Merge branch 'release/0.4.2'

client/components/Reader/TextPage/DrawHelper.js

@@ -103,9 +103,8 @@ export default class DrawHelper {
                 //image: {local: Boolean, inline: Boolean, id: String, imageLine: Number, lineCount: Number, paraIndex: Number},
                 const img = part.image;
                 if (img && img.id && !img.inline && !imageDrawn.has(img.paraIndex)) {
-                    if (img.local) {
-                        const bin = this.parsed.binary[img.id];
-
+                    const bin = this.parsed.binary[img.id];
+                    if (bin) {
                         let imgH = img.lineCount*this.lineHeight;
                         imgH = (imgH <= bin.h ? imgH : bin.h);
                         let imgW = bin.w;
@@ -118,9 +117,11 @@ export default class DrawHelper {
 
                         const left = (this.w - imgW)/2;
                         const top = ((img.lineCount*this.lineHeight - imgH)/2) + (i - img.imageLine)*this.lineHeight;
-                        lineText += `<img src="data:${bin.type};base64,${bin.data}" style="position: absolute; left: ${left}px; top: ${top}px; ${resize}"/>`;
-                    } else {
-                        //
+                        if (img.local) {
+                            lineText += `<img src="data:${bin.type};base64,${bin.data}" style="position: absolute; left: ${left}px; top: ${top}px; ${resize}"/>`;
+                        } else {
+                            lineText += `<img src="${img.id}" style="position: absolute; left: ${left}px; top: ${top}px; ${resize}"/>`;
+                        }
                     }
                     imageDrawn.add(img.paraIndex);
                 }
@@ -128,11 +129,13 @@ export default class DrawHelper {
                 if (img && img.id && img.inline) {
                     if (img.local) {
                         const bin = this.parsed.binary[img.id];
-                        let resize = '';
-                        if (bin.h > this.fontSize) {
-                            resize = `height: ${this.fontSize - 3}px`;
+                        if (bin) {
+                            let resize = '';
+                            if (bin.h > this.fontSize) {
+                                resize = `height: ${this.fontSize - 3}px`;
+                            }
+                            lineText += `<img src="data:${bin.type};base64,${bin.data}" style="${resize}"/>`;
                         }
-                        lineText += `<img src="data:${bin.type};base64,${bin.data}" style="${resize}"/>`;
                     } else {
                         //
                     }

client/components/Reader/share/BookParser.js

@@ -78,6 +78,10 @@ export default class BookParser {
                     resolve();
                 };
 
+                i.onerror = (e) => {
+                    reject(e);
+                };
+
                 i.src = `data:${binaryType};base64,${data}`;
                 await sleep(30*1000);
                 if (!resolved)
@@ -85,6 +89,30 @@ export default class BookParser {
             });
         };
 
+        const getExternalImageDimensions = (src) => {
+            return new Promise (async(resolve, reject) => {
+                const i = new Image();
+                let resolved = false;
+                i.onload = () => {
+                    resolved = true;
+                    this.binary[src] = {
+                        w: i.width,
+                        h: i.height,
+                    };
+                    resolve();
+                };
+
+                i.onerror = (e) => {
+                    reject(e);
+                };
+
+                i.src = src;
+                await sleep(30*1000);
+                if (!resolved)
+                    reject('Не удалось получить размер изображения');
+            });
+        };
+
         const newParagraph = (text, len, addIndex) => {
             paraIndex++;
             let p = {
@@ -147,21 +175,26 @@ export default class BookParser {
 
             if (tag == 'binary') {
                 let attrs = sax.getAttrsSync(tail);
-                binaryType = (attrs['content-type'].value ? attrs['content-type'].value : '');
+                binaryType = (attrs['content-type'] && attrs['content-type'].value ? attrs['content-type'].value : '');
                 if (binaryType == 'image/jpeg' || binaryType == 'image/png')
                     binaryId = (attrs.id.value ? attrs.id.value : '');
             }
 
             if (tag == 'image') {
                 let attrs = sax.getAttrsSync(tail);
-                if (attrs.href.value) {
-                    if (inPara && !this.showInlineImagesInCenter)
-                        growParagraph(`<image-inline href="${attrs.href.value}"></image-inline>`, 0);
-                    else
-                        newParagraph(`<image href="${attrs.href.value}">${' '.repeat(maxImageLineCount)}</image>`, maxImageLineCount);
-
-                    if (inPara && this.showInlineImagesInCenter)
-                        newParagraph(' ', 1);
+                if (attrs.href && attrs.href.value) {
+                    const href = attrs.href.value;
+                    if (href[0] == '#') {//local
+                        if (inPara && !this.showInlineImagesInCenter)
+                            growParagraph(`<image-inline href="${href}"></image-inline>`, 0);
+                        else
+                            newParagraph(`<image href="${href}">${' '.repeat(maxImageLineCount)}</image>`, maxImageLineCount);
+                        if (inPara && this.showInlineImagesInCenter)
+                            newParagraph(' ', 1);
+                    } else {//external
+                        dimPromises.push(getExternalImageDimensions(href));
+                        newParagraph(`<image href="${href}">${' '.repeat(maxImageLineCount)}</image>`, maxImageLineCount);
+                    }
                 }
             }
 
