fix: add text sanitization function to clean user and assistant messages of new parts object

- Introduced a `sanitizeText` function to remove specific HTML elements and content from messages, enhancing the integrity of the streamed text.
- Updated the `streamText` function to utilize `sanitizeText` for both user and assistant messages, ensuring consistent message formatting.
- Adjusted message processing to maintain the structure while applying sanitization.
This commit is contained in:
KevIsDev
2025-07-16 02:29:51 +01:00
parent 2b40b8af52
commit 1af54ecda9
4 changed files with 93 additions and 119 deletions

View File

@@ -26,6 +26,14 @@ export interface StreamingOptions extends Omit<Parameters<typeof _streamText>[0]
const logger = createScopedLogger('stream-text');
function sanitizeText(text: string): string {
let sanitized = text.replace(/<div class=\\"__boltThought__\\">.*?<\/div>/s, '');
sanitized = sanitized.replace(/<think>.*?<\/think>/s, '');
sanitized = sanitized.replace(/<boltAction type="file" filePath="package-lock\.json">[\s\S]*?<\/boltAction>/g, '');
return sanitized.trim();
}
export async function streamText(props: {
messages: Omit<Message, 'id'>[];
env?: Env;
@@ -58,30 +66,25 @@ export async function streamText(props: {
let currentModel = DEFAULT_MODEL;
let currentProvider = DEFAULT_PROVIDER.name;
let processedMessages = messages.map((message) => {
const newMessage = { ...message };
if (message.role === 'user') {
const { model, provider, content } = extractPropertiesFromMessage(message);
currentModel = model;
currentProvider = provider;
return { ...message, content };
newMessage.content = sanitizeText(content);
} else if (message.role == 'assistant') {
let content = message.content;
content = content.replace(/<div class=\\"__boltThought__\\">.*?<\/div>/s, '');
content = content.replace(/<think>.*?<\/think>/s, '');
// Remove package-lock.json content specifically keeping token usage MUCH lower
content = content.replace(
/<boltAction type="file" filePath="package-lock\.json">[\s\S]*?<\/boltAction>/g,
'[package-lock.json content removed]',
);
// Trim whitespace potentially left after removals
content = content.trim();
return { ...message, content };
newMessage.content = sanitizeText(message.content);
}
return message;
// Sanitize all text parts in parts array, if present
if (Array.isArray(message.parts)) {
newMessage.parts = message.parts.map((part) =>
part.type === 'text' ? { ...part, text: sanitizeText(part.text) } : part,
);
}
return newMessage;
});
const provider = PROVIDER_LIST.find((p) => p.name === currentProvider) || DEFAULT_PROVIDER;

View File

@@ -65,6 +65,13 @@ The year is 2025.
: ''
}
${
supabase?.isConnected &&
supabase?.hasSelectedProject &&
supabase?.credentials?.supabaseUrl &&
supabase?.credentials?.anonKey
? `
Create .env file if it doesn't exist${
supabase?.isConnected &&
supabase?.hasSelectedProject &&
@@ -75,10 +82,6 @@ The year is 2025.
VITE_SUPABASE_ANON_KEY=${supabase.credentials.anonKey}`
: '.'
}
NEVER modify Supabase config or .env files apart from creating .env.
Do not generate types for supabase.
DATA PRESERVATION REQUIREMENTS:
- DATA INTEGRITY IS HIGHEST PRIORITY - users must NEVER lose data
- FORBIDDEN: Destructive operations (DROP, DELETE) that could cause data loss
@@ -131,6 +134,9 @@ The year is 2025.
- One migration per logical change
- Use descriptive policy names
- Add indexes for frequently queried columns
`
: ''
}
</database_instructions>
<artifact_instructions>

View File

@@ -597,7 +597,11 @@ export class FilesStore {
// Set up file watcher
webcontainer.internal.watchPaths(
{ include: [`${WORK_DIR}/**`], exclude: ['**/node_modules', '.git'], includeContent: true },
{
include: [`${WORK_DIR}/**`],
exclude: ['**/node_modules', '.git', '**/package-lock.json'],
includeContent: true,
},
bufferWatchEvents(100, this.#processEventBuffer.bind(this)),
);

View File

@@ -151,45 +151,6 @@ export class PreviewsStore {
this._broadcastStorageSync();
});
try {
// Watch for file changes
webcontainer.internal.watchPaths(
{
// Only watch specific file types that affect the preview
include: ['**/*.html', '**/*.css', '**/*.js', '**/*.jsx', '**/*.ts', '**/*.tsx', '**/*.json'],
exclude: ['**/node_modules/**', '**/.git/**', '**/dist/**', '**/build/**', '**/coverage/**'],
},
async (_events) => {
const previews = this.previews.get();
for (const preview of previews) {
const previewId = this.getPreviewId(preview.baseUrl);
if (previewId) {
this.broadcastFileChange(previewId);
}
}
},
);
// Watch for DOM changes that might affect storage
if (typeof window !== 'undefined') {
const observer = new MutationObserver((_mutations) => {
// Broadcast storage changes when DOM changes
this._broadcastStorageSync();
});
observer.observe(document.body, {
childList: true,
subtree: true,
characterData: true,
attributes: true,
});
}
} catch (error) {
console.error('[Preview] Error setting up watchers:', error);
}
// Listen for port events
webcontainer.on('port', (port, type, url) => {
let previewInfo = this.#availablePreviews.get(port);