Commit
·
26bb643
1
Parent(s):
3c5863d
feat: model filtering and UI upgrade for TTD
Browse files- app.py +537 -155
- climateqa/engine/talk_to_data/config.py +33 -0
- climateqa/engine/talk_to_data/main.py +10 -5
- climateqa/engine/talk_to_data/plot.py +26 -17
- climateqa/engine/talk_to_data/sql_query.py +13 -35
- climateqa/engine/talk_to_data/utils.py +44 -30
- climateqa/engine/talk_to_data/workflow.py +38 -34
- style.css +9 -3
app.py
CHANGED
@@ -9,14 +9,14 @@ from climateqa.engine.embeddings import get_embeddings_function
|
|
9 |
from climateqa.engine.llm import get_llm
|
10 |
from climateqa.engine.vectorstore import get_pinecone_vectorstore
|
11 |
from climateqa.engine.reranker import get_reranker
|
12 |
-
from climateqa.engine.graph import make_graph_agent,make_graph_agent_poc
|
13 |
from climateqa.engine.chains.retrieve_papers import find_papers
|
14 |
from climateqa.chat import start_chat, chat_stream, finish_chat
|
15 |
from climateqa.engine.talk_to_data.main import ask_drias, DRIAS_MODELS
|
16 |
from climateqa.engine.talk_to_data.myVanna import MyVanna
|
17 |
|
18 |
-
from front.tabs import
|
19 |
-
from front.tabs import
|
20 |
from front.utils import process_figures
|
21 |
from gradio_modal import Modal
|
22 |
|
@@ -25,14 +25,14 @@ from utils import create_user_id
|
|
25 |
import logging
|
26 |
|
27 |
logging.basicConfig(level=logging.WARNING)
|
28 |
-
os.environ[
|
29 |
logging.getLogger().setLevel(logging.WARNING)
|
30 |
|
31 |
|
32 |
-
|
33 |
# Load environment variables in local mode
|
34 |
try:
|
35 |
from dotenv import load_dotenv
|
|
|
36 |
load_dotenv()
|
37 |
except Exception as e:
|
38 |
pass
|
@@ -63,42 +63,105 @@ share_client = service.get_share_client(file_share_name)
|
|
63 |
user_id = create_user_id()
|
64 |
|
65 |
|
66 |
-
|
67 |
# Create vectorstore and retriever
|
68 |
embeddings_function = get_embeddings_function()
|
69 |
-
vectorstore = get_pinecone_vectorstore(
|
70 |
-
|
71 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
72 |
|
73 |
-
llm = get_llm(provider="openai",max_tokens
|
74 |
if os.environ["GRADIO_ENV"] == "local":
|
75 |
reranker = get_reranker("nano")
|
76 |
-
else
|
77 |
reranker = get_reranker("large")
|
78 |
|
79 |
-
agent = make_graph_agent(
|
80 |
-
|
81 |
-
|
82 |
-
|
83 |
-
|
84 |
-
|
85 |
-
|
86 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
87 |
|
88 |
# def ask_vanna_query(query):
|
89 |
# return ask_vanna(vn, db_vanna_path, query)
|
90 |
|
91 |
-
def ask_drias_query(query: str, index_state: int, drias_model: str):
|
92 |
-
return ask_drias(db_vanna_path, query, index_state, drias_model)
|
93 |
|
94 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
95 |
print("chat cqa - message received")
|
96 |
-
async for event in chat_stream(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
97 |
yield event
|
98 |
-
|
99 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
100 |
print("chat poc - message received")
|
101 |
-
async for event in chat_stream(
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
102 |
yield event
|
103 |
|
104 |
|
@@ -106,14 +169,17 @@ async def chat_poc(query, history, audience, sources, reports, relevant_content_
|
|
106 |
# Gradio
|
107 |
# --------------------------------------------------------------------
|
108 |
|
|
|
109 |
# Function to update modal visibility
|
110 |
def update_config_modal_visibility(config_open):
|
111 |
print(config_open)
|
112 |
new_config_visibility_status = not config_open
|
113 |
return Modal(visible=new_config_visibility_status), new_config_visibility_status
|
114 |
-
|
115 |
|
116 |
-
|
|
|
|
|
|
|
117 |
sources_number = sources_textbox.count("<h2>")
|
118 |
figures_number = figures_cards.count("<h2>")
|
119 |
graphs_number = current_graphs.count("<iframe")
|
@@ -122,9 +188,18 @@ def update_sources_number_display(sources_textbox, figures_cards, current_graphs
|
|
122 |
figures_notif_label = f"Figures ({figures_number})"
|
123 |
graphs_notif_label = f"Graphs ({graphs_number})"
|
124 |
papers_notif_label = f"Papers ({papers_number})"
|
125 |
-
recommended_content_notif_label =
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
126 |
|
127 |
-
return gr.update(label=recommended_content_notif_label), gr.update(label=sources_notif_label), gr.update(label=figures_notif_label), gr.update(label=graphs_notif_label), gr.update(label=papers_notif_label)
|
128 |
|
129 |
# def create_drias_tab():
|
130 |
# with gr.Tab("Beta - Talk to DRIAS", elem_id="tab-vanna", id=6) as tab_vanna:
|
@@ -141,24 +216,112 @@ def update_sources_number_display(sources_textbox, figures_cards, current_graphs
|
|
141 |
# vanna_display = gr.Plot()
|
142 |
# vanna_direct_question.submit(ask_drias_query, [vanna_direct_question], [vanna_sql_query ,vanna_table, vanna_display])
|
143 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
144 |
def create_drias_tab():
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
145 |
with gr.Tab("Beta - Talk to DRIAS", elem_id="tab-vanna", id=6):
|
|
|
|
|
|
|
|
|
146 |
with gr.Row():
|
147 |
-
drias_direct_question = gr.Textbox(
|
148 |
-
|
149 |
-
|
150 |
-
|
151 |
-
|
|
|
|
|
|
|
|
|
|
|
152 |
|
153 |
-
with gr.Accordion(label=
|
154 |
-
|
|
|
|
|
155 |
|
156 |
-
with gr.Accordion(label="Chart"):
|
|
|
|
|
|
|
157 |
drias_display = gr.Plot(elem_id="vanna-plot")
|
158 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
159 |
with gr.Row():
|
160 |
-
prev_button = gr.Button("Previous")
|
161 |
-
next_button = gr.Button("Next")
|
162 |
|
163 |
sql_queries_state = gr.State([])
|
164 |
dataframes_state = gr.State([])
|
@@ -166,96 +329,104 @@ def create_drias_tab():
|
|
166 |
index_state = gr.State(0)
|
167 |
|
168 |
drias_direct_question.submit(
|
169 |
-
ask_drias_query,
|
170 |
-
inputs=[drias_direct_question, index_state
|
171 |
-
outputs=[
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
172 |
)
|
173 |
|
174 |
model_selection.change(
|
175 |
-
|
176 |
-
inputs=[
|
177 |
-
outputs=[
|
178 |
)
|
179 |
|
180 |
def show_previous(index, sql_queries, dataframes, plots):
|
181 |
if index > 0:
|
182 |
index -= 1
|
183 |
-
return
|
|
|
|
|
|
|
|
|
|
|
184 |
|
185 |
def show_next(index, sql_queries, dataframes, plots):
|
186 |
if index < len(sql_queries) - 1:
|
187 |
index += 1
|
188 |
-
return
|
|
|
|
|
|
|
|
|
|
|
189 |
|
190 |
prev_button.click(
|
191 |
-
show_previous,
|
192 |
inputs=[index_state, sql_queries_state, dataframes_state, plots_state],
|
193 |
-
outputs=[drias_sql_query, drias_table, drias_display, index_state]
|
|
|
|
|
|
|
|
|
194 |
)
|
195 |
|
196 |
next_button.click(
|
197 |
-
show_next,
|
198 |
inputs=[index_state, sql_queries_state, dataframes_state, plots_state],
|
199 |
-
outputs=[drias_sql_query, drias_table, drias_display, index_state]
|
|
|
|
|
|
|
|
|
200 |
)
|
201 |
-
|
202 |
-
|
203 |
-
def
|
204 |
-
|
205 |
-
|
206 |
-
with gr.Tab(tab_name):
|
207 |
-
with gr.Row(elem_id="chatbot-row"):
|
208 |
-
# Left column - Chat interface
|
209 |
-
with gr.Column(scale=2):
|
210 |
-
chatbot, textbox, config_button = create_chat_interface(tab_name)
|
211 |
-
|
212 |
-
# Right column - Content panels
|
213 |
-
with gr.Column(scale=2, variant="panel", elem_id="right-panel"):
|
214 |
-
with gr.Tabs(elem_id="right_panel_tab") as tabs:
|
215 |
-
# Examples tab
|
216 |
-
with gr.TabItem("Examples", elem_id="tab-examples", id=0):
|
217 |
-
examples_hidden = create_examples_tab(tab_name)
|
218 |
-
|
219 |
-
# Sources tab
|
220 |
-
with gr.Tab("Sources", elem_id="tab-sources", id=1) as tab_sources:
|
221 |
-
sources_textbox = gr.HTML(show_label=False, elem_id="sources-textbox")
|
222 |
-
|
223 |
-
|
224 |
-
# Recommended content tab
|
225 |
-
with gr.Tab("Recommended content", elem_id="tab-recommended_content", id=2) as tab_recommended_content:
|
226 |
-
with gr.Tabs(elem_id="group-subtabs") as tabs_recommended_content:
|
227 |
-
# Figures subtab
|
228 |
-
with gr.Tab("Figures", elem_id="tab-figures", id=3) as tab_figures:
|
229 |
-
sources_raw, new_figures, used_figures, gallery_component, figures_cards, figure_modal = create_figures_tab()
|
230 |
-
|
231 |
-
# Papers subtab
|
232 |
-
with gr.Tab("Papers", elem_id="tab-citations", id=4) as tab_papers:
|
233 |
-
papers_direct_search, papers_summary, papers_html, citations_network, papers_modal = create_papers_tab()
|
234 |
-
|
235 |
-
# Graphs subtab
|
236 |
-
with gr.Tab("Graphs", elem_id="tab-graphs", id=5) as tab_graphs:
|
237 |
-
graphs_container = gr.HTML(
|
238 |
-
"<h2>There are no graphs to be displayed at the moment. Try asking another question.</h2>",
|
239 |
-
elem_id="graphs-container"
|
240 |
-
)
|
241 |
-
|
242 |
-
|
243 |
-
def config_event_handling(main_tabs_components : list[MainTabPanel], config_componenets : ConfigPanel):
|
244 |
config_open = config_componenets.config_open
|
245 |
config_modal = config_componenets.config_modal
|
246 |
close_config_modal = config_componenets.close_config_modal_button
|
247 |
-
|
248 |
-
for button in [close_config_modal] + [
|
|
|
|
|
249 |
button.click(
|
250 |
fn=update_config_modal_visibility,
|
251 |
inputs=[config_open],
|
252 |
-
outputs=[config_modal, config_open]
|
253 |
-
)
|
254 |
-
|
|
|
255 |
def event_handling(
|
256 |
-
main_tab_components
|
257 |
-
config_components
|
258 |
-
tab_name="ClimateQ&A"
|
259 |
):
|
260 |
chatbot = main_tab_components.chatbot
|
261 |
textbox = main_tab_components.textbox
|
@@ -279,7 +450,7 @@ def event_handling(
|
|
279 |
graphs_container = main_tab_components.graph_container
|
280 |
follow_up_examples = main_tab_components.follow_up_examples
|
281 |
follow_up_examples_hidden = main_tab_components.follow_up_examples_hidden
|
282 |
-
|
283 |
dropdown_sources = config_components.dropdown_sources
|
284 |
dropdown_reports = config_components.dropdown_reports
|
285 |
dropdown_external_sources = config_components.dropdown_external_sources
|
@@ -288,91 +459,302 @@ def event_handling(
|
|
288 |
after = config_components.after
|
289 |
output_query = config_components.output_query
|
290 |
output_language = config_components.output_language
