From 4355fd0c27f17e08d28b8be9b58c651ab7ed9864 Mon Sep 17 00:00:00 2001
From: "[1000i100] Millicent Billette" <git@1000i100.fr>
Date: Wed, 31 Mar 2021 04:16:34 +0200
Subject: [PATCH] WiP: dictionary stuff (dictionary-tree)
---
src/dictionary-parser.mjs | 58 ++++++++++++++++-----
src/dictionary-parser.test.mjs | 2 +-
src/dictionary-tree.mjs | 94 +++++++++++++++++++++-------------
src/dictionary-tree.test.mjs | 32 +++++-------
4 files changed, 116 insertions(+), 70 deletions(-)
diff --git a/src/dictionary-parser.mjs b/src/dictionary-parser.mjs
index b263a66..71f208b 100644
--- a/src/dictionary-parser.mjs
+++ b/src/dictionary-parser.mjs
@@ -1,25 +1,59 @@
+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){
+
+function parseAllLines(lines) {
return lines;
}
-function parseLine(theLine,allLines){
+
+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));
- if(unspecifiedLines.length) idSecPwdLines.push(`${mergeLines(unspecifiedLines)}@@@@${mergeLines(unspecifiedLines)}`)
+ 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 mergeLines(lines){
- console.log('lines:',lines);
- if(!lines.length) return '';
- lines = lines.filter(l=>l.length);
- if(lines.length===1) return lines[0];
+
+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);
+ if (!lines.length) return '';
+ lines = lines.filter(l => l.length);
+ if (lines.length === 1) return lines[0];
return `(${lines.join('|')})`;
}
+
+/*
+const json1 = '(a(b@@@@c|d)|(e|f)|g|h@@@@i)'
+const json2 = {
+ 'a':'a'{
+ 'b@@@@c':'b@@@@c',
+ 'd':'d'
+ },
+ {
+ 'e':'e',
+ 'f':'f'
+ },
+ 'g':'g',
+ 'h@@@@i':'h@@@@i'
+}
+*/
diff --git a/src/dictionary-parser.test.mjs b/src/dictionary-parser.test.mjs
index 1bf239e..fdd9aae 100644
--- a/src/dictionary-parser.test.mjs
+++ b/src/dictionary-parser.test.mjs
@@ -16,5 +16,5 @@ h@@@@(i|j)
`), '(a@@@@(b|c)|h@@@@(i|j)|((d|e)|(f|g))@@@@((d|e)|(f|g)))');
});
test.skip('no @@@@ cases combined with @@@@ case on same line recombine to allways finish with @@@@ case', t => {
- t.is(app.parse('(a|b|c@@@@d)'), '(c@@@@d|(a|b)@@@@(a|b))');
+ t.is(app.parse('(a(b@@@@c|d)|(e|f)|g|h@@@@i)'), '(g@@@@h|(a|b|c)@@@@(a|b|c))');
});
diff --git a/src/dictionary-tree.mjs b/src/dictionary-tree.mjs
index 9e1667e..c07d603 100644
--- a/src/dictionary-tree.mjs
+++ b/src/dictionary-tree.mjs
@@ -2,10 +2,10 @@ export function build(dictionaryString) {
}
export function buildTreeStruct(monoLineString) {
- console.log('srcTree', monoLineString);
+ //console.log('srcTree', monoLineString);
const stringAsArray = monoLineString.split('');
const rawTree = leftParser(stringAsArray);
- console.log('rawTree', JSON.stringify(rawTree));
+ //console.log('rawTree', JSON.stringify(rawTree));
const outOfScope = stringAsArray.length;
if (outOfScope) throw `fail to build tree from : "${monoLineString}" parsed: ${JSON.stringify(rawTree)} unparsed/failed: ${stringAsArray.join('')}`;
//return rawTree;//trivialDedup(flattenTree(rawTree))
@@ -15,70 +15,82 @@ export function buildTreeStruct(monoLineString) {
/**
* leftParser stuff
*/
-function flushNoEmptyString(data){
- if (data.str.length) data.tree[data.tree.length-1].step.push(data.str);
+function flushNoEmptyString(data) {
+ if (data.str.length) data.tree[data.tree.length - 1].step.push(data.str);
data.str = '';
}
-function appendToString(chr,data){
+
+function appendToString(chr, data) {
data.str += chr;
}
-function nextAlt(data){
+
+function nextAlt(data) {
data.tree.push({step: []});
}
-function goIntoNewAlt(data){
- data.tree[data.tree.length-1].step.push(leftParser(data.input));
+
+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)
+ ')': (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,
- str:'',
- tree:[{step: []}],
+ input: inputChrArray,
+ str: '',
+ tree: [{step: []}],
};
while (data.input.length) {
const chr = data.input.shift();
- if(typeof leftParserFunc[chr]!=='undefined') {
