Compare commits

...

54 Commits
0.4.1 ... 0.4.7

Author SHA1 Message Date
Book Pauk
2df88f4280 Merge branch 'release/0.4.7' 2019-02-27 06:21:57 +07:00
Book Pauk
156a1b4aa8 Версия 0.4.7 2019-02-27 06:21:41 +07:00
Book Pauk
962eda7860 Поправки багов 2019-02-27 06:12:59 +07:00
Book Pauk
954ce9e85c Поправки 2019-02-27 06:00:30 +07:00
Book Pauk
ff27ddd442 Оптимизация - улучшение скорости загрузки читалки 2019-02-27 05:52:59 +07:00
Book Pauk
f9cc2ad70a Мелкая поправка 2019-02-27 04:13:56 +07:00
Book Pauk
4bcd45a795 Рефакторинг - вынес методы конвертирования в отдельные классы 2019-02-27 03:59:08 +07:00
Book Pauk
9e7ccd6e20 Мелкий рефакторинг 2019-02-26 18:23:17 +07:00
Book Pauk
631c5930e9 Мелкая поправка 2019-02-25 20:49:12 +07:00
Book Pauk
0ddb182642 Улучшение парсинга fb2 2019-02-25 17:14:44 +07:00
Book Pauk
de2f0e74c8 Небольшая поправка 2019-02-25 15:29:58 +07:00
Book Pauk
ad4ee6ccc9 Merge tag '0.4.6' into develop
0.4.6
2019-02-24 22:02:53 +07:00
Book Pauk
1670df02db Merge branch 'release/0.4.6' 2019-02-24 22:02:43 +07:00
Book Pauk
0268e647cd Версия 0.4.6 2019-02-24 22:02:27 +07:00
Book Pauk
4b9315c13c Поправка бага 2019-02-24 22:00:41 +07:00
Book Pauk
14a9948dd2 Merge tag '0.4.5' into develop
0.4.5
2019-02-24 21:37:21 +07:00
Book Pauk
060ec98b0c Merge branch 'release/0.4.5' 2019-02-24 21:37:11 +07:00
Book Pauk
9c1efe381e Версия 0.4.5 2019-02-24 21:36:49 +07:00
Book Pauk
6a1b052d16 Поправил баг 2019-02-24 21:36:06 +07:00
Book Pauk
d6151d541e Поправил баг 2019-02-24 21:27:18 +07:00
Book Pauk
3e90277e1e Добавил опцию imageFitWidth 2019-02-24 19:32:16 +07:00
Book Pauk
641cbdfe85 Merge tag '0.4.4' into develop
0.4.4
2019-02-24 00:41:59 +07:00
Book Pauk
c36e9b36d8 Merge branch 'release/0.4.4' 2019-02-24 00:41:48 +07:00
Book Pauk
ecc3acce93 Версия 0.4.4 2019-02-24 00:41:35 +07:00
Book Pauk
33a2ca55f0 Поправки 2019-02-24 00:32:19 +07:00
Book Pauk
9d0bbec4b3 Поправка отображения картинок внутри title 2019-02-24 00:00:55 +07:00
Book Pauk
4de0b3cffd Поправки парсера fb2 2019-02-23 23:51:06 +07:00
Book Pauk
06221a474b Мелкая поправка 2019-02-23 23:38:19 +07:00
Book Pauk
e99a42b7af К предыдущему 2019-02-23 23:20:03 +07:00
Book Pauk
37822e8409 Мелкая поправка 2019-02-23 23:18:03 +07:00
Book Pauk
2e477e6c99 Улучшение парсинга СИ 2019-02-23 23:03:01 +07:00
Book Pauk
360ee98d8d Улучшение парсинга fb2 2019-02-23 22:17:16 +07:00
Book Pauk
69afd7720a Поправки форматирования fb2 2019-02-23 20:45:21 +07:00
Book Pauk
a75590c493 Улучшение формирования fb2 2019-02-23 20:15:40 +07:00
Book Pauk
2acb65f6b3 Мелкие поправки 2019-02-23 20:04:41 +07:00
Book Pauk
1e1a58b58c Улучшен convertHtml 2019-02-23 19:47:16 +07:00
Book Pauk
aeadb5aeb8 Улучшено определение кодировки и текстового файла 2019-02-23 19:09:57 +07:00
Book Pauk
3e2f01d56d Мелкие поправки 2019-02-23 18:06:29 +07:00
Book Pauk
cad97e639a Добавил цвет фона 2019-02-23 14:37:40 +07:00
Book Pauk
e2632f1802 Добавил вывод версии в консоль 2019-02-23 14:24:29 +07:00
Book Pauk
9aa0bb2bde Merge tag '0.4.3' into develop
0.4.3
2019-02-23 14:17:13 +07:00
Book Pauk
f015d5f7ed Merge branch 'release/0.4.3' 2019-02-23 14:17:00 +07:00
Book Pauk
74f8f7f9a4 0.4.3 2019-02-23 14:16:43 +07:00
Book Pauk
2598538de9 Улучшил отзывчивость historyPage при открытии 2019-02-23 14:13:41 +07:00
Book Pauk
a78a00df2b Поправка label 2019-02-23 14:03:28 +07:00
Book Pauk
92f6beb64e Merge tag '0.4.2' into develop
0.4.2
2019-02-21 21:39:19 +07:00
Book Pauk
3920b71613 Merge branch 'release/0.4.2' 2019-02-21 21:39:09 +07:00
Book Pauk
d661150665 0.4.2 2019-02-21 21:38:46 +07:00
Book Pauk
ab29c80dab Поправки багов 2019-02-21 21:36:17 +07:00
Book Pauk
e5384e27e5 Поправки багов 2019-02-21 21:02:27 +07:00
Book Pauk
06cdc6eb63 Улучшение парсинга samlib 2019-02-21 20:26:56 +07:00
Book Pauk
da284c793e Добавил загрузку внешних изображений 2019-02-21 20:22:25 +07:00
Book Pauk
c2cef91eb3 Merge tag '0.4.1' into develop
0.4.1
2019-02-20 21:40:10 +07:00
Book Pauk
e272308823 Merge tag '0.4.0' into develop
0.4.0
2019-02-20 21:08:53 +07:00
20 changed files with 870 additions and 554 deletions

View File

@@ -11,7 +11,7 @@ const workerApi = axios.create({
class Reader { class Reader {
async loadBook(url, callback) { async loadBook(url, callback) {
const refreshPause = 200; const refreshPause = 300;
if (!callback) callback = () => {}; if (!callback) callback = () => {};
let response = await api.post('/load-book', {type: 'url', url}); let response = await api.post('/load-book', {type: 'url', url});

View File

@@ -15,8 +15,14 @@
border border
:default-sort = "{prop: 'touchDateTime', order: 'descending'}" :default-sort = "{prop: 'touchDateTime', order: 'descending'}"
:header-cell-style = "headerCellStyle" :header-cell-style = "headerCellStyle"
:row-key = "rowKey"
> >
<el-table-column
type="index"
width="35px"
>
</el-table-column>
<el-table-column <el-table-column
prop="touchDateTime" prop="touchDateTime"
min-width="90px" min-width="90px"
@@ -43,6 +49,7 @@
placeholder="Найти"/--> placeholder="Найти"/-->
<div class="el-input el-input--mini"> <div class="el-input el-input--mini">
<input class="el-input__inner" <input class="el-input__inner"
ref="input"
placeholder="Найти" placeholder="Найти"
style="margin: 0; padding: 0; vertical-align: bottom; margin-top: 20px; padding: 0 10px 0 10px" style="margin: 0; padding: 0; vertical-align: bottom; margin-top: 20px; padding: 0 10px 0 10px"
:value="search" @input="search = $event.target.value" :value="search" @input="search = $event.target.value"
@@ -118,9 +125,16 @@ class HistoryPage extends Vue {
created() { created() {
} }
mounted() { init() {
this.updateTableData(); this.updateTableData();
this.mostRecentBook = bookManager.mostRecentBook(); this.mostRecentBook = bookManager.mostRecentBook();
this.$nextTick(() => {
this.$refs.input.focus();
});
}
rowKey(row) {
return row.key;
} }
updateTableData() { updateTableData() {
@@ -172,13 +186,15 @@ class HistoryPage extends Vue {
} }
const search = this.search; const search = this.search;
this.tableData = result.filter(item => { result = result.filter(item => {
return !search || return !search ||
item.touchTime.includes(search) || item.touchTime.includes(search) ||
item.touchDate.includes(search) || item.touchDate.includes(search) ||
item.desc.title.toLowerCase().includes(search.toLowerCase()) || item.desc.title.toLowerCase().includes(search.toLowerCase()) ||
item.desc.author.toLowerCase().includes(search.toLowerCase()) item.desc.author.toLowerCase().includes(search.toLowerCase())
}); });
this.tableData = result;
} }
headerCellStyle(cell) { headerCellStyle(cell) {
@@ -210,7 +226,8 @@ class HistoryPage extends Vue {
this.updateTableData(); this.updateTableData();
const newRecent = bookManager.mostRecentBook(); const newRecent = bookManager.mostRecentBook();
if (this.mostRecentBook != newRecent)
if (!(this.mostRecentBook && newRecent && this.mostRecentBook.key == newRecent.key))
this.$emit('load-book', newRecent); this.$emit('load-book', newRecent);
this.mostRecentBook = newRecent; this.mostRecentBook = newRecent;

