pandas read_html ValueError: No tables found

Question

I am trying to scrap the historical weather data from the "https://www.wunderground.com/personal-weather-station/dashboard?ID=KMAHADLE7#history/tdata/s20170201/e20170201/mcustom.html" weather underground page. I have the following code:

import pandas as pd 

page_link = 'https://www.wunderground.com/personal-weather-station/dashboard?ID=KMAHADLE7#history/tdata/s20170201/e20170201/mcustom.html'
df = pd.read_html(page_link)
print(df)

I have the following response:

Traceback (most recent call last):
 File "weather_station_scrapping.py", line 11, in <module>
  result = pd.read_html(page_link)
 File "/anaconda3/lib/python3.6/site-packages/pandas/io/html.py", line 987, in read_html
  displayed_only=displayed_only)
 File "/anaconda3/lib/python3.6/site-packages/pandas/io/html.py", line 815, in _parse raise_with_traceback(retained)
 File "/anaconda3/lib/python3.6/site-packages/pandas/compat/__init__.py", line 403, in raise_with_traceback
  raise exc.with_traceback(traceback)
ValueError: No tables found

Although, this page clearly has a table but it is not being picked by the read_html. I have tried using Selenium so that the page can be loaded before I read it.

from selenium import webdriver
from selenium.webdriver.common.keys import Keys

driver = webdriver.Firefox()
driver.get("https://www.wunderground.com/personal-weather-station/dashboard?ID=KMAHADLE7#history/tdata/s20170201/e20170201/mcustom.html")
elem = driver.find_element_by_id("history_table")

head = elem.find_element_by_tag_name('thead')
body = elem.find_element_by_tag_name('tbody')

list_rows = []

for items in body.find_element_by_tag_name('tr'):
    list_cells = []
    for item in items.find_elements_by_tag_name('td'):
        list_cells.append(item.text)
    list_rows.append(list_cells)
driver.close()

Now, the problem is that it cannot find "tr". I would appreciate any suggestions.

Solution 1

You can use requests and avoid opening browser.

You can get current conditions by using:

https://stationdata.wunderground.com/cgi-bin/stationlookup?station=KMAHADLE7&units=both&v=2.0&format=json&callback=jQuery1720724027235122559_1542743885014&_=15

and strip of 'jQuery1720724027235122559_1542743885014(' from the left and ')' from the right. Then handle the json string.

You can get summary and history by calling the API with the following

https://api-ak.wunderground.com/api/606f3f6977348613/history_20170201null/units:both/v:2.0/q/pws:KMAHADLE7.json?callback=jQuery1720724027235122559_1542743885015&_=1542743886276

You then need to strip 'jQuery1720724027235122559_1542743885015(' from the front and ');' from the right. You then have a JSON string you can parse.

Sample of JSON:

You can find these URLs by using F12 dev tools in browser and inspecting the network tab for the traffic created during page load.

An example for current, noting there seems to be a problem with nulls in the JSON so I am replacing with "placeholder":

import requests
import pandas as pd
import json
from pandas.io.json import json_normalize
from bs4 import BeautifulSoup

url = 'https://stationdata.wunderground.com/cgi-bin/stationlookup?station=KMAHADLE7&units=both&v=2.0&format=json&callback=jQuery1720724027235122559_1542743885014&_=15'
res = requests.get(url)
soup = BeautifulSoup(res.content, "lxml")
s = soup.select('html')[0].text.strip('jQuery1720724027235122559_1542743885014(').strip(')')
s = s.replace('null','"placeholder"')
data= json.loads(s)
data = json_normalize(data)
df = pd.DataFrame(data)
print(df)

Solution 2

Here's a solution using selenium for browser automation

from selenium import webdriver
import pandas as pd
driver = webdriver.Chrome(chromedriver)
driver.implicitly_wait(30)

driver.get('https://www.wunderground.com/personal-weather-station/dashboard?ID=KMAHADLE7#history/tdata/s20170201/e20170201/mcustom.html')
    df=pd.read_html(driver.find_element_by_id("history_table").get_attribute('outerHTML'))[0]

Time    Temperature Dew Point   Humidity    Wind    Speed   Gust    Pressure  Precip. Rate. Precip. Accum.  UV  Solar
0   12:02 AM    25.5 °C 18.7 °C 75 %    East    0 kph   0 kph   29.3 hPa    0 mm    0 mm    0   0 w/m²
1   12:07 AM    25.5 °C 19 °C   76 %    East    0 kph   0 kph   29.31 hPa   0 mm    0 mm    0   0 w/m²
2   12:12 AM    25.5 °C 19 °C   76 %    East    0 kph   0 kph   29.31 hPa   0 mm    0 mm    0   0 w/m²
3   12:17 AM    25.5 °C 18.7 °C 75 %    East    0 kph   0 kph   29.3 hPa    0 mm    0 mm    0   0 w/m²
4   12:22 AM    25.5 °C 18.7 °C 75 %    East    0 kph   0 kph   29.3 hPa    0 mm    0 mm    0   0 w/m²

Editing with breakdown of exactly what's happening, since the above one-liner is actually not very good self-documenting code:

After setting up the driver, we select the table with its ID value (Thankfully this site actually uses reasonable and descriptive IDs)

tab=driver.find_element_by_id("history_table")

Then, from that element, we get the HTML instead of the web driver element object

tab_html=tab.get_attribute('outerHTML')

We use pandas to parse the html

tab_dfs=pd.read_html(tab_html)

From the docs:

"read_html returns a list of DataFrame objects, even if there is only a single table contained in the HTML content"

So we index into that list with the only table we have, at index zero

df=tab_dfs[0]

Comments

[Noman Bashir]

Hi, thanks a lot. This works wonders, but I would highly appreciate if you would shed a little light on why did we select an attribute and picked the value at index 0.

[G. Anderson]

Edited with breakdown

[Noman Bashir]

Thanks a lot. It was really helpful.