WiP: dictionary stuff (dictionary-tree)

src/dictionary-parser.mjs

+import * as tree from './dictionary-tree.mjs';
+
 export function parse(dictionaryString) {
 	const allLines = dictionaryString.split('\n');
 	const parsedLines = parseAllLines(allLines);
-	return combineUnspecifiedLines2IdSecPwdLines(parsedLines);
+	return combineUnspecified2IdSecPwd(parsedLines);
 }
+
 function parseAllLines(lines) {
 	return lines;
 }
+
 function parseLine(theLine, allLines) {
 	return theLine;
 }
-function combineUnspecifiedLines2IdSecPwdLines(lines){
-	const unspecifiedLines = [];
+
+function combineUnspecified2IdSecPwd(lines) {
 	const idSecPwdLines = [];
-	lines.forEach(line=>line.includes('@@@@') ? idSecPwdLines.push(line) : unspecifiedLines.push(line));
+	const unspecifiedLines = [];
+	lines.forEach(line => line.includes('@@@@') ? deepUnspecifiedExtractor(line, unspecifiedLines, idSecPwdLines) : unspecifiedLines.push(line));
 	if (unspecifiedLines.length) idSecPwdLines.push(`${mergeLines(unspecifiedLines)}@@@@${mergeLines(unspecifiedLines)}`)
 	return mergeLines(idSecPwdLines);
 }
+
+function deepUnspecifiedExtractor(line, unspecifiedLines, idSecPwdLines) {
+	const hybridTree = tree.buildTreeStruct(line);
+
+	/*
+	buildTree (à créer dans un module à part)
+	parcours en profondeur jusqu'à trouver des branches unspecified.
+	extraction de celles-ci pour les ajouter (avec leur chaine de parents) aux unspecifiedlines
+	ajout de l'arbre restant aux idSecPwdLines
+	 */
+	idSecPwdLines.push(line);
+}
+
 function mergeLines(lines) {
-	console.log('lines:',lines);
+	//console.log('lines:',lines);
 	if (!lines.length) return '';
 	lines = lines.filter(l => l.length);
 	if (lines.length === 1) return lines[0];
 	return `(${lines.join('|')})`;
 }
+
+/*
+const json1 = '(a(b@@@@c|d)|(e|f)|g|h@@@@i)'
+const json2 = {
+	'a':'a'{
+						'b@@@@c':'b@@@@c',
+						'd':'d'
+					},
+	{
+		'e':'e',
+		'f':'f'
+	},
+	'g':'g',
+	'h@@@@i':'h@@@@i'
+}
+*/

src/dictionary-parser.test.mjs

@@ -16,5 +16,5 @@ h@@@@(i|j)
 `), '(a@@@@(b|c)|h@@@@(i|j)|((d|e)|(f|g))@@@@((d|e)|(f|g)))');
 });
 test.skip('no @@@@ cases combined with @@@@ case on same line recombine to allways finish with @@@@ case', t => {
-	t.is(app.parse('(a|b|c@@@@d)'), '(c@@@@d|(a|b)@@@@(a|b))');
+	t.is(app.parse('(a(b@@@@c|d)|(e|f)|g|h@@@@i)'), '(g@@@@h|(a|b|c)@@@@(a|b|c))');
 });

