browser

Documentation

Overview

Package browser drives a real headless Chrome through the DevTools Protocol so JavaScript-built pages are captured as they actually render. kage always goes through here: navigate, let the page settle, then serialise the final DOM — the same markup a human would have seen — which the rest of the pipeline then strips of scripts and localises.

Index

• func LookChrome() (string, bool)
• type Options
  • func DefaultOptions() Options
• type Pool
  • func New(opts Options) *Pool
  • func (p *Pool) Close() error
  • func (p *Pool) Render(ctx context.Context, rawURL string) (RenderResult, error)
• type RenderResult

Functions

func LookChrome

func LookChrome() (string, bool)

LookChrome reports the path of a usable Chrome/Chromium binary and whether one was found, checking KAGE_CHROME, CHROME_BIN, rod's own lookup, and the common system install locations. Tests use it to skip when no browser is present.

Types

type Options

type Options struct {
	Headless      bool          // run Chrome without a window
	Workers       int           // max concurrent pages
	Settle        time.Duration // network-idle quiet period after load
	RenderTimeout time.Duration // hard cap per page render
	Scroll        bool          // auto-scroll to trigger lazy-loaded media
	ChromeBin     string        // explicit binary; empty = autodetect
	ControlURL    string        // attach to an existing Chrome instead of launching
}

Options configure a Pool.

func DefaultOptions

func DefaultOptions() Options

DefaultOptions returns the baseline render settings.

type Pool

type Pool struct {
	// contains filtered or unexported fields
}

Pool owns one Chrome process shared across a run and bounds the number of pages open at once.

func New

func New(opts Options) *Pool

New creates a Pool. Chrome is launched lazily on the first Render.

func (*Pool) Close

func (p *Pool) Close() error

Close shuts down the managed Chrome process.

func (*Pool) Render

func (p *Pool) Render(ctx context.Context, rawURL string) (RenderResult, error)

Render navigates to rawURL, lets it settle, and returns the final rendered HTML. It acquires a page slot from the pool and releases it when done.

type RenderResult

type RenderResult struct {
	HTML     string // the serialised final DOM
	FinalURL string // URL after any client-side redirects
	Title    string
}

RenderResult is the outcome of rendering one page.