View File

@@ -69,7 +69,7 @@
@stop-text-search="stopTextSearch"> @stop-text-search="stopTextSearch">
</SearchPage> </SearchPage>
<CopyTextPage v-if="copyTextActive" ref="copyTextPage" @copy-text-toggle="copyTextToggle"></CopyTextPage> <CopyTextPage v-if="copyTextActive" ref="copyTextPage" @copy-text-toggle="copyTextToggle"></CopyTextPage>
<HistoryPage v-if="historyActive" ref="historyPage" @load-book="loadBook" @history-toggle="historyToggle"></HistoryPage> <HistoryPage v-show="historyActive" ref="historyPage" @load-book="loadBook" @history-toggle="historyToggle"></HistoryPage>
<SettingsPage v-if="settingsActive" ref="settingsPage" @settings-toggle="settingsToggle"></SettingsPage> <SettingsPage v-if="settingsActive" ref="settingsPage" @settings-toggle="settingsToggle"></SettingsPage>
<HelpPage v-if="helpActive" ref="helpPage" @help-toggle="helpToggle"></HelpPage> <HelpPage v-if="helpActive" ref="helpPage" @help-toggle="helpToggle"></HelpPage>
<ClickMapPage v-show="clickMapActive" ref="clickMapPage"></ClickMapPage> <ClickMapPage v-show="clickMapActive" ref="clickMapPage"></ClickMapPage>
@@ -415,6 +415,7 @@ class Reader extends Vue {
this.historyActive = !this.historyActive; this.historyActive = !this.historyActive;
if (this.historyActive) { if (this.historyActive) {
this.closeAllTextPages(); this.closeAllTextPages();
this.$refs.historyPage.init();
this.historyActive = true; this.historyActive = true;
} else { } else {
this.historyActive = false; this.historyActive = false;
@@ -655,7 +656,7 @@ class Reader extends Vue {
// если есть в локальном кэше // если есть в локальном кэше
if (bookParsed) { if (bookParsed) {
await bookManager.setRecentBook(Object.assign({bookPos, bookPosSeen, bookPosPercent}, bookManager.metaOnly(bookParsed))); await bookManager.setRecentBook(Object.assign({bookPos, bookPosSeen, bookPosPercent}, bookParsed));
this.mostRecentBook(); this.mostRecentBook();
this.addAction(bookPos); this.addAction(bookPos);
this.loaderActive = false; this.loaderActive = false;
@@ -698,7 +699,7 @@ class Reader extends Vue {
}); });
// добавляем в историю // добавляем в историю
await bookManager.setRecentBook(Object.assign({bookPos, bookPosSeen, bookPosPercent}, bookManager.metaOnly(addedBook))); await bookManager.setRecentBook(Object.assign({bookPos, bookPosSeen, bookPosPercent}, addedBook));
this.mostRecentBook(); this.mostRecentBook();
this.addAction(bookPos); this.addAction(bookPos);
this.updateRoute(true); this.updateRoute(true);
@@ -831,6 +832,8 @@ class Reader extends Vue {
break; break;
case 'KeyX': case 'KeyX':
this.historyToggle(); this.historyToggle();
event.preventDefault();
event.stopPropagation();
break; break;
case 'KeyS': case 'KeyS':
this.settingsToggle(); this.settingsToggle();

View File

@@ -185,7 +185,7 @@
<el-checkbox v-model="wordWrap">Перенос по слогам</el-checkbox> <el-checkbox v-model="wordWrap">Перенос по слогам</el-checkbox>
</el-form-item> </el-form-item>
<el-form-item label="Обработка"> <el-form-item label="Обработка">
<el-checkbox v-model="cutEmptyParagraphs">Убирать пустые параграфы</el-checkbox> <el-checkbox v-model="cutEmptyParagraphs">Убирать пустые строки</el-checkbox>
</el-form-item> </el-form-item>
<el-form-item label=""> <el-form-item label="">
<el-col :span="12"> <el-col :span="12">
@@ -207,12 +207,17 @@
<template slot="content"> <template slot="content">
Выносить все изображения в центр экрана Выносить все изображения в центр экрана
</template> </template>
<el-checkbox v-model="showInlineImagesInCenter" @change="needReload">Инлайн в центр</el-checkbox> <el-checkbox v-model="showInlineImagesInCenter" @change="needReload" :disabled="!showImages">Инлайн в центр</el-checkbox>
</el-tooltip> </el-tooltip>
</el-col> </el-col>
</el-form-item> </el-form-item>
<el-form-item label=""> <el-form-item label="">
Размер изображения <el-checkbox v-model="imageFitWidth" :disabled="!showImages">Ширина не более размера экрана</el-checkbox>
</el-form-item>
<el-form-item label="">
<el-col :span="12">
Высота не более
</el-col>
<el-tooltip :open-delay="500" effect="light" placement="top"> <el-tooltip :open-delay="500" effect="light" placement="top">
<template slot="content"> <template slot="content">
Определяет высоту изображения количеством строк.<br> Определяет высоту изображения количеством строк.<br>
@@ -220,7 +225,7 @@
уменьшено с сохранением пропорций так, чтобы<br> уменьшено с сохранением пропорций так, чтобы<br>
помещаться в указанное количество строк. помещаться в указанное количество строк.
</template> </template>
<el-input-number v-model="imageHeightLines" :min="1" :max="100"></el-input-number> <el-input-number v-model="imageHeightLines" :min="1" :max="100" :disabled="!showImages"></el-input-number>
</el-tooltip> </el-tooltip>
</el-form-item> </el-form-item>
</el-form> </el-form>
@@ -437,11 +442,12 @@ class SettingsPage extends Vue {
'#000000', '#000000',
'#202020', '#202020',
'#ebe2c9', '#ebe2c9',
'#cfdc99',
'#478355',
'#a6caf0',
'#909080', '#909080',
'#808080', '#808080',
'#c8c8c8', '#c8c8c8',
'#478355',
'#a6caf0',
]; ];
} }

