added utils folder; added several suplamentary modules

1 files changed, 304 insertions, 0 deletions

diff --git a/pieces/retro-search.js b/pieces/retro-search.js
new file mode 100644
index 0000000..ba4d937
--- /dev/null
+++ b/pieces/retro-search.js
@@ -0,0 +1,304 @@
+const kb = require('../utils/kb-util.js');
+const match = require('../utils/match-util.js');
+
+const products = require('../data/products.json');
+
+/** VARIABLES 
+ * may passed as module arguments
+ * -----------------------------------------------------------------------------
+ *//////////////////////////////////////////////////////////////////////////////
+
+
+var _allowFuzzy = true;     // enable|disable fuzzy search
+var n = 2;                  // Ngram base
+var _fuzzyLimit = .5;       // minimum bigram score for being considered a match
+var max_list = 24;
+var tolerance = 42;
+var STORE = { id: 904 };
+
+var _kwlinks;           // keyword links (word-connections; imported via ajax-get)
+var _products = [];     // all products (imported via ajax-get)
+
+// setup options
+var _maxResults = options.max_list;     // limit suggestions
+var _blendProds = 4;      // minimum final-produncts to blend with next-word suggestions
+var _Ngram_base = 2;      // number of N in Ngram spliting algorithm
+var _isReady = false;     // whether the searchbox is ready to be used
+
+// product keywords
+var keywordsURL = options.keywords_json;
+
+var cursor_on = { none: true };     // what product is highlighted; if not on product then { none: true }
+    
+    
+
+
+ 
+  
+/** SUPPLEMENTARY FUNCTIONS
+ * -----------------------------------------------------------------------------
+ *//////////////////////////////////////////////////////////////////////////////
+
+
+
+// callback function for sorting resulrs per r (=rating) property
+function compare_rate(a,b) {
+    return (a.r < b.r);
+}
+
+
+
+/** SEARCH ENGINE
+ * -------------------------------------------------------------------------
+ *//////////////////////////////////////////////////////////////////////////
+
+
+
+
+function matchWordInList(q, list = false) {
+
+    if (list !== false && list.length == 0)     return [];  // no results
+
+    var result = [];
+    var firstPass = false;
+
+    if (list === false) {
+        firstPass = true;       // on first pass
+        list = products;        // list is all products
+    } 
+
+    for(let i = 0 ; i < list.length ; i++) {
+
+        // check exact + rate
+
+        // else check partial + rate
+
+        // else check similarity + rate
+
+            let found = false;
+            let similarity = 0;
+            products[i].kb.split(' ').forEach( w => {
+                let sim = match.resemblance(q, w, 2);
+                if (sim > 0.5) {
+                    found = true;
+                    similarity = (sim > similarity) ? sim : similarity;
+                }    
+            });
+            if (found) {
+                products[i].similarity = similarity;
+                products[i].rating = (firstPass)
+                    ? similarity * 2.0
+                    : products[i].rating + similarity * 2.0
+                result.push(products[i]);
+            } 
+    }
+
+    return result
+}
+
+
+
+// suggestions engine //////////////////////////////////////////////////
+// ---
+function suggestions_engine(query) {
+    var results = [];
+    var pot = [];   pot.length = 0;
+
+    // clean and sanitize and mark links onto q(uery) string
+    var q = kb.keyboardize( kb.sanitizeGR( kb.clean(query.trim()) ) ).trim();
+    
+    // TODO:
+    // construct direct-linked words
+    // = do unequivocally replaces
+    // steps:
+    // 1. replace accented vowels with non accented ones
+    // 2. replace `/some pattern/gi , 'SOME-REPLACE-PATTERN'`
+
+    var qAr = q.split(' ');     // split to words
+    /// if (space_ended) qAr.push(' ');     // if space-end existed, push a space to query array
+    /// 
+    /// if (qAr.slice(-1) == "") {
+    ///     qAr.pop();
+    /// }
+
+
+    pot = _products;      // potential results // NOTE: CRITICAL: BY REFERENCE
+
+    var wi = 0;           // word index (from list)
+    var wc = qAr.length;
+
+    qAr.forEach( w => {
+
+        let sf = [];      // (matches) so far
+        let mi;           // position of match
+        wi++;
+
+        pot.forEach( it => {
+            let matched = false;
+            let tester = ' '+ it.kb + ' ';
+
+            // reset previous history and ratings
+            if (wi == 1) {
+                it.r = 0;
+                it.history = [];
+            }
+
+
+            // rate word-match > start-match > simple-match