src/dictionary-tree.mjs

@@ -2,10 +2,10 @@ export function build(dictionaryString) {
 }
 
 export function buildTreeStruct(monoLineString) {
-	console.log('srcTree', monoLineString);
+	//console.log('srcTree', monoLineString);
 	const stringAsArray = monoLineString.split('');
 	const rawTree = leftParser(stringAsArray);
-	console.log('rawTree', JSON.stringify(rawTree));
+	//console.log('rawTree', JSON.stringify(rawTree));
 	const outOfScope = stringAsArray.length;
 	if (outOfScope) throw `fail to build tree from : "${monoLineString}" parsed: ${JSON.stringify(rawTree)} unparsed/failed: ${stringAsArray.join('')}`;
 	//return rawTree;//trivialDedup(flattenTree(rawTree))
@@ -19,23 +19,33 @@ function flushNoEmptyString(data){
 	if (data.str.length) data.tree[data.tree.length - 1].step.push(data.str);
 	data.str = '';
 }
+
 function appendToString(chr, data) {
 	data.str += chr;
 }
+
 function nextAlt(data) {
 	data.tree.push({step: []});
 }
+
 function goIntoNewAlt(data) {
 	data.tree[data.tree.length - 1].step.push(leftParser(data.input));
 }
-function returnTrue(){return true;}
-function doTheses(){}
+
+function returnTrue() {
+	return true;
+}
+
+function doTheses() {
+}
+
 const leftParserFunc = {
 	')': (chr, data) => returnTrue(flushNoEmptyString(data)),
 	'|': (chr, data) => doTheses(flushNoEmptyString(data), nextAlt(data)),
 	'(': (chr, data) => doTheses(flushNoEmptyString(data), goIntoNewAlt(data)),
 	'default': (chr, data) => appendToString(chr, data)
 }
+
 function leftParser(inputChrArray) {
 	const data = {
 		input: inputChrArray,
@@ -52,6 +62,7 @@ function leftParser(inputChrArray) {
 	flushNoEmptyString(data);
 	return data.tree;
 }
+
 // end leftParser
 function concatStrings(array) {
 	const data = {
@@ -69,6 +80,7 @@ function concatStrings(array){
 	flushNoEmptyString(data)
 	return data.tree[data.tree.length - 1].step;
 }
+
 function flattenTree(tree) {
 	if (typeof tree === 'string') return tree;
 	if (Array.isArray(tree) && tree.length === 0) return '';
@@ -101,6 +113,7 @@ function flattenTree(tree) {
 function isStep(element) {
 	return typeof element === 'object' && typeof element.step !== 'undefined';
 }
+
 function trivialDedup(tree) {
 	if (typeof tree === 'string') return tree;
 	if (isStep(tree)) return {step: tree.step.map(subTree => trivialDedup(subTree))};
@@ -111,3 +124,10 @@ function trivialDedup(tree) {
 	}
 	throw `unknown case : ${tree}`;
 }
+
+export function serialize(treeStruct) {
+	if (typeof treeStruct === 'string') return treeStruct;
+	if (Array.isArray(treeStruct)) return `(${treeStruct.map(serialize).join('|')})`;
+	if (isStep(treeStruct)) return treeStruct.step.map(serialize).join('');
+	throw new Error(`Error: how to serialize ${JSON.stringify(treeStruct)} RAW: ${treeStruct}`);
+}

src/dictionary-tree.test.mjs

@@ -13,16 +13,8 @@ test('(|b) empty choice is choice', t=>t.is(buildTreeStruct('(|b)'),expected([''
 test('(b|b) trivial dedup', t => t.is(buildTreeStruct('(|b||b|)'), expected(['', 'b'])));
 test('a(b|c) ordered part in step obj', t => t.is(buildTreeStruct('a(b|c)'), expected({step: ['a', ['b', 'c']]})));
 test('a(b) flat merge when no alt', t => t.is(buildTreeStruct('a(b)'), expected('ab')));
-
-test('build tree with (|) pattern', t => t.is(buildTreeStruct('(a(|b@@@@c|d|)|(e|f)|g|h@@@@i)'),
-		expected([
-			{step:['a',[
-					'',
-					'b@@@@c',
-					'd'
-				]]},
-			'e',
-			'f',
-			'g',
-			'h@@@@i'
-		])));
+test('build complexe tree with (|) pattern', t => t.is(buildTreeStruct('(a(|b@@@@c|d|)|(e|f)|g|h@@@@i)'),
+	expected([{step: ['a', ['', 'b@@@@c', 'd']]}, 'e', 'f', 'g', 'h@@@@i'])
+));
+test('serialize tree to a(b|c)', t => t.is(app.serialize({step: ['a', ['b', 'c']]}), 'a(b|c)'));
+test('serialize incorrect tree throw', t => t.throws(() => app.serialize({plop: ['a']})));