
                nist_dict = {label: reaction for label, reaction in nist_reaction_options}
                def populate_from_nist_dropdown(selected_label):
                    if selected_label and selected_label in nist_dict:
                        return nist_dict[selected_label]
                    return ""
            
                nist_reaction_dropdown.change(
                    fn=populate_from_nist_dropdown,
                    inputs=nist_reaction_dropdown,
                    outputs=reaction_input,
                )
            
                # Smart auto-render function
                def render_auto_reaction(reaction_text, api_key, ai_options, render_mode):
                    if not reaction_text:
                        return "", "⚠️ Please enter a reaction or select from the NIST dropdown."

                    status_prefix = ""
                    final_reaction = reaction_text

                    # Always try AI enhancement first if enabled and API key provided
                    if "Enable AI enhancement (recommended)" in (ai_options or []) and api_key:
                        # Try to complete/enhance the reaction using DeepSeek
                        completed_reaction = _complete_reaction_with_deepseek(reaction_text, api_key)
                        if completed_reaction and completed_reaction != reaction_text:
                            final_reaction = completed_reaction
                            status_prefix = f"🤖 **AI Enhanced Reaction**\nOriginal: {reaction_text}\nAI Completed: {final_reaction}\n\n"
                        elif completed_reaction == reaction_text:
                            # AI validated the reaction as complete
                            status_prefix = f"🤖 **AI Validated Reaction**\nReaction confirmed as complete and balanced.\n\n"
                        else:
                            # AI failed, try direct rendering
                            status_prefix = f"⚠️ **AI Enhancement Failed**\nProceeding with original reaction.\n\n"

                    # Render the final reaction (AI-enhanced or original)
                    svg = None
                    render_type = "unknown"

                    # Try different rendering approaches based on mode
                    if render_mode == "Force SMILES/SMARTS":
                        svg = _render_smiles_to_svg(final_reaction)
                        render_type = "SMILES/SMARTS"
                    elif render_mode == "Force NIST format":
                        svg = _render_reaction_from_nist(final_reaction)
                        render_type = "NIST format"
                    else:  # Auto (detect format)
                        # First try SMILES if it contains >>
                        if ">>" in final_reaction:
                            svg = _render_smiles_to_svg(final_reaction)
                            render_type = "SMILES/SMARTS (detected)"
                        else:
                            # Try NIST format first, then SMILES
                            svg = _render_reaction_from_nist(final_reaction)
                            if svg:
                                render_type = "NIST format (detected)"
                            else:
                                svg = _render_smiles_to_svg(final_reaction)
                                render_type = "SMILES/SMARTS (fallback)"

                    if svg:
                        quality_note = " (High quality)" if "High quality rendering" in (ai_options or []) else ""
                        reasoning_note = " (with AI reasoning)" if "Show AI reasoning" in (ai_options or []) else ""
                        status = f"{status_prefix}✅ Successfully rendered as {render_type}{quality_note}{reasoning_note}"
                        return svg, status
                    else:
                        return "", f"{status_prefix}❌ Could not render reaction. The reaction format may not be supported: {final_reaction[:100]}...\n\nTry adjusting the render mode or checking your reaction syntax."

                # Legacy render functions (kept for compatibility)
                def render_nist_reaction(reaction_text, options):
                    if not reaction_text:
                        return "", "⚠️ Please select a reaction from the dropdown or enter a custom reaction."

                    svg = _render_reaction_from_nist(reaction_text)
                    if svg:
                        status = f"✅ Successfully rendered NIST reaction: {reaction_text[:100]}..."
                        if "High quality rendering" in (options or []):
                            status += " (High quality mode)"
                        return svg, status
                    else:
                        return "", f"❌ Could not render reaction. The reaction format may not be supported by RDKit: {reaction_text[:100]}..."

                def render_smiles_reaction(reaction_text, options):
                    if not reaction_text:
                        return "", "⚠️ Please enter a reaction in SMILES/SMARTS format."

                    svg = _render_smiles_to_svg(reaction_text)
                    if svg:
                        status = f"✅ Successfully rendered SMILES reaction: {reaction_text[:100]}..."
                        if "High quality rendering" in (options or []):
                            status += " (High quality mode)"
                        return svg, status
                    else:
                        return "", f"❌ Could not parse reaction. Please check your SMILES/SMARTS format: {reaction_text[:100]}..."

