trying to extract file
Browse files- app copy.py +29 -1
app copy.py
CHANGED
|
@@ -5,6 +5,8 @@ import inspect
|
|
| 5 |
import pandas as pd
|
| 6 |
from langchain_core.messages import HumanMessage
|
| 7 |
from agent import build_graph
|
|
|
|
|
|
|
| 8 |
|
| 9 |
# (Keep Constants as is)
|
| 10 |
# --- Constants ---
|
|
@@ -22,8 +24,28 @@ class BasicAgent:
|
|
| 22 |
messages = [HumanMessage(content=question)]
|
| 23 |
result = self.graph.invoke({"messages": messages})
|
| 24 |
answer = result['messages'][-1].content
|
| 25 |
-
print(f"Agent returning answer: {answer}")
|
| 26 |
return answer
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 27 |
|
| 28 |
agent = BasicAgent()
|
| 29 |
questions_url = f"{DEFAULT_API_URL}/questions"
|
|
@@ -34,12 +56,18 @@ for item in questions_data[:5]:
|
|
| 34 |
question_text = item.get("question")
|
| 35 |
if question_text is None:
|
| 36 |
continue
|
|
|
|
|
|
|
|
|
|
|
|
|
| 37 |
output = agent(question_text)
|
| 38 |
print("Q:", question_text)
|
| 39 |
print("A:", output)
|
| 40 |
print("-" * 40)
|
| 41 |
|
| 42 |
|
|
|
|
|
|
|
| 43 |
# def run_and_submit_all( profile: gr.OAuthProfile | None):
|
| 44 |
# """
|
| 45 |
# Fetches all questions, runs the BasicAgent on them, submits all answers,
|
|
|
|
| 5 |
import pandas as pd
|
| 6 |
from langchain_core.messages import HumanMessage
|
| 7 |
from agent import build_graph
|
| 8 |
+
from huggingface_hub import HfApi, hf_hub_download
|
| 9 |
+
from logging import logger
|
| 10 |
|
| 11 |
# (Keep Constants as is)
|
| 12 |
# --- Constants ---
|
|
|
|
| 24 |
messages = [HumanMessage(content=question)]
|
| 25 |
result = self.graph.invoke({"messages": messages})
|
| 26 |
answer = result['messages'][-1].content
|
|
|
|
| 27 |
return answer
|
| 28 |
+
|
| 29 |
+
def file_extract(local_file_path,task_file_paths):
|
| 30 |
+
if local_file_path:
|
| 31 |
+
try:
|
| 32 |
+
# GAIA's file_path is relative to the dataset repo root.
|
| 33 |
+
# Download the file into the allowed cache and get its local path.
|
| 34 |
+
resolved_path = hf_hub_download(
|
| 35 |
+
repo_id="gaia-benchmark/GAIA",
|
| 36 |
+
filename=local_file_path, # e.g. "2023/test/<attachment-id>.pdf"
|
| 37 |
+
repo_type="dataset",
|
| 38 |
+
)
|
| 39 |
+
|
| 40 |
+
task_file_paths[str(task_id)] = resolved_path
|
| 41 |
+
logger.debug(
|
| 42 |
+
f"Stored file path mapping for task_id {task_id}: {resolved_path}"
|
| 43 |
+
)
|
| 44 |
+
except Exception as e:
|
| 45 |
+
logger.warning(
|
| 46 |
+
f"Could not download file '{local_file_path}' for task_id {task_id}: {e}. "
|
| 47 |
+
"Mapping skipped."
|
| 48 |
+
)
|
| 49 |
|
| 50 |
agent = BasicAgent()
|
| 51 |
questions_url = f"{DEFAULT_API_URL}/questions"
|
|
|
|
| 56 |
question_text = item.get("question")
|
| 57 |
if question_text is None:
|
| 58 |
continue
|
| 59 |
+
files_text = item.get("files")
|
| 60 |
+
task_id = item.get("task_id")
|
| 61 |
+
# file = file_extract(,task_id)
|
| 62 |
+
print(files_text,task_id)
|
| 63 |
output = agent(question_text)
|
| 64 |
print("Q:", question_text)
|
| 65 |
print("A:", output)
|
| 66 |
print("-" * 40)
|
| 67 |
|
| 68 |
|
| 69 |
+
|
| 70 |
+
|
| 71 |
# def run_and_submit_all( profile: gr.OAuthProfile | None):
|
| 72 |
# """
|
| 73 |
# Fetches all questions, runs the BasicAgent on them, submits all answers,
|