Spot it. Save it. Free, local-first Chrome extension for images, videos and prompts. Explore 2,000+ curated cases and connect your library to AI Agents via MCP. 提示词导演
JavaScript ai-artai-videochrome-extensionclaude-codecodexhiggsfieldjimenglocal-firstmcpmidjourneymoodboardprompt-engineeringprompt-libraryprompt-managerreference-libraryseedancevisual-bookmarksweb-clipper
from __future__ import annotations
import os
from pathlib import Path
from tempfile import mkdtemp
from playwright.sync_api import expect
from e2e_support import base_entry, extension_session
PNG_BASE64 = "iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNk+A8AAQUBAScY42YAAAAASUVORK5CYII="
def main() -> None:
entry = base_entry(
"captured-x-post",
"Director Name · A practical lighting breakdown",
"A practical lighting breakdown for a midnight chase: keep the street cyan, reserve warm light for the character, and let the wet road carry the contrast.",
"content:prompt:image",
1,
)
entry.update({
"url": "https://x.com/director/status/123",
"sourceFacts": {
"provider": "x",
"pageType": "post",
"itemId": "123",
"author": "Director Name",
"handle": "director",
"publishedAt": "2026-08-16T09:30:00.000Z",
"capturedAt": "2026-08-16T09:31:00.000Z",
"engagement": {"likes": 42, "reposts": 7},
"captureScope": "document",
"extractionMethod": "page",
"status": "complete",
},
"mediaAssets": [
{
"id": "post-image",
"kind": "image",
"usage": "content",
"storageMode": "managed",
"mimeType": "image/png",
"width": 1600,
"height": 1067,
"capturedAt": "2026-08-16T09:31:00.000Z",
"reviewStatus": "verified",
"palette": {"colors": ["#080808"], "source": "screenshot", "version": 2},
},
{
"id": "post-video",
"kind": "video",
"usage": "content",
"storageMode": "reference",
"sourceUrl": "https://x.com/director/status/123/video/1",
"sourceTitle": "Attached clip",
"originalWorkUrl": "https://x.com/director/status/123",
"reference": {
"url": "https://x.com/director/status/123/video/1",
"provider": "x",
"playbackMode": "embed",
},
"reviewStatus": "verified",
},
],
"primaryMediaId": "post-image",
"articleDocument": {
"version": 1,
"blocks": [
{"id": "post-copy", "kind": "paragraph", "text": "A practical lighting breakdown for a midnight chase: keep the street cyan, reserve warm light for the character, and let the wet road carry the contrast.", "sourceOrder": 0},
{"id": "post-image-block", "kind": "image", "assetId": "post-image", "sourceUrl": "https://x.com/post-image.png", "sourceOrder": 1},
{"id": "post-video-block", "kind": "video", "assetId": "post-video", "sourceUrl": "https://x.com/director/status/123/video/1", "sourceOrder": 2},
],
},
})
entry["articleDocument"]["blocks"].extend([
{"id": "quote", "kind": "quote", "text": "引用:先明确人物站位,再描述镜头运动。", "sourceOrder": 3},
{"id": "steps", "kind": "list", "ordered": True, "text": "确定主体位置\n安排运动方向\n检查前后衔接", "sourceOrder": 4},
{"id": "code", "kind": "code", "text": "camera: tracking\n subject: walking", "sourceOrder": 5},
{"id": "link", "kind": "link", "label": "原始参考资料 " + "reference/" * 15,
"sourceUrl": "https://example.com/reference", "sourceOrder": 6},
])
evidence = Path(os.environ.get("PROMPTDIRECTOR_LAB_EVIDENCE_DIR") or mkdtemp(prefix="post-detail-evidence-"))
evidence.mkdir(parents=True, exist_ok=True)
with extension_session("prompt-director-x-post-") as run:
setup = run.open_page("collector.html")
run.seed_storage(setup, {"entries": [entry], "uiPreferences": {"theme": "dark", "locale": "zh-CN", "motion": "reduced"}})
setup.evaluate(
"""async png => {
const {saveMediaBlob} = await import(chrome.runtime.getURL('media-store.js'));
const bytes = Uint8Array.from(atob(png), value => value.charCodeAt(0));
await saveMediaBlob('post-image', new Blob([bytes], {type: 'image/png'}), {checkCapacity: false});
}""", PNG_BASE64)
library = run.open_page("library.html", wait_until="networkidle")
expect(library.locator(".case-card")).to_have_count(1)
library.locator(".case-card").click()
expect(library.locator(".detail-visual-gallery")).to_be_visible(timeout=8000)
expect(library.locator(".original-prompt-panel")).to_contain_text("practical lighting breakdown")
expect(library.locator(".captured-post-view, .article-document-reader")).to_have_count(0)
expect(library.locator("#detail-content")).to_have_class(__import__('re').compile(r'^detail-content\b.*\bhas-primary-media\b'))
before = setup.evaluate("async()=>{const s=await chrome.runtime.sendMessage({type:'GET_STATE'});return s.entries[0]}")
def classify(category):
result = setup.evaluate("""async category => chrome.runtime.sendMessage({
