From 4e00f0866dcdde575b867e0d386f6d50d4b61a8d Mon Sep 17 00:00:00 2001
From: Frank Elsinga <frank@elsinga.de>
Date: Sat, 25 Mar 2023 20:05:05 +0100
Subject: [PATCH] [autoscraping-4] Agressive guessing for roomfinder rooms
 (#463)

* Fixed a runtime bug which sneaked in

* scraped all new data

* made the room guessing more agressive

* added the previously missing room
---
 data/external/results/rooms_roomfinder.json | 23 +++++++++++++++++++++
 data/external/scrapers/roomfinder.py        | 15 +++++++++++---
 2 files changed, 35 insertions(+), 3 deletions(-)

diff --git a/data/external/results/rooms_roomfinder.json b/data/external/results/rooms_roomfinder.json
index de0782e80..36308f17e 100644
--- a/data/external/results/rooms_roomfinder.json
+++ b/data/external/results/rooms_roomfinder.json
@@ -6758,6 +6758,29 @@
     "utm_northing": 5336203.6581,
     "utm_zone": "32"
   },
+  {
+    "b_alias": "U-Trakt",
+    "b_area": "M\u00fcnchen - Stammgel\u00e4nde Nord",
+    "b_id": "0101",
+    "b_name": "N1",
+    "default_map": ["4000", 12, "Stammgel\u00e4nde Basiskarte", 501, 484],
+    "maps": [
+      ["500", 65, "U-Trakt UG", 525, 443],
+      ["4000", 12, "Stammgel\u00e4nde Basiskarte", 501, 484],
+      ["5000", 10, "Lageplan TUM", 657, 951],
+      ["200000", 54, "M\u00fcnchen", 640, 603],
+      ["400000", 156, "M\u00fcnchen und Umgebung", 420, 515],
+      ["100000000", 9, "Weltkarte Norden", 1619, 396]
+    ],
+    "metas": [],
+    "r_alias": "FLUR",
+    "r_id": "NK111B@0101",
+    "r_level": "-1",
+    "r_number": "NK111B",
+    "utm_easting": 691057.0769,
+    "utm_northing": 5336198.9831,
+    "utm_zone": "32"
+  },
   {
     "b_alias": "U-Trakt",
     "b_area": "M\u00fcnchen - Stammgel\u00e4nde Nord",
diff --git a/data/external/scrapers/roomfinder.py b/data/external/scrapers/roomfinder.py
index 0e3ac6fb9..31cf46c4a 100644
--- a/data/external/scrapers/roomfinder.py
+++ b/data/external/scrapers/roomfinder.py
@@ -98,7 +98,7 @@ def scrape_rooms():
 
 
 def _guess_queries(rooms, n_rooms):
-    # First try: all single-digit numbers
+    # First try: any single-digit number?
     for i in range(10):
         if len(rooms) < n_rooms:
             maybe_sleep(0.05)
@@ -106,7 +106,7 @@ def _guess_queries(rooms, n_rooms):
         else:
             return
 
-    # Second try: all double-digit numbers
+    # Second try: any double-digit number?
     for i in range(100):
         if len(rooms) < n_rooms:
             maybe_sleep(0.05)
@@ -114,7 +114,7 @@ def _guess_queries(rooms, n_rooms):
         else:
             return
 
-    # Third try: all characters
+    # Third try: any character?
     for char in string.ascii_lowercase:
         if len(rooms) < n_rooms:
             maybe_sleep(0.05)
@@ -122,6 +122,15 @@ def _guess_queries(rooms, n_rooms):
         else:
             return
 
+    # Fourth try: any character twice?
+    for c1 in string.ascii_lowercase:
+        for c2 in string.ascii_lowercase:
+            if len(rooms) < n_rooms:
+                maybe_sleep(0.05)
+                yield c1 + c2
+            else:
+                return
+
 
 @cached_json("maps_roomfinder.json")
 def scrape_maps():