refactor(WebFetchTool): remove dead Python webtools fallback
Browse filesWebFetch already had a complete pure-TS implementation (turndown +
domino + native fetch). The Python branch spawned scripts/python_webtools.py
which no longer exists, so it always failed and fell back to the TS path.
Remove the dead code and its .venv dependency.
- src/tools/WebFetchTool/utils.ts +0 -101
src/tools/WebFetchTool/utils.ts
CHANGED
|
@@ -137,86 +137,6 @@ async function retryWithBackoff<T>(
|
|
| 137 |
throw lastError
|
| 138 |
}
|
| 139 |
|
| 140 |
-
/**
|
| 141 |
-
* Fetch URL content using Python webtools script
|
| 142 |
-
* Reference: nanobot's web.py implementation
|
| 143 |
-
* Returns markdown formatted content with metadata
|
| 144 |
-
* Returns null if should fall back to direct fetch
|
| 145 |
-
*/
|
| 146 |
-
async function fetchWithPythonWebtools(url: string): Promise<{
|
| 147 |
-
content: string
|
| 148 |
-
contentType: string
|
| 149 |
-
title?: string
|
| 150 |
-
finalUrl?: string
|
| 151 |
-
} | null> {
|
| 152 |
-
console.log(`[WebFetch] Fetching via Python webtools: ${url}`)
|
| 153 |
-
|
| 154 |
-
try {
|
| 155 |
-
const { spawn } = await import('child_process')
|
| 156 |
-
|
| 157 |
-
return new Promise((resolve, reject) => {
|
| 158 |
-
const pythonScript = process.cwd() + '/scripts/python_webtools.py'
|
| 159 |
-
const maxChars = 50000
|
| 160 |
-
|
| 161 |
-
const child = spawn('.venv/bin/python', [pythonScript, 'web_fetch', url, String(50000)], {
|
| 162 |
-
cwd: process.cwd(),
|
| 163 |
-
})
|
| 164 |
-
|
| 165 |
-
let stdout = ''
|
| 166 |
-
let stderr = ''
|
| 167 |
-
|
| 168 |
-
child.stdout.on('data', (data) => {
|
| 169 |
-
stdout += data.toString()
|
| 170 |
-
})
|
| 171 |
-
|
| 172 |
-
child.stderr.on('data', (data) => {
|
| 173 |
-
stderr += data.toString()
|
| 174 |
-
})
|
| 175 |
-
|
| 176 |
-
child.on('close', (code) => {
|
| 177 |
-
if (code !== 0) {
|
| 178 |
-
console.error('[WebFetch] Python script failed:', stderr)
|
| 179 |
-
resolve(null) // Return null to trigger fallback
|
| 180 |
-
return
|
| 181 |
-
}
|
| 182 |
-
|
| 183 |
-
try {
|
| 184 |
-
const result = JSON.parse(stdout)
|
| 185 |
-
|
| 186 |
-
if (!result.success) {
|
| 187 |
-
console.error('[WebFetch] Python fetch failed:', result.error)
|
| 188 |
-
resolve(null) // Return null to trigger fallback
|
| 189 |
-
return
|
| 190 |
-
}
|
| 191 |
-
|
| 192 |
-
console.log(`[WebFetch] Python returned ${result.length} bytes`)
|
| 193 |
-
|
| 194 |
-
resolve({
|
| 195 |
-
content: result.text,
|
| 196 |
-
contentType: 'text/markdown',
|
| 197 |
-
title: undefined, // Python already includes title in text
|
| 198 |
-
finalUrl: result.finalUrl || url,
|
| 199 |
-
})
|
| 200 |
-
} catch (error) {
|
| 201 |
-
console.error('[WebFetch] Failed to parse Python output:', error)
|
| 202 |
-
resolve(null) // Return null to trigger fallback
|
| 203 |
-
}
|
| 204 |
-
})
|
| 205 |
-
|
| 206 |
-
child.on('error', (error) => {
|
| 207 |
-
console.error('[WebFetch] Failed to start Python process:', error)
|
| 208 |
-
resolve(null) // Return null to trigger fallback
|
| 209 |
-
})
|
| 210 |
-
})
|
| 211 |
-
} catch (error) {
|
| 212 |
-
console.error('[WebFetch] Failed to call Python webtools:', error)
|
| 213 |
-
logError('WebFetch: Failed to call Python webtools', error)
|
| 214 |
-
return null // Return null to trigger fallback
|
| 215 |
-
}
|
| 216 |
-
}
|
| 217 |
-
|
| 218 |
-
|
| 219 |
-
|
| 220 |
// Cache for storing fetched URL content
|
| 221 |
type CacheEntry = {
|
| 222 |
bytes: number
|
|
@@ -644,27 +564,6 @@ export async function getURLMarkdownContent(
|
|
| 644 |
console.error('[WebFetch] Local fetch failed:', error)
|
| 645 |
logError('Local fetch failed', error)
|
| 646 |
|
| 647 |
-
// Try Python webtools fallback
|
| 648 |
-
try {
|
| 649 |
-
console.log('[WebFetch] Trying Python webtools fallback')
|
| 650 |
-
const pythonResult = await fetchWithPythonWebtools(upgradedUrl)
|
| 651 |
-
|
| 652 |
-
if (pythonResult) {
|
| 653 |
-
const bytes = Buffer.byteLength(pythonResult.content)
|
| 654 |
-
const entry: CacheEntry = {
|
| 655 |
-
bytes,
|
| 656 |
-
code: 200,
|
| 657 |
-
codeText: 'OK',
|
| 658 |
-
content: pythonResult.content,
|
| 659 |
-
contentType: pythonResult.contentType,
|
| 660 |
-
}
|
| 661 |
-
URL_CACHE.set(url, entry, { size: Math.max(1, bytes) })
|
| 662 |
-
return entry
|
| 663 |
-
}
|
| 664 |
-
} catch (pythonError) {
|
| 665 |
-
console.error('[WebFetch] Python fallback also failed:', pythonError)
|
| 666 |
-
}
|
| 667 |
-
|
| 668 |
throw new Error(`Failed to fetch URL: ${error instanceof Error ? error.message : String(error)}`)
|
| 669 |
}
|
| 670 |
}
|
|
|
|
| 137 |
throw lastError
|
| 138 |
}
|
| 139 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 140 |
// Cache for storing fetched URL content
|
| 141 |
type CacheEntry = {
|
| 142 |
bytes: number
|
|
|
|
| 564 |
console.error('[WebFetch] Local fetch failed:', error)
|
| 565 |
logError('Local fetch failed', error)
|
| 566 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 567 |
throw new Error(`Failed to fetch URL: ${error instanceof Error ? error.message : String(error)}`)
|
| 568 |
}
|
| 569 |
}
|