feat: optimize Gradio 6 CSS visual bubbles, add safe print overrides, and fix scraping matplotlib imports
Browse files- reset_db.py +36 -0
- run_pipeline.py +26 -0
- src/graphBuilder/neo4j/finGraph.py +36 -0
- src/graphBuilder/scrapping/finScrapping.py +50 -46
- src/utils/ui_templates.py +6 -13
reset_db.py
ADDED
|
@@ -0,0 +1,36 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
|
| 3 |
+
import dotenv
|
| 4 |
+
import neo4j
|
| 5 |
+
|
| 6 |
+
dotenv.load_dotenv()
|
| 7 |
+
|
| 8 |
+
|
| 9 |
+
def get_neo4j_driver() -> neo4j.Driver:
|
| 10 |
+
uri = os.getenv("NEO4J_URI", "neo4j://localhost:7687")
|
| 11 |
+
client_id = os.getenv("NEO4J_CLIENT_ID")
|
| 12 |
+
client_secret = os.getenv("NEO4J_CLIENT_SECRET")
|
| 13 |
+
|
| 14 |
+
if client_id and client_secret:
|
| 15 |
+
try:
|
| 16 |
+
d = neo4j.GraphDatabase.driver(uri, auth=(client_id, client_secret))
|
| 17 |
+
d.verify_connectivity()
|
| 18 |
+
return d
|
| 19 |
+
except Exception:
|
| 20 |
+
pass
|
| 21 |
+
|
| 22 |
+
username = os.getenv("NEO4J_USERNAME", "neo4j")
|
| 23 |
+
password = os.getenv("NEO4J_PASSWORD", "password")
|
| 24 |
+
d = neo4j.GraphDatabase.driver(uri, auth=(username, password))
|
| 25 |
+
d.verify_connectivity()
|
| 26 |
+
return d
|
| 27 |
+
|
| 28 |
+
|
| 29 |
+
if __name__ == "__main__":
|
| 30 |
+
print("Connecting to Neo4j to reset the database...")
|
| 31 |
+
driver = get_neo4j_driver()
|
| 32 |
+
with driver.session() as session:
|
| 33 |
+
# DETACH DELETE n deletes all nodes and their relationships
|
| 34 |
+
result = session.run("MATCH (n) DETACH DELETE n")
|
| 35 |
+
print("Database successfully reset. All nodes and relationships deleted.")
|
| 36 |
+
driver.close()
|
run_pipeline.py
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
import os
|
| 2 |
+
import subprocess
|
| 3 |
+
import sys
|
| 4 |
+
|
| 5 |
+
|
| 6 |
+
def run_command(cmd):
|
| 7 |
+
print("\n========================================")
|
| 8 |
+
print(f"Running: {cmd}")
|
| 9 |
+
print("========================================\n")
|
| 10 |
+
env = os.environ.copy()
|
| 11 |
+
env["PYTHONUTF8"] = "1"
|
| 12 |
+
result = subprocess.run(cmd, shell=True, env=env)
|
| 13 |
+
if result.returncode != 0:
|
| 14 |
+
print(f"Command failed with exit code {result.returncode}: {cmd}")
|
| 15 |
+
sys.exit(result.returncode)
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
if __name__ == "__main__":
|
| 19 |
+
print("Starting background rebuilding pipeline...")
|
| 20 |
+
# 1. ν¬λ‘€λ§ (finScrapping.py)
|
| 21 |
+
run_command(r".\.venv\Scripts\python.exe src/graphBuilder/scrapping/finScrapping.py")
|
| 22 |
+
|
| 23 |
+
# 2. μ§μ κ·Έλν λΉλ (finGraph.py)
|
| 24 |
+
run_command(r".\.venv\Scripts\python.exe src/graphBuilder/neo4j/finGraph.py")
|
| 25 |
+
|
| 26 |
+
print("\n[OK] Background rebuilding pipeline completed successfully!")
|
src/graphBuilder/neo4j/finGraph.py
CHANGED
|
@@ -27,6 +27,42 @@ from neo4j_graphrag.llm import OpenAILLM
|
|
| 27 |
dotenv.load_dotenv()
|
| 28 |
|
| 29 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 30 |
def get_neo4j_driver() -> neo4j.Driver:
|
| 31 |
uri = os.getenv("NEO4J_URI", "neo4j://localhost:7687")
|
| 32 |
client_id = os.getenv("NEO4J_CLIENT_ID")
|
|
|
|
| 27 |
dotenv.load_dotenv()
|
| 28 |
|
| 29 |
|
| 30 |
+
# Windows cp949 μΈμ½λ© νκ²½μμ μ΄λͺ¨μ§ μΆλ ₯ μ UnicodeEncodeError λ°©μ§λ₯Ό μν μμ ν print ν¨μ μ μ
|
| 31 |
+
def safe_print(*args, **kwargs) -> None:
|
| 32 |
+
import sys
|
| 33 |
+
try:
|
| 34 |
+
# endλ sep μΈμλ₯Ό μ¬λ°λ₯΄κ² μ²λ¦¬ν μ μλλ‘ λ΄μ₯ printμ κΈ°λ₯ μ€μ
|
| 35 |
+
sep = kwargs.get("sep", " ")
|
| 36 |
+
end = kwargs.get("end", "\n")
|
| 37 |
+
msg = sep.join(map(str, args))
|
| 38 |
+
sys.stdout.write(msg + end)
|
| 39 |
+
sys.stdout.flush()
|
| 40 |
+
except UnicodeEncodeError:
|
| 41 |
+
msg = sep.join(map(str, args))
|
| 42 |
+
cleaned = (
|
| 43 |
+
msg.replace("β
", "[OK]")
|
| 44 |
+
.replace("β οΈ", "[WARN]")
|
| 45 |
+
.replace("π¨", "[ERR]")
|
| 46 |
+
.replace("βοΈ", "[SKIP]")
|
| 47 |
+
.replace("π€", "[AI]")
|
| 48 |
+
.replace("π’", "[COMP]")
|
| 49 |
+
.replace("π", "[GRAPH]")
|
| 50 |
+
.replace("π°", "[NEWS]")
|
| 51 |
+
.replace("π¬", "[TECH]")
|
| 52 |
+
.replace("π", "[LINK]")
|
| 53 |
+
)
|
| 54 |
+
try:
|
| 55 |
+
sys.stdout.write(cleaned + end)
|
| 56 |
+
sys.stdout.flush()
|
| 57 |
+
except Exception:
|
| 58 |
+
ascii_msg = msg.encode("ascii", errors="replace").decode("ascii")
|
| 59 |
+
sys.stdout.write(ascii_msg + end)
|
| 60 |
+
sys.stdout.flush()
|
| 61 |
+
|
| 62 |
+
|
| 63 |
+
print = safe_print
|
| 64 |
+
|
| 65 |
+
|
| 66 |
def get_neo4j_driver() -> neo4j.Driver:
|
| 67 |
uri = os.getenv("NEO4J_URI", "neo4j://localhost:7687")
|
| 68 |
client_id = os.getenv("NEO4J_CLIENT_ID")
|
src/graphBuilder/scrapping/finScrapping.py
CHANGED
|
@@ -245,51 +245,55 @@ print(f"[SAVE] - AI νν
ν¬ κΈ°μ¬: {len(df_filtered)}건")
|
|
| 245 |
|
| 246 |
|
| 247 |
# ββ 4λ¨κ³: ν€μλ λΉλ μκ°ν ββ
|
| 248 |
-
|
| 249 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 250 |
|
| 251 |
-
|
| 252 |
-
|
| 253 |
-
|
| 254 |
-
|
| 255 |
-
|
| 256 |
-
plt.
|
| 257 |
-
|
| 258 |
-
|
| 259 |
-
print("μκ°ν
|
| 260 |
-
else:
|
| 261 |
-
# λΉλμ κ³μ°
|
| 262 |
-
all_kw = [kw for row in filtered_articles for kw in row["matched_keywords"].split(", ")]
|
| 263 |
-
kw_counts = Counter(all_kw)
|
| 264 |
-
|
| 265 |
-
# π λ³κ²½ ν¬μΈνΈ: FINTECH_AI_KEYWORDS μ 체 λͺ©λ‘μ μμλλ‘ κ·Έλνμ κ°μ νμ (0건 ν¬ν¨)
|
| 266 |
-
keywords = FINTECH_AI_KEYWORDS
|
| 267 |
-
counts = [kw_counts.get(kw, 0) for kw in keywords]
|
| 268 |
-
|
| 269 |
-
plt.figure(figsize=(12, 6))
|
| 270 |
-
|
| 271 |
-
# λ§λ κ·Έλν μμ±
|
| 272 |
-
bars = plt.bar(keywords, counts, color="skyblue", edgecolor="white")
|
| 273 |
-
|
| 274 |
-
# λ§λ μμ μ«μ(λΉλμ) νμ
|
| 275 |
-
for bar in bars:
|
| 276 |
-
height = bar.get_height()
|
| 277 |
-
# λ§λμ μ€μ(x), λ§λμ λμ΄(y) μμΉμ ν
μ€νΈλ₯Ό λ°°μΉ
|
| 278 |
-
plt.text(
|
| 279 |
-
bar.get_x() + bar.get_width() / 2.0,
|
| 280 |
-
height,
|
| 281 |
-
f"{height}",
|
| 282 |
-
ha="center",
|
| 283 |
-
va="bottom",
|
| 284 |
-
size=11,
|
| 285 |
-
fontweight="bold",
|
| 286 |
-
color="black",
|
| 287 |
-
)
|
| 288 |
|
| 289 |
-
plt.title("μμ§λ AI νν
ν¬ κΈ°μ¬ ν€μλ μΆν λΉλ (μ 체)", fontsize=15, pad=15)
|
| 290 |
-
plt.xlabel("ν€μλ", fontsize=12)
|
| 291 |
-
plt.ylabel("μΆν νμ (건)", fontsize=12)
|
| 292 |
-
plt.grid(axis="y", linestyle="--", alpha=0.7)
|
| 293 |
-
plt.xticks(rotation=45)
|
| 294 |
-
plt.tight_layout()
|
| 295 |
-
plt.show()
|
|
|
|
| 245 |
|
| 246 |
|
| 247 |
# ββ 4λ¨κ³: ν€μλ λΉλ μκ°ν ββ
|
| 248 |
+
try:
|
| 249 |
+
import platform
|
| 250 |
+
from collections import Counter
|
| 251 |
+
|
| 252 |
+
import matplotlib.pyplot as plt
|
| 253 |
+
|
| 254 |
+
# ν°νΈ κΉ¨μ§ λ°©μ§ (Mac νκ²½: AppleGothic)
|
| 255 |
+
if platform.system() == "Darwin":
|
| 256 |
+
plt.rc("font", family="AppleGothic")
|
| 257 |
+
plt.rcParams["axes.unicode_minus"] = False
|
| 258 |
+
|
| 259 |
+
if not filtered_articles:
|
| 260 |
+
print("μκ°νν λ°μ΄ν°κ° μμ΅λλ€.")
|
| 261 |
+
else:
|
| 262 |
+
# λΉλμ κ³μ°
|
| 263 |
+
all_kw = [kw for row in filtered_articles for kw in row["matched_keywords"].split(", ")]
|
| 264 |
+
kw_counts = Counter(all_kw)
|
| 265 |
+
|
| 266 |
+
# π λ³κ²½ ν¬μΈνΈ: FINTECH_AI_KEYWORDS μ 체 λͺ©λ‘μ μμλλ‘ κ·Έλνμ κ°μ νμ (0건 ν¬ν¨)
|
| 267 |
+
keywords = FINTECH_AI_KEYWORDS
|
| 268 |
+
counts = [kw_counts.get(kw, 0) for kw in keywords]
|
| 269 |
+
|
| 270 |
+
plt.figure(figsize=(12, 6))
|
| 271 |
+
|
| 272 |
+
# λ§λ κ·Έλν μμ±
|
| 273 |
+
bars = plt.bar(keywords, counts, color="skyblue", edgecolor="white")
|
| 274 |
+
|
| 275 |
+
# λ§λ μμ μ«μ(λΉλμ) νμ
|
| 276 |
+
for bar in bars:
|
| 277 |
+
height = bar.get_height()
|
| 278 |
+
# λ§λμ μ€μ(x), λ§λμ λμ΄(y) μμΉμ ν
μ€νΈλ₯Ό λ°°μΉ
|
| 279 |
+
plt.text(
|
| 280 |
+
bar.get_x() + bar.get_width() / 2.0,
|
| 281 |
+
height,
|
| 282 |
+
f"{height}",
|
| 283 |
+
ha="center",
|
| 284 |
+
va="bottom",
|
| 285 |
+
size=11,
|
| 286 |
+
fontweight="bold",
|
| 287 |
+
color="black",
|
| 288 |
+
)
|
| 289 |
|
| 290 |
+
plt.title("μμ§λ AI νν
ν¬ κΈ°μ¬ ν€μλ μΆν λΉλ (μ 체)", fontsize=15, pad=15)
|
| 291 |
+
plt.xlabel("ν€μλ", fontsize=12)
|
| 292 |
+
plt.ylabel("μΆν νμ (건)", fontsize=12)
|
| 293 |
+
plt.grid(axis="y", linestyle="--", alpha=0.7)
|
| 294 |
+
plt.xticks(rotation=45)
|
| 295 |
+
plt.tight_layout()
|
| 296 |
+
plt.show()
|
| 297 |
+
except ImportError:
|
| 298 |
+
print("[INFO] matplotlib λΌμ΄λΈλ¬λ¦¬κ° μ€μΉλμ΄ μμ§ μμ μκ°ν λ¨κ³λ₯Ό 건λλλλ€.")