View File

@@ -26,9 +26,9 @@ export default class DrawHelper {
const font = this.fontByStyle({}); const font = this.fontByStyle({});
const justify = (this.textAlignJustify ? 'text-align: justify; text-align-last: justify;' : ''); const justify = (this.textAlignJustify ? 'text-align: justify; text-align-last: justify;' : '');
let out = `<div style="width: ${this.w}px; height: ${this.h}px;` + let out = `<div style="width: ${this.w}px; height: ${this.h + (isScrolling ? this.lineHeight : 0)}px;` +
` position: absolute; top: ${this.fontSize*this.textShift}px; color: ${this.textColor}; font: ${font}; ${justify}` + ` position: absolute; top: ${this.fontSize*this.textShift}px; color: ${this.textColor}; font: ${font}; ${justify}` +
` line-height: ${this.lineHeight}px;">`; ` line-height: ${this.lineHeight}px; white-space: nowrap; overflow: hidden;">`;
let imageDrawn = new Set(); let imageDrawn = new Set();
let len = lines.length; let len = lines.length;
@@ -91,36 +91,32 @@ export default class DrawHelper {
} else } else
text = part.text; text = part.text;
if (text.trim() == '') if (text && text.trim() == '')
text = `<span style="white-space: pre">${text}</span>`; text = `<span style="white-space: pre">${text}</span>`;
lineText += `${tOpen}${text}${tClose}`; lineText += `${tOpen}${text}${tClose}`;
center = center || part.style.center; center = center || part.style.center;
space = (part.style.space > 0 ? part.style.space : space); space = (part.style.space > space ? part.style.space : space);
//избражения //избражения
//image: {local: Boolean, inline: Boolean, id: String, imageLine: Number, lineCount: Number, paraIndex: Number}, //image: {local: Boolean, inline: Boolean, id: String, imageLine: Number, lineCount: Number, paraIndex: Number, w: Number, h: Number},
const img = part.image; const img = part.image;
if (img && img.id && !img.inline && !imageDrawn.has(img.paraIndex)) { 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;
let resize = ''; let resize = '';
if (bin.h > imgH) { if (bin.h > img.h) {
resize = `height: ${imgH}px`; resize = `height: ${img.h}px`;
imgW = imgW*imgH/bin.h;
} }
const left = (this.w - imgW)/2; const left = (this.w - img.w)/2;
const top = ((img.lineCount*this.lineHeight - imgH)/2) + (i - img.imageLine)*this.lineHeight; const top = ((img.lineCount*this.lineHeight - img.h)/2) + (i - img.imageLine)*this.lineHeight;
if (img.local) {
lineText += `<img src="data:${bin.type};base64,${bin.data}" style="position: absolute; left: ${left}px; top: ${top}px; ${resize}"/>`; lineText += `<img src="data:${bin.type};base64,${bin.data}" style="position: absolute; left: ${left}px; top: ${top}px; ${resize}"/>`;
} else { } else {
// lineText += `<img src="${img.id}" style="position: absolute; left: ${left}px; top: ${top}px; ${resize}"/>`;
}
} }
imageDrawn.add(img.paraIndex); imageDrawn.add(img.paraIndex);
} }
@@ -128,11 +124,13 @@ export default class DrawHelper {
if (img && img.id && img.inline) { if (img && img.id && img.inline) {
if (img.local) { if (img.local) {
const bin = this.parsed.binary[img.id]; const bin = this.parsed.binary[img.id];
if (bin) {
let resize = ''; let resize = '';
if (bin.h > this.fontSize) { if (bin.h > this.fontSize) {
resize = `height: ${this.fontSize - 3}px`; 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 { } else {
// //
} }

View File

@@ -221,6 +221,7 @@ class TextPage extends Vue {
this.parsed.showImages = this.showImages; this.parsed.showImages = this.showImages;
this.parsed.showInlineImagesInCenter = this.showInlineImagesInCenter; this.parsed.showInlineImagesInCenter = this.showInlineImagesInCenter;
this.parsed.imageHeightLines = this.imageHeightLines; this.parsed.imageHeightLines = this.imageHeightLines;
this.parsed.imageFitWidth = this.imageFitWidth;
} }
//statusBar //statusBar

