// Bob Hanson hansonr@stolaf.edu

// create plane and subperiodic (layer, rod, frieze) group data, json, and txt files

targetDir = "c:/temp/bilbao/ite/subgroups/";

// variables for the createJSONFiles call (below), after the functions

itadelay = 0.1;
itaFirst = 0;
itaForceNew = false;
htmlOnly = true;
maxindex = 100;
nsubs = 0;
ntrm = 0;
subtype = "p/";
subtypeName = "plane";
base = "https://www.cryst.ehu.es/"


// only one function to run here:
function createSubJSONFiles(type, typeName, n, nMax, forceNew) {
	itaFirst = n0;
	subtype = type;
	subtypeName = typeName;
	nsubs = 0;
	ntrm = 0;

	// creates sub_<subtypeName>_<i>.json, by space group

	// n option to start at a given number (-30), or only a certain number (30), or all (0)
	// forceNew option true to overwrite local files; false to use them if they are there.

	var allTransforms = [["i", "cleg", "tr_type", "tr_subtype" "det", "index", "conj_class"]]
	var indexList = []
	var allList = []
	for (var i = 1; i <= nMax; i++) {
		if (n < 0 && i < -n || n > 0 && i != n) {
			continue;
		}
		var adata = {"sg":i, "subtype":subtypeName, "created":now("").trim()};
		_getAllSubs(i, adata, allTransforms, indexList);
		allList.push(adata);
		var thedata = adata.format("JSON");
		var fname = targetDir + "jsonx/sub_" + subtypeName + "_" + i + ".json";
		write var thedata @fname;

	}

	if (n == 0 || n == -1) {

		var thedata = allTransforms.join("\t",TRUE);
		var fname = targetDir + "txt/cleg_subgroups_" + subtypeName + ".tab";
		write var thedata @fname;

		var thedata = indexList.format("json");
		var fname = targetDir + "json/sub_" + subtypeName + "_index.json";
		write var thedata @fname;

		var thedata = allList.format("JSON");
		var fname = targetDir + "json/sub_all_" + subtypeName + ".json";
		write var thedata @fname;

		print "DONE " + nsubs + " subGroups; " + ntrm + " cleg";

	}

}

function _getAllSubs(group, adata, allTransforms, indexList) {
	var prog = if(subtype=="p/";"nph-plane_getgen";"nph-sub_maxsub");
	var subs = [];
	adata.subgroups = subs;
	// split by buttons
	var y = _getSubDataHTM(group, 0, "sub", "", forceNew).replace("em>","i>");
	y = y.split(prog);
	adata.hm = subtype + _fixName(y[1].split("<center>")[1].split("roup ")[2].split("(No.")[1]);
	var isub = 0;
	var nt = 0;
	var ipt = 2;
	var list = [];
	indexList.push(list);
	var sglast = -1;
	for (var j = 1; j < y.length; j++) {
		var gdata = y[j + 1];
		var href = base + _getField(gdata, "href", 1000);
		var sindex = 0+_getField(href, "index");
		if (sindex > maxIndex) continue;
		isub++;
		var subgroup = 0+_getField(href, "sub");
		if (subgroup != sglast) {
			list.push(subgroup);
			list.push(sindex);
		}
		sglast = subgroup;
		var stype = _getField(href, "subgrouptype");
		if (!stype) {
			// plane
			stype = gdata.split("<td")[4].split(">")[2].split("</td")[1]
		}
		var trsubtype = stype[2][0];
		var trtype = stype[1];
		var hm = _fixName(gdata.split("<i>")[ipt].split("</td>")[1]);
		var a = {"i":isub,
			"sg":subgroup, 
			"hm":subtype + hm, 
			"trType":trtype, 
			"trSubtype": trsubtype,
			"index":sindex};//, "url": "" + href};
		// name="trm" value="%31%2F%32x%2B%31%2F%32y%2C%2D%31%2F%32x%2B%31%2F%32y%2Cz"
		var sdata = _getSubDataHTM(group, isub, "sub", href, forcenew).replace("> Conj", ">Conj").split('"trm"');
		var transforms = [];
		var classes = [];
		//<small> Conjugacy class a</small>
		var s = sdata[1];
		var sdet;
		var conjClass = "?";
		var n = 0;
		for (var t = 2; t <= sdata.length; t++) {
			nt++;
			n++;
			var ac = s.split(">Conjugacy class ");
			if (ac[2])
				conjClass = ac[2].split("<")[1].trim();
			s = sdata[t];
			var xyz = _fixURI(_getField(s,"value"));
			var tabc = _toTransform(xyz);
			sdet = unitcell(tabc,true)%1%2;
			//var abc = conjClass + "=" + tabc;// + "=" + xyz;
			transforms.push({"conjClass":conjClass, "trm": tabc});
			allTransforms.push([ntrm + nt, "" + subtype + group + ">" + tabc + ">" + subtype + subgroup, stype, subtypeName, sdet, sindex, conjClass]); 
		}
		a["det"] = sdet;
		a["trms"] = transforms;
		a["n"] = n;
		subs.push(a);
	}
	adata.nSubgroups = subs.length;
	adata.nSettings = nt;
	nsubs += subs.length;
	ntrm = += nt;
}

