added timestamp and remove all ';' from data.

diskprices.py

@@ -15,6 +15,7 @@ import argparse
 import html
 import httpx
 from bs4 import BeautifulSoup
+import re
 import csv
 import json
 
@@ -28,7 +29,7 @@ def fetch(date, target="https://diskprices.com", timeout=30.0):
     response = httpx.get(url, follow_redirects=True, timeout=timeout)
     return response.text
 
-def parse(page):
+def parse(page, date=None, timestamp=False):
     """
     parse diskprices html page and convert table to a dict.
     """
@@ -41,6 +42,8 @@ def parse(page):
         headers = line.find_all('th')
         if headers:
             data = list(map(lambda x: string_converter(x.text), headers))
+            if timestamp:
+                data.insert(0, "Date")
             result.append(data)
             next
 
@@ -48,8 +51,11 @@ def parse(page):
         items = line.find_all('td')
         if items:
             data = list(map(lambda x: string_converter(x.text), items))
+            if timestamp:
+                data.insert(0, date)
             result.append(data)
             next
+                
     return result
 
 def string_converter(text):
@@ -59,7 +65,8 @@ def string_converter(text):
     entities = html.unescape(text)
     cleanup = entities.encode("ascii", "ignore")
     utf8 = str(cleanup, 'utf-8')
-    return utf8
+    cleanup_sep = re.sub(";", "", utf8)
+    return cleanup_sep
 
 def to_csv(parsed):
     """
@@ -91,10 +98,13 @@ if __name__ == '__main__':
     )
     parser.add_argument('-d', '--date', required=True)
     parser.add_argument('-o', '--output', default="csv")
+    parser.add_argument('-t', '--timestamp', default=False, action=argparse.BooleanOptionalAction)
     # parser.add_argument('-t', '--timeout')
     args = parser.parse_args()
-    page = fetch(args.date)
-    parsed = parse(page)
+    date = args.date
+    timestamp = args.timestamp
+    page = fetch(date)
+    parsed = parse(page, date=date, timestamp=timestamp)
     if args.output == "json":
         print(to_json(parsed))
     else: