Spaces:
Sleeping
Sleeping
Upload app.py
Browse files
app.py
CHANGED
|
@@ -8,7 +8,6 @@ import time
|
|
| 8 |
import threading
|
| 9 |
import requests
|
| 10 |
import pandas as pd
|
| 11 |
-
import pandas as pd
|
| 12 |
import gradio as gr
|
| 13 |
|
| 14 |
from langchain_core.messages import HumanMessage
|
|
@@ -104,101 +103,6 @@ def submit_answers(url: str, payload: dict) -> dict:
|
|
| 104 |
r.raise_for_status()
|
| 105 |
return r.json()
|
| 106 |
|
| 107 |
-
|
| 108 |
-
def fetch_task_file(task_id: str, question: str = "") -> tuple[str | None, str | None]:
|
| 109 |
-
"""
|
| 110 |
-
Download the file attached to a task from GET /files/{task_id}.
|
| 111 |
-
If that fails, search the question text for embedded file content or links.
|
| 112 |
-
Returns (local_path, injected_text):
|
| 113 |
-
- local_path: saved file path (for binary files like .py the model runs directly)
|
| 114 |
-
- injected_text: pre-read text content to inject into the question (for Excel/CSV)
|
| 115 |
-
Returns (None, None) if no attachment exists.
|
| 116 |
-
"""
|
| 117 |
-
url = f"{DEFAULT_API_URL}/files/{task_id}"
|
| 118 |
-
try:
|
| 119 |
-
r = requests.get(url, timeout=20)
|
| 120 |
-
if r.status_code == 404:
|
| 121 |
-
log(f"File : endpoint /files/{task_id} returned 404 — trying alternate patterns")
|
| 122 |
-
# Fallback: try query param format
|
| 123 |
-
url2 = f"{DEFAULT_API_URL}/files?task_id={task_id}"
|
| 124 |
-
try:
|
| 125 |
-
r = requests.get(url2, timeout=20)
|
| 126 |
-
if r.status_code != 200:
|
| 127 |
-
log(f"File : no attachment found (both endpoints failed)")
|
| 128 |
-
return None, None
|
| 129 |
-
except Exception as e:
|
| 130 |
-
log(f"File : fallback endpoint also failed — {e}")
|
| 131 |
-
return None, None
|
| 132 |
-
|
| 133 |
-
r.raise_for_status()
|
| 134 |
-
|
| 135 |
-
# Determine extension
|
| 136 |
-
ext = ""
|
| 137 |
-
cd = r.headers.get("Content-Disposition", "")
|
| 138 |
-
ct = r.headers.get("Content-Type", "")
|
| 139 |
-
|
| 140 |
-
if "filename=" in cd:
|
| 141 |
-
raw_name = cd.split("filename=")[-1].strip().strip('"').strip("'")
|
| 142 |
-
ext = os.path.splitext(raw_name)[-1].lower()
|
| 143 |
-
elif "spreadsheet" in ct or "excel" in ct or "openxmlformats" in ct:
|
| 144 |
-
ext = ".xlsx"
|
| 145 |
-
elif "csv" in ct:
|
| 146 |
-
ext = ".csv"
|
| 147 |
-
elif "pdf" in ct:
|
| 148 |
-
ext = ".pdf"
|
| 149 |
-
elif "png" in ct:
|
| 150 |
-
ext = ".png"
|
| 151 |
-
elif "jpeg" in ct or "jpg" in ct:
|
| 152 |
-
ext = ".jpg"
|
| 153 |
-
elif "python" in ct or "text/plain" in ct:
|
| 154 |
-
ext = ".py"
|
| 155 |
-
else:
|
| 156 |
-
ext = ""
|
| 157 |
-
|
| 158 |
-
local_path = f"/tmp/{task_id}{ext}"
|
| 159 |
-
with open(local_path, "wb") as f:
|
| 160 |
-
f.write(r.content)
|
| 161 |
-
|
| 162 |
-
log(f"File : downloaded → {local_path} ({len(r.content)} bytes, type={ext})")
|
| 163 |
-
|
| 164 |
-
# For Excel/CSV: read with pandas and inject as text — don't rely on
|
| 165 |
-
# the model downloading it again inside run_python.
|
| 166 |
-
injected_text = None
|
| 167 |
-
if ext in (".xlsx", ".xls"):
|
| 168 |
-
try:
|
| 169 |
-
import pandas as pd
|
| 170 |
-
dfs = pd.read_excel(local_path, sheet_name=None)
|
| 171 |
-
parts = []
|
| 172 |
-
for sheet, df in dfs.items():
|
| 173 |
-
parts.append(f"=== Sheet: {sheet} ===\n{df.to_csv(index=False)}")
|
| 174 |
-
injected_text = "\n\n".join(parts)[:6000]
|
| 175 |
-
log(f"File : Excel read OK — {sum(len(d) for d in dfs.values())} rows total")
|
| 176 |
-
except Exception as e:
|
| 177 |
-
log(f"File : Excel read failed — {e}")
|
| 178 |
-
|
| 179 |
-
elif ext == ".csv":
|
| 180 |
-
try:
|
| 181 |
-
import pandas as pd
|
| 182 |
-
df = pd.read_csv(local_path)
|
| 183 |
-
injected_text = df.to_csv(index=False)[:6000]
|
| 184 |
-
log(f"File : CSV read OK — {len(df)} rows")
|
| 185 |
-
except Exception as e:
|
| 186 |
-
log(f"File : CSV read failed — {e}")
|
| 187 |
-
|
| 188 |
-
elif ext in (".py", ".txt"):
|
| 189 |
-
try:
|
| 190 |
-
with open(local_path, "r", errors="replace") as f:
|
| 191 |
-
injected_text = f.read()[:4000]
|
| 192 |
-
log(f"File : text read OK")
|
| 193 |
-
except Exception as e:
|
| 194 |
-
log(f"File : text read failed — {e}")
|
| 195 |
-
|
| 196 |
-
return local_path, injected_text
|
| 197 |
-
|
| 198 |
-
except Exception as e:
|
| 199 |
-
log(f"File : download failed — {e}")
|
| 200 |
-
return None, None
|
| 201 |
-
|
| 202 |
# ==========================================================
|
| 203 |
# MAIN RUNNER
|
| 204 |
# ==========================================================
|
|
@@ -252,22 +156,6 @@ def run_and_submit_all(profile: gr.OAuthProfile | None):
|
|
| 252 |
log(f"Question : {question[:200]}")
|
| 253 |
log("-" * 80)
|
| 254 |
|
| 255 |
-
# Download any attached file and inject content into the question
|
| 256 |
-
file_path, injected_text = fetch_task_file(task_id, question)
|
| 257 |
-
if injected_text:
|
| 258 |
-
# For Excel/CSV/text: embed the content directly so the model
|
| 259 |
-
# doesn't need to re-download anything
|
| 260 |
-
ext = os.path.splitext(file_path)[-1].lower() if file_path else ""
|
| 261 |
-
if ext in (".xlsx", ".xls", ".csv"):
|
| 262 |
-
question = question + f"\n\n[ATTACHED FILE DATA — use this for calculations, do NOT try to re-read the file]:\n{injected_text}"
|
| 263 |
-
elif ext in (".py", ".txt"):
|
| 264 |
-
question = question + f"\n\n[ATTACHED FILE CONTENT — this is the file to process]:\n{injected_text}"
|
| 265 |
-
else:
|
| 266 |
-
question = question + f"\n[ATTACHED FILE: {file_path}]"
|
| 267 |
-
elif file_path:
|
| 268 |
-
# Binary files (images, PDF) — just pass the path
|
| 269 |
-
question = question + f"\n[ATTACHED FILE: {file_path}]"
|
| 270 |
-
|
| 271 |
start = time.time()
|
| 272 |
submitted_answer = "N/A"
|
| 273 |
tools_used: list[str] = []
|
|
@@ -391,4 +279,4 @@ with gr.Blocks() as demo:
|
|
| 391 |
# ==========================================================
|
| 392 |
|
| 393 |
if __name__ == "__main__":
|
| 394 |
-
demo.launch(debug=True, ssr_mode=False)
|
|
|
|
| 8 |
import threading
|
| 9 |
import requests
|
| 10 |
import pandas as pd
|
|
|
|
| 11 |
import gradio as gr
|
| 12 |
|
| 13 |
from langchain_core.messages import HumanMessage
|
|
|
|
| 103 |
r.raise_for_status()
|
| 104 |
return r.json()
|
| 105 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 106 |
# ==========================================================
|
| 107 |
# MAIN RUNNER
|
| 108 |
# ==========================================================
|
|
|
|
| 156 |
log(f"Question : {question[:200]}")
|
| 157 |
log("-" * 80)
|
| 158 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 159 |
start = time.time()
|
| 160 |
submitted_answer = "N/A"
|
| 161 |
tools_used: list[str] = []
|
|
|
|
| 279 |
# ==========================================================
|
| 280 |
|
| 281 |
if __name__ == "__main__":
|
| 282 |
+
demo.launch(debug=True, ssr_mode=False)
|