type:'CONFIRM_CLASSIFICATION', entryId:'captured-x-post', pathIds:[category], rememberSource:false
})""", category)
assert result['ok'], result
classify('content:reference')
reader = library.locator('.article-document-reader')
expect(reader).to_be_visible(timeout=8000)
expect(library.locator('.detail-visual-gallery, .captured-post-view')).to_have_count(0)
expect(library.locator('.original-prompt-panel')).to_have_count(0)
expect(reader.locator('blockquote')).to_have_text('引用:先明确人物站位,再描述镜头运动。')
expect(reader.locator('ol > li')).to_have_count(3)
expect(reader.locator('pre')).to_have_text('camera: tracking\n subject: walking')
expect(reader.locator('.article-document-image')).to_have_count(1)
expect(reader.locator('.compact-media-link')).to_have_count(1)
assert reader.locator('.compact-media-link').bounding_box()['height'] < 80, '没有封面的视频不能撑成空白大卡片'
expect(reader.locator('iframe')).to_have_count(0)
for theme in ('dark', 'light'):
library.evaluate("async theme=>{const {initializeUi}=await import('./i18n.js');await initializeUi({theme,locale:'zh-CN',motion:'reduced'})}", theme)
for width in (1280, 390, 320):
library.set_viewport_size({'width': width, 'height': 900})
assert reader.evaluate('e=>e.scrollWidth<=e.clientWidth'), (theme, width, 'article overflow')
header = library.locator('.detail-header-section')
assert header.evaluate('e=>e.scrollWidth<=e.clientWidth'), (theme, width, 'header overflow')
title = header.locator('.detail-title').bounding_box()
edit = header.locator('.entry-editor-inline > summary').bounding_box()
assert title['x'] + title['width'] <= edit['x'] + 1, (theme, width, 'title overlap')
resource = reader.locator('.article-document-resource')
assert resource.evaluate('e=>{const t=e.querySelector("strong").getBoundingClientRect(), a=e.querySelector("a").getBoundingClientRect();return a.x>=t.right-1 && a.y<t.bottom && a.bottom>t.y}'), (theme,width,'resource action must stay beside its label')
assert reader.locator('p').first.evaluate('e=>getComputedStyle(e).fontWeight') == '400'
library.screenshot(path=str(evidence / f'article-detail-{theme}-{width}.png'))
# A category change on the open article must remount the existing case template.
classify('content:prompt:video')
expect(library.locator('.detail-visual-gallery')).to_be_visible(timeout=8000)
expect(reader).to_have_count(0)
expect(library.locator('.original-prompt-panel')).to_contain_text('practical lighting breakdown')
library.locator('.detail-visual-thumb').nth(1).click()
expect(library.locator('.detail-visual-gallery')).to_have_class(__import__('re').compile('is-video-detail'))
expect(library.locator('.original-prompt-panel')).to_contain_text('practical lighting breakdown')
classify('content:reference')
expect(reader).to_be_visible(timeout=8000)
after = setup.evaluate("async()=>{const s=await chrome.runtime.sendMessage({type:'GET_STATE'});return s.entries[0]}")
for key in ('text', 'mediaAssets', 'articleDocument', 'sourceFacts'):
assert before[key] == after[key], (key, before[key], after[key])
# List editing stays a plain-text edit, then restores the reading structure.
reader.get_by_role('button', name='编辑正文', exact=True).click()
steps = reader.locator('[data-article-block-id="steps"]')
steps.fill('取消的修改')
reader.get_by_role('button', name='取消', exact=True).click()
expect(reader.locator('ol > li')).to_have_count(3)
reader.get_by_role('button', name='编辑正文', exact=True).click()
steps.fill('新的第一步\n新的第二步')
reader.get_by_role('button', name='保存', exact=True).click()
expect(reader.locator('ol > li')).to_have_count(2)
expect(reader.locator('ol')).to_contain_text('新的第二步')
reader.get_by_role('button', name='复制文字', exact=True).click()
copied = library.evaluate('async()=>navigator.clipboard.readText()')
assert '新的第二步' in copied and '取消的修改' not in copied
assert reader.get_by_role('button', name='复制文字', exact=True).locator('svg').count() == 1
library.reload(wait_until='networkidle')
library.locator('.case-card').click()
expect(library.locator('.article-document-reader ol > li')).to_have_count(2)
stored = setup.evaluate("async()=>{const s=await chrome.runtime.sendMessage({type:'GET_STATE'});return s.entries[0]}")
assert stored['mediaAssets'] == before['mediaAssets']
assert next(b for b in stored['articleDocument']['blocks'] if b['id']=='steps')['text'] == '新的第一步\n新的第二步'
print({'classification_routes_detail': True, 'live_mode_switch': True, 'structured_article': True,
'list_edit_readback': True, 'media_unchanged': True, 'widths': [1280,390,320], 'themes': 2, 'evidence': str(evidence)})
if __name__ == "__main__":
main()