Работа над конвертером Pdf

This commit is contained in:
Book Pauk
2019-02-28 18:58:41 +07:00
parent c7d376adf2
commit 55d5f6524d
3 changed files with 84 additions and 0 deletions

View File

@@ -0,0 +1,81 @@
const fs = require('fs-extra');
const sax = require('./sax');
const utils = require('../utils');
const ConvertHtml = require('./ConvertHtml');
class ConvertPdf extends ConvertHtml {
check(data, opts) {
const {inputFiles} = opts;
return this.config.useExternalBookConverter &&
inputFiles.sourceFileType && inputFiles.sourceFileType.ext == 'pdf';
}
async run(notUsed, opts) {
if (!this.check(notUsed, opts))
return false;
await this.checkExternalConverterPresent();
const {inputFiles, callback} = opts;
const outFile = `${inputFiles.fileListDir}/${utils.randomHexString(10)}.xml`;
//конвертируем в xml
await this.execConverter(this.pdfToHtmlPath, ['-c', '-s', '-xml', inputFiles.sourceFile, outFile]);
callback(50);
const data = await fs.readFile(outFile);
callback(60);
//парсим xml
let lines = [];
let inText = false;
let title = '';
let i = -1;
const onTextNode = (text, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!cutCounter && inText) {
lines[i].text += text + ' ';
if (i < 2)
title += text + ' ';
}
};
const onStartNode = (tag, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (!cutCounter) {
if (tag == 'text' && !inText) {
inText = true;
i++;
let attrs = sax.getAttrsSync(tail);
lines[i] = {
text: '',
top: (attrs.top && attrs.top.value ? attrs.top.value : null),
left: (attrs.left && attrs.left.value ? attrs.left.value : null),
};
}
}
};
const onEndNode = (tag, tail, singleTag, cutCounter, cutTag) => {// eslint-disable-line no-unused-vars
if (tag == 'text')
inText = false;
};
let buf = this.decode(data).toString();
sax.parseSync(buf, {
onStartNode, onEndNode, onTextNode
});
//найдем параграфы и отступы
console.log(lines.length);
//формируем текст
let text = ''
text = title + "\n" + text;
return await super.run(Buffer.from(text), {skipCheck: true, isText: true});
}
}
module.exports = ConvertPdf;

View File

@@ -3,6 +3,7 @@ const FileDetector = require('../FileDetector');
//порядок важен
const convertClassFactory = [
require('./ConvertPdf'),
require('./ConvertRtf'),
require('./ConvertDocX'),
require('./ConvertDoc'),

View File

@@ -15,6 +15,8 @@ class FileDecompressor {
}
async decompressFile(filename, outputDir) {
await fs.ensureDir(outputDir);
const fileType = await this.detector.detectFile(filename);
let result = {