|
291 |
-
|
292 |
new_sources_hmtl = gr.State([])
|
293 |
ttd_data = gr.State([])
|
294 |
|
295 |
-
|
296 |
if tab_name == "ClimateQ&A":
|
297 |
print("chat cqa - message sent")
|
298 |
|
299 |
# Event for textbox
|
300 |
-
(
|
301 |
-
.submit(
|
302 |
-
|
303 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
304 |
)
|
305 |
# Event for examples_hidden
|
306 |
-
(
|
307 |
-
.change(
|
308 |
-
|
309 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
310 |
)
|
311 |
-
(
|
312 |
-
.change(
|
313 |
-
|
314 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
315 |
)
|
316 |
-
|
317 |
elif tab_name == "Beta - POC Adapt'Action":
|
318 |
print("chat poc - message sent")
|
319 |
# Event for textbox
|
320 |
-
(
|
321 |
-
.submit(
|
322 |
-
|
323 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
324 |
)
|
325 |
# Event for examples_hidden
|
326 |
-
(
|
327 |
-
.change(
|
328 |
-
|
329 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
330 |
)
|
331 |
-
(
|
332 |
-
.change(
|
333 |
-
|
334 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
335 |
)
|
336 |
-
|
337 |
-
|
338 |
-
|
339 |
-
|
340 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
341 |
|
342 |
# Update sources numbers
|
343 |
for component in [sources_textbox, figures_cards, current_graphs, papers_html]:
|
344 |
-
component.change(
|
345 |
-
|
|
|
|
|
|
|
|
|
346 |
# Search for papers
|
347 |
for component in [textbox, examples_hidden, papers_direct_search]:
|
348 |
-
component.submit(
|
349 |
-
|
|
|
|
|
|
|
350 |
|
351 |
# if tab_name == "Beta - POC Adapt'Action": # Not untill results are good enough
|
352 |
# # Drias search
|
353 |
# textbox.submit(ask_vanna, [textbox], [vanna_sql_query ,vanna_table, vanna_display])
|
354 |
|
|
|
355 |
def main_ui():
|
356 |
# config_open = gr.State(True)
|
357 |
-
with gr.Blocks(
|
358 |
-
|
359 |
-
|
|
|
|
|
|
|
|
|
|
|
360 |
with gr.Tabs():
|
361 |
-
cqa_components = cqa_tab(tab_name
|
362 |
-
local_cqa_components = cqa_tab(tab_name
|
363 |
create_drias_tab()
|
364 |
-
|
365 |
create_about_tab()
|
366 |
-
|
367 |
-
event_handling(cqa_components, config_components, tab_name
|
368 |
-
event_handling(
|
369 |
-
|
370 |
-
|
371 |
-
|
|
|
|
|
372 |
demo.queue()
|
373 |
-
|
374 |
return demo
|
375 |
|
376 |
-
|
377 |
demo = main_ui()
|
378 |
demo.launch(ssr_mode=False)
|
|
|
9 |
from climateqa.engine.llm import get_llm
|
10 |
from climateqa.engine.vectorstore import get_pinecone_vectorstore
|
11 |
from climateqa.engine.reranker import get_reranker
|
12 |
+
from climateqa.engine.graph import make_graph_agent, make_graph_agent_poc
|
13 |
from climateqa.engine.chains.retrieve_papers import find_papers
|
14 |
from climateqa.chat import start_chat, chat_stream, finish_chat
|
15 |
from climateqa.engine.talk_to_data.main import ask_drias, DRIAS_MODELS
|
16 |
from climateqa.engine.talk_to_data.myVanna import MyVanna
|
17 |
|
18 |
+
from front.tabs import create_config_modal, cqa_tab, create_about_tab
|
19 |
+
from front.tabs import MainTabPanel, ConfigPanel
|
20 |
from front.utils import process_figures
|
21 |
from gradio_modal import Modal
|
22 |
|
|
|
25 |
import logging
|
26 |
|
27 |
logging.basicConfig(level=logging.WARNING)
|
28 |
+
os.environ["TF_CPP_MIN_LOG_LEVEL"] = "2" # Suppresses INFO and WARNING logs
|
29 |
logging.getLogger().setLevel(logging.WARNING)
|
30 |
|
31 |
|
|
|
32 |
# Load environment variables in local mode
|
33 |
try:
|
34 |
from dotenv import load_dotenv
|
35 |
+
|
36 |
load_dotenv()
|
37 |
except Exception as e:
|
38 |
pass
|
|
|
63 |
user_id = create_user_id()
|
64 |
|
65 |
|
|
|
66 |
# Create vectorstore and retriever
|
67 |
embeddings_function = get_embeddings_function()
|
68 |
+
vectorstore = get_pinecone_vectorstore(
|
69 |
+
embeddings_function, index_name=os.getenv("PINECONE_API_INDEX")
|
70 |
+
)
|
71 |
+
vectorstore_graphs = get_pinecone_vectorstore(
|
72 |
+
embeddings_function,
|
73 |
+
index_name=os.getenv("PINECONE_API_INDEX_OWID"),
|
74 |
+
text_key="description",
|
75 |
+
)
|
76 |
+
vectorstore_region = get_pinecone_vectorstore(
|
77 |
+
embeddings_function, index_name=os.getenv("PINECONE_API_INDEX_LOCAL_V2")
|
78 |
+
)
|
79 |
|
80 |
+
llm = get_llm(provider="openai", max_tokens=1024, temperature=0.0)
|
81 |
if os.environ["GRADIO_ENV"] == "local":
|
82 |
reranker = get_reranker("nano")
|
83 |
+
else:
|
84 |
reranker = get_reranker("large")
|
85 |
|
86 |
+
agent = make_graph_agent(
|
87 |
+
llm=llm,
|
88 |
+
vectorstore_ipcc=vectorstore,
|
89 |
+
vectorstore_graphs=vectorstore_graphs,
|
90 |
+
vectorstore_region=vectorstore_region,
|
91 |
+
reranker=reranker,
|
92 |
+
threshold_docs=0.2,
|
93 |
+
)
|
94 |
+
agent_poc = make_graph_agent_poc(
|
95 |
+
llm=llm,
|
96 |
+
vectorstore_ipcc=vectorstore,
|
97 |
+
vectorstore_graphs=vectorstore_graphs,
|
98 |
+
vectorstore_region=vectorstore_region,
|
99 |
+
reranker=reranker,
|
100 |
+
threshold_docs=0,
|
101 |
+
version="v4",
|
102 |
+
) # TODO put back default 0.2
|
103 |
+
|
104 |
+
# Vanna object
|
105 |
+
|
106 |
+
# vn = MyVanna(config = {"temperature": 0, "api_key": os.getenv('THEO_API_KEY'), 'model': os.getenv('VANNA_MODEL'), 'pc_api_key': os.getenv('VANNA_PINECONE_API_KEY'), 'index_name': os.getenv('VANNA_INDEX_NAME'), "top_k" : 4})
|
107 |
+
# db_vanna_path = os.path.join(os.getcwd(), "data/drias/drias.db")
|
108 |
+
# vn.connect_to_sqlite(db_vanna_path)
|
109 |
|
110 |
# def ask_vanna_query(query):
|
111 |
# return ask_vanna(vn, db_vanna_path, query)
|
112 |
|
|
|
|
|
113 |
|
114 |
+
def ask_drias_query(query: str, index_state: int):
|
115 |
+
return ask_drias(query, index_state)
|
116 |
+
|
117 |
+
|
118 |
+
async def chat(
|
119 |
+
query,
|
120 |
+
history,
|
121 |
+
audience,
|
122 |
+
sources,
|
123 |
+
reports,
|
124 |
+
relevant_content_sources_selection,
|
125 |
+
search_only,
|
126 |
+
):
|
127 |
print("chat cqa - message received")
|
128 |
+
async for event in chat_stream(
|
129 |
+
agent,
|
130 |
+
query,
|
131 |
+
history,
|
132 |
+
audience,
|
133 |
+
sources,
|
134 |
+
reports,
|
135 |
+
relevant_content_sources_selection,
|
136 |
+
search_only,
|
137 |
+
share_client,
|
138 |
+
user_id,
|
139 |
+
):
|
140 |
yield event
|
141 |
+
|
142 |
+
|
143 |
+
async def chat_poc(
|
144 |
+
query,
|
145 |
+
history,
|
146 |
+
audience,
|
147 |
+
sources,
|
148 |
+
reports,
|
149 |
+
relevant_content_sources_selection,
|
150 |
+
search_only,
|
151 |
+
):
|
152 |
print("chat poc - message received")
|
153 |
+
async for event in chat_stream(
|
154 |
+
agent_poc,
|
155 |
+
query,
|
156 |
+
history,
|
157 |
+
audience,
|
158 |
+
sources,
|
159 |
+
reports,
|
160 |
+
relevant_content_sources_selection,
|
161 |
+
search_only,
|
162 |
+
share_client,
|
163 |
+
user_id,
|
164 |
+
):
|
165 |
yield event
|
166 |
|
167 |
|
|
|
169 |
# Gradio
|
170 |
# --------------------------------------------------------------------
|
171 |
|
172 |
+
|
173 |
# Function to update modal visibility
|
174 |
def update_config_modal_visibility(config_open):
|
175 |
print(config_open)
|
176 |
new_config_visibility_status = not config_open
|
177 |
return Modal(visible=new_config_visibility_status), new_config_visibility_status
|
|
|
178 |
|
179 |
+
|
180 |
+
def update_sources_number_display(
|
181 |
+
sources_textbox, figures_cards, current_graphs, papers_html
|
182 |
+
):
|
183 |
sources_number = sources_textbox.count("<h2>")
|
184 |
figures_number = figures_cards.count("<h2>")
|
185 |
graphs_number = current_graphs.count("<iframe")
|
|
|
188 |
figures_notif_label = f"Figures ({figures_number})"
|
189 |
graphs_notif_label = f"Graphs ({graphs_number})"
|
190 |
papers_notif_label = f"Papers ({papers_number})"
|
191 |
+
recommended_content_notif_label = (
|
192 |
+
f"Recommended content ({figures_number + graphs_number + papers_number})"
|
193 |
+
)
|
194 |
+
|
195 |
+
return (
|
196 |
+
gr.update(label=recommended_content_notif_label),
|
197 |
+
gr.update(label=sources_notif_label),
|
198 |
+
gr.update(label=figures_notif_label),
|
199 |
+
gr.update(label=graphs_notif_label),
|
200 |
+
gr.update(label=papers_notif_label),
|
201 |
+
)
|
202 |
|
|
|
203 |
|
204 |
# def create_drias_tab():
|
205 |
# with gr.Tab("Beta - Talk to DRIAS", elem_id="tab-vanna", id=6) as tab_vanna:
|
|
|
216 |
# vanna_display = gr.Plot()
|
217 |
# vanna_direct_question.submit(ask_drias_query, [vanna_direct_question], [vanna_sql_query ,vanna_table, vanna_display])
|
218 |
|
219 |
+
|
220 |
+
def show_results(sql_queries_state, dataframes_state, plots_state):
|
221 |
+
if not sql_queries_state or not dataframes_state or not plots_state:
|
222 |
+
# If all results are empty, show "No result"
|
223 |
+
return (
|
224 |
+
gr.update(visible=True),
|
225 |
+
gr.update(visible=False),
|
226 |
+
gr.update(visible=False),
|
227 |
+
gr.update(visible=False),
|
228 |
+
gr.update(visible=False),
|
229 |
+
gr.update(visible=False),
|
230 |
+
gr.update(visible=False),
|
231 |
+
)
|
232 |
+
else:
|
233 |
+
# Show the appropriate components with their data
|
234 |
+
return (
|
235 |
+
gr.update(visible=False),
|
236 |
+
gr.update(visible=True),
|
237 |
+
gr.update(visible=True),
|
238 |
+
gr.update(visible=True),
|
239 |
+
gr.update(visible=True),
|
240 |
+
gr.update(visible=True),
|
241 |
+
gr.update(visible=True),
|
242 |
+
)
|
243 |
+
|
244 |
+
|
245 |
+
def filter_by_model(dataframes, figures, index_state, model_selection):
|
246 |
+
df = dataframes[index_state]
|
247 |
+
if model_selection != "ALL":
|
248 |
+
df = df[df["model"] == model_selection]
|
249 |
+
figure = figures[index_state](df)
|
250 |
+
return df, figure
|
251 |
+
|
252 |
+
|
253 |
+
def update_pagination(index, sql_queries):
|
254 |
+
pagination = f"{index + 1}/{len(sql_queries)}" if sql_queries else ""
|
255 |
+
return pagination
|
256 |
+
|
257 |
+
|
258 |
def create_drias_tab():
|
259 |
+
details_text = """
|
260 |
+
Hi, I'm **Talk to Drias**, designed to answer your questions using [**DRIAS - TRACC 2023**](https://www.drias-climat.fr/accompagnement/sections/401) data.
|
261 |
+
I'll answer by displaying a list of SQL queries, graphs and data most relevant to your question.
|
262 |
+
|
263 |
+
❓ **How to use?**
|
264 |
+
You can ask me anything about these climate indicators: **temperature**, **precipitation** or **drought**.
|
265 |
+
You can specify **location** and/or **year**.
|
266 |
+
You can choose from a list of climate models. By default, we take the **average of each model**.
|
267 |
+
|
268 |
+
For example, you can ask:
|
269 |
+
- What will the temperature be like in Paris?
|
270 |
+
- What will be the total rainfall in France in 2030?
|
271 |
+
- How frequent will extreme events be in Lyon?
|
272 |
+
|
273 |
+
**Example of indicators in the data**:
|
274 |
+
- Mean temperature (annual, winter, summer)
|
275 |
+
- Total precipitation (annual, winter, summer)
|
276 |
+
- Number of days with remarkable precipitations, with dry ground, with temperature above 30°C
|
277 |
+
|
278 |
+
⚠️ **Limitations**:
|
279 |
+
- You can't ask anything that isn't related to **DRIAS - TRACC 2023** data.
|
280 |
+
- You can only ask about **locations in France**.
|
281 |
+
- If you specify a year, there may be **no data for that year for some models**.
|
282 |
+
- You **cannot compare two models**.
|
283 |
+
|
284 |
+
🛈 **Information**
|
285 |
+
Please note that we **log your questions for meta-analysis purposes**, so avoid sharing any sensitive or personal information.
|
286 |
+
"""
|
287 |
with gr.Tab("Beta - Talk to DRIAS", elem_id="tab-vanna", id=6):
|
288 |
+
|
289 |
+
with gr.Accordion(label="Details"):
|
290 |
+
gr.Markdown(details_text)
|
291 |
+
|
292 |
with gr.Row():
|
293 |
+
drias_direct_question = gr.Textbox(
|
294 |
+
label="Direct Question",
|
295 |
+
placeholder="You can write direct question here",
|
296 |
+
elem_id="direct-question",
|
297 |
+
interactive=True,
|
298 |
+
)
|
299 |
+
|
300 |
+
result_text = gr.Textbox(
|
301 |
+
label="", elem_id="no-result-label", interactive=False, visible=True
|
302 |
+
)
|
303 |
|
304 |
+
with gr.Accordion(label="SQL Query Used", visible=False) as query_accordion:
|
305 |
+
drias_sql_query = gr.Textbox(
|
306 |
+
label="", elem_id="sql-query", interactive=False
|
307 |
+
)
|
308 |
|
309 |
+
with gr.Accordion(label="Chart", visible=False) as chart_accordion:
|
310 |
+
model_selection = gr.Dropdown(
|
311 |
+
label="Model", choices=DRIAS_MODELS, value="ALL", interactive=True
|
312 |
+
)
|
313 |
drias_display = gr.Plot(elem_id="vanna-plot")
|
314 |
+
|
315 |
+
with gr.Accordion(
|
316 |
+
label="Data used", open=False, visible=False
|
317 |
+
) as table_accordion:
|
318 |
+
drias_table = gr.DataFrame([], elem_id="vanna-table")
|
319 |
+
|
320 |
+
pagination_display = gr.Markdown(value="", visible=False, elem_id="pagination-display")
|
321 |
+
|
322 |
with gr.Row():
|
323 |
+
prev_button = gr.Button("Previous", visible=False)
|
324 |
+
next_button = gr.Button("Next", visible=False)
|
325 |
|
326 |
sql_queries_state = gr.State([])
|
327 |
dataframes_state = gr.State([])
|
|
|
329 |
index_state = gr.State(0)
|
330 |
|
331 |
drias_direct_question.submit(
|
332 |
+
ask_drias_query,
|
333 |
+
inputs=[drias_direct_question, index_state],
|
334 |
+
outputs=[
|
335 |
+
drias_sql_query,
|
336 |
+
drias_table,
|
337 |
+
drias_display,
|
338 |
+
sql_queries_state,
|
339 |
+
dataframes_state,
|
340 |
+
plots_state,
|
341 |
+
index_state,
|
342 |
+
result_text,
|
343 |
+
],
|
344 |
+
).then(
|
345 |
+
show_results,
|
346 |
+
inputs=[sql_queries_state, dataframes_state, plots_state],
|
347 |
+
outputs=[
|
348 |
+
result_text,
|
349 |
+
query_accordion,
|
350 |
+
table_accordion,
|
351 |
+
chart_accordion,
|
352 |
+
prev_button,
|
353 |
+
next_button,
|
354 |
+
pagination_display
|
355 |
+
],
|
356 |
+
).then(
|
357 |
+
update_pagination,
|
358 |
+
inputs=[index_state, sql_queries_state],
|
359 |
+
outputs=[pagination_display],
|
360 |
)
|
361 |
|
362 |
model_selection.change(
|
363 |
+
filter_by_model,
|
364 |
+
inputs=[dataframes_state, plots_state, index_state, model_selection],
|
365 |
+
outputs=[drias_table, drias_display],
|
366 |
)
|
367 |
|
368 |
def show_previous(index, sql_queries, dataframes, plots):
|
369 |
if index > 0:
|
370 |
index -= 1
|
371 |
+
return (
|
372 |
+
sql_queries[index],
|
373 |
+
dataframes[index],
|
374 |
+
plots[index](dataframes[index]),
|
375 |
+
index,
|
376 |
+
)
|
377 |
|
378 |
def show_next(index, sql_queries, dataframes, plots):
|
379 |
if index < len(sql_queries) - 1:
|
380 |
index += 1
|
381 |
+
return (
|
382 |
+
sql_queries[index],
|
383 |
+
dataframes[index],
|
384 |
+
plots[index](dataframes[index]),
|
385 |
+
index,
|
386 |
+
)
|
387 |
|
388 |
prev_button.click(
|
389 |
+
show_previous,
|
390 |
inputs=[index_state, sql_queries_state, dataframes_state, plots_state],
|
391 |
+
outputs=[drias_sql_query, drias_table, drias_display, index_state],
|
392 |
+
).then(
|
393 |
+
update_pagination,
|
394 |
+
inputs=[index_state, sql_queries_state],
|
395 |
+
outputs=[pagination_display],
|
396 |
)
|
397 |
|
398 |
next_button.click(
|
399 |
+
show_next,
|
400 |
inputs=[index_state, sql_queries_state, dataframes_state, plots_state],
|
401 |
+
outputs=[drias_sql_query, drias_table, drias_display, index_state],
|
402 |
+
).then(
|
403 |
+
update_pagination,
|
404 |
+
inputs=[index_state, sql_queries_state],
|
405 |
+
outputs=[pagination_display],
|
406 |
)
|
407 |
+
|
408 |
+
|
409 |
+
def config_event_handling(
|
410 |
+
main_tabs_components: list[MainTabPanel], config_componenets: ConfigPanel
|
411 |
+
):
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
412 |
config_open = config_componenets.config_open
|
413 |
config_modal = config_componenets.config_modal
|
414 |
close_config_modal = config_componenets.close_config_modal_button
|
415 |
+
|
416 |
+
for button in [close_config_modal] + [
|
417 |
+
main_tab_component.config_button for main_tab_component in main_tabs_components
|
418 |
+
]:
|
419 |
button.click(
|
420 |
fn=update_config_modal_visibility,
|
421 |
inputs=[config_open],
|
422 |
+
outputs=[config_modal, config_open],
|
423 |
+
)
|
424 |
+
|
425 |
+
|
426 |
def event_handling(
|
427 |
+
main_tab_components: MainTabPanel,
|
428 |
+
config_components: ConfigPanel,
|
429 |
+
tab_name="ClimateQ&A",
|
430 |
):
|
431 |
chatbot = main_tab_components.chatbot
|
432 |
textbox = main_tab_components.textbox
|
|
|
450 |
graphs_container = main_tab_components.graph_container
|
451 |
follow_up_examples = main_tab_components.follow_up_examples
|
452 |
follow_up_examples_hidden = main_tab_components.follow_up_examples_hidden
|
453 |
+
|
454 |
dropdown_sources = config_components.dropdown_sources
|
455 |
dropdown_reports = config_components.dropdown_reports
|
456 |
dropdown_external_sources = config_components.dropdown_external_sources
|
|
|
459 |
after = config_components.after
|
460 |
output_query = config_components.output_query
|
461 |
output_language = config_components.output_language
|
462 |
+
|
463 |
new_sources_hmtl = gr.State([])
|
464 |
ttd_data = gr.State([])
|
465 |
|
|
|
466 |
if tab_name == "ClimateQ&A":
|
467 |
print("chat cqa - message sent")
|
468 |
|
469 |
# Event for textbox
|
470 |
+
(
|
471 |
+
textbox.submit(
|
472 |
+
start_chat,
|
473 |
+
[textbox, chatbot, search_only],
|
474 |
+
[textbox, tabs, chatbot, sources_raw],
|
475 |
+
queue=False,
|
476 |
+
api_name=f"start_chat_{textbox.elem_id}",
|
477 |
+
)
|
478 |
+
.then(
|
479 |
+
chat,
|
480 |
+
[
|
481 |
+
textbox,
|
482 |
+
chatbot,
|
483 |
+
dropdown_audience,
|
484 |
+
dropdown_sources,
|
485 |
+
dropdown_reports,
|
486 |
+
dropdown_external_sources,
|
487 |
+
search_only,
|
488 |
+
],
|
489 |
+
[
|
490 |
+
chatbot,
|
491 |
+
new_sources_hmtl,
|
492 |
+
output_query,
|
493 |
+
output_language,
|
494 |
+
new_figures,
|
495 |
+
current_graphs,
|
496 |
+
follow_up_examples.dataset,
|
497 |
+
],
|
498 |
+
concurrency_limit=8,
|
499 |
+
api_name=f"chat_{textbox.elem_id}",
|
500 |
+
)
|
501 |
+
.then(
|
502 |
+
finish_chat, None, [textbox], api_name=f"finish_chat_{textbox.elem_id}"
|
503 |
+
)
|
504 |
)
|
505 |
# Event for examples_hidden
|
506 |
+
(
|
507 |
+
examples_hidden.change(
|
508 |
+
start_chat,
|
509 |
+
[examples_hidden, chatbot, search_only],
|
510 |
+
[examples_hidden, tabs, chatbot, sources_raw],
|
511 |
+
queue=False,
|
512 |
+
api_name=f"start_chat_{examples_hidden.elem_id}",
|
513 |
+
)
|
514 |
+
.then(
|
515 |
+
chat,
|
516 |
+
[
|
517 |
+
examples_hidden,
|
518 |
+
chatbot,
|
519 |
+
dropdown_audience,
|
520 |
+
dropdown_sources,
|
521 |
+
dropdown_reports,
|
522 |
+
dropdown_external_sources,
|
523 |
+
