gatsby-remark-related-posts/gatsby-node.js

Calculate the similarity between posts by tf-idf and Cosine similarity.

var __assign = (this && this.__assign) || function () {
    __assign = Object.assign || function(t) {
        for (var s, i = 1, n = arguments.length; i < n; i++) {
            s = arguments[i];
            for (var p in s) if (Object.prototype.hasOwnProperty.call(s, p))
                t[p] = s[p];
        }
        return t;
    };
    return __assign.apply(this, arguments);
};
var __awaiter = (this && this.__awaiter) || function (thisArg, _arguments, P, generator) {
    function adopt(value) { return value instanceof P ? value : new P(function (resolve) { resolve(value); }); }
    return new (P || (P = Promise))(function (resolve, reject) {
        function fulfilled(value) { try { step(generator.next(value)); } catch (e) { reject(e); } }
        function rejected(value) { try { step(generator["throw"](value)); } catch (e) { reject(e); } }
        function step(result) { result.done ? resolve(result.value) : adopt(result.value).then(fulfilled, rejected); }
        step((generator = generator.apply(thisArg, _arguments || [])).next());
    });
};
var __generator = (this && this.__generator) || function (thisArg, body) {
    var _ = { label: 0, sent: function() { if (t[0] & 1) throw t[1]; return t[1]; }, trys: [], ops: [] }, f, y, t, g;
    return g = { next: verb(0), "throw": verb(1), "return": verb(2) }, typeof Symbol === "function" && (g[Symbol.iterator] = function() { return this; }), g;
    function verb(n) { return function (v) { return step([n, v]); }; }
    function step(op) {
        if (f) throw new TypeError("Generator is already executing.");
        while (_) try {
            if (f = 1, y && (t = op[0] & 2 ? y["return"] : op[0] ? y["throw"] || ((t = y["return"]) && t.call(y), 0) : y.next) && !(t = t.call(y, op[1])).done) return t;
            if (y = 0, t) op = [op[0] & 2, t.value];
            switch (op[0]) {
                case 0: case 1: t = op; break;
                case 4: _.label++; return { value: op[1], done: false };
                case 5: _.label++; y = op[1]; op = [0]; continue;
                case 7: op = _.ops.pop(); _.trys.pop(); continue;
                default:
                    if (!(t = _.trys, t = t.length > 0 && t[t.length - 1]) && (op[0] === 6 || op[0] === 2)) { _ = 0; continue; }
                    if (op[0] === 3 && (!t || (op[1] > t[0] && op[1] < t[3]))) { _.label = op[1]; break; }
                    if (op[0] === 6 && _.label < t[1]) { _.label = t[1]; t = op; break; }
                    if (t && _.label < t[2]) { _.label = t[2]; _.ops.push(op); break; }
                    if (t[2]) _.ops.pop();
                    _.trys.pop(); continue;
            }
            op = body.call(thisArg, _);
        } catch (e) { op = [6, e]; y = 0; } finally { f = t = 0; }
        if (op[0] & 5) throw op[1]; return { value: op[0] ? op[1] : void 0, done: true };
    }
};
var __importDefault = (this && this.__importDefault) || function (mod) {
    return (mod && mod.__esModule) ? mod : { "default": mod };
};
var _this = this;
Object.defineProperty(exports, "__esModule", { value: true });
exports.onPostBootstrap = exports.createSchemaCustomization = void 0;
var kuromoji_1 = __importDefault(require("kuromoji"));
var natural_1 = require("natural");
var path_1 = __importDefault(require("path"));
var crypto_1 = __importDefault(require("crypto"));
var computeCosineSimilarity = require('compute-cosine-similarity');
// settings
var default_option = {
    doc_lang: 'en',
    target_node: 'MarkdownRemark',
    getMarkdown: function (node) { return node.rawMarkdownBody; },
    each_bow_size: 30,
};
// utils
var md5 = function (str) {
    var md5 = crypto_1.default.createHash('md5');
    return md5.update(str, 'binary').digest('hex');
};
var calcVectorSimilarity = function (v1, v2) {
    if (v1.length !== v2.length)
        throw new Error("Both vector's size must be equal");
    return computeCosineSimilarity(v1, v2);
};
var vector_similarity_memo = new Map();
var getMemorizedVectorSimilarity = function (v1, v2) {
    var id = v1.id < v2.id ? "".concat(v1.id, " ").concat(v2.id) : "".concat(v2.id, " ").concat(v1.id);
    var memorized_similarity = vector_similarity_memo.get(id);
    if (memorized_similarity !== undefined)
        return memorized_similarity;
    var similarity = calcVectorSimilarity(v1.vector, v2.vector);
    vector_similarity_memo.set(id, similarity);
    return similarity;
};
var getRelatedPosts = function (id, bow_vectors) {
    var vector = bow_vectors.get(id);
    if (vector === undefined)
        return [];
    var vector_node = {
        id: id,
        vector: vector,
    };
    return Array.from(bow_vectors.entries())
        .sort(function (x, y) {
        var vector_x = {
            id: x[0],
            vector: x[1],
        };
        var vector_y = {
            id: y[0],
            vector: y[1],
        };
        return (getMemorizedVectorSimilarity(vector_y, vector_node) -
            getMemorizedVectorSimilarity(vector_x, vector_node));
    })
        .map(function (x) { return x[0]; });
};
var getTextFromMarkdown = function (markdown) {
    return markdown
        .replace(/```[\s\S]+?```/g, '')
        .replace(/---[\s\S]+?---/g, '')
        .replace(/\$[\s\S]+?\$/g, '')
        .replace(/\$\$[\s\S]+?\$\$/g, '')
        .replace(/^\|.*\|$/gm, '')
        .replace(/<.+?>/g, '')
        .replace(/http[^ ]+/g, '')
        .replace(/[\#\!\(\)\*\_\[\]\|\=\>\+\`\:\-]/g, '');
};
var getKuromojiTokenizer = function () { return __awaiter(_this, void 0, void 0, function () {
    return __generator(this, function (_a) {
        return [2 /*return*/, new Promise(function (resolve, reject) {
                kuromoji_1.default
                    .builder({
                    dicPath: path_1.default.join(path_1.default.dirname(require.resolve('kuromoji')), '../dict'),
                })
                    .build(function (err, tokenizer) {