- const out = leftParserFunc[chr](chr,data);
- if(out) return data.tree;
- } else leftParserFunc['default'](chr,data);
+ if (typeof leftParserFunc[chr] !== 'undefined') {
+ const out = leftParserFunc[chr](chr, data);
+ if (out) return data.tree;
+ } else leftParserFunc['default'](chr, data);
}
flushNoEmptyString(data);
return data.tree;
}
+
// end leftParser
-function concatStrings(array){
+function concatStrings(array) {
const data = {
- input:array,
- str:'',
- tree:[{step: []}],
+ input: array,
+ str: '',
+ tree: [{step: []}],
};
- array.forEach(e=>{
- if(typeof e === 'string') appendToString(e,data);
- else{
+ array.forEach(e => {
+ if (typeof e === 'string') appendToString(e, data);
+ else {
flushNoEmptyString(data)
- data.tree[data.tree.length-1].step.push(e);
+ data.tree[data.tree.length - 1].step.push(e);
}
});
flushNoEmptyString(data)
- return data.tree[data.tree.length-1].step;
+ 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 '';
- if (Array.isArray(tree) && tree.length===1) return flattenTree(tree[0]);
+ if (Array.isArray(tree) && tree.length === 0) return '';
+ if (Array.isArray(tree) && tree.length === 1) return flattenTree(tree[0]);
if (isStep(tree)) {
tree.step = tree.step.map(flattenTree);
if (!Array.isArray(tree.step)) return flattenTree(tree.step);
if (tree.step.length === 1) return flattenTree(tree.step);
if (tree.step.length === 0) return '';
- if(!tree.step.reduce((acc,cur)=>Array.isArray(cur)||acc,false)) return concatStrings(tree.step);
+ if (!tree.step.reduce((acc, cur) => Array.isArray(cur) || acc, false)) return concatStrings(tree.step);
return tree;
}
@@ -93,7 +105,7 @@ function flattenTree(tree) {
//console.log('tree:', JSON.stringify(tree));
///if(isStep(tree[0])) tree[0].step = flattenTree(tree[0].step);
tree = trivialDedup(tree.map(flattenTree));
- tree = [].concat(...tree.map(e=>Array.isArray(e) ? e : [e]));
+ tree = [].concat(...tree.map(e => Array.isArray(e) ? e : [e]));
return tree;
}
@@ -101,13 +113,21 @@ 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))};
- if (Array.isArray(tree)){
+ if (isStep(tree)) return {step: tree.step.map(subTree => trivialDedup(subTree))};
+ if (Array.isArray(tree)) {
const dedupKeys = {};
tree.forEach(v => dedupKeys[JSON.stringify(v)] = v); // eslint-disable-line no-return-assign
- return Object.keys(dedupKeys).map(json=>trivialDedup(JSON.parse(json)));
+ return Object.keys(dedupKeys).map(json => trivialDedup(JSON.parse(json)));
}
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}`);
+}
diff --git a/src/dictionary-tree.test.mjs b/src/dictionary-tree.test.mjs
index 73cf5c5..620877b 100644
--- a/src/dictionary-tree.test.mjs
+++ b/src/dictionary-tree.test.mjs
@@ -2,27 +2,19 @@ import test from 'ava';
import * as app from './dictionary-tree.mjs';
const buildTreeStruct = str => JSON.stringify(app.buildTreeStruct(str));
-const expected= str => JSON.stringify(str);
+const expected = str => JSON.stringify(str);
-test('simple string still simple string',t=>t.is(buildTreeStruct('abc'),expected('abc')));
-test('(a|b) alt as array',t=>t.is(buildTreeStruct('(a|b)'),expected(['a','b'])));
+test('simple string still simple string', t => t.is(buildTreeStruct('abc'), expected('abc')));
+test('(a|b) alt as array', t => t.is(buildTreeStruct('(a|b)'), expected(['a', 'b'])));
//test('a) throw',t=>t.throws(()=>buildTreeStruct('a)')));
//test('(a throw',t=>t.throws(()=>buildTreeStruct('(a')));
//test('a|b throw',t=>t.throws(()=>buildTreeStruct('a|b')));
-test('(|b) empty choice is choice', t=>t.is(buildTreeStruct('(|b)'),expected(['','b'])));
-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('(|b) empty choice is choice', t => t.is(buildTreeStruct('(|b)'), expected(['', 'b'])));
+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 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']})));
--
GitLab