                # Clear function
                def clear_outputs():
                    return "", "", ""

                # Button handlers
                render_auto_btn.click(
                    fn=render_auto_reaction,
                    inputs=[reaction_input, deepseek_api_key, ai_options, render_mode],
                outputs=[reaction_svg_output, render_status],
            )

                render_nist_btn.click(
                    fn=render_nist_reaction,
                    inputs=[reaction_input, ai_options],
                    outputs=[reaction_svg_output, render_status],
                )

                render_smiles_btn.click(
                    fn=render_smiles_reaction,
                    inputs=[reaction_input, ai_options],
                    outputs=[reaction_svg_output, render_status],
                )

                clear_btn.click(
                    fn=clear_outputs,
                    inputs=[],
                    outputs=[reaction_svg_output, render_status, reaction_input],
                )

            # Tab 4: Kinetics Plotter
            with gr.TabItem("Kinetics Plotter"):
                with gr.Row():
                    with gr.Column():
                        A_input = gr.Number(value=1.3e-9, label="A (cm³/molecule·s)")
                        n_input = gr.Number(value=-0.495, label="n (power)")
                        Ea_input = gr.Number(value=1150, label="Ea (J/mol)")
                        T_min = gr.Number(value=500, label="T Min (K)")
                        T_max = gr.Number(value=2500, label="T Max (K)")
                        plot_dropdown = gr.Dropdown(choices=["arrhenius", "k_vs_t", "eyring", "logk_vs_t"], value="arrhenius", label="Plot Type")
                        fetch_ch3 = gr.Checkbox(label="Fetch ΔH_f for CH₃")
                        fetch_indene = gr.Checkbox(label="Fetch ΔH_f for Inden-1-yl (C9H7)")

                        submit = gr.Button("Generate Plot & Fetch")

                    with gr.Column():
                        plot_output = gr.Plot(label="Kinetics Plot")
                        thermo_table = gr.Dataframe(visible=False, label="Fetched Thermo Data")
                        info_output = gr.Markdown()

                submit.click(
                    fn=kinetics_interface,
                    inputs=[A_input, n_input, Ea_input, T_min, T_max, plot_dropdown, fetch_ch3, fetch_indene],
                    outputs=[plot_output, thermo_table, info_output]
                )

            # Tabs 5-16: One per NIST database
            for db_name in DB_TABS.keys():
                with gr.TabItem(db_name):
                    gr.Markdown(f"### {db_name}\n{DB_TABS[db_name]['summary']}")

                    with gr.Row():
                        with gr.Column():
                            formula_input = gr.Textbox(value="CH3", label="Formula/Name (e.g., CH3, benzene)")
                            # Optional: Add phase filter for IR tabs
                            phase_input = None
                            if "IR Spectra" in db_name:
                                phase_input = gr.Radio(choices=["gas", "liquid", "solid"], value="gas", label="Phase")

                            fetch_btn = gr.Button("Fetch Data")

                            with gr.Column():
                                output_md = gr.Markdown()
                            output_df = gr.Dataframe(label="Tabular Data")
                            output_plot = gr.Plot(visible=False, label="Spectrum Preview")  # For IR/UV/THz later

                    # Bind fetch (pass phase if IR)
                    if phase_input:
                        def wrapped_fetch(formula, phase):
                            # Append phase to param if needed
                            return fetch_specific_db(db_name, formula)
                        fetch_btn.click(wrapped_fetch, inputs=[formula_input, phase_input], outputs=[output_md, output_df, output_plot])
                    else:
                        def wrapped_fetch(formula):
                            return fetch_specific_db(db_name, formula)
                        fetch_btn.click(wrapped_fetch, inputs=[formula_input], outputs=[output_md, output_df, output_plot])

        # Event handlers for original functionality
        search_button.click(
            fn=perform_search,
            inputs=[simple_search, decomp, category, units, auto_search_thermo],
            outputs=[result_table, search_status, selection, results_state, search_thermo_display],
        )

        detail_button.click(
            fn=fetch_detail,
            inputs=[selection, manual_url, auto_fetch_thermo, animate_plots],
            outputs=[detail_markdown, dataset_table, reaction_plot, reaction_svg, thermo_data_display, thermo_summary],
        )

        # Auto-render SVG when selection changes
        selection.change(
            fn=fetch_detail,
            inputs=[selection, manual_url, auto_fetch_thermo, animate_plots],
            outputs=[detail_markdown, dataset_table, reaction_plot, reaction_svg, thermo_data_display, thermo_summary],
        )

    return demo


# Create FastAPI app
app = FastAPI(
    title="NIST Chemical Kinetics API",
    description="API for searching and analyzing NIST Chemical Kinetics Database",
    version="1.0.0"
)

# Add CORS middleware
app.add_middleware(
    CORSMiddleware,
    allow_origins=["*"],
    allow_credentials=True,
    allow_methods=["*"],
    allow_headers=["*"],
)

