Merge pull request #18 from claromes/cs-cz-locale

Add cs-CZ locale

README.md

@@ -87,6 +87,7 @@ volleystats [--help] --fed FED (--match MATCH | --comp COMP) [--log]
 
 - pt-BR
 - en-GB
+- cs-CZ
 
 ## Docs
 

setup.py

@@ -15,10 +15,10 @@
     name='volleystats',
     version=version,
     author='Claromes',
-    description='CLI tool to get volleyball statistics from the Data Project Web Competition websites (WCM)',
+    description='Command-line tool to scrape volleyball statistics from Data Project Web Competition websites',
     long_description=long_description,
     long_description_content_type='text/markdown',
-    keywords='volleyball sports cli dataset analytics',
+    keywords='volleyball sports command-line dataset analytics',
     url='https://github.com/claromes/volleystats',
     project_urls={
         'Documentation': 'https://claromes.github.io/volleystats',

volleystats/spiders/match.py

@@ -21,11 +21,14 @@ def parse(self, response):
 
         ptBR = response.xpath("//*[contains(@class, 'RCB_Culture_pt-BR')]/span/input/@value").get()
         enGB = response.xpath("//*[contains(@class, 'RCB_Culture_en-GB')]/span/input/@value").get()
+        csCZ = response.xpath("//*[contains(@class, 'RCB_Culture_cs-CZ')]/span/input/@value").get()
 
         if ptBR == 'PT':
             match_date = parse_ptbr_date(match_date_text)
         elif enGB == 'EN':
             match_date = parse_engb_date(match_date_text)
+        elif csCZ == 'CZ':
+            match_date = parse_cscz_date(match_date_text)
 
         home_team_string = response.xpath("normalize-space(//span[@id='Content_Main_LBL_HomeTeam']/text())").get().replace(' ', '-').lower()
         home_team = re.sub('[^A-Za-z0-9]+', '-', home_team_string)
@@ -80,11 +83,14 @@ def parse(self, response):
 
         ptBR = response.xpath("//*[contains(@class, 'RCB_Culture_pt-BR')]/span/input/@value").get()
         enGB = response.xpath("//*[contains(@class, 'RCB_Culture_en-GB')]/span/input/@value").get()
+        csCZ = response.xpath("//*[contains(@class, 'RCB_Culture_cs-CZ')]/span/input/@value").get()
 
         if ptBR == 'PT':
             match_date = parse_ptbr_date(match_date_text)
         elif enGB == 'EN':
             match_date = parse_engb_date(match_date_text)
+        elif csCZ == 'CZ':
+            match_date = parse_cscz_date(match_date_text)
 
         guest_team_string = response.xpath("normalize-space(//span[@id='Content_Main_LBL_GuestTeam']/text())").get().replace(' ', '-').lower()
         guest_team = re.sub('[^A-Za-z0-9]+', '-', guest_team_string)

volleystats/utils.py

@@ -20,8 +20,6 @@ def parse_ptbr_date(date_string):
     date_re_1 = re.search(r'-(\w+)-', str_4)
     date_re_2 = date_re_1.group(1)
 
-    date_re_en = ''
-
     if date_re_2 == 'janeiro':
         date_re_en = 'january'
     elif date_re_2 == 'fevereiro':
@@ -61,4 +59,46 @@ def parse_engb_date(date_string):
 
     parsed_engb_date = datetime.strptime(str_2, '%d-%B-%Y').date()
 
-    return parsed_engb_date
+    return parsed_engb_date
+
+# cs-CZ: 'čtvrtek 26. října 2023 - 14:00' to 2023-10-26
+def parse_cscz_date(date_string):
+    str_1 = date_string.replace(' ', '-')
+    str_2 = str_1.rsplit('---', 1)[0]
+    str_3 = str_2.replace('.', '')
+    str_4 = str_3.split('-', 1)[1]
+
+    date_re_1 = re.search(r'-(\w+)-', str_4)
+    date_re_2 = date_re_1.group(1)
+
+    if date_re_2 == 'ledna':
+        date_re_en = 'January'
+    elif date_re_2 == 'února':
+        date_re_en = 'February'
+    elif date_re_2 == 'března':
+        date_re_en = 'March'
+    elif date_re_2 == 'dubna':
+        date_re_en = 'April'
+    elif date_re_2 == 'května':
+        date_re_en = 'May'
+    elif date_re_2 == 'června':
+        date_re_en = 'June'
+    elif date_re_2 == 'července':
+        date_re_en = 'July'
+    elif date_re_2 == 'srpna':
+        date_re_en = 'August'
+    elif date_re_2 == 'září':
+        date_re_en = 'September'
+    elif date_re_2 == 'října':
+        date_re_en = 'October'
+    elif date_re_2 == 'listopadu':
+        date_re_en = 'November'
+    elif date_re_2 == 'prosince':
+        date_re_en = 'December'
+
+    var_date_re_2 = f'{re.escape(date_re_2)}-'
+    parsed_date = re.sub(var_date_re_2, f'{date_re_en}-', str_4)
+
+    parsed_cscz_date = datetime.strptime(parsed_date, '%d-%B-%Y').date()
+
+    return parsed_cscz_date

volleystats/version.py

@@ -1 +1 @@
-__version__ = '0.5'
+__version__ = '0.5.1'