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
"""Run historical source directly into current source in one disposable Chrome profile.
Usage: python3 test/s6_direct_upgrade_e2e.py --baseline-ref <release-commit-or-tag>
This verifies data migration and cold-start persistence, not the release installer or Store package.
"""
from __future__ import annotations
import argparse
import copy
import io
import json
import shutil
import subprocess
import tempfile
import zipfile
from contextlib import contextmanager
from pathlib import Path
from playwright.sync_api import expect, sync_playwright
from e2e_support import EXTENSION_DIR, SOURCE_EXTENSION_DIR, base_entry, launch_context, record_page_errors, wait_for_async_condition
SEED = """async fixture => {
const taxonomy = await import('./taxonomy.js');
const {createDefaultFacetCatalog, createFacetNode} = await import('./facets.js');
const {ORGANIZER_VERSION} = await import('./organizer.js');
const {saveMediaBlob} = await import('./media-store.js');
const {migrateLibraryState} = await import('./migration.js');
const defaults = createDefaultFacetCatalog();
const group = defaults.nodes.find(node => !node.parentId && node.status === 'active');
const catalog = createFacetNode(defaults, {id: 'direct-upgrade:manual-tag', name: '人工创作标签',
facetId: group.facetId, parentId: group.id, origin: 'manual'});
const manualNode = catalog.nodes.find(node => node.id === 'direct-upgrade:manual-tag');
const active = {...fixture.active, schemaVersion: taxonomy.SCHEMA_VERSION,
facetAssignments: [{facetId: manualNode.facetId, nodeId: manualNode.id, source: 'manual', status: 'confirmed', confidence: 1}],
mediaPrompts: [{assetId: 'direct-upgrade:image', text: fixture.originalPrompt, source: 'webpage'}],
sourceFacts: {originalPromptAvailable: true, author: 'Fixture creator', originalSourceUrl: fixture.active.url}};
const trashed = {...fixture.trashed, schemaVersion: taxonomy.SCHEMA_VERSION};
const png = Uint8Array.from(atob('iVBORw0KGgoAAAANSUhEUgAAAAEAAAABCAQAAAC1HAwCAAAAC0lEQVR42mNk+A8AAQUBAScY42YAAAAASUVORK5CYII='), c => c.charCodeAt(0));
for (const [entry, id, blob, kind, sourceFormat] of [
[active, 'direct-upgrade:image', new Blob([png], {type: 'image/png'}), 'image', 'png'],
[trashed, 'direct-upgrade:trash-original', new Blob(['original PSD bytes kept in Trash'], {type: 'image/vnd.adobe.photoshop'}), 'attachment', 'psd']
]) {
await saveMediaBlob(id, blob, {checkCapacity: false});
entry.primaryMediaId = id;
entry.mediaAssets = [{id, kind, storageMode: 'managed', mimeType: blob.type, sourceFormat, byteSize: blob.size,
...(kind === 'image' ? {width: 1, height: 1} : {})}];
}
await chrome.storage.local.clear();
// Normalize the synthetic records with the historical implementation before saving them.
