schoenbergerb / noscrape Goto Github PK

Glyphs are sometimes not rendered correctly

demo-data.zip

in 1of 30 tries the letter f is missing in generated font for unknown reasons

Discussed in #44

Description

Extend the Noscrape constructor to allow passing the font as an ArrayBuffer | SharedArrayBuffer in addition to the current option of passing it as a path (string).

Expected Result

• The Noscrape constructor can be called with a path as a string.
  The font is then loaded by the loadSync function of opentype.js.
• The Noscrape constructor can be called with a buffer ArrayBuffer | SharedArrayBuffer.
  In this case, the parse function of opentype.js should be used.

The Readme.md have to be updated to the new object-orientated approach.
Add some use-cases aswell.

Description

The playwright test revealed that on Firefox there were slight rendering differences.

compared with: https://www.diffchecker.com/image-compare/

Expected Result

Text rendered with the original font should match exactly, or as closely as possible, the text rendered with the obfuscation font.

As the new object-oriented approach is much more flexible, the functional is not needed anymore.

if we use full spectrum of its 1294 glyphs from this example.ttf it comes to weird glyph shiftings

example.ttf.zip

Description

Since Playwright installation the "tests" - folder was included into production build. This is wrong and lead to an wrong build path.

As the copy function cannot handle date objects or similar the cloneDeep function from lodash is the better approach here.

src/obfuscate/copy.ts

https://www.geeksforgeeks.org/lodash-_-clonedeep-method/

What should I say... 😅

• add proper ESLint settings
• make sure code is clean and well structured
• add comments!
• and so on

maybe by adding random invisible "dots" within drawn area or something like that

deprecated in 0.5.2, the obfuscate function will be removed in 1.0.0

In this case the u is missing

demodata.zip

project must be build and deployed automatically to npmjs.com via GitHub action

Description

To make sure everything is rendered well and looks like it should (different unicodes, same paths) a visual regression test should be implemented.

Expected Result

• The test should run on any branch on any commit.
• Test should make sure the unicodes are different, e.g. "original string" <> "obfuscated string"
• Test should make sure that visual the obfuscated result is exactly the same as the original
  
  
  

Additional Details

• https://playwright.dev/docs/test-snapshots

reproducible on https://noscrape-example.vercel.app/dynamic by clicking multiple times on reload