Update ui/ui_core.py
Browse files- ui/ui_core.py +67 -45
ui/ui_core.py
CHANGED
@@ -12,9 +12,11 @@ sys.path.insert(0, os.path.abspath(os.path.join(os.path.dirname(__file__), "..",
|
|
12 |
from txagent.txagent import TxAgent
|
13 |
|
14 |
def sanitize_utf8(text: str) -> str:
|
15 |
-
|
16 |
-
|
17 |
-
|
|
|
|
|
18 |
|
19 |
def extract_all_text_from_csv_or_excel(file_path: str, progress=None, index=0, total=1) -> str:
|
20 |
try:
|
@@ -36,11 +38,10 @@ def extract_all_text_from_csv_or_excel(file_path: str, progress=None, index=0, t
|
|
36 |
line = " | ".join(str(cell) for cell in row if pd.notna(cell))
|
37 |
if line:
|
38 |
lines.append(line)
|
39 |
-
|
40 |
-
return sanitize_utf8(content)
|
41 |
|
42 |
except Exception as e:
|
43 |
-
return
|
44 |
|
45 |
def extract_all_text_from_pdf(file_path: str, progress=None, index=0, total=1) -> str:
|
46 |
try:
|
@@ -58,11 +59,10 @@ def extract_all_text_from_pdf(file_path: str, progress=None, index=0, total=1) -
|
|
58 |
progress((index + (i / num_pages)) / total, desc=f"Reading PDF: {os.path.basename(file_path)} ({i+1}/{num_pages})")
|
59 |
except Exception as e:
|
60 |
extracted.append(f"[Error reading page {i+1}]: {str(e)}")
|
61 |
-
|
62 |
-
return sanitize_utf8(content)
|
63 |
|
64 |
except Exception as e:
|
65 |
-
return
|
66 |
|
67 |
def create_ui(agent: TxAgent):
|
68 |
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
@@ -103,46 +103,68 @@ def create_ui(agent: TxAgent):
|
|
103 |
elif path.endswith(".pdf"):
|
104 |
extracted_text += extract_all_text_from_pdf(path, progress, index, total_files) + "\n"
|
105 |
else:
|
106 |
-
extracted_text +=
|
107 |
except Exception as file_error:
|
108 |
-
extracted_text +=
|
109 |
continue
|
110 |
|
111 |
-
|
112 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
)
|
114 |
|
115 |
-
|
116 |
-
|
117 |
-
|
118 |
-
|
119 |
-
|
120 |
-
max_token=8192,
|
121 |
-
call_agent=False,
|
122 |
-
conversation=conversation,
|
123 |
-
uploaded_files=uploaded_files,
|
124 |
-
max_round=30
|
125 |
-
)
|
126 |
-
|
127 |
-
for update in generator:
|
128 |
-
try:
|
129 |
-
if isinstance(update, list):
|
130 |
-
cleaned = [
|
131 |
-
msg for msg in update
|
132 |
-
if hasattr(msg, 'role')
|
133 |
-
and not (
|
134 |
-
msg.role == "assistant"
|
135 |
-
and hasattr(msg, 'content')
|
136 |
-
and sanitize_utf8(msg.content.strip()).startswith("π§ ")
|
137 |
-
)
|
138 |
-
]
|
139 |
-
if cleaned:
|
140 |
-
yield cleaned
|
141 |
-
elif isinstance(update, str) and not sanitize_utf8(update.strip()).startswith("π§ "):
|
142 |
-
yield sanitize_utf8(update)
|
143 |
-
except Exception as update_error:
|
144 |
-
print(f"Error processing update: {update_error}")
|
145 |
-
continue
|
146 |
|
147 |
except Exception as chat_error:
|
148 |
print(f"Chat handling error: {chat_error}")
|
@@ -158,4 +180,4 @@ def create_ui(agent: TxAgent):
|
|
158 |
["Is there anything abnormal in the attached blood work report?"]