                    if (err)
                        reject();
                    resolve(tokenizer);
                });
            })];
    });
}); };
var kuromoji_tokenizer = null;
var getSpaceSeparatedDoc = {
    en: function (doc) { return __awaiter(_this, void 0, void 0, function () {
        return __generator(this, function (_a) {
            return [2 /*return*/, doc.toLowerCase().split(' ')];
        });
    }); },
    ja: function (doc) { return __awaiter(_this, void 0, void 0, function () {
        return __generator(this, function (_a) {
            switch (_a.label) {
                case 0:
                    if (!(kuromoji_tokenizer === null)) return [3 /*break*/, 2];
                    return [4 /*yield*/, getKuromojiTokenizer()];
                case 1:
                    kuromoji_tokenizer = _a.sent();
                    _a.label = 2;
                case 2: return [2 /*return*/, kuromoji_tokenizer
                        .tokenize(doc)
                        .filter(function (x) {
                        return x.pos === '名詞' &&
                            ['一般', '固有名詞'].indexOf(x.pos_detail_1) !== -1;
                    })
                        .map(function (x) { return (x.basic_form !== '*' ? x.basic_form : x.surface_form); })];
            }
        });
    }); },
};
// gatsby api
var createSchemaCustomization = function (_a, user_option) {
    var actions = _a.actions;
    var option = __assign(__assign({}, default_option), user_option);
    actions.createTypes("\n    type related".concat(option.target_node, "s implements Node {\n      posts: [").concat(option.target_node, "]\n    }\n  "));
};
exports.createSchemaCustomization = createSchemaCustomization;
var onPostBootstrap = function (_a, user_option) {
    var actions = _a.actions, getNode = _a.getNode, getNodesByType = _a.getNodesByType, createNodeId = _a.createNodeId, reporter = _a.reporter, cache = _a.cache;
    return __awaiter(_this, void 0, void 0, function () {
        var option, nodes, docs, tfidf, _i, docs_1, doc, key, cached_ssd, ssd, doc_terms, all_keywords, tfidf_map_for_each_doc, bow_vectors;
        return __generator(this, function (_b) {
            switch (_b.label) {
                case 0:
                    option = __assign(__assign({}, default_option), user_option);
                    nodes = getNodesByType(option.target_node);
                    docs = nodes.map(function (node) { return ({
                        id: node.id,
                        text: option.getMarkdown(node),
                    }); });
                    tfidf = new natural_1.TfIdf();
                    _i = 0, docs_1 = docs;
                    _b.label = 1;
                case 1:
                    if (!(_i < docs_1.length)) return [3 /*break*/, 6];
                    doc = docs_1[_i];
                    key = "related-posts-ssd-".concat(md5(doc.text));
                    return [4 /*yield*/, cache.get(key)];
                case 2:
                    cached_ssd = _b.sent();
                    if (cached_ssd !== undefined) {
                        tfidf.addDocument(cached_ssd);
                        return [3 /*break*/, 5];
                    }
                    return [4 /*yield*/, getSpaceSeparatedDoc[option.doc_lang](getTextFromMarkdown(doc.text))];
                case 3:
                    ssd = _b.sent();
                    tfidf.addDocument(ssd);
                    return [4 /*yield*/, cache.set(key, ssd)];
                case 4:
                    _b.sent();
                    _b.label = 5;
                case 5:
                    _i++;
                    return [3 /*break*/, 1];
                case 6:
                    doc_terms = docs.map(function (_, i) {
                        return tfidf.listTerms(i)
                            .map(function (x) { return (__assign(__assign({}, x), { tfidf: x.tf * x.idf })); })
                            .sort(function (x, y) { return y.tfidf - x.tfidf; });
                    });
                    all_keywords = new Set();
                    tfidf_map_for_each_doc = [];
                    doc_terms.forEach(function (x, i) {
                        tfidf_map_for_each_doc[i] = new Map();
                        x.slice(0, option.each_bow_size).forEach(function (x) {
                            all_keywords.add(x.term);
                            tfidf_map_for_each_doc[i].set(x.term, x.tfidf);
                        });
                    });
                    bow_vectors = new Map();
                    docs.forEach(function (x, i) {
                        if (bow_vectors === null)
                            return;
                        bow_vectors.set(x.id, Array.from(all_keywords)
                            .map(function (x) { return tfidf_map_for_each_doc[i].get(x); })
                            .map(function (x) { return (x === undefined ? 0 : x); }));
                    });
                    reporter.info("[related-posts] bow vectors generated, dimention: ".concat(all_keywords.size));
                    // create related nodes
                    nodes.forEach(function (node) {
                        var related_nodes = getRelatedPosts(node.id, bow_vectors)
                            .slice(1)
                            .map(function (id) { return getNode(id); });
                        var digest = "".concat(node.id, " >>> related").concat(option.target_node, "s");
                        actions.createNode({
                            id: createNodeId(digest),
                            parent: node.id,
                            internal: {
                                type: "related".concat(option.target_node, "s"),
                                contentDigest: digest,
                            },
                            posts: related_nodes,
                        });
                    });
                    return [2 /*return*/];
            }
        });
    });
};
exports.onPostBootstrap = onPostBootstrap;