search_only,
|
524 |
+
],
|
525 |
+
[
|
526 |
+
chatbot,
|
527 |
+
new_sources_hmtl,
|
528 |
+
output_query,
|
529 |
+
output_language,
|
530 |
+
new_figures,
|
531 |
+
current_graphs,
|
532 |
+
follow_up_examples.dataset,
|
533 |
+
],
|
534 |
+
concurrency_limit=8,
|
535 |
+
api_name=f"chat_{examples_hidden.elem_id}",
|
536 |
+
)
|
537 |
+
.then(
|
538 |
+
finish_chat,
|
539 |
+
None,
|
540 |
+
[textbox],
|
541 |
+
api_name=f"finish_chat_{examples_hidden.elem_id}",
|
542 |
+
)
|
543 |
)
|
544 |
+
(
|
545 |
+
follow_up_examples_hidden.change(
|
546 |
+
start_chat,
|
547 |
+
[follow_up_examples_hidden, chatbot, search_only],
|
548 |
+
[follow_up_examples_hidden, tabs, chatbot, sources_raw],
|
549 |
+
queue=False,
|
550 |
+
api_name=f"start_chat_{examples_hidden.elem_id}",
|
551 |
+
)
|
552 |
+
.then(
|
553 |
+
chat,
|
554 |
+
[
|
555 |
+
follow_up_examples_hidden,
|
556 |
+
chatbot,
|
557 |
+
dropdown_audience,
|
558 |
+
dropdown_sources,
|
559 |
+
dropdown_reports,
|
560 |
+
dropdown_external_sources,
|
561 |
+
search_only,
|
562 |
+
],
|
563 |
+
[
|
564 |
+
chatbot,
|
565 |
+
new_sources_hmtl,
|
566 |
+
output_query,
|
567 |
+
output_language,
|
568 |
+
new_figures,
|
569 |
+
current_graphs,
|
570 |
+
follow_up_examples.dataset,
|
571 |
+
],
|
572 |
+
concurrency_limit=8,
|
573 |
+
api_name=f"chat_{examples_hidden.elem_id}",
|
574 |
+
)
|
575 |
+
.then(
|
576 |
+
finish_chat,
|
577 |
+
None,
|
578 |
+
[textbox],
|
579 |
+
api_name=f"finish_chat_{follow_up_examples_hidden.elem_id}",
|
580 |
+
)
|
581 |
)
|
582 |
+
|
583 |
elif tab_name == "Beta - POC Adapt'Action":
|
584 |
print("chat poc - message sent")
|
585 |
# Event for textbox
|
586 |
+
(
|
587 |
+
textbox.submit(
|
588 |
+
start_chat,
|
589 |
+
[textbox, chatbot, search_only],
|
590 |
+
[textbox, tabs, chatbot, sources_raw],
|
591 |
+
queue=False,
|
592 |
+
api_name=f"start_chat_{textbox.elem_id}",
|
593 |
+
)
|
594 |
+
.then(
|
595 |
+
chat_poc,
|
596 |
+
[
|
597 |
+
textbox,
|
598 |
+
chatbot,
|
599 |
+
dropdown_audience,
|
600 |
+
dropdown_sources,
|
601 |
+
dropdown_reports,
|
602 |
+
dropdown_external_sources,
|
603 |
+
search_only,
|
604 |
+
],
|
605 |
+
[
|
606 |
+
chatbot,
|
607 |
+
new_sources_hmtl,
|
608 |
+
output_query,
|
609 |
+
output_language,
|
610 |
+
new_figures,
|
611 |
+
current_graphs,
|
612 |
+
],
|
613 |
+
concurrency_limit=8,
|
614 |
+
api_name=f"chat_{textbox.elem_id}",
|
615 |
+
)
|
616 |
+
.then(
|
617 |
+
finish_chat, None, [textbox], api_name=f"finish_chat_{textbox.elem_id}"
|
618 |
+
)
|
619 |
)
|
620 |
# Event for examples_hidden
|
621 |
+
(
|
622 |
+
examples_hidden.change(
|
623 |
+
start_chat,
|
624 |
+
[examples_hidden, chatbot, search_only],
|
625 |
+
[examples_hidden, tabs, chatbot, sources_raw],
|
626 |
+
queue=False,
|
627 |
+
api_name=f"start_chat_{examples_hidden.elem_id}",
|
628 |
+
)
|
629 |
+
.then(
|
630 |
+
chat_poc,
|
631 |
+
[
|
632 |
+
examples_hidden,
|
633 |
+
chatbot,
|
634 |
+
dropdown_audience,
|
635 |
+
dropdown_sources,
|
636 |
+
dropdown_reports,
|
637 |
+
dropdown_external_sources,
|
638 |
+
search_only,
|
639 |
+
],
|
640 |
+
[
|
641 |
+
chatbot,
|
642 |
+
new_sources_hmtl,
|
643 |
+
output_query,
|
644 |
+
output_language,
|
645 |
+
new_figures,
|
646 |
+
current_graphs,
|
647 |
+
],
|
648 |
+
concurrency_limit=8,
|
649 |
+
api_name=f"chat_{examples_hidden.elem_id}",
|
650 |
+
)
|
651 |
+
.then(
|
652 |
+
finish_chat,
|
653 |
+
None,
|
654 |
+
[textbox],
|
655 |
+
api_name=f"finish_chat_{examples_hidden.elem_id}",
|
656 |
+
)
|
657 |
)
|
658 |
+
(
|
659 |
+
follow_up_examples_hidden.change(
|
660 |
+
start_chat,
|
661 |
+
[follow_up_examples_hidden, chatbot, search_only],
|
662 |
+
[follow_up_examples_hidden, tabs, chatbot, sources_raw],
|
663 |
+
queue=False,
|
664 |
+
api_name=f"start_chat_{examples_hidden.elem_id}",
|
665 |
+
)
|
666 |
+
.then(
|
667 |
+
chat,
|
668 |
+
[
|
669 |
+
follow_up_examples_hidden,
|
670 |
+
chatbot,
|
671 |
+
dropdown_audience,
|
672 |
+
dropdown_sources,
|
673 |
+
dropdown_reports,
|
674 |
+
dropdown_external_sources,
|
675 |
+
search_only,
|
676 |
+
],
|
677 |
+
[
|
678 |
+
chatbot,
|
679 |
+
new_sources_hmtl,
|
680 |
+
output_query,
|
681 |
+
output_language,
|
682 |
+
new_figures,
|
683 |
+
current_graphs,
|
684 |
+
follow_up_examples.dataset,
|
685 |
+
],
|
686 |
+
concurrency_limit=8,
|
687 |
+
api_name=f"chat_{examples_hidden.elem_id}",
|
688 |
+
)
|
689 |
+
.then(
|
690 |
+
finish_chat,
|
691 |
+
None,
|
692 |
+
[textbox],
|
693 |
+
api_name=f"finish_chat_{follow_up_examples_hidden.elem_id}",
|
694 |
+
)
|
695 |
)
|
696 |
+
|
697 |
+
new_sources_hmtl.change(
|
698 |
+
lambda x: x, inputs=[new_sources_hmtl], outputs=[sources_textbox]
|
699 |
+
)
|
700 |
+
current_graphs.change(
|
701 |
+
lambda x: x, inputs=[current_graphs], outputs=[graphs_container]
|
702 |
+
)
|
703 |
+
new_figures.change(
|
704 |
+
process_figures,
|
705 |
+
inputs=[sources_raw, new_figures],
|
706 |
+
outputs=[sources_raw, figures_cards, gallery_component],
|
707 |
+
)
|
708 |
|
709 |
# Update sources numbers
|
710 |
for component in [sources_textbox, figures_cards, current_graphs, papers_html]:
|
711 |
+
component.change(
|
712 |
+
update_sources_number_display,
|
713 |
+
[sources_textbox, figures_cards, current_graphs, papers_html],
|
714 |
+
[tab_recommended_content, tab_sources, tab_figures, tab_graphs, tab_papers],
|
715 |
+
)
|
716 |
+
|
717 |
# Search for papers
|
718 |
for component in [textbox, examples_hidden, papers_direct_search]:
|
719 |
+
component.submit(
|
720 |
+
find_papers,
|
721 |
+
[component, after, dropdown_external_sources],
|
722 |
+
[papers_html, citations_network, papers_summary],
|
723 |
+
)
|
724 |
|
725 |
# if tab_name == "Beta - POC Adapt'Action": # Not untill results are good enough
|
726 |
# # Drias search
|
727 |
# textbox.submit(ask_vanna, [textbox], [vanna_sql_query ,vanna_table, vanna_display])
|
728 |
|
729 |
+
|
730 |
def main_ui():
|
731 |
# config_open = gr.State(True)
|
732 |
+
with gr.Blocks(
|
733 |
+
title="Climate Q&A",
|
734 |
+
css_paths=os.getcwd() + "/style.css",
|
735 |
+
theme=theme,
|
736 |
+
elem_id="main-component",
|
737 |
+
) as demo:
|
738 |
+
config_components = create_config_modal()
|
739 |
+
|
740 |
with gr.Tabs():
|
741 |
+
cqa_components = cqa_tab(tab_name="ClimateQ&A")
|
742 |
+
local_cqa_components = cqa_tab(tab_name="Beta - POC Adapt'Action")
|
743 |
create_drias_tab()
|
744 |
+
|
745 |
create_about_tab()
|
746 |
+
|
747 |
+
event_handling(cqa_components, config_components, tab_name="ClimateQ&A")
|
748 |
+
event_handling(
|
749 |
+
local_cqa_components, config_components, tab_name="Beta - POC Adapt'Action"
|
750 |
+
)
|
751 |
+
|
752 |
+
config_event_handling([cqa_components, local_cqa_components], config_components)
|
753 |
+
|
754 |
demo.queue()
|
755 |
+
|
756 |
return demo
|
757 |
|
758 |
+
|
759 |
demo = main_ui()
|
760 |
demo.launch(ssr_mode=False)
|
climateqa/engine/talk_to_data/config.py
ADDED
@@ -0,0 +1,33 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
DRIAS_TABLES = [
|
2 |
+
"total_winter_precipitation",
|
3 |
+
"total_summer_precipiation",
|
4 |
+
"total_annual_precipitation",
|
5 |
+
"total_remarkable_daily_precipitation",
|
6 |
+
"frequency_of_remarkable_daily_precipitation",
|
7 |
+
"extreme_precipitation_intensity",
|
8 |
+
"mean_winter_temperature",
|
9 |
+
"mean_summer_temperature",
|
10 |
+
"mean_annual_temperature",
|
11 |
+
"number_of_tropical_nights",
|
12 |
+
"maximum_summer_temperature",
|
13 |
+
"number_of_days_with_tx_above_30",
|
14 |
+
"number_of_days_with_tx_above_35",
|
15 |
+
"number_of_days_with_a_dry_ground",
|
16 |
+
]
|
17 |
+
|
18 |
+
INDICATOR_COLUMNS_PER_TABLE = {
|
19 |
+
"total_winter_precipitation": "total_winter_precipitation",
|
20 |
+
"total_summer_precipiation": "total_summer_precipitation",
|
21 |
+
"total_annual_precipitation": "total_annual_precipitation",
|
22 |
+
"total_remarkable_daily_precipitation": "total_remarkable_daily_precipitation",
|
23 |
+
"frequency_of_remarkable_daily_precipitation": "frequency_of_remarkable_daily_precipitation",
|
24 |
+
"extreme_precipitation_intensity": "extreme_precipitation_intensity",
|
25 |
+
"mean_winter_temperature": "mean_winter_temperature",
|
26 |
+
"mean_summer_temperature": "mean_summer_temperature",
|
27 |
+
"mean_annual_temperature": "mean_annual_temperature",
|
28 |
+
"number_of_tropical_nights": "number_tropical_nights",
|
29 |
+
"maximum_summer_temperature": "maximum_summer_temperature",
|
30 |
+
"number_of_days_with_tx_above_30": "number_of_days_with_tx_above_30",
|
31 |
+
