#! /usr/bin/env python3
#
# Copyright 2009-2021 The VOTCA Development Team (http://www.votca.org)
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
#     http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
#

import argparse
import importlib
import multiprocessing
import os
import shutil
import subprocess
import xml.etree.ElementTree as ET
from collections import defaultdict
from pathlib import Path
from typing import Any, Dict, List, NamedTuple
from xml.dom import minidom
from rdkit import Chem
import rdkit.Chem.Descriptors
import numpy as np

try:
    importlib.import_module("rdkit")
except ModuleNotFoundError:
    exc = ModuleNotFoundError(
        """'xtp_autogen_mapping' requires the 'rdkit' package: https://anaconda.org/conda-forge/rdkit"""
    )
    exc.__cause__ = None
    raise exc
from rdkit import Chem


VERSION = '@PROJECT_VERSION@ #VOTCA_GIT_ID#'
PROGTITLE = 'THE VOTCA::XTP auto mapping generator'
VOTCAHEADER = f"""
==================================================
========   VOTCA (http://www.votca.org)   ========
==================================================

{PROGTITLE}

please read and cite: @PROJECT_CITATION@
and submit bugs to @PROJECT_CONTACT@
votca_autogen_mapping, version {VERSION}
"""
DESCRIPTION = f"""{VOTCAHEADER}
Automatically generate a mapping file"""

# Hartree to eV
H2EV = 27.21138


class JobMetadata(NamedTuple):
    """Create a namedtuple with the data to call Orca."""
    name: str
    path_orca: str
    functional: str
    basis: str
    RI: bool
    threads: int


def search_cores() -> int:
    """Try to find the number of cores in the system."""
    cpuinfo = Path("/proc/cpuinfo")
    if not cpuinfo.exists():
        return multiprocessing.cpu_count()

    with open(cpuinfo, 'r') as handler:
        lines = handler.readlines()

    # Number of available CPUs
    ncpus = len(list(filter(lambda x: "processor" in x, lines)))

    # Look for hyperthreading
    flags = next(filter(lambda x: "flags" in x, lines), None)
    # ht is the flag for hyperthreading
    if flags is not None and 'ht' in flags:
        ncpus = ncpus / 2

    return ncpus


def run_cmd(cmd: str):
    """Run a shell command."""
    print("running: ", cmd)
    result = subprocess.run(cmd, stdout=subprocess.PIPE,
                            stderr=subprocess.PIPE, shell=True)
    if result.stderr:
        print("command error: ", result.stderr)


def write_input(
        input_file_name: str, crg_spin_coord: str, meta: JobMetadata, optimize: bool = True):
    """Write Orca input file."""
    # number of cores to use
    if meta.threads == "":
        ncores = search_cores()
    else:
        ncores = meta.threads
    
    # geometry optimization starts from MD structure for n, and n structure of e/h
    ri_string = ' RIJCOSX' if meta.RI else ""
    opt = "opt" if optimize else ""
    chelpg = "! CHELPG" if optimize else ""

    # Polar calculation
    polar = "% elprop\npolar 1\nSolver CG\nend\n" if optimize else ""

    # Input file
    inp = f"""{crg_spin_coord}

%pal
nprocs {ncores}
end

! DFT {opt} {meta.functional} {meta.basis} {ri_string} SlowConv
! D3BJ
{chelpg}

{polar}
"""
    with open(input_file_name, "w") as handler:
        handler.write(inp)


def read_energies(logfile_name: str) -> List[float]:
    """parse SINGLE POINT ENERGIES."""
    energies = []
    with open(logfile_name, 'r') as logfile:
        for line in logfile.readlines():
            if 'FINAL SINGLE POINT ENERGY' in line:
                energies.append(float(line.split()[-1]))

    return energies


def optimize_geometry(state: str, meta: JobMetadata) -> List[float]:
    """Perform a geometry optimization using Orca."""
    root = f"{meta.name}_{state}"
    input_file_name = f"{root}.inp"
    logfile_name = f"{root}.log"
    errfile_name = f"{root}.err"

    if state == 'n':
        crg_spin_coord = f'* xyzfile 0 1 {meta.name}_MD.xyz'
    elif state == 'e':
        crg_spin_coord = f'* xyzfile -1 2 {meta.name}_n.xyz'
    elif state == 'h':
        crg_spin_coord = f'* xyzfile 1 2 {meta.name}_n.xyz'

