The Dormouse’s story

Parsing HTML Tables in Python with BeautifulSoup and pandas

Something that seems daunting at first when switching from R to Python is replacing all the ready-made functions R has. For example, R has a nice CSV reader out of the box. Python users will eventually find pandas, but what about other R libraries like their HTML Table Reader from the xml package? That’s very helpful for scraping web pages, but in Python it might take a little more work. So in this post, we’re going to write a brief but robust HTML table parser.

Writing a Table Scraper

Our parser is going to be built on top of the Python package BeautifulSoup. It’s a convenient package and easy to use. Our use will focus on the “find_all” function, but before we start parsing, you need to understand the basics of HTML terminology.

An HTML object consists of a few fundamental pieces: a tag. The format that defines a tag is

and it could have attributes which consistes of a property and a value. A tag we are interested in is the table tag, which defined a table in a website. This table tag has many elements. An element is a component of the page which typically contains content. For a table in HTML, they consist of rows designated by elements within the tr tags, and then column content inside the td tags. A typical example is

It turns out that most sites keep data you’d like to scrape in tables, and so we’re going to learn to parse them.

Parsing a Table in BeautifulSoup

To parse the table, we are going to use the Python library BeautifulSoup. It constructs a tree from the HTML and gives you an API to access different elements of the webpage.

Let’s say we already have our table object returned from BeautifulSoup. To parse the table, we’d like to grab a row, take the data from its columns, and then move on to the next row ad nauseam. In the next bit of code, we define a website that is simply the HTML for a table. We load it into BeautifulSoup and parse it, returning a pandas data frame of the contents.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
 
import pandas as pd from bs4 import BeautifulSoup html_string = ''' 
 
 
Hello!
 
Table
 
 
 ''' soup = BeautifulSoup(html_string, 'lxml') # Parse the HTML as a string table = soup.find_all('table')[0] # Grab the first table new_table = pd.DataFrame(columns=range(0,2), index = [0]) # I know the size row_marker = 0 for row in table.find_all('tr'): column_marker = 0 columns = row.find_all('td') for column in columns: new_table.iat[row_marker,column_marker] = column.get_text() column_marker += 1 new_table
 
As you can see, we grab all the tr elements from the table, followed by grabbing the td elements one at a time. We use the “get_text()” method from the td element (called a column in each iteration) and put it into our python object representing a table (it will eventually be a pandas dataframe).
 
Using Requests to Access a Web Content
 
Now, that we have our plan to parse a table, we probably need to figure out how to get to that point. That’s actually easier! We’re going to use the requests package in Python.
 
import requests url = "https://www.fantasypros.com/nfl/reports/leaders/qb.php?year=2015" response = requests.get(url) response.text[:100] # Access the HTML with the text property
 
'\r\n \n lang="en">\n\n \n Fantasy Football Leaders Weeks 1 to 17 - QB' 
 
The Parser Object
 
So, now we can define our HTML table parser object. You’ll notice we added more bells and whistles to the html table parser. To summarize the functionality outside of basic parsing:
 
1. We take th elements and use them as column names. 2. We cast any column with numbers to float. 3. We also return a list of tuples for each table in the page. 
 
The tuples we return are in the form (table id, parsed table) for every table in the document.
 
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
 
import requests import pandas as pd from bs4 import BeautifulSoup class HTMLTableParser: def parse_url(self, url): response = requests.get(url) soup = BeautifulSoup(response.text, 'lxml') return [(table['id'],self.parse_html_table(table))\ for table in soup.find_all('table')] def parse_html_table(self, table): n_columns = 0 n_rows=0 column_names = [] # Find number of rows and columns # we also find the column titles if we can for row in table.find_all('tr'): # Determine the number of rows in the table td_tags = row.find_all('td') if len(td_tags) > 0: n_rows+=1 if n_columns == 0: # Set the number of columns for our table n_columns = len(td_tags) # Handle column names if we find them th_tags = row.find_all('th') if len(th_tags) > 0 and len(column_names) == 0: for th in th_tags: column_names.append(th.get_text()) # Safeguard on Column Titles if len(column_names) > 0 and len(column_names) != n_columns: raise Exception("Column titles do not match the number of columns") columns = column_names if len(column_names) > 0 else range(0,n_columns) df = pd.DataFrame(columns = columns, index= range(0,n_rows)) row_marker = 0 for row in table.find_all('tr'): column_marker = 0 columns = row.find_all('td') for column in columns: df.iat[row_marker,column_marker] = column.get_text() column_marker += 1 if len(columns) > 0: row_marker += 1 # Convert to float if possible for col in df: try: df[col] = df[col].astype(float) except ValueError: pass return df
 