View File

@@ -45,6 +45,9 @@ export default class BookParser {
let italic = false; let italic = false;
let space = 0; let space = 0;
let inPara = false; let inPara = false;
let isFirstBody = true;
let isFirstSection = true;
let isFirstTitlePara = false;
this.binary = {}; this.binary = {};
let binaryId = ''; let binaryId = '';
@@ -78,6 +81,10 @@ export default class BookParser {
resolve(); resolve();
}; };
i.onerror = (e) => {
reject(e);
};
i.src = `data:${binaryType};base64,${data}`; i.src = `data:${binaryType};base64,${data}`;
await sleep(30*1000); await sleep(30*1000);
if (!resolved) if (!resolved)
@@ -85,6 +92,30 @@ 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) => { const newParagraph = (text, len, addIndex) => {
paraIndex++; paraIndex++;
let p = { let p = {
@@ -147,33 +178,47 @@ export default class BookParser {
if (tag == 'binary') { if (tag == 'binary') {
let attrs = sax.getAttrsSync(tail); 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') if (binaryType == 'image/jpeg' || binaryType == 'image/png')
binaryId = (attrs.id.value ? attrs.id.value : ''); binaryId = (attrs.id.value ? attrs.id.value : '');
} }
if (tag == 'image') { if (tag == 'image') {
let attrs = sax.getAttrsSync(tail); let attrs = sax.getAttrsSync(tail);
if (attrs.href.value) { if (attrs.href && attrs.href.value) {
if (inPara && !this.showInlineImagesInCenter) const href = attrs.href.value;
growParagraph(`<image-inline href="${attrs.href.value}"></image-inline>`, 0); if (href[0] == '#') {//local
if (inPara && !this.showInlineImagesInCenter && !center)
growParagraph(`<image-inline href="${href}"></image-inline>`, 0);
else else
newParagraph(`<image href="${attrs.href.value}">${' '.repeat(maxImageLineCount)}</image>`, maxImageLineCount); newParagraph(`<image href="${href}">${' '.repeat(maxImageLineCount)}</image>`, maxImageLineCount);
if (inPara && this.showInlineImagesInCenter) if (inPara && this.showInlineImagesInCenter)
newParagraph(' ', 1); newParagraph(' ', 1);
} else {//external
dimPromises.push(getExternalImageDimensions(href));
newParagraph(`<image href="${href}">${' '.repeat(maxImageLineCount)}</image>`, maxImageLineCount);
}
} }
} }
if (path.indexOf('/fictionbook/body') == 0) { if (path.indexOf('/fictionbook/body') == 0) {
if (tag == 'body') {
if (!isFirstBody)
newParagraph(' ', 1);
isFirstBody = false;
}
if (tag == 'title') { if (tag == 'title') {
newParagraph(' ', 1); newParagraph(' ', 1);
isFirstTitlePara = true;
bold = true; bold = true;
center = true; center = true;
} }
if (tag == 'section') { if (tag == 'section') {
if (!isFirstSection)
newParagraph(' ', 1); newParagraph(' ', 1);
isFirstSection = false;
} }
if (tag == 'emphasis' || tag == 'strong') { if (tag == 'emphasis' || tag == 'strong') {
@@ -181,13 +226,17 @@ export default class BookParser {
} }
if ((tag == 'p' || tag == 'empty-line' || tag == 'v')) { if ((tag == 'p' || tag == 'empty-line' || tag == 'v')) {
if (!(tag == 'p' && isFirstTitlePara))
newParagraph(' ', 1); newParagraph(' ', 1);
if (tag == 'p') if (tag == 'p') {
inPara = true; inPara = true;
isFirstTitlePara = false;
}
} }
if (tag == 'subtitle') { if (tag == 'subtitle') {
newParagraph(' ', 1); newParagraph(' ', 1);
isFirstTitlePara = true;
bold = true; bold = true;
} }
@@ -215,6 +264,7 @@ export default class BookParser {
if (path.indexOf('/fictionbook/body') == 0) { if (path.indexOf('/fictionbook/body') == 0) {
if (tag == 'title') { if (tag == 'title') {
isFirstTitlePara = false;
bold = false; bold = false;
center = false; center = false;
} }
@@ -228,6 +278,7 @@ export default class BookParser {
} }
if (tag == 'subtitle') { if (tag == 'subtitle') {
isFirstTitlePara = false;
bold = false; bold = false;
} }
@@ -260,13 +311,13 @@ export default class BookParser {
text = text.replace(/>/g, '&gt;'); text = text.replace(/>/g, '&gt;');
text = text.replace(/</g, '&lt;'); text = text.replace(/</g, '&lt;');
if (text != ' ' && text.trim() == '') if (text && text.trim() == '')
text = text.trim(); text = (text.indexOf(' ') >= 0 ? ' ' : '');
if (text == '') if (!text)
return; return;
text = text.replace(/[\t\n\r]/g, ' '); text = text.replace(/[\t\n\r\xa0]/g, ' ');
switch (path) { switch (path) {
case '/fictionbook/description/title-info/author/first-name': case '/fictionbook/description/title-info/author/first-name':
@@ -315,13 +366,7 @@ export default class BookParser {
} }
if (path.indexOf('/fictionbook/body/section') == 0) { if (path.indexOf('/fictionbook/body/section') == 0) {
switch (tag) {
case 'p':
growParagraph(`${tOpen}${text}${tClose}`, text.length); growParagraph(`${tOpen}${text}${tClose}`, text.length);
break;
default:
growParagraph(`${tOpen}${text}${tClose}`, text.length);
}
} }
if (binaryId) { if (binaryId) {
@@ -409,14 +454,14 @@ export default class BookParser {
break; break;
case 'space': { case 'space': {
let attrs = sax.getAttrsSync(tail); let attrs = sax.getAttrsSync(tail);
if (attrs.w.value) if (attrs.w && attrs.w.value)
style.space = attrs.w.value; style.space = attrs.w.value;
break; break;
} }
case 'image': { case 'image': {
let attrs = sax.getAttrsSync(tail); let attrs = sax.getAttrsSync(tail);
if (attrs.href && attrs.href.value) {
let id = attrs.href.value; let id = attrs.href.value;
if (id) {
let local = false; let local = false;
if (id[0] == '#') { if (id[0] == '#') {
id = id.substr(1); id = id.substr(1);
@@ -428,8 +473,8 @@ export default class BookParser {
} }
case 'image-inline': { case 'image-inline': {
let attrs = sax.getAttrsSync(tail); let attrs = sax.getAttrsSync(tail);
if (attrs.href && attrs.href.value) {
let id = attrs.href.value; let id = attrs.href.value;
if (id) {
let local = false; let local = false;
if (id[0] == '#') { if (id[0] == '#') {
id = id.substr(1); id = id.substr(1);
@@ -564,7 +609,8 @@ export default class BookParser {
para.parsed.cutEmptyParagraphs === this.cutEmptyParagraphs && para.parsed.cutEmptyParagraphs === this.cutEmptyParagraphs &&
para.parsed.addEmptyParagraphs === this.addEmptyParagraphs && para.parsed.addEmptyParagraphs === this.addEmptyParagraphs &&
para.parsed.showImages === this.showImages && para.parsed.showImages === this.showImages &&
para.parsed.imageHeightLines === this.imageHeightLines para.parsed.imageHeightLines === this.imageHeightLines &&
para.parsed.imageFitWidth === this.imageFitWidth
) )
return para.parsed; return para.parsed;
@@ -578,6 +624,7 @@ export default class BookParser {
addEmptyParagraphs: this.addEmptyParagraphs, addEmptyParagraphs: this.addEmptyParagraphs,
showImages: this.showImages, showImages: this.showImages,
imageHeightLines: this.imageHeightLines, imageHeightLines: this.imageHeightLines,
imageFitWidth: this.imageFitWidth,
visible: !( visible: !(
(this.cutEmptyParagraphs && para.cut) || (this.cutEmptyParagraphs && para.cut) ||
(para.addIndex > this.addEmptyParagraphs) (para.addIndex > this.addEmptyParagraphs)
@@ -593,7 +640,7 @@ export default class BookParser {
last: Boolean, last: Boolean,
parts: array of { parts: array of {
style: {bold: Boolean, italic: Boolean, center: Boolean}, style: {bold: Boolean, italic: Boolean, center: Boolean},
image: {local: Boolean, inline: Boolean, id: String, imageLine: Number, lineCount: Number, paraIndex: Number}, image: {local: Boolean, inline: Boolean, id: String, imageLine: Number, lineCount: Number, paraIndex: Number, w: Number, h: Number},
text: String, text: String,
} }
}*/ }*/
@@ -608,7 +655,6 @@ export default class BookParser {
let style = {}; let style = {};
let ofs = 0;//смещение от начала параграфа para.offset let ofs = 0;//смещение от начала параграфа para.offset
let imgW = 0; let imgW = 0;
// тут начинается самый замес, перенос по слогам и стилизация, а также изображения // тут начинается самый замес, перенос по слогам и стилизация, а также изображения
for (const part of parts) { for (const part of parts) {
style = part.style; style = part.style;
@@ -616,11 +662,26 @@ export default class BookParser {
//изображения //изображения
if (part.image.id && !part.image.inline) { if (part.image.id && !part.image.inline) {
parsed.visible = this.showImages; parsed.visible = this.showImages;
const bin = this.binary[part.image.id]; let bin = this.binary[part.image.id];
if (!bin)
bin = {h: 1, w: 1};
let lineCount = this.imageHeightLines; let lineCount = this.imageHeightLines;
const c = Math.ceil(bin.h/this.lineHeight); let c = Math.ceil(bin.h/this.lineHeight);
const maxH = lineCount*this.lineHeight;
let maxH2 = maxH;
if (this.imageFitWidth && bin.w > this.w) {
maxH2 = bin.h*this.w/bin.w;
c = Math.ceil(maxH2/this.lineHeight);
}
lineCount = (c < lineCount ? c : lineCount); lineCount = (c < lineCount ? c : lineCount);
let imageHeight = (maxH2 < maxH ? maxH2 : maxH);
imageHeight = (imageHeight <= bin.h ? imageHeight : bin.h);
let imageWidth = (bin.h > imageHeight ? bin.w*imageHeight/bin.h : bin.w);
let i = 0; let i = 0;
for (; i < lineCount - 1; i++) { for (; i < lineCount - 1; i++) {
line.end = para.offset + ofs; line.end = para.offset + ofs;
@@ -632,7 +693,9 @@ export default class BookParser {
id: part.image.id, id: part.image.id,
imageLine: i, imageLine: i,
lineCount, lineCount,
paraIndex paraIndex,
w: imageWidth,
h: imageHeight,
}}); }});
lines.push(line); lines.push(line);
line = {begin: line.end + 1, parts: []}; line = {begin: line.end + 1, parts: []};
@@ -642,17 +705,22 @@ export default class BookParser {
line.first = (j == 0); line.first = (j == 0);
line.last = true; line.last = true;
line.parts.push({style, text: ' ', line.parts.push({style, text: ' ',
image: {local: part.image.local, inline: false, id: part.image.id, imageLine: i, lineCount, paraIndex}}); image: {local: part.image.local, inline: false, id: part.image.id,
imageLine: i, lineCount, paraIndex, w: imageWidth, h: imageHeight}
});
continue; continue;
} }
if (part.image.id && part.image.inline && this.showImages) { if (part.image.id && part.image.inline && this.showImages) {
const bin = this.binary[part.image.id]; const bin = this.binary[part.image.id];
if (bin) {
let imgH = (bin.h > this.fontSize ? this.fontSize : bin.h); let imgH = (bin.h > this.fontSize ? this.fontSize : bin.h);
imgW += bin.w*imgH/bin.h; imgW += bin.w*imgH/bin.h;
line.parts.push({style, text: '', line.parts.push({style, text: '',
image: {local: part.image.local, inline: true, id: part.image.id}}); image: {local: part.image.local, inline: true, id: part.image.id}});
} }
}
let words = part.text.split(' '); let words = part.text.split(' ');

View File

@@ -5,6 +5,10 @@ import BookParser from './BookParser';
const maxDataSize = 500*1024*1024;//chars, not bytes const maxDataSize = 500*1024*1024;//chars, not bytes
const bmCacheStore = localForage.createInstance({
name: 'bmCacheStore'
});
const bmMetaStore = localForage.createInstance({ const bmMetaStore = localForage.createInstance({
name: 'bmMetaStore' name: 'bmMetaStore'
}); });
@@ -20,11 +24,31 @@ const bmRecentStore = localForage.createInstance({
class BookManager { class BookManager {
async init(settings) { async init(settings) {
this.settings = settings; this.settings = settings;
this.books = {};
this.recent = {}; //this.booksCached нужен только для ускорения загрузки читалки
this.booksCached = await bmCacheStore.getItem('books');
if (!this.booksCached)
this.booksCached = {};
this.recent = await bmCacheStore.getItem('recent');
this.books = Object.assign({}, this.booksCached);
this.recentChanged1 = true; this.recentChanged1 = true;
this.recentChanged2 = true; this.recentChanged2 = true;
if (!this.books || !this.recent) {
this.books = {};
this.recent = {};
await this.loadMeta(true);
} else {
this.loadMeta(false);
}
}
async loadMeta(immediate) {
if (!immediate)
await utils.sleep(2000);
let len = await bmMetaStore.length(); let len = await bmMetaStore.length();
for (let i = 0; i < len; i++) { for (let i = 0; i < len; i++) {
const key = await bmMetaStore.key(i); const key = await bmMetaStore.key(i);
@@ -45,6 +69,12 @@ class BookManager {
} }
await this.cleanBooks(); await this.cleanBooks();
this.booksCached = {};
for (const key in this.books) {
this.booksCached[key] = this.metaOnly(this.books[key]);
}
await bmCacheStore.setItem('books', this.booksCached);
} }
async cleanBooks() { async cleanBooks() {
@@ -80,9 +110,11 @@ class BookManager {
const result = await this.parseBook(meta, newBook.data, callback); const result = await this.parseBook(meta, newBook.data, callback);
this.books[meta.key] = result; this.books[meta.key] = result;
this.booksCached[meta.key] = this.metaOnly(result);
await bmMetaStore.setItem(`bmMeta-${meta.key}`, this.metaOnly(result)); await bmMetaStore.setItem(`bmMeta-${meta.key}`, this.metaOnly(result));
await bmDataStore.setItem(`bmData-${meta.key}`, result.data); await bmDataStore.setItem(`bmData-${meta.key}`, result.data);
await bmCacheStore.setItem('books', this.booksCached);
return result; return result;
} }
@@ -127,11 +159,15 @@ class BookManager {
await bmDataStore.removeItem(`bmData-${meta.key}`); await bmDataStore.removeItem(`bmData-${meta.key}`);
delete this.books[meta.key]; delete this.books[meta.key];
delete this.booksCached[meta.key];
await bmCacheStore.setItem('books', this.booksCached);
} }
async parseBook(meta, data, callback) { async parseBook(meta, data, callback) {
if (!this.books) if (!this.books)
await this.init(); await this.init();
const parsed = new BookParser(this.settings); const parsed = new BookParser(this.settings);
const parsedMeta = await parsed.parse(data, callback); const parsedMeta = await parsed.parse(data, callback);
@@ -159,7 +195,7 @@ class BookManager {
async setRecentBook(value, noTouch) { async setRecentBook(value, noTouch) {
if (!this.recent) if (!this.recent)
await this.init(); await this.init();
const result = Object.assign({}, value); const result = this.metaOnly(value);
if (!noTouch) if (!noTouch)
Object.assign(result, {touchTime: Date.now()}); Object.assign(result, {touchTime: Date.now()});
@@ -170,6 +206,7 @@ class BookManager {
await bmRecentStore.setItem(result.key, result); await bmRecentStore.setItem(result.key, result);
await this.cleanRecentBooks(); await this.cleanRecentBooks();
await bmCacheStore.setItem('recent', this.recent);
this.recentChanged1 = true; this.recentChanged1 = true;
this.recentChanged2 = true; this.recentChanged2 = true;
@@ -188,6 +225,8 @@ class BookManager {
await bmRecentStore.removeItem(value.key); await bmRecentStore.removeItem(value.key);
delete this.recent[value.key]; delete this.recent[value.key];
await bmCacheStore.setItem('recent', this.recent);
this.recentChanged1 = true; this.recentChanged1 = true;
this.recentChanged2 = true; this.recentChanged2 = true;
} }

View File

@@ -164,6 +164,7 @@ const settingDefaults = {
showImages: true, showImages: true,
showInlineImagesInCenter: false, showInlineImagesInCenter: false,
imageHeightLines: 100, imageHeightLines: 100,
imageFitWidth: true,
fontShifts: {}, fontShifts: {},
}; };

View File

@@ -1,6 +1,6 @@
{ {
"name": "Liberama", "name": "Liberama",
"version": "0.4.1", "version": "0.4.7",
"engines": { "engines": {
"node": ">=10.0.0" "node": ">=10.0.0"
}, },

View File

@@ -0,0 +1,96 @@
const iconv = require('iconv-lite');
const chardet = require('chardet');
const textUtils = require('./textUtils');
class ConvertBase {
constructor(config) {
this.config = config;
}
run(data, opts) {// eslint-disable-line no-unused-vars
//override
}
decode(data) {
let selected = textUtils.getEncoding(data);
if (selected == 'ISO-8859-5') {
const charsetAll = chardet.detectAll(data.slice(0, 20000));
for (const charset of charsetAll) {
if (charset.name.indexOf('ISO-8859') < 0) {
selected = charset.name;
break;
}
}
}
if (selected.toLowerCase() != 'utf-8')
return iconv.decode(data, selected);
else
return data;
}
repSpaces(text) {
return text.replace(/&nbsp;|[\t\n\r]/g, ' ');
}
formatFb2(fb2) {
let out = '<?xml version="1.0" encoding="utf-8"?>';
out += '<FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:l="http://www.w3.org/1999/xlink">';
out += this.formatFb2Node(fb2);
out += '</FictionBook>';
return out;
}
formatFb2Node(node, name) {
let out = '';
if (Array.isArray(node)) {
for (const n of node) {
out += this.formatFb2Node(n);
}
} else if (typeof node == 'string') {
if (name)
out += `<${name}>${this.repSpaces(node)}</${name}>`;
else
out += this.repSpaces(node);
} else {
if (node._n)
name = node._n;
let attrs = '';
if (node._attrs) {
for (let attrName in node._attrs) {
attrs += ` ${attrName}="${node._attrs[attrName]}"`;
}
}
let tOpen = '';
let tBody = '';
let tClose = '';
if (name)
tOpen += `<${name}${attrs}>`;
if (node.hasOwnProperty('_t'))
tBody += this.repSpaces(node._t);
for (let nodeName in node) {
if (nodeName && nodeName[0] == '_' && nodeName != '_a')
continue;
const n = node[nodeName];
tBody += this.formatFb2Node(n, nodeName);
}
if (name)
tClose += `</${name}>`;
if (attrs == '' && name == 'p' && tBody.trim() == '')
out += '<empty-line/>'
else
out += `${tOpen}${tBody}${tClose}`;
}
return out;
}
}
module.exports = ConvertBase;

View File

@@ -0,0 +1,41 @@
const ConvertBase = require('./ConvertBase');
const iconv = require('iconv-lite');
class ConvertFb2 extends ConvertBase {
check(data, opts) {
const {fileType} = opts;
return (fileType && fileType.ext == 'xml' && data.toString().indexOf('<FictionBook') >= 0);
}
run(data, opts) {
if (!this.check(data, opts))
return false;
return this.checkEncoding(data);
}
checkEncoding(data) {
let result = data;
const left = data.indexOf('<?xml version="1.0"');
if (left >= 0) {
const right = data.indexOf('?>', left);
if (right >= 0) {
const head = data.slice(left, right + 2).toString();
const m = head.match(/encoding="(.*)"/);
if (m) {
let encoding = m[1].toLowerCase();
if (encoding != 'utf-8') {
result = iconv.decode(data, encoding);
result = Buffer.from(result.toString().replace(m[0], 'encoding="utf-8"'));
}
}
}
}
return result;
}
}
module.exports = ConvertFb2;

