feat(web): attach images from the composer

The (+) door and the clipboard door both stage files as removable
chips, the claw's own budget enforced client-side (eight a message,
5 MiB decoded each, errors surfaced, never silent drops). The sent
row renders local previews at once; the stream's user_message echo
carries the stored ids and dedupes against it, so a replayed message
renders its thumbnails through the same bearer-guarded fetch as an
attachment card. Model options mark vision when the listing knows.
This commit is contained in:
vhaudiquet committed 2026-10-26 23:24:41 -05:00
1 parent cfb349508f
commit 6efb69331d
6 files changed
+396 -39

No files matched your search

+230 -11
View File
@@ -13,7 +13,7 @@ import {
pasteParts,
} from '~/composables/useTranscript'
import type { ClawSession, ConsentAnswer, StreamStatus } from '~/composables/useClaw'
import type { PastePart, TranscriptItem, TurnBlock } from '~/composables/useTranscript'
import type { PastePart, SentImage, StagedImage, TranscriptItem, TurnBlock } from '~/composables/useTranscript'
const props = defineProps<{
items: TranscriptItem[]
@@ -31,8 +31,15 @@ const props = defineProps<{
const agentName = computed(() => props.identity?.name?.trim() || 'rhema')
/** One send as the panel hands it up: the text (pastes embedded) and
* the staged images, previews included for the optimistic row. */
interface SendPayload {
text: string
images: StagedImage[]
}
const emit = defineEmits<{
send: [text: string]
send: [message: SendPayload]
openProviders: []
consentAnswer: [id: string, answer: ConsentAnswer]
consentDeny: [id: string]
@@ -74,10 +81,15 @@ watch(
function submit(): void {
const text = draft.value.trim()
// The draft stays the gate: the claw 422s an empty text even with
// images aboard.
if (!text || !props.canSend || props.busy) return
emit('send', embedPastes(text, pastes.value))
emit('send', { text: embedPastes(text, pastes.value), images: stagedImages.value })
stageGeneration++
draft.value = ''
pastes.value = []
stagedImages.value = []
imageError.value = null
pinned.value = true
void nextTick(resizePrompt)
}
@@ -119,14 +131,22 @@ function onPaste(event: ClipboardEvent): void {
if (absorbPaste(text)) event.preventDefault()
return
}
// A file paste reads as its text when it claims a textual type; the
// rest the composer ignores.
const file = transfer.files[0]
if (!file || !file.type.startsWith('text/')) return
event.preventDefault()
void file.text().then((text) => {
if (!absorbPaste(text)) insertInline(text)
})
// A file paste: text files read inline or chip, images stage for the
// next send — the same door the (+) button opens; anything else the
// composer still ignores.
let handled = false
for (const file of Array.from(transfer.files)) {
if (file.type.startsWith('text/')) {
handled = true
void file.text().then((pasted) => {
if (!absorbPaste(pasted)) insertInline(pasted)
})
} else if (mediaTypeOf(file) !== null) {
stageImage(file)
handled = true
}
}
if (handled) event.preventDefault()
}
/** A long paste becomes one chip; whether it did is the answer. */
@@ -149,6 +169,119 @@ function insertInline(text: string): void {
resizePrompt()
}
// --- staged images ----------------------------------------------------
// The (+) button and the clipboard both land here: files held for the
// next send, previewed from object URLs, read to base64 at stage time
// so a bad file surfaces before submit, not inside it. The budget is
// the claw's own (http.rs): eight images, 5 MiB decoded each.
const MAX_IMAGES = 8
const MAX_IMAGE_BYTES = 5 * 1024 * 1024
const IMAGE_TYPES: readonly string[] = ['image/png', 'image/jpeg', 'image/gif', 'image/webp']
const IMAGE_ACCEPT = IMAGE_TYPES.join(',')
const stagedImages = ref<StagedImage[]>([])
const imageError = ref<string | null>(null)
const fileInput = ref<HTMLInputElement | null>(null)
/** Bumped on submit: a read still in flight when the draft cleared
* must not stage into a message that already left. */
let stageGeneration = 0
/** The wire media type a file stages as, or null when it is not one
* the claw accepts. The type usually rides the file; browsers that
* cannot name one leave it empty and the extension decides. */
function mediaTypeOf(file: File): string | null {
if (IMAGE_TYPES.includes(file.type)) return file.type
const byExt: Record<string, string> = {
png: 'image/png',
jpg: 'image/jpeg',
jpeg: 'image/jpeg',
gif: 'image/gif',
webp: 'image/webp',
}
return byExt[file.name.split('.').pop()?.toLowerCase() ?? ''] ?? null
}
/** B / KB / MB, one decimal past 1 KB — a chip label, not precision. */
function formatBytes(size: number): string {
if (size < 1024) return `${size} B`
if (size < 1024 * 1024) return `${(size / 1024).toFixed(1)} KB`
return `${(size / (1024 * 1024)).toFixed(1)} MB`
}
function pickImages(): void {
fileInput.value?.click()
}
function onPick(event: Event): void {
const input = event.target as HTMLInputElement
for (const file of Array.from(input.files ?? [])) stageImage(file)
// Cleared so picking the same file again still fires `change`.
input.value = ''
}