"number_of_days_with_tx_above_35": "number_of_days_with_tx_above_35",
|
32 |
+
"number_of_days_with_a_dry_ground": "number_of_days_with_dry_ground"
|
33 |
+
}
|
climateqa/engine/talk_to_data/main.py
CHANGED
@@ -13,8 +13,8 @@ def ask_llm_column_names(sql_query, llm):
|
|
13 |
columns_list = ast.literal_eval(columns.strip("```python\n").strip())
|
14 |
return columns_list
|
15 |
|
16 |
-
def ask_drias(
|
17 |
-
final_state = drias_workflow(
|
18 |
sql_queries = []
|
19 |
result_dataframes = []
|
20 |
figures = []
|
@@ -28,10 +28,15 @@ def ask_drias(db_drias_path:str, query:str, index_state: int = 0, drias_model: s
|
|
28 |
if 'dataframe' in table_state and table_state['dataframe'] is not None:
|
29 |
result_dataframes.append(table_state['dataframe'])
|
30 |
if 'figure' in table_state and table_state['figure'] is not None:
|
31 |
-
figures.append(table_state['figure']
|
32 |
|
33 |
-
|
34 |
-
|
|
|
|
|
|
|
|
|
|
|
35 |
|
36 |
DRIAS_MODELS = [
|
37 |
'ALL',
|
|
|
13 |
columns_list = ast.literal_eval(columns.strip("```python\n").strip())
|
14 |
return columns_list
|
15 |
|
16 |
+
def ask_drias(query:str, index_state: int = 0):
|
17 |
+
final_state = drias_workflow(query)
|
18 |
sql_queries = []
|
19 |
result_dataframes = []
|
20 |
figures = []
|
|
|
28 |
if 'dataframe' in table_state and table_state['dataframe'] is not None:
|
29 |
result_dataframes.append(table_state['dataframe'])
|
30 |
if 'figure' in table_state and table_state['figure'] is not None:
|
31 |
+
figures.append(table_state['figure'])
|
32 |
|
33 |
+
if "error" in final_state and final_state["error"] != "":
|
34 |
+
return None, None, None, [], [], [], 0, final_state["error"]
|
35 |
+
|
36 |
+
sql_query = sql_queries[index_state]
|
37 |
+
dataframe = result_dataframes[index_state]
|
38 |
+
figure = figures[index_state](dataframe)
|
39 |
+
return sql_query, dataframe, figure, sql_queries, result_dataframes, figures, index_state, ""
|
40 |
|
41 |
DRIAS_MODELS = [
|
42 |
'ALL',
|
climateqa/engine/talk_to_data/plot.py
CHANGED
@@ -29,7 +29,6 @@ def plot_indicator_evolution_at_location(params: dict) -> Callable[..., Figure]:
|
|
29 |
Callable[..., Figure]: Function which can be call to create the figure with the associated dataframe
|
30 |
"""
|
31 |
indicator = params["indicator_column"]
|
32 |
-
model = params["model"]
|
33 |
location = params["location"]
|
34 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
35 |
|
@@ -43,7 +42,7 @@ def plot_indicator_evolution_at_location(params: dict) -> Callable[..., Figure]:
|
|
43 |
Figure: Plotly figure
|
44 |
"""
|
45 |
fig = go.Figure()
|
46 |
-
if model
|
47 |
df_avg = df.groupby("year", as_index=False)[indicator].mean()
|
48 |
|
49 |
# Transform to list to avoid pandas encoding
|
@@ -58,8 +57,10 @@ def plot_indicator_evolution_at_location(params: dict) -> Callable[..., Figure]:
|
|
58 |
.astype(float)
|
59 |
.tolist()
|
60 |
)
|
|
|
|
|
61 |
else:
|
62 |
-
df_model = df
|
63 |
|
64 |
# Transform to list to avoid pandas encoding
|
65 |
indicators = df_model[indicator].astype(float).tolist()
|
@@ -73,6 +74,8 @@ def plot_indicator_evolution_at_location(params: dict) -> Callable[..., Figure]:
|
|
73 |
.astype(float)
|
74 |
.tolist()
|
75 |
)
|
|
|
|
|
76 |
|
77 |
# Indicator per year plot
|
78 |
fig.add_scatter(
|
@@ -93,7 +96,7 @@ def plot_indicator_evolution_at_location(params: dict) -> Callable[..., Figure]:
|
|
93 |
marker=dict(color="#d62728"),
|
94 |
)
|
95 |
fig.update_layout(
|
96 |
-
title=f"Plot of {indicator_label} in {location} {
|
97 |
xaxis_title="Year",
|
98 |
yaxis_title=indicator_label,
|
99 |
template="plotly_white",
|
@@ -125,7 +128,6 @@ def plot_indicator_number_of_days_per_year_at_location(
|
|
125 |
"""
|
126 |
|
127 |
indicator = params["indicator_column"]
|
128 |
-
model = params["model"]
|
129 |
location = params["location"]
|
130 |
|
131 |
def plot_data(df: pd.DataFrame) -> Figure:
|
@@ -138,19 +140,21 @@ def plot_indicator_number_of_days_per_year_at_location(
|
|
138 |
Figure: Plotly figure
|
139 |
"""
|
140 |
fig = go.Figure()
|
141 |
-
if model
|
142 |
df_avg = df.groupby("year", as_index=False)[indicator].mean()
|
143 |
|
144 |
# Transform to list to avoid pandas encoding
|
145 |
indicators = df_avg[indicator].astype(float).tolist()
|
146 |
years = df_avg["year"].astype(int).tolist()
|
|
|
147 |
|
148 |
else:
|
149 |
-
df_model = df
|
150 |
-
|
151 |
# Transform to list to avoid pandas encoding
|
152 |
indicators = df_model[indicator].astype(float).tolist()
|
153 |
years = df_model["year"].astype(int).tolist()
|
|
|
|
|
154 |
|
155 |
# Bar plot
|
156 |
fig.add_trace(
|
@@ -165,7 +169,7 @@ def plot_indicator_number_of_days_per_year_at_location(
|
|
165 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
166 |
|
167 |
fig.update_layout(
|
168 |
-
title=f"{indicator_label} in {location} {
|
169 |
xaxis_title="Year",
|
170 |
yaxis_title=indicator,
|
171 |
yaxis=dict(range=[0, max(indicators)]),
|
@@ -199,7 +203,6 @@ def plot_distribution_of_indicator_for_given_year(
|
|
199 |
Callable[..., Figure]: Function which can be call to create the figure with the associated dataframe
|
200 |
"""
|
201 |
indicator = params["indicator_column"]
|
202 |
-
model = params["model"]
|
203 |
year = params["year"]
|
204 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
205 |
|
@@ -213,18 +216,22 @@ def plot_distribution_of_indicator_for_given_year(
|
|
213 |
Figure: Plotly figure
|
214 |
"""
|
215 |
fig = go.Figure()
|
216 |
-
if
|
217 |
df_avg = df.groupby(["latitude", "longitude"], as_index=False)[
|
218 |
indicator
|
219 |
].mean()
|
220 |
|
221 |
# Transform to list to avoid pandas encoding
|
222 |
indicators = df_avg[indicator].astype(float).tolist()
|
|
|
|
|
223 |
else:
|
224 |
-
df_model = df
|
225 |
|
226 |
# Transform to list to avoid pandas encoding
|
227 |
indicators = df_model[indicator].astype(float).tolist()
|
|
|
|
|
228 |
|
229 |
fig.add_trace(
|
230 |
go.Histogram(
|
@@ -236,7 +243,7 @@ def plot_distribution_of_indicator_for_given_year(
|
|
236 |
)
|
237 |
|
238 |
fig.update_layout(
|
239 |
-
title=f"Distribution of {indicator_label} in {year} {
|
240 |
xaxis_title=indicator_label,
|
241 |
yaxis_title="Frequency",
|
242 |
plot_bgcolor="rgba(0, 0, 0, 0)",
|
@@ -270,13 +277,12 @@ def plot_map_of_france_of_indicator_for_given_year(
|
|
270 |
"""
|
271 |
|
272 |
indicator = params["indicator_column"]
|
273 |
-
model = params["model"]
|
274 |
year = params["year"]
|
275 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
276 |
|
277 |
def plot_data(df: pd.DataFrame) -> Figure:
|
278 |
fig = go.Figure()
|
279 |
-
if model
|
280 |
df_avg = df.groupby(["latitude", "longitude"], as_index=False)[
|
281 |
indicator
|
282 |
].mean()
|
@@ -284,14 +290,17 @@ def plot_map_of_france_of_indicator_for_given_year(
|
|
284 |
indicators = df_avg[indicator].astype(float).tolist()
|
285 |
latitudes = df_avg["latitude"].astype(float).tolist()
|
286 |
longitudes = df_avg["longitude"].astype(float).tolist()
|
|
|
287 |
|
288 |
else:
|
289 |
-
df_model = df
|
290 |
|
291 |
# Transform to list to avoid pandas encoding
|
292 |
indicators = df_model[indicator].astype(float).tolist()
|
293 |
latitudes = df_model["latitude"].astype(float).tolist()
|
294 |
longitudes = df_model["longitude"].astype(float).tolist()
|
|
|
|
|
295 |
|
296 |
fig.add_trace(
|
297 |
go.Scattermapbox(
|
@@ -314,7 +323,7 @@ def plot_map_of_france_of_indicator_for_given_year(
|
|
314 |
mapbox_zoom=3,
|
315 |
mapbox_center={"lat": 46.6, "lon": 2.0},
|
316 |
coloraxis_colorbar=dict(title=f"{indicator_label}"), # Add legend
|
317 |
-
title=f"{indicator_label} in {year} in France {
|
318 |
)
|
319 |
return fig
|
320 |
|
|
|
29 |
Callable[..., Figure]: Function which can be call to create the figure with the associated dataframe
|
30 |
"""
|
31 |
indicator = params["indicator_column"]
|
|
|
32 |
location = params["location"]
|
33 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
34 |
|
|
|
42 |
Figure: Plotly figure
|
43 |
"""
|
44 |
fig = go.Figure()
|
45 |
+
if df['model'].nunique() != 1:
|
46 |
df_avg = df.groupby("year", as_index=False)[indicator].mean()
|
47 |
|
48 |
# Transform to list to avoid pandas encoding
|
|
|
57 |
.astype(float)
|
58 |
.tolist()
|
59 |
)
|
60 |
+
model_label = "Model Average"
|
61 |
+
|
62 |
else:
|
63 |
+
df_model = df
|
64 |
|
65 |
# Transform to list to avoid pandas encoding
|
66 |
indicators = df_model[indicator].astype(float).tolist()
|
|
|
74 |
.astype(float)
|
75 |
.tolist()
|
76 |
)
|
77 |
+
model_label = f"Model : {df['model'].unique()[0]}"
|
78 |
+
|
79 |
|
80 |
# Indicator per year plot
|
81 |
fig.add_scatter(