View File

@@ -0,0 +1,162 @@
const ConvertBase = require('./ConvertBase');
const sax = require('./sax');
const textUtils = require('./textUtils');
class ConvertHtml extends ConvertBase {
check(data, opts) {
const {fileType} = opts;
if (fileType && (fileType.ext == 'html' || fileType.ext == 'xml'))
return {isText: false};
//может это чистый текст?
if (textUtils.checkIfText(data)) {
return {isText: true};
}
return false;
}
run(data, opts) {
const checkResult = this.check(data, opts);
if (!checkResult)
return false;
let {isText} = checkResult;
let titleInfo = {};
let desc = {_n: 'description', 'title-info': titleInfo};
let pars = [];
let body = {_n: 'body', section: {_a: []}};
let fb2 = [desc, body];
let title = '';
let inTitle = false;
let spaceCounter = [];
const repCrLfTab = (text) => text.replace(/[\n\r]/g, '').replace(/\t/g, ' ');
const newParagraph = () => {
pars.push({_n: 'p', _t: ''});
};
const growParagraph = (text) => {
if (!pars.length)
newParagraph();
const l = pars.length;
if (pars[l - 1]._t == '')
text = text.trimLeft();
pars[l - 1]._t += text;
//посчитаем отступы у текста, чтобы выделить потом параграфы
const lines = text.split('\n');
for (let line of lines) {
line = repCrLfTab(line)
let l = 0;
while (l < line.length && line[l] == ' ') {
l++;
}
if (!spaceCounter[l])
spaceCounter[l] = 0;
spaceCounter[l]++;
}
};
const newPara = new Set(['tr', 'br', 'br/', 'dd', 'p', 'title', '/title', 'h1', 'h2', 'h3', '/h1', '/h2', '/h3']);
const onTextNode = (text, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!cutCounter) {
growParagraph(text);
}
if (inTitle && !title)
title = text;
};
const onStartNode = (tag, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!cutCounter) {
if (newPara.has(tag))
newParagraph();
}
if (tag == 'title')
inTitle = true;
};
const onEndNode = (tag, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (tag == 'title')
inTitle = false;
};
let buf = this.decode(data).toString();
sax.parseSync(buf, {
onStartNode, onEndNode, onTextNode,
innerCut: new Set(['head', 'script', 'style', 'binary'])
});
titleInfo['book-title'] = title;
//подозрение на чистый текст, надо разбить на параграфы
if (isText || pars.length < buf.length/2000) {
let total = 0;
for (let i = 0; i < spaceCounter.length; i++) {
total += (spaceCounter[i] ? spaceCounter[i] : 0);
}
total /= 10;
let i = spaceCounter.length - 1;
while (i > 0 && (!spaceCounter[i] || spaceCounter[i] < total)) i--;
const parIndent = (i > 0 ? i : 0);
let newPars = [];
const newPar = () => {
newPars.push({_n: 'p', _t: ''});
};
const growPar = (text) => {
if (!newPars.length)
newPar();
const l = newPars.length;
newPars[l - 1]._t += text;
}
i = 0;
for (const par of pars) {
if (i > 0)
newPar();
i++;
const lines = par._t.split('\n');
for (let line of lines) {
line = repCrLfTab(line);
let l = 0;
while (l < line.length && line[l] == ' ') {
l++;
}
if (l >= parIndent)
newPar();
growPar(line.trim() + ' ');
}
}
body.section._a[0] = newPars;
} else {
body.section._a[0] = pars;
}
//убираем лишнее
for (let i = 0; i < pars.length; i++)
pars[i]._t = this.repSpaces(pars[i]._t).trim();
return this.formatFb2(fb2);
}
}
module.exports = ConvertHtml;

View File

@@ -0,0 +1,275 @@
const _ = require('lodash');
const URL = require('url').URL;
const sax = require('./sax');
const ConvertBase = require('./ConvertBase');
class ConvertSamlib extends ConvertBase {
check(data, opts) {
const {url} = opts;
const parsedUrl = new URL(url);
if (parsedUrl.hostname == 'samlib.ru' ||
parsedUrl.hostname == 'budclub.ru' ||
parsedUrl.hostname == 'zhurnal.lib.ru') {
return {hostname: parsedUrl.hostname};
}
return false;
}
run(data, opts) {
const checkResult = this.check(data, opts);
if (!checkResult)
return false;
const {hostname} = checkResult;
let titleInfo = {};
let desc = {_n: 'description', 'title-info': titleInfo};
let pars = [];
let body = {_n: 'body', section: {_a: pars}};
let fb2 = [desc, body];
let inSubtitle = false;
let inJustify = true;
let inImage = false;
let isFirstPara = 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, attrs) => {
if (name == 'p')
inPara = true;
let n = {_n: name, _attrs: attrs, _a: [], _p: node};
node._a.push(n);
node = n;
};
const closeTag = (name) => {
if (name == 'p')
inPara = false;
if (node._p) {
const exact = (node._n == name);
node = node._p;
if (!exact)
closeTag(name);
}
};
const growParagraph = (text) => {
if (!node._p) {
if (text.trim() != '')
openTag('p');
else
return;
}
if (node._n == 'p' && node._a.length == 0)
text = text.trimLeft();
node._a.push({_t: text});
};
const onStartNode = (elemName, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (elemName == '')
return;
if (!inText) {
path += '/' + elemName;
tag = elemName;
} else {
switch (elemName) {
case 'li':
case 'p':
case 'dd':
case 'br':
if (!(inSubtitle && isFirstPara)) {
if (inPara)
closeTag('p');
openTag('p');
}
isFirstPara = false;
break;
case 'h1':
case 'h2':
case 'h3':
if (inPara)
closeTag('p');
openTag('p');
bold = true;
break;
case 'i':
case 'em':
italic = true;
break;
case 'b':
case 'strong':
bold = true;
break;
case 'div':
if (inPara)
closeTag('p');
if (tail.indexOf('align="center"') >= 0) {
openTag('subtitle');
inSubtitle = true;
isFirstPara = true;
}
if (tail.indexOf('align="justify"') >= 0) {
openTag('p');
inJustify = true;
}
break;
case 'img': {
if (inPara)
closeTag('p');
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;
}
}
}
};
const onEndNode = (elemName, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!inText) {
const oldPath = path;
let t = '';
do {
let i = path.lastIndexOf('/');
t = path.substr(i + 1);
path = path.substr(0, i);
} while (t != elemName && path);
if (t != elemName) {
path = oldPath;
}
let i = path.lastIndexOf('/');
tag = path.substr(i + 1);
} else {
switch (elemName) {
case 'li':
case 'p':
case 'dd':
closeTag('p');
break;
case 'h1':
case 'h2':
case 'h3':
closeTag('p');
bold = false;
break;
case 'i':
case 'em':
italic = false;
break;
case 'b':
case 'strong':
bold = false;
break;
case 'div':
if (inSubtitle) {
closeTag('subtitle');
inSubtitle = false;
isFirstPara = false;
}
if (inJustify) {
closeTag('p');
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 -------------') {
inText = true;
textFound = true;
}
if (text == '-----------------------------------------------')
inText = false;
};
const onTextNode = (text) => {// eslint-disable-line no-unused-vars
if (text && text.trim() == '')
text = (text.indexOf(' ') >= 0 ? ' ' : '');
if (!text)
return;
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;
}
let tOpen = (bold ? '<strong>' : '');
tOpen += (italic ? '<emphasis>' : '');
let tClose = (italic ? '</emphasis>' : '');
tClose += (bold ? '</strong>' : '');
if (inText)
growParagraph(`${tOpen}${text}${tClose}`);
};
sax.parseSync(this.decode(data).toString().replace(/&nbsp;/g, ' '), {
onStartNode, onEndNode, onTextNode, onComment,
innerCut: new Set(['head', 'script', 'style'])
});
//текст не найден на странице, обработать корректно не получилось
if (!textFound)
return false;
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);
}
}
module.exports = ConvertSamlib;

