Refs and Click Handling Reference

Overview

Refs give you direct access to rendered elements from your event handlers. Attach a ref to an element in .gsx, and after the render pass completes, the ref holds a pointer to that element. From there you can read layout bounds, control scroll position, or wire up click handlers.

go-tui provides three ref types to cover different use cases:

• Ref -- a single element reference
• RefList -- an ordered collection built from for loops
• RefMap[K] -- a keyed collection built from for loops with a key attribute

All three are thread-safe.

Ref

A reference to one element. Declare it as a struct field, create it in the constructor, attach it in .gsx with ref={...}, and read it in handlers.

NewRef

func NewRef() *Ref

Creates a new empty ref. The framework populates it during each render pass.

type myApp struct {
    header *tui.Ref
}

func MyApp() *myApp {
    return &myApp{
        header: tui.NewRef(),
    }
}

Set

func (r *Ref) Set(v *Element)

Stores an element in the ref. You rarely call this yourself since the generated code calls it during rendering. Each render pass overwrites the previous value, so the ref always points to the latest element.

El

func (r *Ref) El() *Element

Returns the stored element, or nil if the ref hasn't been set yet. Always check for nil before using the result, since the ref may be empty during the first render or if the element is conditionally hidden.

el := s.header.El()
if el != nil {
    _, maxY := el.MaxScroll()
    // use maxY
}

IsSet

func (r *Ref) IsSet() bool

Returns true if the ref holds a non-nil element.

if s.header.IsSet() {
    s.header.El().ScrollToBottom()
}

Usage in .gsx

Attach a ref to any element with the ref attribute:

templ (s *myApp) Render() {
    <div ref={s.header} class="border-rounded p-1">
        <span>Header content</span>
    </div>
}

The generated code calls s.header.Set(element) after creating the element, making it available in HandleMouse, KeyMap, and other handler methods.

RefList

An ordered collection of element references, populated automatically when you use ref= inside a for loop without a key attribute.

NewRefList

func NewRefList() *RefList

Creates an empty ref list.

type listApp struct {
    items    *tui.State[[]string]
    itemRefs *tui.RefList
}

func ListApp() *listApp {
    return &listApp{
        items:    tui.NewState([]string{"alpha", "beta", "gamma"}),
        itemRefs: tui.NewRefList(),
    }
}

Append

func (r *RefList) Append(el *Element)

Adds an element to the list. Called by the generated code for each iteration of a for loop where the element has a ref attribute.

All

func (r *RefList) All() []*Element

Returns a copy of all stored elements. The returned slice is safe to modify without affecting the ref list.

for _, el := range s.itemRefs.All() {
    fmt.Println(el.Text())
}

At

func (r *RefList) At(i int) *Element

Returns the element at the given index, or nil if the index is out of bounds.

el := s.itemRefs.At(2)
if el != nil {
    fmt.Println(el.Text())
}

Len

func (r *RefList) Len() int

Returns the number of elements in the list.

Usage in .gsx

Use ref= inside a for loop without a key attribute. The analyzer detects the loop context and generates Append calls instead of Set:

templ (s *listApp) Render() {
    <div class="flex-col">
        for _, item := range s.items.Get() {
            <span ref={s.itemRefs} class="p-1">{item}</span>
        }
    </div>
}

The same ref variable is used on every iteration. The generated code calls s.itemRefs.Append(el) each time through the loop.

RefMap[K]

A keyed collection of element references, populated when you use ref= inside a for loop with a key attribute. The key tells the generator to use Put(key, el) instead of Append(el), so you can look up a specific element by its key rather than by position.

NewRefMap

func NewRefMap[K comparable]() *RefMap[K]

Creates an empty ref map. The type parameter K must be comparable (strings, ints, and most value types work).

type tabApp struct {
    tabs    *tui.State[[]string]
    tabRefs *tui.RefMap[string]
}

func TabApp() *tabApp {
    return &tabApp{
        tabs:    tui.NewState([]string{"home", "settings", "help"}),
        tabRefs: tui.NewRefMap[string](),
    }
}

Put

func (r *RefMap[K]) Put(key K, el *Element)

Stores an element under the given key. Called by the generated code for each iteration of a for loop where the element has both ref and key attributes.

Get

func (r *RefMap[K]) Get(key K) *Element

Returns the element stored under the key, or nil if the key doesn't exist.

el := s.tabRefs.Get("settings")
if el != nil {
    // do something with the settings tab element
}

All

func (r *RefMap[K]) All() map[K]*Element

Returns a copy of all keyed elements. Safe to iterate over without affecting the ref map.

Len

func (r *RefMap[K]) Len() int

Returns the number of entries in the map.

Usage in .gsx

Use ref= together with key= inside a for loop. The key attribute provides the map key for each element:

templ (s *tabApp) Render() {
    <div class="flex gap-1">
        for _, name := range s.tabs.Get() {
            <button ref={s.tabRefs} key={name} class="px-1">{name}</button>
        }
    </div>
}

The generated code calls s.tabRefs.Put(name, el) each iteration. Without key, the same loop would generate Append calls and treat the ref as a RefList instead.