|
159 |
], inputs=message_input)
|
160 |
|
161 |
-
return demo
|
|
|
12 |
from txagent.txagent import TxAgent
|
13 |
|
14 |
def sanitize_utf8(text: str) -> str:
|
15 |
+
return re.sub(r'[\ud800-\udfff]', '', text)
|
16 |
+
|
17 |
+
def chunk_text(text: str, chunk_size: int = 8000) -> List[str]:
|
18 |
+
words = text.split()
|
19 |
+
return [" ".join(words[i:i + chunk_size]) for i in range(0, len(words), chunk_size)]
|
20 |
|
21 |
def extract_all_text_from_csv_or_excel(file_path: str, progress=None, index=0, total=1) -> str:
|
22 |
try:
|
|
|
38 |
line = " | ".join(str(cell) for cell in row if pd.notna(cell))
|
39 |
if line:
|
40 |
lines.append(line)
|
41 |
+
return f"π {os.path.basename(file_path)}\n\n" + "\n".join(lines)
|
|
|
42 |
|
43 |
except Exception as e:
|
44 |
+
return f"[Error reading {os.path.basename(file_path)}]: {str(e)}"
|
45 |
|
46 |
def extract_all_text_from_pdf(file_path: str, progress=None, index=0, total=1) -> str:
|
47 |
try:
|
|
|
59 |
progress((index + (i / num_pages)) / total, desc=f"Reading PDF: {os.path.basename(file_path)} ({i+1}/{num_pages})")
|
60 |
except Exception as e:
|
61 |
extracted.append(f"[Error reading page {i+1}]: {str(e)}")
|
62 |
+
return f"π {os.path.basename(file_path)}\n\n" + "\n\n".join(extracted)
|
|
|
63 |
|
64 |
except Exception as e:
|
65 |
+
return f"[Error reading PDF {os.path.basename(file_path)}]: {str(e)}"
|
66 |
|
67 |
def create_ui(agent: TxAgent):
|
68 |
with gr.Blocks(theme=gr.themes.Soft()) as demo:
|
|
|
103 |
elif path.endswith(".pdf"):
|
104 |
extracted_text += extract_all_text_from_pdf(path, progress, index, total_files) + "\n"
|
105 |
else:
|
106 |
+
extracted_text += f"(Uploaded file: {os.path.basename(path)})\n"
|
107 |
except Exception as file_error:
|
108 |
+
extracted_text += f"[Error processing file: {os.path.basename(path)}] β {str(file_error)}\n"
|
109 |
continue
|
110 |
|
111 |
+
sanitized_text = sanitize_utf8(extracted_text.strip())
|
112 |
+
chunks = chunk_text(sanitized_text)
|
113 |
+
|
114 |
+
for i, chunk in enumerate(chunks):
|
115 |
+
message_chunk = (
|
116 |
+
f"{context}\n\n--- Uploaded File Chunk {i + 1} ---\n\n{chunk}\n\n--- End of Chunk ---\n\nNow begin your reasoning:"
|
117 |
+
)
|
118 |
+
generator = agent.run_gradio_chat(
|
119 |
+
message=message_chunk,
|
120 |
+
history=history,
|
121 |
+
temperature=0.3,
|
122 |
+
max_new_tokens=1024,
|
123 |
+
max_token=8192,
|
124 |
+
call_agent=False,
|
125 |
+
conversation=conversation,
|
126 |
+
uploaded_files=uploaded_files,
|
127 |
+
max_round=30
|
128 |
+
)
|
129 |
+
|
130 |
+
for update in generator:
|
131 |
+
try:
|
132 |
+
if isinstance(update, list):
|
133 |
+
cleaned = [
|
134 |
+
msg for msg in update
|
135 |
+
if hasattr(msg, 'role') and not (
|
136 |
+
msg.role == "assistant"
|
137 |
+
and hasattr(msg, 'content')
|
138 |
+
and msg.content.strip().startswith("π§ ")
|
139 |
+
)
|
140 |
+
]
|
141 |
+
if cleaned:
|
142 |
+
yield cleaned
|
143 |
+
elif isinstance(update, str) and not update.strip().startswith("π§ "):
|
144 |
+
yield sanitize_utf8(update.encode("utf-8", "replace").decode("utf-8"))
|
145 |
+
except Exception as update_error:
|
146 |
+
print(f"Error processing update: {update_error}")
|
147 |
+
continue
|
148 |
+
|
149 |
+
else:
|
150 |
+
# Fallback for message-only interactions
|
151 |
+
generator = agent.run_gradio_chat(
|
152 |
+
message=message,
|
153 |
+
history=history,
|
154 |
+
temperature=0.3,
|
155 |
+
max_new_tokens=1024,
|
156 |
+
max_token=8192,
|
157 |
+
call_agent=False,
|
158 |
+
conversation=conversation,
|
159 |
+
uploaded_files=uploaded_files,
|
160 |
+
max_round=30
|
161 |
)
|
162 |
|
163 |
+
for update in generator:
|
164 |
+
if isinstance(update, str):
|
165 |
+
yield sanitize_utf8(update.encode("utf-8", "replace").decode("utf-8"))
|
166 |
+
else:
|
167 |
+
yield update
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
168 |
|
169 |
except Exception as chat_error:
|
170 |
print(f"Chat handling error: {chat_error}")
|
|
|
180 |
["Is there anything abnormal in the attached blood work report?"]
|
181 |
], inputs=message_input)
|
182 |
|
183 |
+
return demo
|