@@ -409,14 +442,14 @@ export default class BookParser {
                     break;
                 case 'space': {
                     let attrs = sax.getAttrsSync(tail);
-                    if (attrs.w.value)
+                    if (attrs.w && attrs.w.value)
                         style.space = attrs.w.value;
                     break;
                 }
                 case 'image': {
                     let attrs = sax.getAttrsSync(tail);
-                    let id = attrs.href.value;
-                    if (id) {
+                    if (attrs.href && attrs.href.value) {
+                        let id = attrs.href.value;
                         let local = false;
                         if (id[0] == '#') {
                             id = id.substr(1);
@@ -428,8 +461,8 @@ export default class BookParser {
                 }
                 case 'image-inline': {
                     let attrs = sax.getAttrsSync(tail);
-                    let id = attrs.href.value;
-                    if (id) {
+                    if (attrs.href && attrs.href.value) {
+                        let id = attrs.href.value;
                         let local = false;
                         if (id[0] == '#') {
                             id = id.substr(1);
@@ -616,7 +649,9 @@ export default class BookParser {
             //изображения
             if (part.image.id && !part.image.inline) {
                 parsed.visible = this.showImages;
-                const bin = this.binary[part.image.id];
+                let bin = this.binary[part.image.id];
+                if (!bin)
+                    bin = {h: 0, w: 0};
 
                 let lineCount = this.imageHeightLines;
                 const c = Math.ceil(bin.h/this.lineHeight);
@@ -643,15 +678,18 @@ export default class BookParser {
                 line.last = true;
                 line.parts.push({style, text: ' ',
                     image: {local: part.image.local, inline: false, id: part.image.id, imageLine: i, lineCount, paraIndex}});
+                
                 continue;
             }
 
             if (part.image.id && part.image.inline && this.showImages) {
                 const bin = this.binary[part.image.id];
-                let imgH = (bin.h > this.fontSize ? this.fontSize : bin.h);
-                imgW += bin.w*imgH/bin.h;
-                line.parts.push({style, text: '',
-                    image: {local: part.image.local, inline: true, id: part.image.id}});
+                if (bin) {
+                    let imgH = (bin.h > this.fontSize ? this.fontSize : bin.h);
+                    imgW += bin.w*imgH/bin.h;
+                    line.parts.push({style, text: '',
+                        image: {local: part.image.local, inline: true, id: part.image.id}});
+                }
             }
 
             let words = part.text.split(' ');

package.json

@@ -1,6 +1,6 @@
 {
   "name": "Liberama",
-  "version": "0.4.1",
+  "version": "0.4.2",
   "engines": {
     "node": ">=10.0.0"
   },

server/core/BookConverter/index.js

@@ -9,6 +9,7 @@ const textUtils = require('./textUtils');
 const FileDetector = require('../FileDetector');
 
 const repSpaces = (text) => text.replace(/ |[\t\n\r]/g, ' ');
+const repSpaces2 = (text) => text.replace(/[\n\r]/g, '');
 
 class BookConverter {
     constructor() {
@@ -31,7 +32,7 @@ class BookConverter {
             if (parsedUrl.hostname == 'samlib.ru' ||
                 parsedUrl.hostname == 'budclub.ru' ||
                 parsedUrl.hostname == 'zhurnal.lib.ru') {
-                await fs.writeFile(outputFile, this.convertSamlib(data));
+                await fs.writeFile(outputFile, this.convertSamlib(data, parsedUrl.hostname));
                 return;
             }
 
@@ -216,7 +217,7 @@ class BookConverter {
         return this.formatFb2(fb2);
     }
 
-    convertSamlib(data) {
+    convertSamlib(data, hostname) {
         let titleInfo = {};
         let desc = {_n: 'description', 'title-info': titleInfo};
         let pars = [];
@@ -225,20 +226,22 @@ class BookConverter {
 
         let inSubtitle = false;
         let inJustify = true;
+        let inImage = false;
         let path = '';
         let tag = '';// eslint-disable-line no-unused-vars
 
         let inText = false;
+        let textFound = false;
         let node = {_a: pars};
 
         let inPara = false;
         let italic = false;
         let bold = false;
 
