Reputation: 5122
I am using bash to POST to a website that requires that I be logged in first, so I need to send the request with my login cookie. I tried logging in and keeping the cookies, but it doesn't work because the site uses javascript to hash the password in a really weird fashion, so instead I'm going to just take my login cookies for the site from Chrome. How do get the cookies from Chrome and format them for Curl?
I'm trying to do this:
curl --request POST -d "a=X&b=Y" -b "what goes here?" "site.com/a.php"
Upvotes: 89
Views: 162469
Reputation: 1682
Can't believe no one has mentioned this. Here's the easiest and quickest way.
Simply open up your browser's Developer Tools
, click on the Console
tab, and lastly within the console, simply type the following & press ENTER...
console.log(document.cookie)
The results will be immediately printed in proper syntax. Simply highlight it and copy it.
Upvotes: 17
Reputation: 23592
For anyone that wants all of the cookies for a site, but doesn't want to use an extension:
Ctrl
/Cmd
-A
Ctrl
/Cmd
-C
Now you have a TSV (tab-separated value) string of cookie data. You can process this in any language you want, but in Python (for example):
import io
import pandas as pd
cookie_str = """[paste cookie str here]"""
# Copied from the developer tools window.
cols = ['name', 'value', 'domain', 'path', 'max_age', 'size', 'http_only', 'secure', 'same_party', 'priority']
# Parse into a dataframe.
df = pd.read_csv(io.StringIO(cookie_str), sep='\t', names=cols, index_col=False)
Now you can export them in Netscape format:
# Fill in NaNs and format True/False for cookies.txt.
df = df.fillna(False).assign(flag=True).replace({True: 'TRUE', False: 'FALSE'})
# Get unix timestamp from max_age
max_age = (
df.max_age
.replace({'Session': np.nan})
.pipe(pd.to_datetime))
start = pd.Timestamp("1970-01-01", tz='UTC')
max_age = (
((max_age - start) // pd.Timedelta('1s'))
.fillna(0) # Session expiry are 0s
.astype(int)) # Floats end with ".0"
df = df.assign(max_age=max_age)
cookie_file_cols = ['domain', 'flag', 'path', 'secure', 'max_age', 'name', 'value']
with open('cookies.txt') as fh:
# Python's cookiejar wants this header.
fh.write('# Netscape HTTP Cookie File\n')
df[cookie_file_cols].to_csv(fh, sep='\t', index=False, header=False)
And finally, back to the shell:
# Get user agent from navigator.userAgent in devtools
wget -U $USER_AGENT --load-cookies cookies.txt $YOUR_URL
Upvotes: 0
Reputation: 123510
This will give you the curl command for the action you triggered, fully populated with cookies and all. You can of course also copy the flags as a basis for new curl commands.
Upvotes: 149
Reputation: 1416
I was curious if others were reporting that chrome doesn't allow "copy as curl" feature to have cookies anymore.
It then occurred to me that this is like a security idea. If you visit example.com
, copying requests as curl to example.com
will have cookies. However, copying requests to other domains or subdomains will sanitize the cookies. a.example.com
or test.com
will not have cookies for example.
Upvotes: 4
Reputation: 202
In Chrome:
In the terminal
--cookie "cookiename=cookievalue"
to your curl request.Upvotes: 17