Доделал в начальном виде распознавание текста страницы самлиб

server/core/BookConverter/index.js

@@ -1,7 +1,11 @@
 const fs = require('fs-extra');
-const FileDetector = require('../FileDetector');
 const URL = require('url').URL;
-const EasySAXParser = require('./easysax');
+const iconv = require('iconv-lite');
+const chardet = require('chardet');
+const _ = require('lodash');
+
+const FileDetector = require('../FileDetector');
+const EasySAXParser = require('./easysaxmod');
 
 class BookConverter {
     constructor() {
@@ -12,9 +16,9 @@ class BookConverter {
         const fileType = await this.detector.detectFile(inputFile);
         
         if (fileType && (fileType.ext == 'html' || fileType.ext == 'xml')) {
-            const data = await fs.readFile(inputFile, 'utf8');
+            const data = await fs.readFile(inputFile);
 
-            if (data.indexOf('<FictionBook') >= 0) {            
+            if (data.toString().indexOf('<FictionBook') >= 0) {            
                 await fs.writeFile(outputFile, data);
                 return;
             }
@@ -26,7 +30,6 @@ class BookConverter {
                 return;
             }
 
-
             //Заглушка
             await fs.writeFile(outputFile, data);
             callback(100);
@@ -39,12 +42,30 @@ class BookConverter {
     }
 
     async convertSamlib(data) {
-        let fb2 = [{parentName: 'description'}];
+        let titleInfo = {};
+        let desc = {_n: 'description', 'title-info': titleInfo};
+        let pars = [];
+        let body = {_n: 'body', section: {_a: [pars]}};
+        let fb2 = [desc, body];
+
         let path = '';
-        let tag = '';
+        let tag = '';// eslint-disable-line no-unused-vars
 
         let inText = false;
 
+        const newParagraph = () => {
+            pars.push({_n: 'p', _t: ''});
+        };
+
+        const growParagraph = (text) => {
+            const l = pars.length;
+            if (l) {
+                if (pars[l - 1]._t == '')
+                    text = text.trimLeft();
+                pars[l - 1]._t += text;
+            }
+        };
+
         const parser = new EasySAXParser();
 
         parser.on('error', (msgError) => {// eslint-disable-line no-unused-vars
@@ -57,7 +78,10 @@ class BookConverter {
             if (!inText) {
                 path += '/' + elemName;
                 tag = elemName;
-console.log(path);
+            } else {
+                if (elemName == 'p' || elemName == 'dd') {
+                    newParagraph();
+                }
             }
 
         });
@@ -78,14 +102,28 @@ console.log(path);
 
                 let i = path.lastIndexOf('/');
                 tag = path.substr(i + 1);
-
-    console.log('cl', elemName);            
-    console.log('tag', tag);            
-    console.log(path);
             }
         });
 
         parser.on('textNode', (text) => {// eslint-disable-line no-unused-vars
+            switch (path) {
+                case '/html/body/center/h2':
+                    titleInfo['book-title'] = text;
+                    return;
+                case '/html/body/div/h3':
+                    if (!titleInfo.author)
+                        titleInfo.author = {};
+                    text = text.replace(':', '').trim().split(' ');
+                    if (text[0])
+                        titleInfo.author['last-name'] = text[0];
+                    if (text[1])
+                        titleInfo.author['first-name'] = text[1];
+                    if (text[2])
+                        titleInfo.author['middle-name'] = text[2];
+                    return;
+            }
+            if (inText)
+                growParagraph(text);
         });
 
         parser.on('cdata', (data) => {// eslint-disable-line no-unused-vars
@@ -100,7 +138,22 @@ console.log(path);
         });
         */
 
-        await parser.parse(data);
+        await parser.parse(iconv.decode(data, chardet.detect(data)));
+
+        const title = (titleInfo['book-title'] ? titleInfo['book-title'] : '');
+        let author = '';
+        if (titleInfo.author) {
+            author = _.compact([
+                (titleInfo.author['last-name'] ? titleInfo.author['last-name'] : ''),
+                (titleInfo.author['first-name'] ? titleInfo.author['first-name'] : ''),
+                (titleInfo.author['middle-name'] ? titleInfo.author['middle-name'] : ''),
+            ]).join(' ');
+        }
+
+        pars.unshift({_n: 'title', _a: [
+            {_n: 'p', _t: author}, {_n: 'p', _t: ''},
+            {_n: 'p', _t: title}, {_n: 'p', _t: ''},
+        ]})
 
         return this.formatFb2(fb2);
     }
@@ -120,20 +173,26 @@ console.log(out);
             for (const n of node) {
                 out += this.formatFb2Node(n);
             }
+        } else if (typeof node == 'string') {
+            out += `<${name}>${node}</${name}>`;
         } else {
-            if (node.parentName)
+            if (node._n)
-                name = node.parentName;
+                name = node._n;
             if (!name)
                 throw new Error(`malformed fb2 object`);
 
             out += `<${name}>`;
+            if (node.hasOwnProperty('_t')) {
+                out += node._t;
+            } else {
                 for (let nodeName in node) {
-                if (nodeName == 'parentName')
+                    if (nodeName == '_n')
                         continue;
 
                     const n = node[nodeName];
                     out += this.formatFb2Node(n, nodeName);
                 }
+            }
             out += `</${name}>`;
         }
         return out;