-        const openTag = (name) => {
+        const openTag = (name, attrs) => {
             if (name == 'p')
                 inPara = true;
-            let n = {_n: name, _a: [], _p: node};
+            let n = {_n: name, _attrs: attrs, _a: [], _p: node};
             node._a.push(n);
             node = n;
         };
@@ -269,7 +272,7 @@ class BookConverter {
                 path += '/' + elemName;
                 tag = elemName;
             } else {
-                if (inPara && elemName != 'i' && elemName != 'b')
+                if (inPara && elemName != 'i' && elemName != 'b' && elemName != 'em' && elemName != 'strong' && elemName != 'img')
                     closeTag('p');
 
                 switch (elemName) {
@@ -279,12 +282,15 @@ class BookConverter {
                     case 'h1':
                     case 'h2':
                     case 'h3':
+                    case 'br':
                         openTag('p');
                         break;
                     case 'i':
+                    case 'em':
                         italic = true;
                         break;
                     case 'b':
+                    case 'strong':
                         bold = true;
                         break;
                     case 'div':
@@ -299,6 +305,17 @@ class BookConverter {
                         }
 
                         break;
+                    case 'img': {
+                        const attrs = sax.getAttrsSync(tail);
+                        if (attrs.src && attrs.src.value) {
+                            let href = attrs.src.value;
+                            if (href[0] == '/')
+                                href = `http://${hostname}${href}`;
+                            openTag('image', {href});
+                            inImage = true;
+                        }
+                        break;
+                    }
                 }
             }
         };
@@ -330,9 +347,11 @@ class BookConverter {
                         closeTag('p');
                         break;
                     case 'i':
+                    case 'em':
                         italic = false;
                         break;
                     case 'b':
+                    case 'strong':
                         bold = false;
                         break;
                     case 'div':
@@ -346,13 +365,20 @@ class BookConverter {
                             inJustify = false;
                         }
                         break;
+                    case 'img':
+                        if (inImage)
+                            closeTag('image');
+                        inImage = false;
+                        break;
                 }
             }
         };
 
         const onComment = (text) => {// eslint-disable-line no-unused-vars
-            if (text == '--------- Собственно произведение -------------')
+            if (text == '--------- Собственно произведение -------------') {
                 inText = true;
+                textFound = true;
+            }
             if (text == '-----------------------------------------------')
                 inText = false;
         };
@@ -390,11 +416,15 @@ class BookConverter {
                 growParagraph(`${tOpen}${text}${tClose}`);
         };
 
-        sax.parseSync(repSpaces(this.decode(data).toString()), {
+        sax.parseSync(repSpaces(repSpaces2(this.decode(data).toString())), {
             onStartNode, onEndNode, onTextNode, onComment,
             innerCut: new Set(['head', 'script', 'style'])
         });
 
+        //текст не найден на странице, обрабатываем как html
+        if (!textFound)
+            return this.convertHtml(data);
+
         const title = (titleInfo['book-title'] ? titleInfo['book-title'] : '');
         let author = '';
         if (titleInfo.author) {
@@ -437,8 +467,15 @@ class BookConverter {
             if (node._n)
                 name = node._n;
 
+            let attrs = '';
+            if (node._attrs) {
+                for (let attrName in node._attrs) {
+                    attrs += ` ${attrName}="${node._attrs[attrName]}"`;
+                }
+            }
+
             if (name)
-                out += `<${name}>`;
+                out += `<${name}${attrs}>`;
             if (node.hasOwnProperty('_t'))
                 out += repSpaces(node._t);
 

server/core/FileDownloader.js

@@ -8,8 +8,14 @@ class FileDownloader {
 
     async load(url, callback) {
         let errMes = '';
+        const options = {
+            encoding: null,
+            headers: {
+                'user-agent': 'Mozilla/5.0 (X11; HasCodingOs 1.0; Linux x64) AppleWebKit/637.36 (KHTML, like Gecko) Chrome/70.0.3112.101 Safari/637.36 HasBrowser/5.0'
+            }
+        };
 
-        const response = await got(url, {method: 'HEAD'});
+        const response = await got(url, Object.assign({}, options, {method: 'HEAD'}));
 
         let estSize = 0;
         if (response.headers['content-length']) {
@@ -17,7 +23,7 @@ class FileDownloader {
         }
 
         let prevProg = 0;
-        const request = got(url, {encoding: null}).on('downloadProgress', progress => {
+        const request = got(url, options).on('downloadProgress', progress => {
             if (progress.transferred > maxDownloadSize) {
                 errMes = 'file too big';
                 request.cancel();