You can then look up specific elements in handlers:

el := s.tabRefs.Get("settings")
if el != nil && el.ContainsPoint(me.X, me.Y) {
    s.activateTab("settings")
}

Or iterate over all entries:

for name, el := range s.tabRefs.All() {
    if el != nil && el.ContainsPoint(me.X, me.Y) {
        s.activateTab(name)
        return true
    }
}

Click Handling

go-tui pairs refs with handler functions so you don't have to compare mouse coordinates against layout bounds yourself. You list which ref maps to which action, and the framework does the hit testing.

ClickBinding

type ClickBinding struct {
    Ref *Ref
    Fn  func()
}

Pairs a ref with a handler function. Created by Click and consumed by HandleClicks.

Click

func Click(ref *Ref, fn func()) ClickBinding

Links a ref to a callback.

tui.Click(s.saveBtn, s.save)

HandleClicks

func HandleClicks(me MouseEvent, bindings ...ClickBinding) bool

Tests a mouse event against a list of click bindings. Returns true if a binding matched and its handler was called.

How it works:

1. Ignores everything except left-button press events (MouseLeft + MousePress). Right clicks, releases, drags, and scroll wheel events return false immediately.
2. Iterates through bindings in order.
3. For each binding, checks that the ref is set (non-nil element) and that the click coordinates fall within the element's layout bounds via ContainsPoint.
4. Calls the handler of the first match and returns true.
5. If nothing matches, returns false.

Use it in your HandleMouse implementation:

func (c *colorMixer) HandleMouse(me tui.MouseEvent) bool {
    return tui.HandleClicks(me,
        tui.Click(c.redUpBtn, func() { c.adjustColor(&c.red, 16) }),
        tui.Click(c.redDnBtn, func() { c.adjustColor(&c.red, -16) }),
        tui.Click(c.greenUpBtn, func() { c.adjustColor(&c.green, 16) }),
        tui.Click(c.greenDnBtn, func() { c.adjustColor(&c.green, -16) }),
    )
}

Binding order matters. If two elements overlap, the first matching binding wins. Place more specific bindings before general ones.

Enabling Mouse Support

Mouse events only fire if you enable mouse support on the app:

app, err := tui.NewApp(
    tui.WithRootComponent(MyApp()),
    tui.WithMouse(),
)

Without WithMouse(), your HandleMouse method never gets called.

Practical Patterns

Scroll control with refs

Use refs to read layout information for scroll clamping:

type logViewer struct {
    lines    *tui.State[[]string]
    scrollY  *tui.State[int]
    content  *tui.Ref
}

func LogViewer() *logViewer {
    return &logViewer{
        lines:   tui.NewState([]string{}),
        scrollY: tui.NewState(0),
        content: tui.NewRef(),
    }
}

func (v *logViewer) scrollBy(delta int) {
    el := v.content.El()
    if el == nil {
        return
    }
    _, maxY := el.MaxScroll()
    newY := v.scrollY.Get() + delta
    if newY < 0 {
        newY = 0
    } else if newY > maxY {
        newY = maxY
    }
    v.scrollY.Set(newY)
}

templ (v *logViewer) Render() {
    <div
        ref={v.content}
        class="flex-col border-single p-1"
        scrollable={tui.ScrollVertical}
        scrollOffset={0, v.scrollY.Get()}
        flexGrow={1.0}
    >
        for _, line := range v.lines.Get() {
            <span>{line}</span>
        }
    </div>
}

Clickable buttons with visual feedback

Combine refs, state, and click handling for interactive buttons:

templ (s *myApp) Render() {
    <div class="flex gap-2">
        <button ref={s.incBtn} class="px-2 border-rounded text-green">{" + "}</button>
        <span class="font-bold">{fmt.Sprintf("%d", s.count.Get())}</span>
        <button ref={s.decBtn} class="px-2 border-rounded text-red">{" - "}</button>
    </div>
}

func (s *myApp) HandleMouse(me tui.MouseEvent) bool {
    return tui.HandleClicks(me,
        tui.Click(s.incBtn, func() {
            s.count.Update(func(v int) int { return v + 1 })
        }),
        tui.Click(s.decBtn, func() {
            s.count.Update(func(v int) int { return v - 1 })
        }),
    )
}

Conditional refs

When an element is conditionally rendered, its ref may be nil on some frames. Always guard access:

func (s *myApp) doSomething() {
    if el := s.panel.El(); el != nil {
        el.ScrollToTop()
    }
}

Thread Safety

All ref types use sync.RWMutex internally. El(), IsSet(), All(), At(), Get(), and Len() acquire a read lock. Set(), Append(), and Put() acquire a write lock.

In practice, the framework writes refs during the render pass (single-threaded) and you read them in handlers (also single-threaded on the event loop). The mutex is there for safety, not because concurrent access is common.