+            // ... up to 8 points
+
+            if (tester.indexOf(' '+ w +' ') != -1) {
+                it.r += 9;
+                it.history.push({ w: w, rate: 9 });
+                matched = true;
+            }
+            else if (tester.indexOf(' '+ w) != -1) {
+                it.r += 5;
+                it.history.push({ w: w, rate: 5 });
+                matched = true;
+            }
+            else if (tester.indexOf(w) != -1) {
+                it.r += 2;
+                it.history.push({ w: w, rate: 2 });
+                matched = true;
+            }
+            
+            // rate `near-to-start` matching .. up to 7p
+            // rate `earlyness` of word in query .. up to 7p
+
+            if ((mi = tester.indexOf(' '+w)) != -1) {
+                let fc1 = 100 - ((mi < 99) ? mi : 99);    // near-to-start factor
+                let fc2 = wc - wi + 1;                    // query earlyness factor
+                let r1 = Math.floor(7*fc1/100);
+                let r2 = Math.floor(7*fc2/wc);
+
+                it.r += (r1 + r2);
+                it.history.push({ w: w, left: [fc1, r1], early: [fc2, r2] });
+            }
+            if (matched) sf.push(it);
+        });
+
+        if ((sf.length > (_maxResults + Math.floor(_maxResults/2)))
+        || (wi == 1) ) {
+            // ..if pot has a fair amount (= max + 50%) of results
+            // ..or these are results of '1st-query-word'
+            // set sf as new source
+            pot.lenght = 0; pot = [];
+            pot = JSON.parse(JSON.stringify(sf));   // copy by value
+
+        } else {
+            // else.. keep the source list and increase of 'so-far rating'
+            // console.log('found small list', sf, pot)
+            pot.forEach( it => {
+                sf.forEach( si => {
+                    if (it.id == si.id) {
+                        it.r += 10;
+                        it.history.push({ w: w, plus: '+10'});
+                    }
+                });
+            });
+        }
+    });
+
+    // sort results, get max-list of best rated
+    results = (pot.length > _maxResults)
+        ? pot.sort(compare_rate).slice(0, _maxResults)
+        : pot.sort(compare_rate)
+
+    if (options.debug) console.log(results);
+
+    return results;
+}
+
+
+// sub-module (start)
+////////////////////////////////////////////////////////////////////////////
+
+function update_common_search_results(q, results) {
+    let queries = getSessionObj('sr');
+    let newSRlist = [];
+    let isnewQ = true;
+    if (queries === null) {
+        setSessionObj('sr', [{
+            q: q,
+            result: result,
+            t: + new Date()
+        }]);
+        return true;
+
+    } else {
+
+        queries.forEach(it => {
+            if (it.q == q) {
+                newSRlist.push({
+                    q:q,
+                    result: result,
+                    t: + new Date()
+                });
+                isnewQ = false;
+            } else { newSRlist.push(it); }
+        });
+
+        if (isnewQ) {
+            newSRlist.push({
+                q:q,
+                result: result,
+                t: + new Date()
+            });
+        }
+
+        return true;
+    }
+}
+
+
+
+////////////////////////////////////////////////////////////////////////////
+// sub-module (end)
+
+
+function common_search(query) {
+    // clear ; sanitize ; split
+    var qAr = keyboardize( sanitize_GR( clean_text(query) ) ).toLowerCase().split(' ');
+
+    // if last item is empty, remove it
+    if ((qAr.slice(-1) == ' ') || (qAr.slice(-1) == '')) qAr.pop()
+
+    var results = _products;
+
+    // for each key fitler results
+    qAr.forEach( key => {
+        results = key_sublist(key, results)
+    });
+
+    // echo products (and prepare list to POST)
+    var list_ = [];
+    results.forEach( item => {
+        if (options.debug) console.log(item.id, ':', item.w);
+        list_.push(item.id)
+    })
+
+    // *** TODO: keep results in local storage (or on session storage)
+
+    // update_common_search_results(query, list_);
+
+    let l = list_.join(',');
+    var url = encodeURI(`${options.visualize_search_results_url}?search=${query}&eys_code=${l}`);
+
+    console.log('common search: search query > location = search')
+    window.location.href = encodeURI(`${options.visualize_search_results_url}?search=${query}`);
+
+}
+
+/** return from list only items that include 'key'
+ */
+function key_sublist(key, list) {
+    var result = [];
+    list.forEach( item => {
+        if (item.kb.includes(key)) {
+            result.push(item);
+        }
+    });
+    
+    return result;
+}