README.md

@@ -3,10 +3,6 @@
 `apt install python3-pydantic python3-httpx python3-bs4 python3-tqdm locales-all && sudo locale-gen`
 
 # Windows build
-1. Install wine
-2. `winecfg` and configure windows 10
-3. Install [python](https://www.python.org/downloads/windows/)
-4. Install pyinstaller in wine `wine pip install pyinstaller`
-5. Create venv and `pip install .`
-6. `wine pyinstaller --paths=venv/lib/python3.10/site-packages/ --onefile scraperoog/scrape.py`
-7. .exe is found under dist/
+1. install wine
+2. create venv and `pip install .`
+3. `wine pyinstaller --paths=venv/lib/python3.10/site-packages/ --collect-submodules=lxml --onefile scraperoog/scrape.py`

config.ini

@@ -3,4 +3,3 @@ haupt_url = https://www.spiekeroog-vermieter.de/suche/monatskalenderSite.htm?woh
 data_url = https://www.spiekeroog-vermieter.de/suche/monatskalender.htm?wohnids=
 von_id = 200
 bis_id = 300
-id_fehler = nicht aktiv

pyproject.toml

@@ -1,9 +1,10 @@
 [project]
 name = "scraperoog"
-version = "0.0.5"
+version = "0.0.2"
 dependencies = [
     "tqdm",
     "bs4",
+    "lxml",
     "httpx",
     "pydantic",
 ]

scraperoog/scrape.py

@@ -1,11 +1,10 @@
 import asyncio
 import csv
+import json
 import locale
+locale.setlocale(locale.LC_TIME, "German") # dates on that page are German
+import pickle
 import platform
-if platform.system() == "Windows":
-    locale.setlocale(locale.LC_TIME, "German")
-else:
-    locale.setlocale(locale.LC_TIME, "de_DE.utf_8")
 from configparser import ConfigParser
 from datetime import datetime
 from typing import List, Optional, Tuple
@@ -24,7 +23,6 @@ MAIN_URL = config['Allgemein'].get('haupt_url')
 DATA_URL = config['Allgemein'].get('data_url')
 FROM = config['Allgemein'].getint('von_id')
 TO = config['Allgemein'].getint('bis_id')
-ID_NONEXISTANT = config['Allgemein'].get('id_fehler')
 DATEFORMAT = "%Y-%m-%d"
 STATUS_MAPPING = {"DayF": 0, "DayB": 1, "DayFB": 0.5, "DayBF": 0.5}
 
@@ -83,34 +81,35 @@ def convert_to_datestring(day: str, month: str, year: str) -> datetime:
 
 async def request_data(index: int, client: AsyncClient) -> Optional[Entry]:
     response_data = await client.get(DATA_URL + str(index), timeout=20.0)
-    if ID_NONEXISTANT in response_data.text:
-        return Entry(index=0)
-    response_title = await client.get(MAIN_URL + str(index), timeout=20.0)
-    title_soup = BeautifulSoup(response_title.text, "html.parser")
-    apartment = (
-        title_soup.body.header.h2.get_text()
-        .replace("\xa0", " ")
-        .replace("Wohneinheit: ", "")
-    )
-
-    data_soup = BeautifulSoup(response_data.text, "html.parser")
-    valid_element = data_soup.find_all("td", attrs={"data-daynum": True})
-    availabilities = []
-    for elm in valid_element:
-        date_raw = elm.parent.parent.parent.thead.tr.td.get_text(strip=True)
-        status = STATUS_MAPPING[elm["class"][0]]
-        date = convert_to_datestring(
-            elm.get_text(strip=True),
-            date_raw.split(" ")[0],
-            date_raw.split(" ")[1],
+    if "Die Darstellung ist derzeit deaktiviert" not in response_data.text:
+        response_title = await client.get(MAIN_URL + str(index), timeout=20.0)
+        title_soup = BeautifulSoup(response_title.text, "lxml")
+        apartment = (
+            title_soup.body.header.h2.get_text()
+            .replace("\xa0", " ")
+            .replace("Wohneinheit: ", "")
         )
-        availabilities.append(Availability(date=date, status=status))
-    return Entry(
-        index=index,
-        haus=title_soup.body.header.h1.get_text().encode("utf-8"),
-        wohneinheit=apartment.encode("utf-8"),
-        availabilities=availabilities,
-    )
+
+        data_soup = BeautifulSoup(response_data.text, "lxml")
+        valid_element = data_soup.find_all("td", attrs={"data-daynum": True})
+        availabilities = []
+        for elm in valid_element:
+            date_raw = elm.parent.parent.parent.thead.tr.td.get_text(strip=True)
+            status = STATUS_MAPPING[elm["class"][0]]
+            date = convert_to_datestring(
+                elm.get_text(strip=True),
+                date_raw.split(" ")[0],
+                date_raw.split(" ")[1],
+            )
+            availabilities.append(Availability(date=date, status=status))
+        return Entry(
+            index=index,
+            haus=title_soup.body.header.h1.get_text().encode("utf-8"),
+            wohneinheit=apartment.encode("utf-8"),
+            availabilities=availabilities,
+        )
+    else:
+        return Entry(index=0)
 
 
 async def extract_results() -> None: