Spaces:
Sleeping
Sleeping
meirk-brd
commited on
Commit
·
7ee2792
1
Parent(s):
92b8ebf
read temp file
Browse files
tool.py
CHANGED
|
@@ -106,7 +106,7 @@ class BrightDataDatasetTool(Tool):
|
|
| 106 |
return json.dumps(data, indent=2)
|
| 107 |
except requests.exceptions.RequestException as exc:
|
| 108 |
details = exc.response.text if getattr(exc, "response", None) is not None else ""
|
| 109 |
-
return json.dumps({"error": str(exc), "details": details, "payload": payload})
|
| 110 |
|
| 111 |
def _build_params(
|
| 112 |
self,
|
|
@@ -216,6 +216,16 @@ class BrightDataDatasetTool(Tool):
|
|
| 216 |
return self._ensure_scheme(raw)
|
| 217 |
|
| 218 |
if isinstance(raw, dict):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 219 |
orig_name = raw.get("orig_name")
|
| 220 |
if (
|
| 221 |
isinstance(orig_name, str)
|
|
@@ -225,8 +235,20 @@ class BrightDataDatasetTool(Tool):
|
|
| 225 |
return self._ensure_scheme(orig_name)
|
| 226 |
|
| 227 |
url_value = raw.get("url")
|
| 228 |
-
if isinstance(url_value, str)
|
| 229 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 230 |
|
| 231 |
return None
|
| 232 |
|
|
|
|
| 106 |
return json.dumps(data, indent=2)
|
| 107 |
except requests.exceptions.RequestException as exc:
|
| 108 |
details = exc.response.text if getattr(exc, "response", None) is not None else ""
|
| 109 |
+
return json.dumps({"error": str(exc), "details": details, "payload": payload, "coerced_url": url})
|
| 110 |
|
| 111 |
def _build_params(
|
| 112 |
self,
|
|
|
|
| 216 |
return self._ensure_scheme(raw)
|
| 217 |
|
| 218 |
if isinstance(raw, dict):
|
| 219 |
+
path_value = raw.get("path")
|
| 220 |
+
if isinstance(path_value, str) and os.path.isfile(path_value):
|
| 221 |
+
try:
|
| 222 |
+
with open(path_value, "r", encoding="utf-8", errors="ignore") as fh:
|
| 223 |
+
text = fh.read().strip()
|
| 224 |
+
if text:
|
| 225 |
+
return self._ensure_scheme(text)
|
| 226 |
+
except OSError:
|
| 227 |
+
pass
|
| 228 |
+
|
| 229 |
orig_name = raw.get("orig_name")
|
| 230 |
if (
|
| 231 |
isinstance(orig_name, str)
|
|
|
|
| 235 |
return self._ensure_scheme(orig_name)
|
| 236 |
|
| 237 |
url_value = raw.get("url")
|
| 238 |
+
if isinstance(url_value, str):
|
| 239 |
+
if url_value.startswith("/gradio_api/file="):
|
| 240 |
+
file_path = url_value.split("=", 1)[-1]
|
| 241 |
+
if os.path.isfile(file_path):
|
| 242 |
+
try:
|
| 243 |
+
with open(file_path, "r", encoding="utf-8", errors="ignore") as fh:
|
| 244 |
+
text = fh.read().strip()
|
| 245 |
+
if text:
|
| 246 |
+
return self._ensure_scheme(text)
|
| 247 |
+
except OSError:
|
| 248 |
+
pass
|
| 249 |
+
return None
|
| 250 |
+
if url_value.startswith(("http://", "https://")):
|
| 251 |
+
return url_value
|
| 252 |
|
| 253 |
return None
|
| 254 |
|