    # Write optimization Orca input
    write_input(input_file_name, crg_spin_coord, meta)

    # Call orca
    orca_command = f"{meta.path_orca} {input_file_name} > {logfile_name} 2> {errfile_name}"
    run_cmd(orca_command)

    return read_energies(logfile_name)


def run_single_point(state: str, meta: JobMetadata):
    """Run a single point calculation with Orca."""
    crg_spin_coord = f'* xyzfile 0 1 {meta.name}_{state}.xyz'
    root = f"{meta.name}_{state}"

    input_n_file_name = f"{root}_n.inp"
    write_input(input_n_file_name, crg_spin_coord, meta, optimize=False)
    # run ORCA again
    logfile_n_name = f"{root}_n.log"
    errfile_n_name = f"{root}_n.err"
    orca_command2 = f"{meta.path_orca} {input_n_file_name} > {logfile_n_name} 2> {errfile_n_name}"
    run_cmd(orca_command2)

    # parse SINGLE POINT ENERGIES
    return read_energies(logfile_n_name)


def process_charges_and_polarization(root: str) -> None:
    """Read charges and compute polarization."""
    # parse CHELPG charges
    logfile_name = f"{root}_log2mps.log"
    errfile_name = f"{root}_log2mps.err"
    log2mps_command = f"xtp_tools -e log2mps -c job_name=\"{root}\" > {logfile_name} 2> {errfile_name}"
    run_cmd(log2mps_command)

    # prep molpol.xml
    options = ET.Element('options')
    molpol = ET.SubElement(options, 'molpol')
    mode = ET.SubElement(molpol, 'mode')
    mode.text = 'qmpackage'
    logname = ET.SubElement(molpol, 'logfile')
    logname.text = f"{root}.log"
    molpol_xml_str = minidom.parseString(
        ET.tostring(options)).toprettyxml(indent="   ")
    with open("molpol.xml", "w") as f:
        f.write(molpol_xml_str)
    # run molpol
    logfile_name = f"{root}_molpol.log"
    errfile_name = f"{root}_molpol.err"
    molpol_command = f"xtp_tools -e molpol -o molpol.xml -c job_name=\"{root}\" > {logfile_name} 2> {errfile_name}"
    os.system(molpol_command)
    # copy optimized MPS_FILE
    os.makedirs('../MP_FILES', exist_ok=True)
    mpsfile = f"{root}_polar.mps"
    mpsfile_store = f"../MP_FILES/{mpsfile}"
    shutil.copy(mpsfile, mpsfile_store)


def optimize(state: str, meta: JobMetadata):
    """Optimize geometry, get CHELPG charges, polarizability tensor."""
    root = f"{meta.name}_{state}"
    energies = optimize_geometry(state, meta)
    E_initial = energies[0]
    E_final = energies[-1]

    # copy optimized geometries
    os.makedirs('../QC_FILES', exist_ok=True)
    geofile = f"{root}.xyz"
    geofile_store = f'../QC_FILES/{geofile}'
    shutil.copy(geofile, geofile_store)

    process_charges_and_polarization(root)

    # if state is h or e, calculate a single energy for the neural molecule in the optimized geometry
    if state != 'n':
        energies = run_single_point(state, meta)
        E_cross = energies[0]
    else:
        E_cross = 0.0

    names = [f"E_{state}_{kind}" for kind in ("init", "final", "cross")]

    return {name: value for name, value in zip(names, (E_initial, E_final, E_cross))}


def optimize_geometry_in_state(
        mol: Chem.rdchem.Mol, states: List[str], meta: JobMetadata) -> Dict[str, float]:
    """Optimize the molecular geometry in the given staten."""

    os.makedirs('generate', exist_ok=True)
    basedir = os.getcwd()
    os.chdir('generate')

    # read the initial PDB file
    mdXYZ_file_name = f'{meta.name}_MD.xyz'
    Chem.MolToXYZFile(mol, mdXYZ_file_name)
    # neutral
    energies = optimize('n', meta)
    # cation, if requested
    if 'h' in states:
        energies_h = optimize('h', meta)
        energies.update(energies_h)

    # anion, if requested
    if 'e' in states:
        energies_e = optimize('e', meta)
        energies.update(energies_e)
    os.chdir(basedir)

    return energies


def generate_mapping(args: argparse.Namespace) -> None:
    """Generate the mapping files."""