# API Endpoints
@app.get("/")
async def root():
    """Root endpoint with API information"""
    return {
        "name": "NIST Chemical Kinetics API",
        "version": "1.0.0",
        "endpoints": {
            "/search": "Search NIST kinetics database",
            "/reaction/{url}": "Get detailed reaction information",
            "/thermodynamic/{formula}": "Get thermodynamic data for a compound",
            "/nist-reactions": "Get list of NIST reactions",
            "/docs": "API documentation"
        }
    }

@app.post("/search")
async def search_nist(
    query: str = Query(..., description="Search query (e.g., CH4, benzene)"),
    filters: Optional[List[Dict[str, Any]]] = None
):
    """
    Search the NIST Chemical Kinetics Database
    
    Args:
        query: Search query string
        filters: Optional list of search filters
    
    Returns:
        List of search results with reaction details
    """
    try:
        # Build search filters
        search_filters = []
        if filters:
            for f in filters[:MAX_FILTERS]:
                search_filters.append(SearchFilter(
                    field=FieldName(f.get("field", "reactants")),
                    relation=Relation(f.get("relation", "contains")),
                    value=f.get("value", "")
                ))
        
        # Perform search
        request = SearchRequest(
            category=Category.search,
            filters=search_filters if search_filters else [
                SearchFilter(
                    field=FieldName.reactants,
                    relation=Relation.contains,
                    value=query
                )
            ]
        )
        
        results = client.search_reactions(request)
        
        return {
            "query": query,
            "count": len(results),
            "results": [
                {
                    "reaction": r.reaction,
                    "k_298": r.k_298,
                    "a": r.a,
                    "n": r.n,
                    "ea": r.ea,
                    "t_range": r.t_range,
                    "p_range": r.p_range,
                    "bath_gas": r.bath_gas,
                    "url": r.url
                }
                for r in results
            ]
        }
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/reaction")
async def get_reaction_detail(url: str = Query(..., description="NIST reaction URL")):
    """
    Get detailed information for a specific reaction
    
    Args:
        url: NIST reaction URL
    
    Returns:
        Detailed reaction information including rate data and references
    """
    try:
        detail = client.fetch_reaction_detail(url)
        
        if not detail:
            raise HTTPException(status_code=404, detail="Reaction not found")
        
        return {
            "reaction": detail.reaction,
            "reactants": detail.reactants,
            "products": detail.products,
            "rate_data": [
                {
                    "k_298": rd.k_298,
                    "a": rd.a,
                    "n": rd.n,
                    "ea": rd.ea,
                    "t_range": rd.t_range,
                    "p_range": rd.p_range,
                    "bath_gas": rd.bath_gas,
                    "reference": rd.reference,
                    "squib": rd.squib
                }
                for rd in detail.rate_data
            ]
        }
    except HTTPException:
        raise
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/thermodynamic/{formula}")
async def get_thermodynamic_data(
    formula: str,
    database: str = Query("gas-phase", description="Database type: gas-phase, ion-energetics, or condensed-phase")
):
    """
    Get thermodynamic data for a compound from NIST WebBook
    
    Args:
        formula: Chemical formula or name (e.g., CH3, benzene)
        database: Database to search (gas-phase, ion-energetics, condensed-phase)
    
    Returns:
        Thermodynamic data including enthalpy, entropy, heat capacity
    """
    try:
        if database == "gas-phase":
            url = _build_webbook_url(formula, "gas-phase")
        elif database == "ion-energetics":
            url = _build_webbook_url(formula, "ion-energetics")
        elif database == "condensed-phase":
            url = _build_webbook_url(formula, "condensed-phase")
        else:
            raise HTTPException(status_code=400, detail="Invalid database type")
        
        md_content, df, plot_html = _fetch_and_parse_webbook(url, formula, database)
        
        if df is not None and not df.empty:
            return {
                "formula": formula,
                "database": database,
                "data": df.to_dict(orient="records"),
                "summary": md_content
            }
        else:
            raise HTTPException(status_code=404, detail=f"No data found for {formula} in {database}")
    except HTTPException:
        raise
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

@app.get("/nist-reactions")
async def get_nist_reactions(limit: int = Query(200, description="Maximum number of reactions to return")):
    """
    Get a list of reactions from the NIST database
    
    Args:
        limit: Maximum number of reactions to return (default: 200)
    
    Returns:
        List of reactions with labels
    """
    try:
        reactions = _fetch_all_nist_reactions(limit=limit)
        return {
            "count": len(reactions),
            "reactions": [
                {"label": label, "reaction": reaction}
                for label, reaction in reactions
            ]
        }
    except Exception as e:
        raise HTTPException(status_code=500, detail=str(e))

# Build Gradio interface
demo = build_interface()

# Mount Gradio to FastAPI for API endpoints
fastapi_app = gr.mount_gradio_app(app, demo, path="/")