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
"""Cases edited from another page update the open library in place. Isolated synthetic fixture.
The incremental result must equal a fresh page load for the same stored library, unchanged cards keep
their DOM nodes (no flicker), and search sees the new text without re-reading the whole library.
"""
import random
from e2e_support import extension_session
COUNT = 240
SEED = """async (count) => {
const [{SCHEMA_VERSION}, {createDefaultFacetCatalog}, {createDefaultOrganizerState}] = await Promise.all([
import('./taxonomy.js'), import('./facets.js'), import('./organizer.js')]);
const entries = Array.from({length: count}, (_, i) => ({schemaVersion: SCHEMA_VERSION, id: `case-${i}`,
title: `增量案例${i}`, text: `第${i}条参考内容`, textRevision: 1,
savedAt: new Date(Date.UTC(2026, 8, 1, 0, 0, i)).toISOString(),
classification: {pathIds: ['content:image-case'], status: 'confirmed', source: 'manual'},
mediaAssets: [], customLabels: ['参考'], timeNotes: []}));
const {getLibraryStorage}=await import('./library-storage.js');
await getLibraryStorage().set({schemaVersion: SCHEMA_VERSION, entries, facetCatalog: createDefaultFacetCatalog(),
organizerState: createDefaultOrganizerState(), compoundCases: [], trashState: {version: 1, items: []}});
}"""
SNAPSHOT = """() => ({
cards: [...document.querySelectorAll('#case-list .case-card')].map(card => `${card.dataset.entryId}|${card.textContent.replace(/\\s+/g, ' ').trim()}`),
count: document.querySelector('#result-count')?.textContent
})"""
def search(page, query):
page.locator("#search-input").fill(query)
page.wait_for_timeout(300)
result = page.evaluate("() => [...document.querySelectorAll('#case-list .case-card')].map(card => card.dataset.entryId)")
page.locator("#search-input").fill("")
page.wait_for_timeout(300)
return result
def main():
rng = random.Random(20261006)
with extension_session("pd-incremental-refresh-", viewport={"width": 1440, "height": 900}) as run:
editor = run.open_page("collector.html")
editor.evaluate(SEED, COUNT)
library = run.open_page("library.html")
library.wait_for_selector('body[data-library-state="ready"]')
library.wait_for_function("() => document.querySelectorAll('#case-list .case-card').length >= 24")
library.evaluate("""() => { window.pdFullReads = 0;
const get = chrome.storage.local.get.bind(chrome.storage.local);
chrome.storage.local.get = keys => { if ([keys].flat().includes('entries')) window.pdFullReads++; return get(keys); }; }""")
library.wait_for_timeout(1500) # let the opening refresh settle before marking card nodes
library.evaluate("""() => { window.pdSerializedCases = [];
const stringify = JSON.stringify;
JSON.stringify = function(value, ...rest) {
if (value && typeof value === 'object' && typeof value.id === 'string'
&& value.id.startsWith('case-') && typeof value.text === 'string') window.pdSerializedCases.push(value.id);
return stringify.call(JSON, value, ...rest);
};
}""")
library.evaluate("() => document.querySelectorAll('#case-list .case-card').forEach(card => { card.pdOriginal = true; })")
edited = set()
for step in range(12):
index = rng.randrange(0, 24)
entry_id = f"case-{COUNT - 1 - index}"
kind = rng.choice(["labels", "title"])
message = ({"type": "UPDATE_ENTRY_CUSTOM_LABELS", "entryId": entry_id, "addLabels": [f"新标签{step}"], "removeLabels": []}
if kind == "labels" else {"type": "UPDATE_ENTRY_TITLE", "entryId": entry_id, "title": f"改名案例{step}"})
assert editor.evaluate("message => chrome.runtime.sendMessage(message)", message)["ok"], message
edited.add(entry_id)
if kind == "title":
library.wait_for_function("title => document.querySelector('#case-list').textContent.includes(title)", arg=f"改名案例{step}")
else:
library.wait_for_timeout(250)
library.wait_for_timeout(400)
incremental = library.evaluate(SNAPSHOT)
kept = library.evaluate("""edited => [...document.querySelectorAll('#case-list .case-card')]
.filter(card => !edited.includes(card.dataset.entryId)).every(card => card.pdOriginal === true)""", sorted(edited))
full_reads = library.evaluate("window.pdFullReads")
serialized = library.evaluate("window.pdSerializedCases")
incremental_search = [search(library, "改名案例"), search(library, "新标签3")]
fresh = run.open_page("library.html")
fresh.wait_for_selector('body[data-library-state="ready"]')
fresh.wait_for_function("() => document.querySelectorAll('#case-list .case-card').length >= 24")
expected = fresh.evaluate(SNAPSHOT)
expected_search = [search(fresh, "改名案例"), search(fresh, "新标签3")]
assert incremental == expected, {"incremental": incremental, "fresh": expected}
assert incremental_search == expected_search, (incremental_search, expected_search)
assert kept, "unchanged cards must keep their nodes and loaded previews"
assert not (set(serialized) - edited), f"untouched original cases must not be serialized during precise updates: {len(serialized)} serializations"
assert full_reads == 0, f"case-only changes must not re-read the whole library ({full_reads})"
print({"edits": 12, "equal_to_fresh_load": True, "unchanged_cards_kept": True, "library_rereads": full_reads,
"search_results": [len(item) for item in incremental_search]})
if __name__ == "__main__":
main()