|
|
|
96 |
marker=dict(color="#d62728"),
|
97 |
)
|
98 |
fig.update_layout(
|
99 |
+
title=f"Plot of {indicator_label} in {location} ({model_label})",
|
100 |
xaxis_title="Year",
|
101 |
yaxis_title=indicator_label,
|
102 |
template="plotly_white",
|
|
|
128 |
"""
|
129 |
|
130 |
indicator = params["indicator_column"]
|
|
|
131 |
location = params["location"]
|
132 |
|
133 |
def plot_data(df: pd.DataFrame) -> Figure:
|
|
|
140 |
Figure: Plotly figure
|
141 |
"""
|
142 |
fig = go.Figure()
|
143 |
+
if df['model'].nunique() != 1:
|
144 |
df_avg = df.groupby("year", as_index=False)[indicator].mean()
|
145 |
|
146 |
# Transform to list to avoid pandas encoding
|
147 |
indicators = df_avg[indicator].astype(float).tolist()
|
148 |
years = df_avg["year"].astype(int).tolist()
|
149 |
+
model_label = "Model Average"
|
150 |
|
151 |
else:
|
152 |
+
df_model = df
|
|
|
153 |
# Transform to list to avoid pandas encoding
|
154 |
indicators = df_model[indicator].astype(float).tolist()
|
155 |
years = df_model["year"].astype(int).tolist()
|
156 |
+
model_label = f"Model : {df['model'].unique()[0]}"
|
157 |
+
|
158 |
|
159 |
# Bar plot
|
160 |
fig.add_trace(
|
|
|
169 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
170 |
|
171 |
fig.update_layout(
|
172 |
+
title=f"{indicator_label} in {location} ({model_label})",
|
173 |
xaxis_title="Year",
|
174 |
yaxis_title=indicator,
|
175 |
yaxis=dict(range=[0, max(indicators)]),
|
|
|
203 |
Callable[..., Figure]: Function which can be call to create the figure with the associated dataframe
|
204 |
"""
|
205 |
indicator = params["indicator_column"]
|
|
|
206 |
year = params["year"]
|
207 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
208 |
|
|
|
216 |
Figure: Plotly figure
|
217 |
"""
|
218 |
fig = go.Figure()
|
219 |
+
if df['model'].nunique() != 1:
|
220 |
df_avg = df.groupby(["latitude", "longitude"], as_index=False)[
|
221 |
indicator
|
222 |
].mean()
|
223 |
|
224 |
# Transform to list to avoid pandas encoding
|
225 |
indicators = df_avg[indicator].astype(float).tolist()
|
226 |
+
model_label = "Model Average"
|
227 |
+
|
228 |
else:
|
229 |
+
df_model = df
|
230 |
|
231 |
# Transform to list to avoid pandas encoding
|
232 |
indicators = df_model[indicator].astype(float).tolist()
|
233 |
+
model_label = f"Model : {df['model'].unique()[0]}"
|
234 |
+
|
235 |
|
236 |
fig.add_trace(
|
237 |
go.Histogram(
|
|
|
243 |
)
|
244 |
|
245 |
fig.update_layout(
|
246 |
+
title=f"Distribution of {indicator_label} in {year} ({model_label})",
|
247 |
xaxis_title=indicator_label,
|
248 |
yaxis_title="Frequency",
|
249 |
plot_bgcolor="rgba(0, 0, 0, 0)",
|
|
|
277 |
"""
|
278 |
|
279 |
indicator = params["indicator_column"]
|
|
|
280 |
year = params["year"]
|
281 |
indicator_label = " ".join([word.capitalize() for word in indicator.split("_")])
|
282 |
|
283 |
def plot_data(df: pd.DataFrame) -> Figure:
|
284 |
fig = go.Figure()
|
285 |
+
if df['model'].nunique() != 1:
|
286 |
df_avg = df.groupby(["latitude", "longitude"], as_index=False)[
|
287 |
indicator
|
288 |
].mean()
|
|
|
290 |
indicators = df_avg[indicator].astype(float).tolist()
|
291 |
latitudes = df_avg["latitude"].astype(float).tolist()
|
292 |
longitudes = df_avg["longitude"].astype(float).tolist()
|
293 |
+
model_label = "Model Average"
|
294 |
|
295 |
else:
|
296 |
+
df_model = df
|
297 |
|
298 |
# Transform to list to avoid pandas encoding
|
299 |
indicators = df_model[indicator].astype(float).tolist()
|
300 |
latitudes = df_model["latitude"].astype(float).tolist()
|
301 |
longitudes = df_model["longitude"].astype(float).tolist()
|
302 |
+
model_label = f"Model : {df['model'].unique()[0]}"
|
303 |
+
|
304 |
|
305 |
fig.add_trace(
|
306 |
go.Scattermapbox(
|
|
|
323 |
mapbox_zoom=3,
|
324 |
mapbox_center={"lat": 46.6, "lon": 2.0},
|
325 |
coloraxis_colorbar=dict(title=f"{indicator_label}"), # Add legend
|
326 |
+
title=f"{indicator_label} in {year} in France ({model_label}) " # Title
|
327 |
)
|
328 |
return fig
|
329 |
|
climateqa/engine/talk_to_data/sql_query.py
CHANGED
@@ -1,41 +1,23 @@
|
|
1 |
-
import
|
2 |
-
|
|
|
3 |
|
4 |
-
|
5 |
-
class SqlQueryOutput(TypedDict):
|
6 |
-
labels: list[str]
|
7 |
-
data: list[list[Any]]
|
8 |
-
|
9 |
-
|
10 |
-
def execute_sql_query(db_path: str, sql_query: str) -> SqlQueryOutput:
|
11 |
"""Execute the SQL Query on the sqlite database
|
12 |
|
13 |
Args:
|
14 |
-
db_ (str): path to the sqlite database
|
15 |
sql_query (str): sql query to execute
|
16 |
|
17 |
Returns:
|
18 |
SqlQueryOutput: labels of the selected column and fetched data
|
19 |
"""
|
20 |
|
21 |
-
# Connect to sqlite3 database
|
22 |
-
conn = sqlite3.connect(db_path)
|
23 |
-
cursor = conn.cursor()
|
24 |
|
25 |
# Execute the query
|
26 |
-
|
27 |
-
|
28 |
-
# Fetch labels of selected columns
|
29 |
-
labels = [desc[0] for desc in cursor.description]
|
30 |
|
31 |
-
#
|
32 |
-
|
33 |
-
conn.close()
|
34 |
-
|
35 |
-
return {
|
36 |
-
"labels": labels,
|
37 |
-
"data": data,
|
38 |
-
}
|
39 |
|
40 |
|
41 |
class IndicatorPerYearAtLocationQueryParams(TypedDict, total=False):
|
@@ -60,15 +42,13 @@ def indicator_per_year_at_location_query(
|
|
60 |
indicator_column = params.get("indicator_column")
|
61 |
latitude = params.get("latitude")
|
62 |
longitude = params.get("longitude")
|
63 |
-
model = params.get('model')
|
64 |
|
65 |
if indicator_column is None or latitude is None or longitude is None: # If one parameter is missing, returns an empty query
|
66 |
return ""
|
67 |
|
68 |
-
|
69 |
-
|
70 |
-
|
71 |
-
sql_query = f"SELECT year, {indicator_column}, model\nFROM {table}\nWHERE latitude = {latitude} \nAnd longitude = {longitude} \nAnd model = '{model}' \nOrder by Year"
|
72 |
|
73 |
return sql_query
|
74 |
|
@@ -91,12 +71,10 @@ def indicator_for_given_year_query(
|
|
91 |
"""
|
92 |
indicator_column = params.get("indicator_column")
|
93 |
year = params.get('year')
|
94 |
-
model = params.get('model')
|
95 |
if year is None or indicator_column is None: # If one parameter is missing, returns an empty query
|
96 |
return ""
|
97 |
|
98 |
-
|
99 |
-
|
100 |
-
|
101 |
-
sql_query = f"Select {indicator_column}, latitude, longitude, model\nFrom {table}\nWhere year = {year}\nAnd model = '{model}'"
|
102 |
return sql_query
|
|
|
1 |
+
from typing import TypedDict
|
2 |
+
import duckdb
|
3 |
+
import pandas as pd
|
4 |
|
5 |
+
def execute_sql_query(sql_query: str) -> pd.DataFrame:
|
|
|
|
|
|
|
|
|
|
|
|
|
6 |
"""Execute the SQL Query on the sqlite database
|
7 |
|
8 |
Args:
|
|
|
9 |
sql_query (str): sql query to execute
|
10 |
|
11 |
Returns:
|
12 |
SqlQueryOutput: labels of the selected column and fetched data
|
13 |
"""
|
14 |
|
|
|
|
|
|
|
15 |
|
16 |
# Execute the query
|
17 |
+
results = duckdb.sql(sql_query)
|
|
|
|
|
|
|
18 |
|
19 |
+
# return fetched data
|
20 |
+
return results.fetchdf()
|
|
|
|
|
|
|
|
|
|
|
|
|
21 |
|
22 |
|
23 |
class IndicatorPerYearAtLocationQueryParams(TypedDict, total=False):
|
|
|
42 |
indicator_column = params.get("indicator_column")
|
43 |
latitude = params.get("latitude")
|
44 |
longitude = params.get("longitude")
|
|
|
45 |
|
46 |
if indicator_column is None or latitude is None or longitude is None: # If one parameter is missing, returns an empty query
|
47 |
return ""
|
48 |
|
49 |
+
table = f"'hf://datasets/timeki/drias_db/{table.lower()}.parquet'"
|
50 |
+
|
51 |
+
sql_query = f"SELECT year, {indicator_column}, model\nFROM {table}\nWHERE latitude = {latitude} \nAnd longitude = {longitude} \nOrder by Year"
|
|
|
52 |
|
53 |
return sql_query
|
54 |
|
|
|
71 |
"""
|
72 |
indicator_column = params.get("indicator_column")
|
73 |
year = params.get('year')
|
|
|
74 |
if year is None or indicator_column is None: # If one parameter is missing, returns an empty query
|
75 |
return ""
|
76 |
|
77 |
+
table = f"'hf://datasets/timeki/drias_db/{table.lower()}.parquet'"
|
78 |
+
|
79 |
+
sql_query = f"Select {indicator_column}, latitude, longitude, model\nFrom {table}\nWhere year = {year}"
|
|
|
80 |
return sql_query
|
climateqa/engine/talk_to_data/utils.py
CHANGED
@@ -1,11 +1,10 @@
|
|
1 |
import re
|
2 |
from typing import Annotated, TypedDict
|
3 |
-
|
4 |
-
from sympy import use
|
5 |
from geopy.geocoders import Nominatim
|
6 |
-
import sqlite3
|
7 |
import ast
|
8 |
from climateqa.engine.llm import get_llm
|
|
|
9 |
from climateqa.engine.talk_to_data.plot import PLOTS, Plot
|
10 |
from langchain_core.prompts import ChatPromptTemplate
|
11 |
|
@@ -35,7 +34,7 @@ class ArrayOutput(TypedDict):
|
|
35 |
|
36 |
array: Annotated[str, ..., "Syntactically valid python array."]
