Spaces:
Runtime error
Runtime error
hellorahulk
commited on
Commit
·
7ef4196
1
Parent(s):
08e59e6
Create parse_website.py
Browse files- steps/parse_website.py +41 -0
steps/parse_website.py
ADDED
@@ -0,0 +1,41 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
from abc import ABC, abstractmethod
|
2 |
+
import pandas as pd
|
3 |
+
|
4 |
+
class ParseWebsite(ABC):
|
5 |
+
'''
|
6 |
+
Abstract Class for parsing website
|
7 |
+
'''
|
8 |
+
|
9 |
+
@abstractmethod
|
10 |
+
def get_table(html: str) -> pd.DataFrame:
|
11 |
+
'''
|
12 |
+
Args:
|
13 |
+
html: string
|
14 |
+
Returns:
|
15 |
+
pd.DataFrame: Competition records
|
16 |
+
'''
|
17 |
+
pass
|
18 |
+
|
19 |
+
|
20 |
+
class ParseCraft(ParseWebsite):
|
21 |
+
'''
|
22 |
+
Parse craft.co website for competitor intel
|
23 |
+
'''
|
24 |
+
|
25 |
+
def get_table(self,html: str) -> pd.DataFrame:
|
26 |
+
html_tables = pd.read_html(html, index_col=0)[0]
|
27 |
+
return html_tables
|
28 |
+
|
29 |
+
|
30 |
+
class ParseLink:
|
31 |
+
'''
|
32 |
+
Parses Website
|
33 |
+
'''
|
34 |
+
def __init__(self,website_type: str,html: str):
|
35 |
+
self.website_type = website_type
|
36 |
+
self.html = html
|
37 |
+
|
38 |
+
def parse(self):
|
39 |
+
if(self.website_type == "craft"):
|
40 |
+
df = ParseCraft().get_table(self.html)
|
41 |
+
return df
|