    # Read  the molecule
    mol = Chem.MolFromPDBFile(args.pdbfile, removeHs=False)
    seg_name = Path(args.pdbfile).stem
    if not args.mdname:
        args.mdname = Chem.rdMolDescriptors.CalcMolFormula(mol)
        msg = """No *mdname* (-m, --mdname) has been specified, setting it to the molecular
formula. If you used GROMACS, replace the entry in the generated mapping.xml with the name in
topol.top (or similar).

"""
        print(msg)

    if args.optimize:
        segment_name = Path(args.pdbfile).stem
        meta = JobMetadata(
            name=segment_name, path_orca=args.orca, functional=args.functional,
            basis=args.basis, RI=args.nori,threads=args.threads)
        energies = optimize_geometry_in_state(mol, args.states, meta)
    else:
        # The default energy for any key is 0
        energies = defaultdict(lambda: 0.0)
        msg = """WARNING **The user didn't request an optimization**, therefore:
1.  `qmcoords` and `multipoles` tags are going to be added to the mapping file but
    the *xyz and *mps files, and the QC_FILES and MP_FILES directories
    are not automatically generated, you must add them yourself.
2.  Excitation and reorganization energies are all set to 0. You need to change
    those values.
"""
        print(msg)

    # determine the rotatable bonds
    rotable_bond = Chem.MolFromSmarts('[!$(*#*)&!D1]-&!@[!$(*#*)&!D1]')
    rbonds = mol.GetSubstructMatches(rotable_bond)
    bonds = [mol.GetBondBetweenAtoms(x, y).GetIdx() for x, y in rbonds]
    mol1 = Chem.Mol(mol)

    fragments = []
    # fragment molecule based on the rotatable bonds
    if bonds == []:
        fragments.append(mol1)
    else:
        new_mol = Chem.FragmentOnBonds(mol1, bonds)
        fragments = Chem.GetMolFrags(new_mol, asMols=True)

    generate_xml_tree(mol, args.states, fragments,
                      seg_name, energies, args.mdname)


def generate_xml_tree(
        mol: Chem.rdchem.Mol, states: List[str], fragments: List[Any],
        seg_name: str, energies: Dict[str, float], md_name: str) -> None:
    topology = ET.Element('topology')
    molecules = ET.SubElement(topology, 'molecules')
    molecule = ET.SubElement(molecules, 'molecule')
    name = ET.SubElement(molecule, 'name')
    name.text = seg_name
    mdname = ET.SubElement(molecule, 'mdname')
    mdname.text = md_name
    segments = ET.SubElement(molecule, 'segments')
    segment = ET.SubElement(segments, 'segment')
    segment_name = ET.SubElement(segment, 'name')
    segment_name.text = seg_name

    # Add qmcoords, xyz, multipoles and polar files
    add_state_files('n', segment, seg_name)

    if 'h' in states:
        add_state_files('h', segment, seg_name)
        add_energies('h', segment, energies)
    if 'e' in states:
        add_state_files('e', segment, seg_name)
        add_energies('e', segment, energies)

    map2md = ET.SubElement(segment, 'map2md')
    map2md.text = '0'
    xmlfragments = ET.SubElement(segment, 'fragments')