View File

@@ -1,459 +1,45 @@
const fs = require('fs-extra'); const fs = require('fs-extra');
const URL = require('url').URL;
const iconv = require('iconv-lite');
const chardet = require('chardet');
const _ = require('lodash');
const sax = require('./sax');
const textUtils = require('./textUtils');
const FileDetector = require('../FileDetector'); const FileDetector = require('../FileDetector');
const repSpaces = (text) => text.replace(/&nbsp;|[\t\n\r]/g, ' '); //порядок важен
const convertClassFactory = [
require('./ConvertFb2'),
require('./ConvertSamlib'),
require('./ConvertHtml'),
];
class BookConverter { class BookConverter {
constructor() { constructor(config) {
this.detector = new FileDetector(); this.detector = new FileDetector();
this.convertFactory = [];
for (const convertClass of convertClassFactory) {
this.convertFactory.push(new convertClass(config));
}
} }
async convertToFb2(inputFile, outputFile, url, callback) { async convertToFb2(inputFile, outputFile, url, callback) {
const fileType = await this.detector.detectFile(inputFile); const fileType = await this.detector.detectFile(inputFile);
const data = await fs.readFile(inputFile); const data = await fs.readFile(inputFile);
callback(100); let result = false;
for (const convert of this.convertFactory) {
if (fileType && (fileType.ext == 'html' || fileType.ext == 'xml')) { result = convert.run(data, {inputFile, url, callback, fileType});
if (data.toString().indexOf('<FictionBook') >= 0) { if (result) {
await fs.writeFile(outputFile, this.checkEncoding(data)); await fs.writeFile(outputFile, result);
return; break;
}
} }
const parsedUrl = new URL(url); if (!result) {
if (parsedUrl.hostname == 'samlib.ru' ||
parsedUrl.hostname == 'budclub.ru' ||
parsedUrl.hostname == 'zhurnal.lib.ru') {
await fs.writeFile(outputFile, this.convertSamlib(data));
return;
}
await fs.writeFile(outputFile, this.convertHtml(data));
return;
} else {
if (fileType) if (fileType)
throw new Error(`Этот формат файла не поддерживается: ${fileType.mime}`); throw new Error(`Этот формат файла не поддерживается: ${fileType.mime}`);
else { else {
//может это чистый текст?
if (textUtils.checkIfText(data)) {
await fs.writeFile(outputFile, this.convertHtml(data, true));
return;
}
throw new Error(`Не удалось определить формат файла: ${url}`); throw new Error(`Не удалось определить формат файла: ${url}`);
} }
} }
}
decode(data) { callback(100);
let selected = textUtils.getEncoding(data);
if (selected == 'ISO-8859-5') {
const charsetAll = chardet.detectAll(data.slice(0, 20000));
for (const charset of charsetAll) {
if (charset.name.indexOf('ISO-8859') < 0) {
selected = charset.name;
break;
}
}
}
return iconv.decode(data, selected);
}
checkEncoding(data) {
let result = data;
const left = data.indexOf('<?xml version="1.0"');
if (left >= 0) {
const right = data.indexOf('?>', left);
if (right >= 0) {
const head = data.slice(left, right + 2).toString();
const m = head.match(/encoding="(.*)"/);
if (m) {
let encoding = m[1].toLowerCase();
if (encoding != 'utf-8') {
result = iconv.decode(data, encoding);
result = Buffer.from(result.toString().replace(m[0], 'encoding="utf-8"'));
}
}
}
}
return result;
}
convertHtml(data, isText) {
let titleInfo = {};
let desc = {_n: 'description', 'title-info': titleInfo};
let pars = [];
let body = {_n: 'body', section: {_a: []}};
let fb2 = [desc, body];
let title = '';
let inTitle = false;
let spaceCounter = [];
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;
}
//посчитаем отступы у текста, чтобы выделить потом параграфы
const lines = text.split('\n');
for (const line of lines) {
const sp = line.split(' ');
let l = 0;
while (l < sp.length && sp[l].trim() == '') {
l++;
}
if (!spaceCounter[l])
spaceCounter[l] = 0;
spaceCounter[l]++;
}
};
newParagraph();
const newPara = new Set(['tr', 'br', 'br/', 'dd', 'p', 'title', '/title', 'h1', 'h2', 'h3', '/h1', '/h2', '/h3']);
const onTextNode = (text, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!cutCounter) {
growParagraph(text);
}
if (inTitle && !title)
title = text;
};
const onStartNode = (tag, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!cutCounter) {
if (newPara.has(tag))
newParagraph();
}
if (tag == 'title')
inTitle = true;
};
const onEndNode = (tag, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (tag == 'title')
inTitle = false;
};
let buf = this.decode(data).toString();
sax.parseSync(buf, {
onStartNode, onEndNode, onTextNode,
innerCut: new Set(['head', 'script', 'style'])
});
titleInfo['book-title'] = title;
//подозрение на чистый текст, надо разбить на параграфы
if (isText || pars.length < buf.length/2000) {
let total = 0;
for (let i = 0; i < spaceCounter.length; i++) {
total += (spaceCounter[i] ? spaceCounter[i] : 0);
}
total /= 10;
let i = spaceCounter.length - 1;
while (i > 0 && (!spaceCounter[i] || spaceCounter[i] < total)) i--;
const parIndent = (i > 0 ? i : 0);
let newPars = [];
const newPar = () => {
newPars.push({_n: 'p', _t: ''});
};
const growPar = (text) => {
const l = newPars.length;
if (l) {
newPars[l - 1]._t += text;
}
}
for (const par of pars) {
newPar();
const lines = par._t.split('\n');
for (const line of lines) {
const sp = line.split(' ');
let l = 0;
while (l < sp.length && sp[l].trim() == '') {
l++;
}
if (l >= parIndent)
newPar();
growPar(line.trim() + ' ');
}
}
body.section._a[0] = newPars;
} else {
body.section._a[0] = pars;
}
//убираем лишнее
for (let i = 0; i < pars.length; i++)
pars[i]._t = repSpaces(pars[i]._t).trim();
return this.formatFb2(fb2);
}
convertSamlib(data) {
let titleInfo = {};
let desc = {_n: 'description', 'title-info': titleInfo};
let pars = [];
let body = {_n: 'body', section: {_a: pars}};
let fb2 = [desc, body];
let inSubtitle = false;
let inJustify = true;
let path = '';
let tag = '';// eslint-disable-line no-unused-vars
let inText = false;
let node = {_a: pars};
let inPara = false;
let italic = false;
let bold = false;
const openTag = (name) => {
if (name == 'p')
inPara = true;
let n = {_n: name, _a: [], _p: node};
node._a.push(n);
node = n;
};
const closeTag = (name) => {
if (name == 'p')
inPara = false;
if (node._p) {
const exact = (node._n == name);
node = node._p;
if (!exact)
closeTag(name);
}
};
const growParagraph = (text) => {
if (node._n == 'p' && node._a.length == 0)
text = text.trimLeft();
node._a.push({_t: text});
};
openTag('p');
const onStartNode = (elemName, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (elemName == '')
return;
if (!inText) {
path += '/' + elemName;
tag = elemName;
} else {
if (inPara && elemName != 'i' && elemName != 'b')
closeTag('p');
switch (elemName) {
case 'li':
case 'p':
case 'dd':
case 'h1':
case 'h2':
case 'h3':
openTag('p');
break;
case 'i':
italic = true;
break;
case 'b':
bold = true;
break;
case 'div':
if (tail.indexOf('align="center"') >= 0) {
openTag('subtitle');
inSubtitle = true;
}
if (tail.indexOf('align="justify"') >= 0) {
openTag('p');
inJustify = true;
}
break;
}
}
};
const onEndNode = (elemName, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!inText) {
const oldPath = path;
let t = '';
do {
let i = path.lastIndexOf('/');
t = path.substr(i + 1);
path = path.substr(0, i);
} while (t != elemName && path);
if (t != elemName) {
path = oldPath;
}
let i = path.lastIndexOf('/');
tag = path.substr(i + 1);
} else {
switch (elemName) {
case 'li':
case 'p':
case 'dd':
case 'h1':
case 'h2':
case 'h3':
closeTag('p');
break;
case 'i':
italic = false;
break;
case 'b':
bold = false;
break;
case 'div':
if (inSubtitle) {
closeTag('subtitle');
inSubtitle = false;
}
if (inJustify) {
closeTag('p');
inJustify = false;
}
break;
}
}
};
const onComment = (text) => {// eslint-disable-line no-unused-vars
if (text == '--------- Собственно произведение -------------')
inText = true;
if (text == '-----------------------------------------------')
inText = false;
};
const onTextNode = (text) => {// eslint-disable-line no-unused-vars
if (text != ' ' && text.trim() == '')
text = text.trim();
if (text == '')
return;
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;
}
let tOpen = (bold ? '<strong>' : '');
tOpen += (italic ? '<emphasis>' : '');
let tClose = (italic ? '</emphasis>' : '');
tClose += (bold ? '</strong>' : '');
if (inText)
growParagraph(`${tOpen}${text}${tClose}`);
};
sax.parseSync(repSpaces(this.decode(data).toString()), {
onStartNode, onEndNode, onTextNode, onComment,
innerCut: new Set(['head', 'script', 'style'])
});
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);
}
formatFb2(fb2) {
let out = '<?xml version="1.0" encoding="utf-8"?>';
out += '<FictionBook xmlns="http://www.gribuser.ru/xml/fictionbook/2.0" xmlns:l="http://www.w3.org/1999/xlink">';
out += this.formatFb2Node(fb2);
out += '</FictionBook>';
return out;
}
formatFb2Node(node, name) {
let out = '';
if (Array.isArray(node)) {
for (const n of node) {
out += this.formatFb2Node(n);
}
} else if (typeof node == 'string') {
if (name)
out += `<${name}>${repSpaces(node)}</${name}>`;
else
out += repSpaces(node);
} else {
if (node._n)
name = node._n;
if (name)
out += `<${name}>`;
if (node.hasOwnProperty('_t'))
out += repSpaces(node._t);
for (let nodeName in node) {
if (nodeName && nodeName[0] == '_' && nodeName != '_a')
continue;
const n = node[nodeName];
out += this.formatFb2Node(n, nodeName);
}
if (name)
out += `</${name}>`;
}
return out;
} }
} }

