deadcxap/liberama
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Улучшена эвристика определения текста

Browse Source
This commit is contained in:
Book Pauk
2019-02-17 14:27:49 +07:00
parent 73bfc07082
commit db4bc2afb2
1 changed files with 11 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

13
server/core/BookConverter/textUtils.js
View File

@@ -70,13 +70,22 @@ function getEncoding(buf) {
function checkIfText(buf) { function checkIfText(buf) {
let spaceCount = 0; let spaceCount = 0;
let crCount = 0;
let lfCount = 0;
for (let i = 0; i < buf.length; i++) { for (let i = 0; i < buf.length; i++) {
if (buf[i] == 32) if (buf[i] == 32)
spaceCount++; spaceCount++;
if (buf[i] == 13)
crCount++;
if (buf[i] == 10)
lfCount++;
} }
const freq = spaceCount/(buf.length + 1);
return (freq > 0.1); const spaceFreq = spaceCount/(buf.length + 1);
const crFreq = crCount/(buf.length + 1);
const lfFreq = lfCount/(buf.length + 1);
return (spaceFreq > 0.1 || crFreq > 0.03 || lfFreq > 0.03);
} }
module.exports = { module.exports = {