await chrome.storage.local.set(migrateLibraryState({schemaVersion: taxonomy.SCHEMA_VERSION,
entries: [active], taxonomy: taxonomy.createDefaultTaxonomy(), facetCatalog: catalog,
classificationRules: [], compoundCases: [],
organizerState: {version: ORGANIZER_VERSION, collections: [{id: 'direct-upgrade:project',
name: '升级归属项目', parentId: null, order: 0, entryIds: [active.id]}]},
trashState: {version: 1, items: [{id: 'direct-upgrade:trash', kind: 'entry', targetId: trashed.id,
deletedAt: '2026-08-26T00:10:00.000Z', snapshot: trashed, relationships: {collections: []}}]}
}).state);
const state = await chrome.runtime.sendMessage({type: 'GET_STATE'});
if (!state.ok) throw new Error(state.message);
}"""
SNAPSHOT = """async () => {
const state = await chrome.runtime.sendMessage({type: 'GET_STATE'});
if (!state.ok) throw new Error(state.message);
const {getMediaBlob} = await import('./media-store.js');
const stripSchema = entry => {const {schemaVersion, ...content} = entry; return content;};
const media = [];
for (const entry of [...state.entries, ...state.trashState.items.map(item => item.snapshot)]) {
for (const asset of entry.mediaAssets) {
const blob = await getMediaBlob(asset.id);
if (!blob) throw new Error(`Missing original ${asset.id}`);
media.push({id: asset.id, type: blob.type, bytes: Array.from(new Uint8Array(await blob.arrayBuffer()))});
}
}
return {schema: state.schemaVersion, version: chrome.runtime.getManifest().version,
content: {entries: state.entries.map(stripSchema),
trash: state.trashState.items.map(item => ({...item, snapshot: stripSchema(item.snapshot)})),
projects: state.organizerState, facets: state.facetCatalog, compounds: state.compoundCases, media}};
}"""
@contextmanager
def opened_library(playwright, profile: Path, installation: Path, *, reload_extension=False):
context = launch_context(playwright, str(profile), viewport={"width": 1280, "height": 900},
accept_downloads=True, extension_dir=installation)
errors = []
try:
worker = context.service_workers[0]
extension_id = worker.url.split('/')[2]
if reload_extension:
# The source build keeps its manifest version; Chrome can retain the previous worker
# across browser restarts until the user reloads the unpacked extension.
status_page = context.new_page()
status_page.goto('chrome://extensions')
if not status_page.locator('#devMode').evaluate('element => element.checked'):
status_page.locator('#devMode').click()
status_page.evaluate("async id => chrome.developerPrivate.reload(id, {failQuietly: false})", extension_id)
wait_for_async_condition(status_page, """async id =>
(await chrome.developerPrivate.getExtensionsInfo({includeDisabled:true,includeTerminated:true}))
.some(extension => extension.id === id && extension.state === 'ENABLED' && !extension.disableReasons.reloading)
""", arg=extension_id)
status_page.close()
page = context.new_page()
record_page_errors(page, errors)
page.goto(f"chrome-extension://{extension_id}/library.html", wait_until="networkidle")
page.wait_for_selector('body[data-library-state="ready"]')
yield page, extension_id
assert not errors, errors
finally:
context.close()
def main() -> None:
parser = argparse.ArgumentParser(description=__doc__)
parser.add_argument('--baseline-ref', required=True, help='Historical Git ref containing the pre-S6 extension')
args = parser.parse_args()
repo = SOURCE_EXTENSION_DIR.parent
baseline_commit = subprocess.check_output(['git', 'rev-parse', '--verify', f'{args.baseline_ref}^{{commit}}'], cwd=repo, text=True).strip()
archive = subprocess.check_output(['git', 'archive', '--format=zip', baseline_commit, 'extension'], cwd=repo)
active = base_entry('direct-upgrade:active', '直接升级案例', '正文第一段。\n\n正文第二段:标点、空行和完整说明应保留。', 'content:image-case')
active['customLabels'] = ['人工标签', '导演选材']
trashed = base_entry('direct-upgrade:trashed', '回收站原件', '回收站正文不丢失', 'content:image-case')
trashed['customLabels'] = ['回收站人工标签']
original_prompt = 'Original prompt, line one.\n原始提示词第二行 — 保留完整内容。'
with tempfile.TemporaryDirectory(prefix='pd-direct-upgrade-') as temporary, sync_playwright() as playwright:
root = Path(temporary)
with zipfile.ZipFile(io.BytesIO(archive)) as source:
source.extractall(root)
installation, profile = root / 'extension', root / 'profile'
with opened_library(playwright, profile, installation) as (page, extension_id):
page.evaluate(SEED, {'active': active, 'trashed': trashed, 'originalPrompt': original_prompt})
baseline = page.evaluate(SNAPSHOT)
raw_before = page.evaluate("() => chrome.storage.local.get(['entries', 'caseIndex', 'upgradeBackup'])")
assert 'caseIndex' not in raw_before, 'The baseline must precede per-case storage'
assert 'upgradeBackup' not in raw_before, 'This exercise must start without an intermediate guard version'
entry = baseline['content']['entries'][0]
assert entry['text'] == active['text'] and entry['customLabels'] == active['customLabels']
assert entry['mediaPrompts'][0]['text'] == original_prompt
assert entry['facetAssignments'][0]['source'] == 'manual'
assert baseline['content']['trash'][0]['snapshot']['text'] == trashed['text']
assert baseline['content']['projects']['collections'][0]['entryIds'] == [active['id']]