Usage Example
 
Let’s do an example where we scrape a table from a website. We initialize the parser object and grab the table using our code above:
 
hp = HTMLTableParser() table = hp.parse_url(url)[0][1] # Grabbing the table from the tuple table.head()
 
 
 
Rank
 
Player
 
Team
 
Points
 
Games
 
Avg
 
  
 
 
0
 
1
 
Cam Newton
 
CAR
 
389.1
 
16
 
24.3
 
 
 
1
 
2
 
Tom Brady
 
NE
 
343.7
 
16
 
21.5
 
 
 
2
 
3
 
Russell Wilson
 
SEA
 
336.4
 
16
 
21.0
 
 
 
3
 
4
 
Blake Bortles
 
JAC
 
316.1
 
16
 
19.8
 
 
 
4
 
5
 
Carson Palmer
 
ARI
 
309.2
 
16
 
19.3
 
 
 
 
If you had looked at the URL above, you’d have seen that we were parsing QB stats from the 2015 season off of FantasyPros.com. Our data has been prepared in such a way that we can immediately start an analysis.
 
%matplotlib inline import matplotlib.pyplot as plt plt.figure() avg=table['Avg'].values plt.hist(avg, bins = 50) plt.title('Average QB Points Per Game in 2015')
 
 
Final Thoughts
 
As you can see, this code may find it’s way into some scraper scripts once Football season starts again, but it’s perfectly capable of scraping any page with an HTML table. The code actually will scrape every table on a page, and you can just select the one you want from the resulting list. Happy scraping!
 
Источник
 
How to Convert HTML to Text with Python and Pandas
 
In this short guide, we'll see how to convert HTML to raw text with Python and Pandas. It is also known as text extraction from HTML tags.
 
2. Setup
 
In this Python guide, we'll use the following DataFrame, which consists of two columns. Column html contains HTML tags and text inside the tags:
 
import pandas as pd data = Hello", "
Use applymap
", "
Code:
 df.head()" ]> df = pd.DataFrame.from_dict(data) 
 
 
 
title
 
html
 
  
 
 
0
 
Intro
 
Hello
 
 
 
1
 
Code
 
 
Use applymap
 
 
 
 
2
 
Code Text
 
 
Code:
 
 df.head()
 
 
 
 
We would like to extract the raw text from the column without the HTML tags with Python:
 
 
 
html
 
  
 
 
0
 
Hello
 
 
 
1
 
Use applymap
 
 
 
2
 
Code: df.head()
 
 
 
 
Step 1: Install Beautiful Soup library
 
First we will need to install Python library - beautifulsoup4 by:
 
pip install beautifulsoup4 
 
The official documentation of the library is available on this link: Beautiful Soup Documentation.
 
The Beautiful Soup is described as:
 
 
Beautiful Soup is a Python library for pulling data out of HTML and XML files.
 
 
Step 2: Extract text from HTML tags by Python
 
Now let's check how we can extract the text from HTML code or tags in Python. We will demonstrate the extraction in simple example.
 
Suppose we have the following HTML document:
 
html_doc = """ 
 
 

We can extract the raw text from the HTML tags by:
from bs4 import BeautifulSoup soup = BeautifulSoup(html_doc, 'html.parser') print(soup.get_text()) 
The Dormouse's story The Dormouse's story Once upon a time there were three little sisters; and their names were 
Or we can pretty print the HTML code by:
Step 3: HTML to raw text in Pandas
In order to convert HTML to raw text we will apply BeautifulSoup library to Pandas column.
To apply the BeautifulSoup function soup.get_text() to Pandas column we can use the following code:
df[['html']].applymap(lambda text: BeautifulSoup(text, 'html.parser').get_text()) 


html




0
Hello


1
Use applymap


2
Code: df.head()



How does it work? We are applying the function .get_text() with html.parser to each row from the DataFrame - df[['html']] - in this case it has only a single column.
If we pass non HTML column or NaNs we will get errors.
Conclusion
In this post, we saw how to convert HTML tags or document into raw text with Python and Pandas.
We've learn also how to apply BeautifulSoup library function to Pandas DataFrame.
The image below shows all the steps which we covered:

By using DataScientYst - Data Science Simplified, you agree to our Cookie Policy. 
Источник