View File

@@ -1,4 +1,4 @@
function getEncoding(buf) { function getEncoding(buf, returnAll) {
const lowerCase = 3; const lowerCase = 3;
const upperCase = 1; const upperCase = 1;
@@ -8,6 +8,7 @@ function getEncoding(buf) {
'd': 'cp866', 'd': 'cp866',
'i': 'ISO-8859-5', 'i': 'ISO-8859-5',
'm': 'maccyrillic', 'm': 'maccyrillic',
'u': 'utf-8',
}; };
let charsets = { let charsets = {
@@ -15,19 +16,27 @@ function getEncoding(buf) {
'w': 0, 'w': 0,
'd': 0, 'd': 0,
'i': 0, 'i': 0,
'm': 0 'm': 0,
'u': 0,
}; };
const len = buf.length; const len = buf.length;
const blockSize = (len > 5*3000 ? 3000 : len); const blockSize = (len > 5*3000 ? 3000 : len);
let counter = 0; let counter = 0;
let i = 0; let i = 0;
let totalChecked = 0;
while (i < len) { while (i < len) {
const char = buf[i]; const char = buf[i];
const nextChar = (i < len - 1 ? buf[i + 1] : 0);
totalChecked++;
i++; i++;
//non-russian characters //non-russian characters
if (char < 128 || char > 256) if (char < 128 || char > 256)
continue; continue;
//UTF-8
if ((char == 208 || char == 209) && nextChar >= 128 && nextChar <= 190)
charsets['u'] += lowerCase;
else {
//CP866 //CP866
if ((char > 159 && char < 176) || (char > 223 && char < 242)) charsets['d'] += lowerCase; if ((char > 159 && char < 176) || (char > 223 && char < 242)) charsets['d'] += lowerCase;
if ((char > 127 && char < 160)) charsets['d'] += upperCase; if ((char > 127 && char < 160)) charsets['d'] += upperCase;
@@ -47,6 +56,7 @@ function getEncoding(buf) {
//ISO-8859-5 //ISO-8859-5
if (char > 207 && char < 240) charsets['i'] += lowerCase; if (char > 207 && char < 240) charsets['i'] += lowerCase;
if (char > 175 && char < 208) charsets['i'] += upperCase; if (char > 175 && char < 208) charsets['i'] += upperCase;
}
counter++; counter++;
@@ -57,18 +67,24 @@ function getEncoding(buf) {
} }
let sorted = Object.keys(charsets).map(function(key) { let sorted = Object.keys(charsets).map(function(key) {
return { codePage: codePage[key], c: charsets[key] }; return { codePage: codePage[key], c: charsets[key], totalChecked };
}); });
sorted.sort((a, b) => b.c - a.c); sorted.sort((a, b) => b.c - a.c);
if (sorted[0].c > 0) if (returnAll)
return sorted;
else if (sorted[0].c > 0)
return sorted[0].codePage; return sorted[0].codePage;
else else
return 'ISO-8859-5'; return 'ISO-8859-5';
} }
function checkIfText(buf) { function checkIfText(buf) {
const enc = getEncoding(buf, true);
if (enc[0].c > enc[0].totalChecked*0.9)
return true;
let spaceCount = 0; let spaceCount = 0;
let crCount = 0; let crCount = 0;
let lfCount = 0; let lfCount = 0;

View File

@@ -9,18 +9,18 @@ detect.addSignature(
"rules": [ "rules": [
{ "type": "or", "rules": { "type": "or", "rules":
[ [
{ "type": "contains", "bytes": "3c68746d6c" }, { "type": "equal", "end": 5, "bytes": "3c68746d6c" },
{ "type": "contains", "bytes": "3c00680074006d006c00" }, { "type": "equal", "end": 10, "bytes": "3c00680074006d006c00" },
{ "type": "contains", "bytes": "3c21646f6374797065" }, { "type": "equal", "end": 9, "bytes": "3c21646f6374797065" },
{ "type": "contains", "bytes": "3c626f6479" }, { "type": "equal", "end": 5, "bytes": "3c626f6479" },
{ "type": "contains", "bytes": "3c68656164" }, { "type": "equal", "end": 5, "bytes": "3c68656164" },
{ "type": "contains", "bytes": "3c696672616d65" }, { "type": "equal", "end": 7, "bytes": "3c696672616d65" },
{ "type": "contains", "bytes": "3c696d67" }, { "type": "equal", "end": 4, "bytes": "3c696d67" },
{ "type": "contains", "bytes": "3c6f626a656374" }, { "type": "equal", "end": 7, "bytes": "3c6f626a656374" },
{ "type": "contains", "bytes": "3c736372697074" }, { "type": "equal", "end": 7, "bytes": "3c736372697074" },
{ "type": "contains", "bytes": "3c7461626c65" }, { "type": "equal", "end": 6, "bytes": "3c7461626c65" },
{ "type": "contains", "bytes": "3c7469746c65" }, { "type": "equal", "end": 6, "bytes": "3c7469746c65" },
] ]
} }
] ]
@@ -36,7 +36,7 @@ detect.addSignature(
"rules": [ "rules": [
{ "type": "or", "rules": { "type": "or", "rules":
[ [
{ "type": "contains", "bytes": "3c3f786d6c2076657273696f6e3d22312e3022" }, { "type": "equal", "end": 19, "bytes": "3c3f786d6c2076657273696f6e3d22312e3022" },
] ]
} }
] ]

View File

@@ -8,8 +8,14 @@ class FileDownloader {
async load(url, callback) { async load(url, callback) {
let errMes = ''; 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; let estSize = 0;
if (response.headers['content-length']) { if (response.headers['content-length']) {
@@ -17,7 +23,7 @@ class FileDownloader {
} }
let prevProg = 0; let prevProg = 0;
const request = got(url, {encoding: null}).on('downloadProgress', progress => { const request = got(url, options).on('downloadProgress', progress => {
if (progress.transferred > maxDownloadSize) { if (progress.transferred > maxDownloadSize) {
errMes = 'file too big'; errMes = 'file too big';
request.cancel(); request.cancel();

View File

@@ -22,7 +22,7 @@ class ReaderWorker {
this.down = new FileDownloader(); this.down = new FileDownloader();
this.decomp = new FileDecompressor(); this.decomp = new FileDecompressor();
this.bookConverter = new BookConverter(); this.bookConverter = new BookConverter(this.config);
if (!singleCleanExecute) { if (!singleCleanExecute) {
this.periodicCleanDir(this.config.tempPublicDir, this.config.maxTempPublicDirSize, 60*60*1000);//1 раз в час this.periodicCleanDir(this.config.tempPublicDir, this.config.maxTempPublicDirSize, 60*60*1000);//1 раз в час
@@ -66,14 +66,14 @@ class ReaderWorker {
const decompFilename = await this.decomp.decompressFile(downloadedFilename, decompDir); const decompFilename = await this.decomp.decompressFile(downloadedFilename, decompDir);
wState.set({progress: 100}); wState.set({progress: 100});
//parse book //конвертирование в fb2
wState.set({state: 'convert', step: 3, progress: 0}); wState.set({state: 'convert', step: 3, progress: 0});
convertFilename = `${this.config.tempDownloadDir}/${tempFilename2}`; convertFilename = `${this.config.tempDownloadDir}/${tempFilename2}`;
await this.bookConverter.convertToFb2(decompFilename, convertFilename, url, progress => { await this.bookConverter.convertToFb2(decompFilename, convertFilename, url, progress => {
wState.set({progress}); wState.set({progress});
}); });
//compress file to tmp dir, if not exists with the same hashname //сжимаем файл в tmp, если там уже нет с тем же именем-sha256
const compFilename = await this.decomp.gzipFileIfNotExists(convertFilename, `${this.config.tempPublicDir}`); const compFilename = await this.decomp.gzipFileIfNotExists(convertFilename, `${this.config.tempPublicDir}`);
wState.set({progress: 100}); wState.set({progress: 100});

View File

@@ -1,7 +1,7 @@
const config = require('./config'); const config = require('./config');
const {initLogger, getLog} = require('./core/getLogger'); const logger = require('./core/getLogger');
initLogger(config); logger.initLogger(config);
const log = getLog(); const log = logger.getLog();
const configSaver = require('./config/configSaver'); const configSaver = require('./config/configSaver');
const argv = require('minimist')(process.argv.slice(2)); const argv = require('minimist')(process.argv.slice(2));
@@ -31,6 +31,7 @@ async function init() {
} }
async function main() { async function main() {
log(`${config.name} v${config.version}`);
log('Initializing'); log('Initializing');
await init(); await init();