From: Christian Heller <c.heller@plomlompom.de>
Date: Sun, 26 Jul 2020 17:52:18 +0000 (+0200)
Subject: Grow scraper date range.
X-Git-Url: https://plomlompom.com/repos/%7B%7Bprefix%7D%7D/process?a=commitdiff_plain;h=6238ffaeaaacf95a632b691ec1529966cd92a381;p=berlin-corona-table

Grow scraper date range.
---

diff --git a/scrape.py b/scrape.py
index e449b99..de458f4 100755
--- a/scrape.py
+++ b/scrape.py
@@ -25,8 +25,9 @@ abbrevs = {
   'sum': {'Summe', 'Berlin'},
 }
 
-# Here only image files are available for the table data.
-unparsable_graphics_fallback = {
+# some pre-filled values
+data = {
+   # For these, only image files are available for the table data.
     datetime.datetime(2020, 7, 2): {
         'CW': {'growth': 4, 'total': 851},
         'FK': {'growth': 10, 'total': 681},
@@ -86,7 +87,24 @@ unparsable_graphics_fallback = {
         'TS': {'growth': 41, 'total': 362},
         'TK': {'growth': 14, 'total': 183},
         'sum': {'growth': 263, 'total': 3486}
-    }
+    },
+   # This one has no press release but can be reconstructed from
+   # the neighbour ones.
+   datetime.datetime(2020, 3, 13): {
+        'CW': {'growth': 16, 'total': 47},
+        'FK': {'growth': 8, 'total': 22},
+        'Li': {'growth': 2, 'total': 8},
+        'MH': {'growth': 1, 'total': 4},
+        'Mi': {'growth': 9, 'total': 29},
+        'Ne': {'growth': 6, 'total': 16},
+        'Pa': {'growth': 11, 'total': 26},
+        'Re': {'growth': 0, 'total': 11},
+        'Sp': {'growth': 1, 'total': 9},
+        'SZ': {'growth': 0, 'total': 20},
+        'TS': {'growth': 1, 'total': 17},
+        'TK': {'growth': 3, 'total': 7},
+        'sum': {'growth': 58, 'total': 216}
+   }
 }
 fixes = {
    # Here the official total is 215, while the summation of district
@@ -146,11 +164,10 @@ for i in range(max_page):
         day_urls += [link['href']]
 
 # Collect infection data.
-data = {}
 first_run = True
 districts_sorted = []
 # TODO: Push limit further back (might need more data fixes for that).
-date_limit = datetime.datetime(2020, 3, 16)
+date_limit = datetime.datetime(2020, 3, 12)
 for path in day_urls:
     url = url_prefix + path
     with urllib.request.urlopen(url) as response:
@@ -162,14 +179,16 @@ for path in day_urls:
     date = datetime.datetime.strptime(date_formatted , '%d.%m.%Y')
     if date_limit > date:
         break
-    if date in data:
-        raise Exception('Double date %s', date)
-        #date -= datetime.timedelta(days=1)
-    data[date] = {}
+    # On that day, two press releases were released, for that and the prev day.
+    if date == datetime.datetime(2020, 3, 15) and date in data:
+       date = datetime.datetime(2020, 3, 14)
+    # From here on, press releases describe numbers from prev day.
+    if date <= datetime.datetime(2020, 3, 13):
+       date = date - datetime.timedelta(days=1)
     table = soup.find('table')
-    if table is None:
-        data[date] = unparsable_graphics_fallback[date]
+    if table is None and date in data:
         continue
+    data[date] = {}
     for tr in [tr for tr in table.children if type(tr) == bs4.element.Tag][1:]:
         printable_tds = []
         for td in [td for td in tr.children if type(td) == bs4.element.Tag][:2]:
@@ -206,7 +225,7 @@ for date in dates_sorted:
               raise Exception('Dates not contiguous: %s missing', prev_date)
            else:
               continue
-        prev_total = data[date - datetime.timedelta(days=1)][district]['total']
+        prev_total = data[prev_date][district]['total']
         cur_total = data[date][district]['total']
         if cur_total - data[date][district]['growth'] != prev_total:
             raise Exception('Questionable district infection total in %s/%s' % (district, date))