/** Stages one file, or surfaces why not: never a silent drop. */
function stageImage(file: File): void {
const label = file.name || 'clipboard image'
imageError.value = null
if (stagedImages.value.length >= MAX_IMAGES) {
imageError.value = `${label}: at most ${MAX_IMAGES} images a message`
return
}
const media_type = mediaTypeOf(file)
if (!media_type) {
imageError.value = `${label}: not a png, jpeg, gif, or webp image`
return
}
const generation = stageGeneration
const reader = new FileReader()
reader.onload = () => {
if (generation !== stageGeneration) return
// The data URL's head carries the mime back; the payload after the
// first comma is the bare base64 the body wants.
const dataUrl = typeof reader.result === 'string' ? reader.result : ''
const comma = dataUrl.indexOf(',')
const data = comma === -1 ? '' : dataUrl.slice(comma + 1)
if (data === '') {
imageError.value = `${label}: could not be read`
return
}
// Four base64 characters decode to three bytes — the same budget
// check the claw runs, no decode needed.
if (data.length * 3 / 4 > MAX_IMAGE_BYTES) {
imageError.value = `${label}: over the 5 MiB decoded budget`
return
}
stagedImages.value.push({ media_type, data, url: URL.createObjectURL(file), name: label })
}
reader.onerror = () => {
if (generation === stageGeneration) imageError.value = `${label}: could not be read`
}
reader.readAsDataURL(file)
}
function removeImage(index: number): void {
const [gone] = stagedImages.value.splice(index, 1)
if (gone) URL.revokeObjectURL(gone.url)
imageError.value = null
}
/** A stored image opens the same download path an attachment card
* uses; a local preview has nothing behind it to download. */
function downloadImage(image: SentImage): void {
if (image.id === null) return
emit('downloadAttachment', image.id, storedImageName(image))
}
/** A stored image's download name: the staged file's when the row is
* local, derived from the id when only the claw knows the bytes. */
function storedImageName(image: SentImage): string {
if (image.name) return image.name
const ext = image.media_type.replace('image/', '')
return `image-${image.id}.${ext === 'jpeg' ? 'jpg' : ext}`
}
/** One render piece of a user message: literal text or a paste chip. */
type MessageSegment = PastePart | { kind: 'chip'; label: string }
@@ -224,6 +357,18 @@ function streamTone(status: StreamStatus): 'ok' | 'warn' | 'error' {
<div class="avatar user" aria-hidden="true">·</div>
<div class="msg-body">
<div class="who">you</div>
<div v-if="item.images.length > 0" class="sent-images">
<button
v-for="(image, index) in item.images"
:key="index"
class="sent-image"
:disabled="image.url === null || image.id === null"
:title="image.url !== null && image.id !== null ? `download ${storedImageName(image)}` : (image.name ?? 'attached image')"
@click="downloadImage(image)"
>
<img v-if="image.url" :src="image.url" :alt="image.name ?? 'attached image'" />
</button>
</div>
<div class="reply plain">
<template v-for="(segment, index) in messageSegments(item.text)" :key="index">
<span v-if="segment.kind === 'chip'" class="badge paste-chip">{{ segment.label }}</span>
@@ -300,7 +445,22 @@ function streamTone(status: StreamStatus): 'ok' | 'warn' | 'error' {
</button>
</span>
</div>
<div v-if="stagedImages.length > 0" class="paste-chips">
<span v-for="(image, index) in stagedImages" :key="image.url" class="badge paste-chip">
▣{{ index + 1 }} {{ image.name }} {{ formatBytes(Math.round((image.data.length * 3) / 4)) }}
<button class="chip-x" :title="`drop ${image.name}`" @click="removeImage(index)">
×
</button>
</span>
</div>
<p v-if="imageError" class="image-error" role="alert">
{{ imageError }}
<button class="chip-x" title="dismiss" @click="imageError = null">×</button>
</p>
<div class="input-row">
<button class="ghost attach" title="attach images" :disabled="!canSend" @click="pickImages">
+
</button>
<textarea
ref="prompt"
v-model="draft"
@@ -314,6 +474,14 @@ function streamTone(status: StreamStatus): 'ok' | 'warn' | 'error' {
<button class="primary send" :disabled="!canSend || busy || draft.trim() === ''" title="send" @click="submit">
↑
</button>
<input
ref="fileInput"
class="file-input"
type="file"
:accept="IMAGE_ACCEPT"
multiple
@change="onPick"
/>
</div>
</footer>
</section>
@@ -646,6 +814,57 @@ function streamTone(status: StreamStatus): 'ok' | 'warn' | 'error' {
gap: 8px;
}
.attach {
width: 38px;
padding: 0;
font-size: 16px;
border-radius: 9px;
flex: none;
}
.file-input {
display: none;
}
.image-error {
display: flex;
align-items: center;
gap: 6px;
margin: 0 0 8px;
color: var(--error);
font-size: 12.5px;
}
/* Images a user message carried: the attachment-card visual, sized to
the sent row. A stored image downloads on click; a local preview has
nothing behind it. */
.sent-images {
display: flex;
flex-wrap: wrap;
gap: 8px;
margin-bottom: 6px;
}
.sent-image {
display: block;
padding: 0;
border: none;
background: none;
cursor: pointer;
}
.sent-image:disabled {
cursor: default;
}
.sent-image img {
display: block;
max-width: 100%;
max-height: 240px;
border: 1px solid var(--border);
border-radius: 8px;
}
.input-row textarea {
flex: 1;
min-width: 0;