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 299 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
src/utils/ui_templates.py
CHANGED
|
@@ -225,7 +225,6 @@ body, .gradio-container {
|
|
| 225 |
/* ββ μ±λ΄ 컨ν
μ΄λ ν
λ리 μ κ±° ββ */
|
| 226 |
div[data-testid="chatbot"], .chatbot-container, .chatbot {
|
| 227 |
border: none !important;
|
| 228 |
-
overflow: visible !important;
|
| 229 |
}
|
| 230 |
|
| 231 |
/* ββ μ±λ΄ λ΄λΆ Placeholder(μκ°κΈ μμ) μλ¨ μ§€λ¦Ό μꡬ μ°¨λ¨ ββ */
|
|
@@ -408,21 +407,16 @@ label.svelte-1ipelgc, span.svelte-1ipelgc {
|
|
| 408 |
}
|
| 409 |
|
| 410 |
/* ββ μ¬μ©μ λ²λΈ (λ€ν¬κ·Έλ μ΄ ν리미μ ν
λ§) ββ */
|
| 411 |
-
.message.user
|
| 412 |
background-color: #111827 !important;
|
| 413 |
border-radius: 12px 12px 0 12px !important;
|
| 414 |
-
padding:
|
| 415 |
margin: 2px 0 !important;
|
| 416 |
border: none !important;
|
| 417 |
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.08) !important;
|
| 418 |
-
|
| 419 |
-
}
|
| 420 |
-
[data-testid="user"] > div, .bubble-wrap [data-testid="user"], .message-wrap.user > div, .message-row.user {
|
| 421 |
-
background: transparent !important;
|
| 422 |
-
background-color: transparent !important;
|
| 423 |
-
border: none !important;
|
| 424 |
}
|
| 425 |
-
|
| 426 |
color: #ffffff !important;
|
| 427 |
line-height: 1.4 !important;
|
| 428 |
margin: 0 !important;
|
|
@@ -430,7 +424,7 @@ label.svelte-1ipelgc, span.svelte-1ipelgc {
|
|
| 430 |
}
|
| 431 |
|
| 432 |
/* ββ λ΄ λ²λΈ (νμ΄νΈ & κ·Έλ μ΄ κ²½κ³μ ν
λ§) ββ */
|
| 433 |
-
.message.bot
|
| 434 |
background-color: #ffffff !important;
|
| 435 |
color: #1f2937 !important;
|
| 436 |
border: 1px solid #e5e7eb !important;
|
|
@@ -438,8 +432,7 @@ label.svelte-1ipelgc, span.svelte-1ipelgc {
|
|
| 438 |
padding: 16px 20px !important;
|
| 439 |
box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.05) !important;
|
| 440 |
}
|
| 441 |
-
.message.bot
|
| 442 |
-
.message.bot li, .message.bot div {
|
| 443 |
color: #1e3a5f !important;
|
| 444 |
background: transparent !important;
|
| 445 |
}
|
|
|
|
| 225 |
/* ββ μ±λ΄ 컨ν
μ΄λ ν
λ리 μ κ±° ββ */
|
| 226 |
div[data-testid="chatbot"], .chatbot-container, .chatbot {
|
| 227 |
border: none !important;
|
|
|
|
| 228 |
}
|
| 229 |
|
| 230 |
/* ββ μ±λ΄ λ΄λΆ Placeholder(μκ°κΈ μμ) μλ¨ μ§€λ¦Ό μꡬ μ°¨λ¨ ββ */
|
|
|
|
| 407 |
}
|
| 408 |
|
| 409 |
/* ββ μ¬μ©μ λ²λΈ (λ€ν¬κ·Έλ μ΄ ν리미μ ν
λ§) ββ */
|
| 410 |
+
.message.user {
|
| 411 |
background-color: #111827 !important;
|
| 412 |
border-radius: 12px 12px 0 12px !important;
|
| 413 |
+
padding: 10px 14px !important;
|
| 414 |
margin: 2px 0 !important;
|
| 415 |
border: none !important;
|
| 416 |
box-shadow: 0 2px 5px rgba(0, 0, 0, 0.08) !important;
|
| 417 |
+
color: #ffffff !important;
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 418 |
}
|
| 419 |
+
.message.user * {
|
| 420 |
color: #ffffff !important;
|
| 421 |
line-height: 1.4 !important;
|
| 422 |
margin: 0 !important;
|
|
|
|
| 424 |
}
|
| 425 |
|
| 426 |
/* ββ λ΄ λ²λΈ (νμ΄νΈ & κ·Έλ μ΄ κ²½κ³μ ν
λ§) ββ */
|
| 427 |
+
.message.bot {
|
| 428 |
background-color: #ffffff !important;
|
| 429 |
color: #1f2937 !important;
|
| 430 |
border: 1px solid #e5e7eb !important;
|
|
|
|
| 432 |
padding: 16px 20px !important;
|
| 433 |
box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.05) !important;
|
| 434 |
}
|
| 435 |
+
.message.bot * {
|
|
|
|
| 436 |
color: #1e3a5f !important;
|
| 437 |
background: transparent !important;
|
| 438 |
}
|