// all other functions are "private" to this script

function _getSubDataHTM(i, j, what, url0, forceNew) {

//https://www.cryst.ehu.es/cgi-bin/subperiodic/programs/nph-sub_maxsub?gnum=14&subtype=layer
//https://www.cryst.ehu.es/cgi-bin/cryst/programs/nph-tranmax?way=up&super=14&sub=2&index=2&client=submax&what=gp&path=&subtype=layer&subgrouptype=t&conj=all
//"chbas" https://www.cryst.ehu.es/cgi-bin/cryst/programs/nph-trgen?client=submax&super=14&sub=2&from=submax&type=layer&what=gp&way=up&trm=x%252Cy%252Cz

//
	var localFile = targetDir + "data/" + what + "_" + subtypeName + "_" + i + if(j;"_"+j+".html";".htm");
	if (url0) {
	  var url = url0;
	} else {
	  var url = if(subtype=="/p";base + "cgi-bin/plane/programs/nph-plane_maxsub?type=plane";
			base + "cgi-bin/subperiodic/programs/nph-sub_maxsub?subtype=" + subtypeName);
	}
	url = if(url0;url0;url + "&gnum=" + i);
	url = if(forceNew ; url;  localFile);

	print url;
	delay @itadelay;
	var gdata = load(url);
	if (forceNew) {
		write var gdata @localFile;
	} else if (gdata.find("FileNotFound")) {
		gdata = _getSubDataHTM(i, j, what, url0, true);
	}
	return gdata;
}


function _getField(gdata, field, max) {
	
	if (!max) max = 0;

	// get a &... field form a URL

	var ret;
	var i = gdata.find(field + "=");
	if (i == 0) return "";
	gdata = gdata[i + field.length + 1][0];
	if (gdata[1] == '"') {
		return gdata.split('"')[2];	
	} 
	return (gdata[1][max] + "&").split("&")[1];
}

// utility methods

function _fixName(s) {

	// remove HTML markings

	return (s.trim()
		.replace("<i>","").replace("</i>","")
		.replace("<b>","").replace("</b>","")
		.replace("<sub>","").replace("</sub>","")
		.replace("&#124;","|")).replace("  "," ");
}


function _fixURI(s) {
	s = s.replace('%3A',':').replace('%3B',';').replace('%3','').replace('%2F','/').replace('%2B','+').replace('%2C',',')
	return s.replace('%2D','-').replace('%2E','.').replace('%28','(').replace('%29',')').replace('%20',' ')
}

m40 = [[0 0 0 0][0 0 0 0][0 0 0 0][0 0 0 1]];

function _toTransform(xyz){ 
	// 1 0 -1 | 0
	// 0 1  0 | 0
	// 0 0  2 | 1/2
	// 
	// x-z,y,2z+1/2 to "a,b,-a+2c;0,0,1/2 (transposed)
	// a,b,-a+2c+1/2 also to "a,b,-a+2c;0,0,1/2 (not transposed)

	// the function unitcell() runs org.jmol.symmetry.UnitCell.getMatrixAndUnitCell()
	// unitcell() removes any embedded translations and puts the rotation in xyz-row,abc column format
	// we get any embedded translation using symop()

	var m = -unitcell(xyz, true);
	var abc = symop(m, "xyz").replace('x','a').replace('y','b').replace('z','c');

	var t = symop(xyz, "matrix")%2;
	if (t == 0) {
		return abc;
	}
	return abc + ";" + symop(m40 + t, "xyz");
}

function _toXYZ(abc, asTrm){

	// getGen needs the transposed matrix, in xyz format. 
	// this does the trick, as -m4 is the transpose of m4, and symop(matrix, "xyz") 
 	// will accept a,b,c and return x,y,z
	// check for translation and don't transpose that

	var a = symop(abc, "matrix");
	var t = a%2; // get translation vector
	if (1.0 * t == 0) {
		// transpose if no translation
		a = -a;
	} else {
		// remove translation from 4x4, transpose it, then return translation
		a = -(a + -t) + t;
	}
	return symop(a, "xyz");
}




createSubJSONFiles("l/","layer", 0, 80, iteForceNew);
createSubJSONFiles("f/","frieze", 0, 7, iteForceNew);
createSubJSONFiles("p/","plane", 0, 17, iteForceNew);
createSubJSONFiles("r/","rod", 0, 75, iteForceNew);

/**
last update 2024.10.10 "HM" and "subgroupHM" to "hm"; adds cleg subtype prefix to hm