    # add the fragments into the xml tree
    add_fragments(mol, fragments, xmlfragments)

    xml_str = minidom.parseString(
        ET.tostring(topology)).toprettyxml(indent="   ")
    with open("mapping.xml", "w") as f:
        f.write(xml_str)


def add_state_files(state: str, segment: ET.Element, seg_name: str) -> None:
    """Add the files for each state."""
    segment_coord = ET.SubElement(segment, f'qmcoords_{state}')
    segment_coord.text = f"QC_FILES/{seg_name}_{state}.xyz"
    segment_mpoles = ET.SubElement(segment, f'multipoles_{state}')
    segment_mpoles.text = f"MP_FILES/{seg_name}_{state}_polar.mps"


def add_energies(state: str, segment: ET.Element, energies: Dict[str, float]) -> None:
    """Add the energies for the states."""
    # adiabatic excitation energy
    segment_U_xX_nN = ET.SubElement(segment, f'U_xX_nN_{state}')

    segment_U_xX_nN.text = format_energy(
        energies, f"E_{state}_final", "E_n_final")
    # reorg energy deexcitation
    segment_U_nX_nN = ET.SubElement(
        segment, f'U_nX_nN_{state}')
    segment_U_nX_nN.text = format_energy(
        energies, f"E_{state}_cross", "E_n_final")

    # reorg energy excitation
    segment_U_xN_xX = ET.SubElement(segment, f'U_xN_xX_{state}')

    segment_U_xN_xX.text = format_energy(
        energies, f"E_{state}_init", f"E_{state}_final")


def add_fragments(
        mol: Chem.rdchem.Mol, fragments: List[Chem.rdchem.Mol],
        xmlfragments: ET.Element) -> None:
    """Add fragments to the XML tree."""
    for i, fragment in enumerate(fragments):
        i += 1

        filename = f"fragment{i}.pdb"
        Chem.MolToPDBFile(fragment, filename)
        conf = fragment.GetConformer()

        xmlfragment = ET.SubElement(xmlfragments, 'fragment')
        frag_name = ET.SubElement(xmlfragment, 'name')
        frag_name.text = f'fragment{i}'
        mdatoms = ET.SubElement(xmlfragment, 'mdatoms')
        qmatoms = ET.SubElement(xmlfragment, 'qmatoms')
        mpoles = ET.SubElement(xmlfragment, 'mpoles')
        weights = ET.SubElement(xmlfragment, 'weights')
        localframe = ET.SubElement(xmlfragment, 'localframe')

        mdatoms_str = ""
        qmatoms_str = ""
        weights_str = ""
        localframe_str = ""
        idx = np.sort(check_collinear(fragment))
        for atom in range(fragment.GetNumAtoms()):
            this_element = fragment.GetAtoms()[atom].GetSymbol()
            if this_element != "*":
                this_pos = conf.GetAtomPosition(atom)
                for index_atom in range(mol.GetNumAtoms()):
                    full_atom = mol.GetAtoms()[index_atom]
                    full_element = full_atom.GetPDBResidueInfo().GetName().replace(" ", "")
                    full_element_qm = full_atom.GetSymbol()
                    full_pos = mol.GetConformer().GetAtomPosition(index_atom)
                    if all(getattr(full_pos, x) == getattr(this_pos, x) for x in {'x', 'y', 'z'}):
                        mdatoms_str += f"0:{full_element}:{index_atom} "
                        qmatoms_str += f"{index_atom}:{full_element_qm} "
                        weights_str += f"{full_atom.GetMass()} "
        mdatoms.text = mdatoms_str
        qmatoms.text = qmatoms_str
        mpoles.text = qmatoms_str
        weights.text = weights_str

        for label in ["".join(c for c in qmatoms_str.split()[i] if c.isdigit()) for i in idx]:
            localframe_str += f"{label} "
        localframe.text = localframe_str


def random_selection(fragment) -> np.ndarray:
    n_atoms = fragment.GetNumAtoms()
    # Choose 3 random idx (aka choose three random atoms in the fragment, excluding ghost atoms with symbol *)
    atomic_list = [a for a in range(n_atoms) if fragment.GetAtoms()[
        a].GetSymbol() != '*']
    # Divide this list in H-list and rest_of_the_atoms-list
    h_list = [a for a in atomic_list if fragment.GetAtoms()[
        a].GetSymbol() == 'H']
    rest_list = [a for a in atomic_list if fragment.GetAtoms()[
        a].GetSymbol() != 'H']

