ocr-train/index.js

const fs = require('fs')
const path = require('path')
const Canvas = require('canvas')
const exec = require('child_process').execSync
const sharp = require('sharp')
Canvas.registerFont('./pubg.ttf', { family: 'pubg' })
var canvas0 = Canvas.createCanvas(320, 320)
var ctx0 = canvas0.getContext('2d')
ctx0.font = '30px pubg'
let count = 0;
const genGtData = (str) => {
    let metrics = ctx0.measureText(str)
    var canvas = Canvas.createCanvas(metrics.width + 10, metrics.actualBoundingBoxAscent + metrics.actualBoundingBoxDescent + 10)
    var ctx = canvas.getContext('2d')
    ctx.fillStyle = '#ffffff'
    ctx.fillRect(0, 0, metrics.width + 10, metrics.actualBoundingBoxAscent + metrics.actualBoundingBoxDescent + 10)
    ctx.font = '30px pubg'
    ctx.fillStyle = '#000000'
    ctx.fillText(str, 5, 5 + metrics.actualBoundingBoxAscent)

    var img_src = canvas.toBuffer().toString('base64')
    const base64Data = new Buffer(img_src.replace(/^data:image\/\w+;base64,/, ""), 'base64')
    sharp(base64Data)
        .tiff({
            compression: 'lzw',
            squash: true
        })
        .toFile(`./data/ground-truth/${count}.tif`)
        .then(res => {
            console.log(res)
        })
    fs.writeFileSync(`./data/ground-truth/${count}.gt.txt`, str + '\n');
    count++
}

const genTiffs = () => {
    for (i = 0; i < 1000; i++) {
        let str = ''
        for (let j = 0; j < 1 + Math.round(Math.random() * 10); j++) {
            switch (Math.round(Math.random() * 7)) {
                case 0:
                    let num = (Math.random() * 40).toFixed(1);
                    if (num.endsWith('.0')) {
                        num = num.replace('.0', '')
                    }
                    str += `${num}分钟 `
                    break
                case 1:
                    str += `第${parseInt(Math.random() * 100 + 1)} `
                    break
                case 2:
                case 3:
                case 4:
                    let s = i.toFixed(1);
                    if (s.endsWith('.0')) {
                        s = s.replace('.0', '')
                    }
                    str += `${s} `
                    break
                case 5:
                case 6:
                case 7:
                    str += `/${parseInt(Math.random() * 100 + 1)} `
                    break
            }
        }
        genGtData(str)
    }
}

const genTiff = () => {
    exec("find . -name 'pubg*' ! -name 'pubg.ttf' -d 1 -exec rm -f {} +")
    let chars = "1234567890./分钟第剩余淘汰"
    let trainingText = ''
    let width = 0, height = 0
    for (i = 0; i < 200; i++) {
        let line = ''
        for (j = 0; j < 50; j++) {
            line += chars[Math.round(Math.random() * (chars.length - 1))]
        }
        trainingText += line + '\n'
        let metrics = ctx0.measureText(line)
        if (metrics.width > width) {
            width = metrics.width
        }
        height += 40
    }
    var canvas = Canvas.createCanvas(width + 10, height + 10)
    var ctx = canvas.getContext('2d')
    ctx.fillStyle = '#ffffff'
    ctx.fillRect(0, 0, width + 10, height + 10)
    ctx.font = '30px pubg'
    ctx.fillStyle = '#000000'

    let boxInfo = ''
    let x = 5, y = 35, cx = 0, cy = 0, lastAscent = 0, lastDecent = 0, line = ''
    for (i in trainingText) {
        if (trainingText[i] == '\n') {
            let metrics = ctx0.measureText(line)
            boxInfo += `\t 5 ${Math.round(height + 10 - y - metrics.actualBoundingBoxDescent)} ${Math.round(metrics.width + 10)} ${Math.round(height + 10 - y + metrics.actualBoundingBoxAscent)} 0\n`
            y += 40
            x = 5
            line = ''
        } else {
            line += trainingText[i]
            ctx.fillText(trainingText[i], x, y)
            let metrics = ctx0.measureText(trainingText[i])
            boxInfo += `${trainingText[i]} ${Math.round(x)} ${Math.round(height + 10 - y - metrics.actualBoundingBoxDescent)} ${Math.round(x + metrics.width)} ${Math.round(height + 10 - y + metrics.actualBoundingBoxAscent)} 0\n`
            lastAscent = metrics.actualBoundingBoxAscent
            lastDecent = metrics.actualBoundingBoxDescent
            x += metrics.width
        }
    }
    fs.writeFileSync('pubg.font.exp0.box', boxInfo)
    fs.writeFileSync('font_properties', 'pubg 0 0 0 0 0')

    var img_src = canvas.toBuffer().toString('base64')
    const base64Data = new Buffer(img_src.replace(/^data:image\/\w+;base64,/, ""), 'base64')
    sharp(base64Data)
        .tiff({
            compression: 'lzw',
            squash: true
        })
        .toFile(`pubg.font.exp0.tif`)
        .then(res => {
            console.log(exec('tesseract pubg.font.exp0.tif pubg.font.exp0 nobatch box.train').toString())
            console.log(exec('unicharset_extractor pubg.font.exp0.box').toString())
            console.log(exec('shapeclustering -F font_properties -U unicharset -O pubg.unicharset pubg.font.exp0.tr').toString())
            console.log(exec('mftraining -F font_properties -U unicharset -O pubg.unicharset pubg.font.exp0.tr').toString())
            console.log(exec('cntraining pubg.font.exp0.tr').toString())
            console.log(exec('mv inttemp pubg.inttemp').toString())
            console.log(exec('mv pffmtable pubg.pffmtable').toString())
            console.log(exec('mv shapetable pubg.shapetable').toString())
            console.log(exec('mv normproto pubg.normproto').toString())
            console.log(exec('combine_tessdata pubg.').toString())
        })

}

genTiff()