python.django.security.injection.tainted-url-host.tainted-url-host

profile photo of semgrepsemgrep
Author
unknown
Download Count*

User data flows into the host portion of this manually-constructed URL. This could allow an attacker to send data to their own server, potentially exposing sensitive data such as cookies or authorization information sent with this request. They could also probe internal servers or other resources that the server runnig this code can access. (This is called server-side request forgery, or SSRF.) Do not allow arbitrary hosts. Instead, create an allowlist for approved hosts hardcode the correct host.

Run Locally

Run in CI

Defintion

rules:
  - id: tainted-url-host
    languages:
      - python
    message: User data flows into the host portion of this manually-constructed URL.
      This could allow an attacker to send data to their own server, potentially
      exposing sensitive data such as cookies or authorization information sent
      with this request. They could also probe internal servers or other
      resources that the server runnig this code can access. (This is called
      server-side request forgery, or SSRF.) Do not allow arbitrary hosts.
      Instead, create an allowlist for approved hosts hardcode the correct host.
    metadata:
      cwe:
        - "CWE-918: Server-Side Request Forgery (SSRF)"
      owasp:
        - A10:2021 - Server-Side Request Forgery (SSRF)
      references:
        - https://cheatsheetseries.owasp.org/cheatsheets/Server_Side_Request_Forgery_Prevention_Cheat_Sheet.html
      category: security
      technology:
        - flask
      license: Commons Clause License Condition v1.0[LGPL-2.1-only]
      cwe2022-top25: true
      cwe2021-top25: true
      subcategory:
        - audit
      impact: MEDIUM
      likelihood: LOW
      confidence: LOW
      vulnerability_class:
        - Server-Side Request Forgery (SSRF)
    mode: taint
    pattern-sinks:
      - patterns:
          - pattern-either:
              - patterns:
                  - pattern: '"$URLSTR" % ...'
                  - metavariable-pattern:
                      metavariable: $URLSTR
                      language: generic
                      patterns:
                        - pattern-either:
                            - pattern: $SCHEME://%s
                            - pattern: $SCHEME://%r
              - patterns:
                  - pattern: '"$URLSTR".format(...)'
                  - metavariable-pattern:
                      metavariable: $URLSTR
                      language: generic
                      pattern: $SCHEME:// { ... }
              - patterns:
                  - pattern: '"$URLSTR" + ...'
                  - metavariable-regex:
                      metavariable: $URLSTR
                      regex: .*://$
              - patterns:
                  - pattern: f"$URLSTR{...}..."
                  - metavariable-regex:
                      metavariable: $URLSTR
                      regex: .*://$
              - patterns:
                  - pattern-inside: |
                      $URL = "$URLSTR"
                      ...
                  - pattern: $URL += ...
                  - metavariable-regex:
                      metavariable: $URLSTR
                      regex: .*://$
    pattern-sources:
      - patterns:
          - pattern: request.$ANYTHING
          - pattern-not: request.build_absolute_uri
    severity: WARNING

Examples

tainted-url-host.py

from django.http import HttpResponse
import requests

class Person(models.Model):
    first_name = models.CharField(...)
    last_name = models.CharField(...)
    birth_date = models.DateField(...)

##### True Positives #########
def ex1(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ruleid: tainted-url-host
  user_age = requests.get("https://%s/%s/age" % (env, user_name))
  return HttpResponse(user_age)

def ex2(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ruleid: tainted-url-host
  user_age = requests.get("https://{}/{}/age".format(env, user_name))
  return HttpResponse(user_age)

def ex3(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ruleid: tainted-url-host
  user_age = requests.get(f"https://{env}/{user_name}/age")
  return HttpResponse(user_age)

def ex4(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ruleid: tainted-url-host
  user_age = requests.get(f"https://" + env + "/" + user_name + "/age")
  return HttpResponse(user_age)

def ex5(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ruleid: tainted-url-host
  url = "https://{}/{}/age".format(env, user_name)
  user_age = requests.get(url)
  return HttpResponse(user_age)

def ex6(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ruleid: tainted-url-host
  url = "https://{}/{}/age".format(env, user_name)
  user_age = requests.get(url)
  return HttpResponse(user_age)

def ex7(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  url = "https://%s/%s/age"
  # ruleid: tainted-url-host
  user_age = requests.get(url % (env, user_name))
  return HttpResponse(user_age)

def ex8(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  url = "https://{}/{}/age"
  # ruleid: tainted-url-host
  user_age = requests.get(url.format(env, user_name))
  return HttpResponse(user_age)

##### True Negatives #########
def ok1(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ok: tainted-url-host
  user_age = requests.get("https://example.com/%s/%s/age" % (env, user_name))
  return HttpResponse(user_age)

def ok2(request):
  env = request.POST.get('env')
  user_name = request.POST.get('user_name')
  # ok: tainted-url-host
  user_age = requests.get("https://example.com/%s/%s/age".format(env, user_name))
  return HttpResponse(user_age)