Add caching support

.gitignore

@@ -1,3 +1,6 @@
+# Cache files
+url_map.pickle
+
 # Output graphs
 output/
 

src/__main__.py

@@ -1,5 +1,7 @@
 import asyncio
+import pickle
 import re
+from pathlib import Path
 from time import perf_counter
 
 import httpx
@@ -10,6 +12,7 @@ from src.visualization import display_top_urls
 
 URL = httpx.URL("https://ailab.fai.utb.cz")
 ALLOWED_HOSTS_RE = re.compile(r"(?:.*\.)?utb\.cz")
+CACHE_RESULTS_DIR = Path.cwd() / "url_map.pickle"
 
 
 async def main() -> None:
@@ -21,6 +24,11 @@ async def main() -> None:
     print("Scraping...")
     start = perf_counter()
 
+    if CACHE_RESULTS_DIR.exists():
+        print("> Using cached URLs")
+        with CACHE_RESULTS_DIR.open("rb") as fp:
+            url_map = pickle.load(fp)  # noqa: S301
+    else:
         async with httpx.AsyncClient() as client:
             url_map = await get_urlmap(
                 client,
@@ -30,6 +38,9 @@ async def main() -> None:
                 suppress_exception=standard_urlmap_exception_suppressor,
             )
 
+        with CACHE_RESULTS_DIR.open("wb") as fp:
+            pickle.dump(url_map, fp)
+
     took = perf_counter() - start
     print(f"Took: {round(took, 2)} seconds")