|
37 |
|
38 |
-
def detect_year_with_openai(sentence: str):
|
39 |
"""
|
40 |
Detects years in a sentence using OpenAI's API via LangChain.
|
41 |
"""
|
@@ -56,7 +55,7 @@ def detect_year_with_openai(sentence: str):
|
|
56 |
if len(years_list) > 0:
|
57 |
return years_list[0]
|
58 |
else:
|
59 |
-
return
|
60 |
|
61 |
|
62 |
def detectTable(sql_query):
|
@@ -81,24 +80,26 @@ def coords2loc(coords: tuple):
|
|
81 |
return "Unknown Location"
|
82 |
|
83 |
|
84 |
-
def nearestNeighbourSQL(
|
85 |
-
conn = sqlite3.connect(db)
|
86 |
long = round(location[1], 3)
|
87 |
lat = round(location[0], 3)
|
88 |
-
|
89 |
-
|
|
|
|
|
90 |
f"SELECT latitude, longitude FROM {table} WHERE latitude BETWEEN {lat - 0.3} AND {lat + 0.3} AND longitude BETWEEN {long - 0.3} AND {long + 0.3}"
|
91 |
-
)
|
|
|
|
|
|
|
92 |
# cursor.execute(f"SELECT latitude, longitude FROM {table} WHERE latitude BETWEEN {lat - 0.3} AND {lat + 0.3} AND longitude BETWEEN {long - 0.3} AND {long + 0.3}")
|
93 |
-
results
|
94 |
-
return results[0]
|
95 |
|
96 |
|
97 |
-
def detect_relevant_tables(
|
98 |
"""Detect relevant tables regarding the plot and the user input
|
99 |
|
100 |
Args:
|
101 |
-
db (str): database path
|
102 |
user_question (str): initial user input
|
103 |
plot (Plot): plot object for which we wanna plot
|
104 |
llm (_type_): LLM
|
@@ -106,19 +107,21 @@ def detect_relevant_tables(db: str, user_question: str, plot: Plot, llm) -> list
|
|
106 |
Returns:
|
107 |
list[str]: list of table names
|
108 |
"""
|
109 |
-
conn = sqlite3.connect(db)
|
110 |
-
cursor = conn.cursor()
|
111 |
|
112 |
# Get all table names
|
113 |
-
|
114 |
-
table_names_list = cursor.fetchall()
|
115 |
|
116 |
prompt = (
|
117 |
f"You are helping to build a plot following this description : {plot['description']}."
|
|
|
118 |
f"Based on the description of the plot, which table are appropriate for that kind of plot."
|
119 |
-
f"
|
120 |
-
f"
|
|
|
|
|
121 |
)
|
|
|
|
|
122 |
table_names = ast.literal_eval(
|
123 |
llm.invoke(prompt).content.strip("```python\n").strip()
|
124 |
)
|
@@ -141,17 +144,28 @@ def detect_relevant_plots(user_question: str, llm):
|
|
141 |
plots_description += " - Description: " + plot["description"] + "\n"
|
142 |
|
143 |
prompt = (
|
144 |
-
f"You are helping to answer a
|
145 |
-
f"
|
146 |
-
f"
|
147 |
-
f"
|
148 |
-
f"
|
149 |
-
f"
|
150 |
-
f"
|
151 |
)
|
152 |
-
|
153 |
-
|
154 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
155 |
|
156 |
|
157 |
# Next Version
|
|
|
1 |
import re
|
2 |
from typing import Annotated, TypedDict
|
3 |
+
import duckdb
|
|
|
4 |
from geopy.geocoders import Nominatim
|
|
|
5 |
import ast
|
6 |
from climateqa.engine.llm import get_llm
|
7 |
+
from climateqa.engine.talk_to_data.config import DRIAS_TABLES
|
8 |
from climateqa.engine.talk_to_data.plot import PLOTS, Plot
|
9 |
from langchain_core.prompts import ChatPromptTemplate
|
10 |
|
|
|
34 |
|
35 |
array: Annotated[str, ..., "Syntactically valid python array."]
|
36 |
|
37 |
+
def detect_year_with_openai(sentence: str) -> str:
|
38 |
"""
|
39 |
Detects years in a sentence using OpenAI's API via LangChain.
|
40 |
"""
|
|
|
55 |
if len(years_list) > 0:
|
56 |
return years_list[0]
|
57 |
else:
|
58 |
+
return ""
|
59 |
|
60 |
|
61 |
def detectTable(sql_query):
|
|
|
80 |
return "Unknown Location"
|
81 |
|
82 |
|
83 |
+
def nearestNeighbourSQL(location: tuple, table: str) -> tuple[str, str]:
|
|
|
84 |
long = round(location[1], 3)
|
85 |
lat = round(location[0], 3)
|
86 |
+
|
87 |
+
table = f"'hf://datasets/timeki/drias_db/{table.lower()}.parquet'"
|
88 |
+
|
89 |
+
results = duckdb.sql(
|
90 |
f"SELECT latitude, longitude FROM {table} WHERE latitude BETWEEN {lat - 0.3} AND {lat + 0.3} AND longitude BETWEEN {long - 0.3} AND {long + 0.3}"
|
91 |
+
).fetchdf()
|
92 |
+
|
93 |
+
if len(results) == 0:
|
94 |
+
return "", ""
|
95 |
# cursor.execute(f"SELECT latitude, longitude FROM {table} WHERE latitude BETWEEN {lat - 0.3} AND {lat + 0.3} AND longitude BETWEEN {long - 0.3} AND {long + 0.3}")
|
96 |
+
return results['latitude'].iloc[0], results['longitude'].iloc[0]
|
|
|
97 |
|
98 |
|
99 |
+
def detect_relevant_tables(user_question: str, plot: Plot, llm) -> list[str]:
|
100 |
"""Detect relevant tables regarding the plot and the user input
|
101 |
|
102 |
Args:
|
|
|
103 |
user_question (str): initial user input
|
104 |
plot (Plot): plot object for which we wanna plot
|
105 |
llm (_type_): LLM
|
|
|
107 |
Returns:
|
108 |
list[str]: list of table names
|
109 |
"""
|
|
|
|
|
110 |
|
111 |
# Get all table names
|
112 |
+
table_names_list = DRIAS_TABLES
|
|
|
113 |
|
114 |
prompt = (
|
115 |
f"You are helping to build a plot following this description : {plot['description']}."
|
116 |
+
f"You are given a list of tables and a user question."
|
117 |
f"Based on the description of the plot, which table are appropriate for that kind of plot."
|
118 |
+
f"Write the 3 most relevant tables to use. Answer only a python list of table name."
|
119 |
+
f"### List of tables : {table_names_list}"
|
120 |
+
f"### User question : {user_question}"
|
121 |
+
f"### List of table name : "
|
122 |
)
|
123 |
+
|
124 |
+
|
125 |
table_names = ast.literal_eval(
|
126 |
llm.invoke(prompt).content.strip("```python\n").strip()
|
127 |
)
|
|
|
144 |
plots_description += " - Description: " + plot["description"] + "\n"
|
145 |
|
146 |
prompt = (
|
147 |
+
f"You are helping to answer a quesiton with insightful visualizations."
|
148 |
+
f"You are given an user question and a list of plots with their name and description."
|
149 |
+
f"Based on the descriptions of the plots, which plot is appropriate to answer to this question."
|
150 |
+
f"Write the most relevant tables to use. Answer only a python list of plot name."
|
151 |
+
f"### Descriptions of the plots : {plots_description}"
|
152 |
+
f"### User question : {user_question}"
|
153 |
+
f"### Name of the plot : "
|
154 |
)