# Replace only this disposable installation after closing Chromium. No intermediate source runs.
shutil.rmtree(installation)
shutil.copytree(EXTENSION_DIR, installation)
with opened_library(playwright, profile, installation, reload_extension=True) as (page, current_id):
assert current_id == extension_id, 'Upgrading must keep extension identity and its existing profile'
upgraded = page.evaluate(SNAPSHOT)
assert upgraded['schema'] > baseline['schema'], (baseline['schema'], upgraded['schema'])
assert upgraded['content'] == baseline['content'], {key: {'before': baseline['content'][key], 'after': upgraded['content'][key]} for key in baseline['content'] if upgraded['content'][key] != baseline['content'][key]}
layout = page.evaluate("() => chrome.storage.local.get(['entries', 'caseIndex', 'legacyEntries', 'upgradeBackup'])")
assert 'entries' not in layout and layout['caseIndex']['ids'] == [active['id']], layout.keys()
assert layout['legacyEntries'] == raw_before['entries'], 'The initial recovery copy must preserve the historical records'
backup = layout['upgradeBackup']
assert [backup['fromSchemaVersion'], backup['toSchemaVersion']] == [baseline['schema'], upgraded['schema']]
assert backup['state']['entries'] == raw_before['entries']
for message in [
{'type': 'UPDATE_ENTRY_TITLE', 'entryId': active['id'], 'title': '升级后新的标题'},
{'type': 'UPDATE_ENTRY_CUSTOM_LABELS', 'entryId': active['id'], 'addLabels': ['升级后新增标签'], 'removeLabels': []},
]:
result = page.evaluate('message => chrome.runtime.sendMessage(message)', message)
assert result.get('ok'), result
edited = page.evaluate(SNAPSHOT)
expected_content = copy.deepcopy(baseline['content'])
expected_content['entries'][0]['title'] = '升级后新的标题'
expected_content['entries'][0]['customLabels'] = [*active['customLabels'], '升级后新增标签']
# Edits also advance timestamps/revisions; every unaffected user field stays exact.
for field, value in expected_content['entries'][0].items():
if field not in {'updatedAt', 'libraryUpdatedAt'}:
assert edited['content']['entries'][0][field] == value, field
for field in ['trash', 'projects', 'facets', 'compounds', 'media']:
assert edited['content'][field] == expected_content[field], field
expect(page.locator('#case-list')).to_contain_text('升级后新的标题')
with opened_library(playwright, profile, installation) as (page, reopened_id):
assert reopened_id == extension_id
assert page.evaluate(SNAPSHOT) == edited, 'New edits must survive a cold browser/background restart'
expect(page.locator('#case-list')).to_contain_text('升级后新的标题')
retained = page.evaluate("() => chrome.storage.local.get(['legacyEntries', 'upgradeBackup'])")
assert retained['legacyEntries'] == raw_before['entries']
assert retained['upgradeBackup'] == backup
print(json.dumps({'baselineCommit': baseline_commit, 'versions': [baseline['version'], upgraded['version']],
'schemaUpgrade': [baseline['schema'], upgraded['schema']], 'extensionIdentityKept': True,
'historicalContentAndOriginalBytesKept': True, 'coldRestartEditReadback': True,
'releaseInstallerTested': False}, ensure_ascii=False))
if __name__ == '__main__':
main()