    size_sample = 0
    # If the rest of the atoms are 3 or less
    if len(rest_list) < 4:
        # If there are no Hydrogens it spits out a sample size equal to the rest of the atoms list
        if len(h_list) < 1:
            size_sample = len(rest_list)
        # If there are Hydrogens and the rest of the atoms are less than 3 (2 or 1)
        if len(h_list) > 0 and len(rest_list) < 3:
            # If there are at least 3 H
            if len(h_list) > 2:
                new_items = np.random.choice(
                    h_list, size=3 - len(rest_list), replace=False)
                for i in new_items:
                    rest_list.append(i)
                size_sample = len(rest_list)
            # Otherwise just spit out the all atomic list
            # (this because you can have max 1 H and 2 rest of the atoms)
            else:
                rest_list = atomic_list
                size_sample = len(rest_list)
    # Else the rest of the atoms are more than 3 then sampling randomly
    # three of them it is fine
    else:
        size_sample = 3
    idx = np.random.choice(rest_list, size=size_sample, replace=False)
    return idx


def angle_fragment(idx: np.ndarray, fragment: Chem.rdchem.Mol) -> float:
    try:
        conformer = fragment.GetConformer()
        if len(idx)>2:
            pos_1, pos_2, pos_3 = [
                conformer.GetAtomPosition(idx[x].item()) for x in range(3)]
            vec1 = np.array([pos_1.x - pos_2.x, pos_1.y -
                            pos_2.y, pos_1.z - pos_2.z])
            vec2 = np.array([pos_1.x - pos_3.x, pos_1.y -
                            pos_3.y, pos_1.z - pos_3.z])
        # Sin(theta) with theta angle between vec1 and vec2
            cp = np.linalg.norm(np.cross(vec1, vec2)) / \
                (np.linalg.norm(vec1) * np.linalg.norm(vec2))
        else:
            cp = 1 
    except ZeroDivisionError:
        # Just throw out a big value for the cp so check_collinear whill break immediatly
        cp = 1
    return cp


def check_collinear(fragment: Chem.rdchem.Mol) -> int:
    """Check if three random selected points are collinear."""
    while True:
        idx = random_selection(fragment)
        cp_norm = angle_fragment(idx, fragment)
        if cp_norm > 0.6:
            break
    return idx


def format_energy(energies: Dict[str, float], first: str, second: str) -> str:
    """Format the energy difference."""
    diff = H2EV * (energies[first] - energies[second])
    return str(diff)


def main():
    """Read the command line arguments."""
    parser = argparse.ArgumentParser("xtp_autogen_mapping", usage=argparse.SUPPRESS,
                                     description=DESCRIPTION, formatter_class=argparse.RawDescriptionHelpFormatter)

    # Add the arguments to the parser
    parser.add_argument("-pdb", "--pdbfile", required=True, type=exists,
                        help="PDB coordinate file")
    parser.add_argument("-opt", "--optimize", action="store_true",
                        help="optimization of XYZ and MPS files")
    parser.add_argument("-s", "--states", default="n", nargs="+",
                        help="which states to optimize", choices=["n", "e", "h"])
    parser.add_argument("-orca", "--orca", default="/opt/orca-5.0.3/orca",
                        help="full path to ORCA executable")
    parser.add_argument("-f", "--functional", default="PBE0",
                        help="DFT functional")
    parser.add_argument(
        "-b", "--basis", default="def2-tzvp", help="basis set")
    parser.add_argument("--nori", action="store_false",
                        help="Do not to use the Resolution of Identity approximation")
    parser.add_argument("-m", "--mdname", default="",
                        help="Name of the molecule in the MD topology")
    parser.add_argument("-t", "--threads", default="",
                        help="Number of threads to use in ORCA")
    parser.add_argument('--version', action='version',
                        version=f"%(prog)s {VERSION}")
    args = parser.parse_args()

    generate_mapping(args)


def exists(input_file: str) -> str:
    """Check if the input file exists."""
    path = Path(input_file)
    if not path.exists():
        raise argparse.ArgumentTypeError(f"{input_file} doesn't exist!")

    return path.absolute().as_posix()


if __name__ == "__main__":
    main()