|
155 |
+
# prompt = (
|
156 |
+
# f"You are helping to answer a question with insightful visualizations. "
|
157 |
+
# f"Given a list of plots with their name and description: "
|
158 |
+
# f"{plots_description} "
|
159 |
+
# f"The user question is: {user_question}. "
|
160 |
+
# f"Choose the most relevant plots to answer the question. "
|
161 |
+
# f"The answer must be a Python list with the names of the relevant plots, and nothing else. "
|
162 |
+
# f"Ensure the response is in the exact format: ['PlotName1', 'PlotName2']."
|
163 |
+
# )
|
164 |
+
|
165 |
+
plot_names = ast.literal_eval(
|
166 |
+
llm.invoke(prompt).content.strip("```python\n").strip()
|
167 |
+
)
|
168 |
+
return plot_names
|
169 |
|
170 |
|
171 |
# Next Version
|
climateqa/engine/talk_to_data/workflow.py
CHANGED
@@ -5,6 +5,7 @@ import pandas as pd
|
|
5 |
|
6 |
from plotly.graph_objects import Figure
|
7 |
from climateqa.engine.llm import get_llm
|
|
|
8 |
from climateqa.engine.talk_to_data.plot import PLOTS, Plot
|
9 |
from climateqa.engine.talk_to_data.sql_query import execute_sql_query
|
10 |
from climateqa.engine.talk_to_data.utils import (
|
@@ -37,12 +38,12 @@ class State(TypedDict):
|
|
37 |
user_input: str
|
38 |
plots: list[str]
|
39 |
plot_states: dict[str, PlotState]
|
|
|
40 |
|
41 |
-
def drias_workflow(
|
42 |
"""Performs the complete workflow of Talk To Drias : from user input to sql queries, dataframes and figures generated
|
43 |
|
44 |
Args:
|
45 |
-
db_drias_path (str): path to the drias database
|
46 |
user_input (str): initial user input
|
47 |
|
48 |
Returns:
|
@@ -60,8 +61,12 @@ def drias_workflow(db_drias_path: str, user_input: str, model: str) -> State:
|
|
60 |
state['plots'] = plots
|
61 |
|
62 |
if not state['plots']:
|
|
|
63 |
return state
|
64 |
|
|
|
|
|
|
|
65 |
for plot_name in state['plots']:
|
66 |
|
67 |
plot = next((p for p in PLOTS if p['name'] == plot_name), None) # Find the associated plot object
|
@@ -76,21 +81,23 @@ def drias_workflow(db_drias_path: str, user_input: str, model: str) -> State:
|
|
76 |
|
77 |
plot_state['plot_name'] = plot_name
|
78 |
|
79 |
-
relevant_tables = find_relevant_tables_per_plot(state, plot,
|
|
|
|
|
80 |
|
81 |
plot_state['tables'] = relevant_tables
|
82 |
|
83 |
-
for table in plot_state['tables']:
|
|
|
|
|
|
|
84 |
table_state: TableState = {
|
85 |
'table_name': table,
|
86 |
'params': {},
|
87 |
'status': 'OK'
|
88 |
}
|
89 |
-
table_state['params'] = {
|
90 |
-
'model': model
|
91 |
-
}
|
92 |
for param_name in plot['params']:
|
93 |
-
param = find_param(state, param_name, table
|
94 |
if param:
|
95 |
table_state['params'].update(param)
|
96 |
|
@@ -99,17 +106,30 @@ def drias_workflow(db_drias_path: str, user_input: str, model: str) -> State:
|
|
99 |
if sql_query == "":
|
100 |
table_state['status'] = 'ERROR'
|
101 |
continue
|
|
|
|
|
102 |
|
103 |
table_state['sql_query'] = sql_query
|
104 |
-
|
|
|
|
|
|
|
105 |
|
106 |
-
df = pd.DataFrame(results['data'], columns=results['labels'])
|
107 |
figure = plot['plot_function'](table_state['params'])
|
108 |
table_state['dataframe'] = df
|
109 |
table_state['figure'] = figure
|
110 |
plot_state['table_states'][table] = table_state
|
111 |
|
112 |
state['plot_states'][plot_name] = plot_state
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
113 |
return state
|
114 |
|
115 |
|
@@ -118,26 +138,25 @@ def find_relevant_plots(state: State, llm) -> list[str]:
|
|
118 |
relevant_plots = detect_relevant_plots(state['user_input'], llm)
|
119 |
return relevant_plots
|
120 |
|
121 |
-
def find_relevant_tables_per_plot(state: State, plot: Plot,
|
122 |
print(f"---- Find relevant tables for {plot['name']} ----")
|
123 |
-
relevant_tables = detect_relevant_tables(
|
124 |
return relevant_tables
|
125 |
|
126 |
|
127 |
-
def find_param(state: State, param_name:str, table: str
|
128 |
"""Perform the good method to retrieve the desired parameter
|
129 |
|
130 |
Args:
|
131 |
state (State): state of the workflow
|
132 |
param_name (str): name of the desired parameter
|
133 |
table (str): name of the table
|
134 |
-
db_path (str): path to the databse
|
135 |
|
136 |
Returns:
|
137 |
dict[str, Any] | None:
|
138 |
"""
|
139 |
if param_name == 'location':
|
140 |
-
location = find_location(state['user_input'], table
|
141 |
return location
|
142 |
if param_name == 'indicator_column':
|
143 |
indicator_column = find_indicator_column(table)
|
@@ -153,13 +172,13 @@ class Location(TypedDict):
|
|
153 |
latitude: NotRequired[str]
|
154 |
longitude: NotRequired[str]
|
155 |
|
156 |
-
def find_location(user_input: str, table: str
|
157 |
print(f"---- Find location in table {table} ----")
|
158 |
location = detect_location_with_openai(user_input)
|
159 |
output: Location = {'location' : location}
|
160 |
if location:
|
161 |
coords = loc2coords(location)
|
162 |
-
neighbour = nearestNeighbourSQL(
|
163 |
output.update({
|
164 |
"latitude": neighbour[0],
|
165 |
"longitude": neighbour[1],
|
@@ -182,23 +201,8 @@ def find_indicator_column(table: str) -> str:
|
|
182 |
"""
|
183 |
|
184 |
print(f"---- Find indicator column in table {table} ----")
|
185 |
-
|
186 |
-
|
187 |
-
"total_summer_precipiation": "total_summer_precipitation",
|
188 |
-
"total_annual_precipitation": "total_annual_precipitation",
|
189 |
-
"total_remarkable_daily_precipitation": "total_remarkable_daily_precipitation",
|
190 |
-
"frequency_of_remarkable_daily_precipitation": "frequency_of_remarkable_daily_precipitation",
|
191 |
-
"extreme_precipitation_intensity": "extreme_precipitation_intensity",
|
192 |
-
"mean_winter_temperature": "mean_winter_temperature",
|
193 |
-
"mean_summer_temperature": "mean_summer_temperature",
|
194 |
-
"mean_annual_temperature": "mean_annual_temperature",
|
195 |
-
"number_of_tropical_nights": "number_tropical_nights",
|
196 |
-
"maximum_summer_temperature": "maximum_summer_temperature",
|
197 |
-
"number_of_days_with_tx_above_30": "number_of_days_with_tx_above_30",
|
198 |
-
"number_of_days_with_tx_above_35": "number_of_days_with_tx_above_35",
|
199 |
-
"number_of_days_with_a_dry_ground": "number_of_days_with_dry_ground"
|
200 |
-
}
|
201 |
-
return indicator_columns_per_table[table]
|
202 |
|
203 |
|
204 |
# def make_write_query_node():
|
|
|
5 |
|
6 |
from plotly.graph_objects import Figure
|
7 |
from climateqa.engine.llm import get_llm
|
8 |
+
from climateqa.engine.talk_to_data.config import INDICATOR_COLUMNS_PER_TABLE
|
9 |
from climateqa.engine.talk_to_data.plot import PLOTS, Plot
|
10 |
from climateqa.engine.talk_to_data.sql_query import execute_sql_query
|
11 |
from climateqa.engine.talk_to_data.utils import (
|
|
|
38 |
user_input: str
|
39 |
plots: list[str]
|
40 |
plot_states: dict[str, PlotState]
|
41 |
+
error: NotRequired[str]
|
42 |
|
43 |
+
def drias_workflow(user_input: str) -> State:
|
44 |
"""Performs the complete workflow of Talk To Drias : from user input to sql queries, dataframes and figures generated
|
45 |
|
46 |
Args:
|
|
|
47 |
user_input (str): initial user input
|
48 |
|
49 |
Returns:
|
|
|
61 |
state['plots'] = plots
|
62 |
|
63 |
if not state['plots']:
|
64 |
+
state['error'] = 'There is no plot to answer to the question'
|
65 |
return state
|
66 |
|
67 |
+
have_relevant_table = False
|
68 |
+
have_sql_query = False
|
69 |
+
have_dataframe = False
|
70 |
for plot_name in state['plots']:
|
71 |
|
72 |
plot = next((p for p in PLOTS if p['name'] == plot_name), None) # Find the associated plot object
|
|
|
81 |
|
82 |
plot_state['plot_name'] = plot_name
|
83 |
|
84 |
+
relevant_tables = find_relevant_tables_per_plot(state, plot, llm)
|
85 |
+
if len(relevant_tables) > 0 :
|
86 |
+
have_relevant_table = True
|
87 |
|
88 |
plot_state['tables'] = relevant_tables
|
89 |
|
90 |
+
for n, table in enumerate(plot_state['tables']):
|
91 |
+
if n > 2:
|
92 |
+
break
|
93 |
+
|
94 |
table_state: TableState = {
|
95 |
'table_name': table,
|
96 |
'params': {},
|
97 |
'status': 'OK'
|
98 |
}
|
|
|
|
|
|
|
99 |
for param_name in plot['params']:
|
100 |
+
param = find_param(state, param_name, table)
|
101 |
if param:
|
102 |
table_state['params'].update(param)
|
103 |
|
|
|
106 |
if sql_query == "":
|
107 |
table_state['status'] = 'ERROR'
|
108 |
continue
|
109 |
+
else :
|
110 |
+
have_sql_query = True
|
111 |
|
112 |
table_state['sql_query'] = sql_query
|
113 |
+
df = execute_sql_query(sql_query)
|
114 |
+
|
115 |
+
if len(df) > 0:
|
116 |
+
have_dataframe = True
|
117 |
|
|
|
118 |
figure = plot['plot_function'](table_state['params'])
|
119 |
table_state['dataframe'] = df
|
120 |
table_state['figure'] = figure
|
121 |
plot_state['table_states'][table] = table_state
|
122 |
|
123 |
state['plot_states'][plot_name] = plot_state
|
124 |
+
|
125 |
+
if not have_relevant_table:
|
126 |
+
state['error'] = "There is no relevant table in the our database to answer your question"
|
127 |
+
elif not have_sql_query:
|
128 |
+
state['error'] = "There is no relevant sql query on our database that can help to answer your question"
|
129 |
+
elif not have_dataframe:
|
130 |
+
state['error'] = "There is no data in our table that can answer to your question"
|
131 |
+
|
132 |
+
|
133 |
return state
|
134 |
|
135 |
|
|
|
138 |
relevant_plots = detect_relevant_plots(state['user_input'], llm)
|
139 |
return relevant_plots
|
140 |
|
141 |
+
def find_relevant_tables_per_plot(state: State, plot: Plot, llm) -> list[str]:
|
142 |
print(f"---- Find relevant tables for {plot['name']} ----")
|
143 |
+
relevant_tables = detect_relevant_tables(state['user_input'], plot, llm)
|
144 |
return relevant_tables
|
145 |
|
146 |
|
147 |
+
def find_param(state: State, param_name:str, table: str) -> dict[str, Any] | None:
|
148 |
"""Perform the good method to retrieve the desired parameter
|
149 |
|
150 |
Args:
|
151 |
state (State): state of the workflow
|
152 |
param_name (str): name of the desired parameter
|
153 |
table (str): name of the table
|
|
|
154 |
|
155 |
Returns:
|
156 |
dict[str, Any] | None:
|
157 |
"""
|
158 |
if param_name == 'location':
|
159 |
+
location = find_location(state['user_input'], table)
|
160 |
return location
|
161 |
if param_name == 'indicator_column':
|
162 |
indicator_column = find_indicator_column(table)
|
|
|
172 |
latitude: NotRequired[str]
|
173 |
longitude: NotRequired[str]
|
174 |
|
175 |
+
def find_location(user_input: str, table: str) -> Location:
|
176 |
print(f"---- Find location in table {table} ----")
|
177 |
location = detect_location_with_openai(user_input)
|
178 |
output: Location = {'location' : location}
|
179 |
if location:
|
180 |
coords = loc2coords(location)
|
181 |
+
neighbour = nearestNeighbourSQL(coords, table)
|
182 |
output.update({
|
183 |
"latitude": neighbour[0],
|
184 |
"longitude": neighbour[1],
|
|
|
201 |
"""
|
202 |
|
203 |
print(f"---- Find indicator column in table {table} ----")
|
204 |
+
|
205 |
+
return INDICATOR_COLUMNS_PER_TABLE[table]
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
206 |
|
207 |
|
208 |
# def make_write_query_node():
|
style.css
CHANGED
@@ -644,17 +644,23 @@ a {
|
|
644 |
overflow-y:scroll;
|
645 |
}
|
646 |
|
|
|
|
|
|
|
|
|
647 |
#sql-query span{
|
648 |
display: none;
|
649 |
}
|
650 |
div#tab-vanna{
|
651 |
max-height: 100¨vh;
|
652 |
-
overflow-y:
|
653 |
}
|
654 |
#vanna-plot{
|
655 |
max-height:500px
|
656 |
}
|
657 |
|
658 |
-
#
|
659 |
-
|
|
|
|
|
660 |
}
|
|
|
644 |
overflow-y:scroll;
|
645 |
}
|
646 |
|
647 |
+
#sql-query textarea{
|
648 |
+
min-height: 100px !important;
|
649 |
+
}
|
650 |
+
|
651 |
#sql-query span{
|
652 |
display: none;
|
653 |
}
|
654 |
div#tab-vanna{
|
655 |
max-height: 100¨vh;
|
656 |
+
overflow-y: hidden;
|
657 |
}
|
658 |
#vanna-plot{
|
659 |
max-height:500px
|
660 |
}
|
661 |
|
662 |
+
#pagination-display{
|
663 |
+
text-align: center;
|
664 |
+
font-weight: bold;
|
665 |
+
font-size: 16px;
|
666 |
}
|