How thu03092004/rpa_lucidchart works
This repository appears to support natural-language-driven browser automation: instructions are parsed into actions, elements can be located through text or image-based methods, and Selenium actions are executed or emitted as Java. Visual utilities also detect shapes and compare screenshots. The excerpts establish the parser, generator, browser/image routines, and canvas dispatcher, but do not show enough wiring to assert how every locator or visual subsystem is called; those links are omitted where uncertain.
Main entry points: generator.py Action generator · find_element.py Element locator · by_image.py Image locator · step_parser.py Step parser
Components 16
Instruction and execution
- Step parser
step_parser.py - Action generator
generator.py - Generated Java
GeneratedTest.java
Element location
- Element locator
find_element.py - Text locator
by_text.py - Image locator
by_image.py - Image helpers
by_image_helper.py
Visual analysis
- Canvas shape dispatcher
detect_shapes/canvas/check_canvas.py - Canvas detectors
detect_shapes/canvas - Element detectors
detect_shapes/element - Screenshot differ
detect_shapes/diff_ui.py - Detection outputs
detect_img
Browser interaction
- Chrome browser
- Screenshot files
screenshots
Other components
- Automation user
- Web page
Connections 16
- Automation user → Step parser: provides steps
- Step parser → Action generator: parsed steps
- Action generator → Element locator: locates elements
- Action generator → Chrome browser: executes actions
- Action generator → Generated Java: emits Java
- Element locator → Text locator: delegates lookup
- Element locator → Image locator: delegates lookup
- Text locator → Chrome browser: reads page DOM
- Image locator → Chrome browser: captures screenshots
- Image locator → Image helpers: uses helpers
- Chrome browser → Web page: opens page
- Chrome browser → Screenshot files: saves captures
- Canvas shape dispatcher → Canvas detectors: dispatches detection
- Canvas detectors → Detection outputs: writes annotated images
- Screenshot differ → Detection outputs: writes difference mask
- Screenshot files → Screenshot differ: compares images