I am converting some of my web-scraping code from R to Python (I can't get geckodriver to work with R, but it's working with Python). Anyways, I am trying to understand how to parse and read HTML tables with Python. Quick background, here is my code for R:
doc <- htmlParse(remDr$getPageSource()[[1]],ignoreBlanks=TRUE, replaceEntities = FALSE, trim=TRUE, encoding="UTF-8")
WebElem <- readHTMLTable(doc, stringsAsFactors = FALSE)[[7]]
I would parse the HTML page to the doc object. Then I would start with doc[[1]]
, and move through higher numbers until I saw the data I wanted. In this case I got to doc[[7]]
and saw the data I wanted. I then would read that HTML table and assign it to the WebElem object. Eventually I would turn this into a dataframe and play with it.
So what I am doing in Python is this:
html = None
doc = None
html = driver.page_source
doc = BeautifulSoup(html)
Then I started to play with doc.get_text
but I don't really know how to get just the data I want to see. The data I want to see is like a 10x10 matrix. When I used R, I would just use doc[[7]]
and that matrix would almost be in a perfect structure for me to convert it to a dataframe. However, I just can't seem to do that with Python. Any advice would be much appreciated.
UPDATE:
I have been able to get the data I want using Python--I followed this blog for creating a dataframe with python: Python Web-Scraping. Here is the website that we are scraping in that blog: Most Popular Dog Breeds. In that blog post, you have to work your way through the elements, create a dict, loop through each row of the table and store the data in each column, and then you are able to create a dataframe.
With R, the only code I had to write was:
doc <- htmlParse(remDr$getPageSource()[[1]],ignoreBlanks=TRUE, replaceEntities = FALSE, trim=TRUE, encoding="UTF-8")
df <- as.data.frame(readHTMLTable(doc, stringsAsFactors = FALSE)
With just that, I have a pretty nice dataframe that I only need to adjust the column names and data types--it looks like this with just that code:
NULL.V1 NULL.V2 NULL.V3 NULL.V4
1 BREED 2015 2014 2013
2 Retrievers (Labrador) 1 1 1
3 German Shepherd Dogs 2 2 2
4 Retrievers (Golden) 3 3 3
5 Bulldogs 4 4 5
6 Beagles 5 5 4
7 French Bulldogs 6 9 11
8 Yorkshire Terriers 7 6 6
9 Poodles 8 7 8
10 Rottweilers 9 10 9
Is there not something available in Python to make this a bit simpler, or is this just simpler in R because R is more built for dataframes(at least that's how it seems to me, but I could be wrong)?
Ok, after some hefty digging around I feel like I came to good solution--matching that of R. If you are looking at the HTML provided in the link above, Dog Breeds, and you have the web driver running for that link you can run the following code:
tbl = driver.find_element_by_xpath("//html/body/main/article/section[2]/div/article/table").get_attribute('outerHTML')
df = pd.read_html(tbl)
Then you are looking a pretty nice dataframe after only a couple lines of code:
In [145]: df
Out[145]:
[ 0 1 2 3
0 BREED 2015 2014 2013.0
1 Retrievers (Labrador) 1 1 1.0
2 German Shepherd Dogs 2 2 2.0
3 Retrievers (Golden) 3 3 3.0
4 Bulldogs 4 4 5.0
5 Beagles 5 5 4.0
I feel like this is much easier than working through the tags, creating a dict, and looping through each row of data as the blog suggests. It might not be the most correct way of doing things, I'm new to Python, but it gets the job done quickly. I hope this helps out some fellow web-scrapers.