{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "6f7c9cf7-93ca-4921-9783-d6377f2606b7",
   "metadata": {},
   "source": [
    "This notebook selects the non-duplicates and moves them one folder upwards"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "4fc7ed5a-d1a9-4169-a2f1-465955af619b",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "6e4d6adc-1307-4524-8ddb-3b41ea4c372a",
   "metadata": {},
   "outputs": [],
   "source": [
    "d_reg_num_name={}\n",
    "d_reg_num_name['01'] = 'Alaska'\n",
    "d_reg_num_name['02'] = 'Western Canada & USA'\n",
    "d_reg_num_name['03'] = 'Arctic Canada North'\n",
    "d_reg_num_name['04'] = 'Arctic Canada South'\n",
    "d_reg_num_name['05'] = 'Greenland'\n",
    "d_reg_num_name['06'] = 'Iceland'\n",
    "d_reg_num_name['07'] = 'Svalbard and Jan Mayen'\n",
    "d_reg_num_name['08'] = 'Scandinavia'\n",
    "d_reg_num_name['09'] = 'Russian Arctic'\n",
    "d_reg_num_name['10'] = 'North Asia'\n",
    "d_reg_num_name['11'] = 'Central Europe'\n",
    "d_reg_num_name['12'] = 'Caucasus and Middle East'\n",
    "d_reg_num_name['13'] = 'Central Asia'\n",
    "d_reg_num_name['14'] = 'South Asia West'\n",
    "d_reg_num_name['15'] = 'South Asia East'\n",
    "d_reg_num_name['16'] = 'Low Latitudes'\n",
    "d_reg_num_name['17'] = 'Southern Andes'\n",
    "d_reg_num_name['18'] = 'New Zealand'\n",
    "d_reg_num_name['19'] = 'Antarctic and Subantarctic'\n",
    "\n",
    "###\n",
    "d_reg_name_sh_num={}\n",
    "d_reg_name_sh_num['Alaska'] = '01'\n",
    "d_reg_name_sh_num['WesternCanada'] = '02'\n",
    "d_reg_name_sh_num['ArcticCanadaN'] = '03'\n",
    "d_reg_name_sh_num['ArcticCanadaS']='04'\n",
    "d_reg_name_sh_num['Greenland']='05'\n",
    "d_reg_name_sh_num['Iceland']='06'\n",
    "d_reg_name_sh_num['Svalbard']='07'\n",
    "d_reg_name_sh_num['Scandinavia']='08'\n",
    "d_reg_name_sh_num['RussianArctic']='09'\n",
    "d_reg_name_sh_num['NorthAsia']='10' \n",
    "d_reg_name_sh_num['CentralEurope']='11'\n",
    "d_reg_name_sh_num['Caucasus']='12'\n",
    "d_reg_name_sh_num['CentralAsia']='13'\n",
    "d_reg_name_sh_num['SouthAsiaWest']='14'\n",
    "d_reg_name_sh_num['SouthAsiaEast']='15'\n",
    "d_reg_name_sh_num['LowLatitudes']='16'\n",
    "d_reg_name_sh_num['SouthernAndes']='17'\n",
    "d_reg_name_sh_num['NewZealand']='18'\n",
    "d_reg_name_sh_num['Antarctic']='19'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "a80e4d57-03b1-45a5-938a-625fdc713e00",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_actual_runs = pd.read_csv('GloGEM_GlacierMIP3_actual_runs_Lily_list.csv')\n",
    "pd_actual_runs = pd_actual_runs.melt()\n",
    "pd_actual_runs['value'] = pd_actual_runs.value.astype(str)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "ca1b449e-d944-4987-ac51-6d6e8d127ec6",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "mv: cannot stat 'Huss_rgi*': No such file or directory\n"
     ]
    }
   ],
   "source": [
    "mv Huss_rgi* duplicates/."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "d5e6ff5c-8555-4c93-a02c-a1bc52b8e116",
   "metadata": {},
   "outputs": [],
   "source": [
    "rgi_regs = []\n",
    "for rgi_reg in np.arange(1,20,1):\n",
    "    if rgi_reg < 10:\n",
    "        rgi_reg = '0'+str(rgi_reg)\n",
    "    else:\n",
    "        rgi_reg = str(rgi_reg)\n",
    "    rgi_regs.append(rgi_reg)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "f4c12c03-26db-4cb1-85e6-8b94c2ad542f",
   "metadata": {},
   "outputs": [],
   "source": [
    "run=False\n",
    "if run:\n",
    "    rgi_reg_l = []\n",
    "    gcm_l = []\n",
    "    ssp_l = []\n",
    "    period_l = []\n",
    "\n",
    "    for ssp in ['ssp126','ssp370','ssp585']:\n",
    "        for gcm in ['gfdl-esm4','ipsl-cm6a-lr','mpi-esm1-2-hr','mri-esm2-0','ukesm1-0-ll']:\n",
    "\n",
    "            if ssp == 'ssp126':\n",
    "                periods=['1851-1870','1901-1920','1951-1970','1995-2014','2021-2040','2041-2060','2061-2080','2081-2100']\n",
    "            else:\n",
    "                periods=['2021-2040','2041-2060','2061-2080','2081-2100']\n",
    "\n",
    "            for period in periods:\n",
    "                for region in rgi_regs:\n",
    "\n",
    "                    rgi_reg_l.append(region)\n",
    "                    gcm_l.append(gcm)\n",
    "                    if period in ['1851-1870','1901-1920','1951-1970','1995-2014']:\n",
    "                        ssp_t = 'hist'\n",
    "                    else:\n",
    "                        ssp_t = ssp\n",
    "                    ssp_l.append(ssp_t)\n",
    "                    period_l.append(period)\n",
    "\n",
    "    pd_actual_runs['gcm'] = gcm_l\n",
    "    pd_actual_runs['ssp'] = ssp_l\n",
    "    pd_actual_runs['rgi_reg'] = rgi_reg_l\n",
    "    pd_actual_runs['period'] = period_l\n",
    "    \n",
    "else:\n",
    "    for l in pd_actual_runs.index:\n",
    "        val = pd_actual_runs['value'].loc[l] #:\n",
    "        if val =='nan':\n",
    "            pass\n",
    "        else:\n",
    "            _,_,_,_,_,region,gcm,ssp,period_long=val.split('/')\n",
    "            pd_actual_runs.loc[l,'region'] = region\n",
    "            pd_actual_runs.loc[l,'rgi_reg'] = d_reg_name_sh_num[region]\n",
    "            pd_actual_runs.loc[l,'gcm'] = gcm\n",
    "            pd_actual_runs.loc[l,'period'] = period_long[:-7]\n",
    "            if period_long[:-7] in ['1851-1870','1901-1920','1951-1970','1995-2014']:\n",
    "                pd_actual_runs.loc[l,'ssp'] = 'hist'\n",
    "            else:\n",
    "                pd_actual_runs.loc[l,'ssp'] = ssp\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "27a323be-3b40-4bf6-bf72-b9a8136b0c9c",
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_path_d(folder_path='/home/www/lschuster/glacierMIP3_analysis',\n",
    "             rgi_reg='01',\n",
    "             model_author='Rounce',\n",
    "             period='1851-1870', gcm='gfdl-esm4', ssp='hist'):\n",
    "    ''' output the regional simulation file path for the respective rgi_reg, model_author, period, gcm & ssp '''\n",
    "    if (model_author == 'Rounce'):\n",
    "        path = f'/home/www/drounce/regional-filled-norunaway/{rgi_reg}/{model_author}_rgi{rgi_reg}_sum_{period}_{gcm}_{ssp}.nc'\n",
    "    elif ('OGGM' in model_author):\n",
    "        # was corrected for missing glaciers ... (by Lilian Schuster with method of Fabien)\n",
    "        if 'OGGM' in model_author and ssp == 'hist':\n",
    "            ssp = 'historical'\n",
    "            path =  f'{folder_path}/{model_author[:4]}/regional_filled/{rgi_reg}/{model_author}_rgi{rgi_reg}_sum_{period}_{gcm}_{ssp}_filled.nc'\n",
    "        else:\n",
    "            path =  f'{folder_path}/{model_author[:4]}/regional_filled/{rgi_reg}/{model_author}_rgi{rgi_reg}_sum_{period}_{gcm}_{ssp}_filled.nc'\n",
    "    else:\n",
    "        if gcm=='ipsl-cm6a-lr':\n",
    "            if model_author=='Compagno':\n",
    "                gcm='psl_cm6a-lr' # no 'i' and '_' instead of '-'\n",
    "            if model_author == 'Huss':\n",
    "                # also Huss has some spelling \n",
    "                gcm = 'ipsl_cm6a-lr'\n",
    "        if model_author=='Kraaijenbrink':\n",
    "            if ssp == 'hist':\n",
    "                ssp='historical'\n",
    "        path = f'{folder_path}/{model_author}/duplicates/{model_author}_rgi{rgi_reg}_sum_{period}_{gcm}_{ssp}.nc'\n",
    "    return path\n",
    "\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "id": "16df2021-4d29-4592-b789-e2b0e373d536",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/scratch_net/vierzack06_fourth/mhuss/GlacierMIP/Alaska/ukesm1-0-ll/ssp585/2081-2100_stable'"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_sel = pd_actual_runs.loc[pd_actual_runs.gcm=='ukesm1-0-ll']\n",
    "pd_sel.loc[pd_sel.rgi_reg=='01'].value.iloc[7]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "id": "62378c63-0bd1-4810-8907-990d68ebfcb5",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/scratch_net/vierzack06_fourth/mhuss/GlacierMIP/Alaska/ukesm1-0-ll/ssp585/2021-2040_stable'"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_sel.loc[pd_sel.rgi_reg=='01'].value.iloc[4]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "fc09175c-2ab2-4450-b65e-5f26c1cdd4f6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/scratch_net/vierzack06_fourth/mhuss/GlacierMIP/ArcticCanadaS/ukesm1-0-ll/ssp126/1851-1870_stable'"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_sel.loc[pd_sel.rgi_reg=='04'].value.iloc[0]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 34,
   "id": "8912a7c3-4730-400e-9c29-b42089ee67f7",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/scratch_net/vierzack06_fourth/mhuss/GlacierMIP/ArcticCanadaS/ukesm1-0-ll/ssp126/1995-2014_stable'"
      ]
     },
     "execution_count": 34,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_sel.loc[pd_sel.rgi_reg=='04'].value.iloc[3]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 35,
   "id": "bc713e1e-ccfd-467b-8c27-b7bace2ddf3e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/scratch_net/vierzack06_fourth/mhuss/GlacierMIP/Greenland/ukesm1-0-ll/ssp126/1851-1870_stable'"
      ]
     },
     "execution_count": 35,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_sel.loc[pd_sel.rgi_reg=='05'].value.iloc[0]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 36,
   "id": "62336182-87de-4937-b37c-249ff455b202",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "'/scratch_net/vierzack06_fourth/mhuss/GlacierMIP/Greenland/ukesm1-0-ll/ssp126/1995-2014_stable'"
      ]
     },
     "execution_count": 36,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd_sel.loc[pd_sel.rgi_reg=='05'].value.iloc[3]\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "ad8d4246-1fdd-4893-aa85-02b3bc4234f9",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d6792030-70ae-47c3-be19-9223546a1250",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "bb72dc51-9d5c-4474-aba9-50f6537ce881",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "960"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# amount of working runs\n",
    "len(pd_actual_runs.where(pd_actual_runs.value != 'nan').dropna())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "id": "baf2b3c2-680c-4e8b-95d3-22fa5b222920",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "560"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# amount of missing runs\n",
    "len(pd_actual_runs.where(pd_actual_runs.value == 'nan').dropna(how='all'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "b45f4f6a-c416-4658-af20-72daffae9849",
   "metadata": {},
   "outputs": [],
   "source": [
    "pd_working_runs = pd_actual_runs.where(pd_actual_runs.value != 'nan').dropna()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "c5887e10-1919-40a7-8ca7-ba7500b27830",
   "metadata": {},
   "outputs": [],
   "source": [
    "import shutil\n",
    "for j in pd_working_runs.index:\n",
    "    f = get_path_d(rgi_reg=pd_working_runs.loc[j]['rgi_reg'],\n",
    "             model_author='Huss', period=pd_working_runs.loc[j]['period'],\n",
    "             gcm=pd_working_runs.loc[j]['gcm'], ssp = pd_working_runs.loc[j]['ssp'])    \n",
    "\n",
    "    # absolute path\n",
    "    src_path = f\n",
    "    dst_path = '/home/www/lschuster/glacierMIP3_analysis/Huss'\n",
    "    shutil.move(src_path, dst_path)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "fa93513f-1945-4a0d-be81-816ac8b43828",
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "\n",
    "_, _, files = next(os.walk(\"/home/www/lschuster/glacierMIP3_analysis/Huss/duplicates\"))\n",
    "file_count = len(files)\n",
    "assert file_count == len(pd_actual_runs.where(pd_actual_runs.value == 'nan').dropna(how='all'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "7852a20a-30bf-4eba-ae9e-e7752c1f8888",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "4946eafd-77ec-4f58-b058-7e5e6183a188",
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "markdown",
   "id": "54c4ee79-378e-4143-84a9-dd3795e580bb",
   "metadata": {
    "tags": []
   },
   "source": [
    "### OLD"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d1fc4af6-e82b-414b-ba68-9ebcb04d538d",
   "metadata": {},
   "outputs": [],
   "source": [
    "for l in pd_actual_runs.index:\n",
    "    val = pd_actual_runs['value'].loc[l] #:\n",
    "    assert val[-16:-7] == pd_actual_runs['period'].loc[l]\n",
    "    assert pd_actual_runs['gcm'].loc[l] in val\n",
    "    if pd_actual_runs['ssp'].loc[l] == 'hist':\n",
    "        ssp_test = 'ssp126'\n",
    "    else:\n",
    "        ssp_test = pd_actual_runs['ssp'].loc[l]\n",
    "    assert val[-23:-17] == ssp_test\n",
    "    a = d_reg_num_name[pd_actual_runs['rgi_reg'].loc[l]].split()\n",
    "    assert a[0] in val\n",
    "    if len(a)>1:\n",
    "        assert a[1] in val\n",
    "\n"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.10.8"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}