From a29c16b770aa0bb53c632eecc1764caf02789fe7 Mon Sep 17 00:00:00 2001
From: Muriithi Frederick Muriuki
Date: Fri, 9 Mar 2018 18:15:19 +0300
Subject: Add link_checker module

* Add the module that will hold the code to test the links on the
  system.
---
 test/requests/link_checker.py | 9 +++++++++
 1 file changed, 9 insertions(+)
 create mode 100644 test/requests/link_checker.py

(limited to 'test/requests/link_checker.py')

diff --git a/test/requests/link_checker.py b/test/requests/link_checker.py
new file mode 100644
index 00000000..f171c12a
--- /dev/null
+++ b/test/requests/link_checker.py
@@ -0,0 +1,9 @@
+import requests
+
+def check_links(args_obj, parser):
+    print("")
+    print("Checking links")
+    print("########################")
+    print("Not implemented yet.")
+    print("This is supposed to check all links in the system.")
+    print("########################")
-- 
cgit v1.2.3


From dd5ec1d3080b74fa065620f6915eeacf2cca8a2d Mon Sep 17 00:00:00 2001
From: Muriithi Frederick Muriuki
Date: Mon, 12 Mar 2018 13:00:03 +0300
Subject: Add tests to check links.

---
 test/requests/link_checker.py | 62 ++++++++++++++++++++++++++++++++++++++++---
 1 file changed, 58 insertions(+), 4 deletions(-)

(limited to 'test/requests/link_checker.py')

diff --git a/test/requests/link_checker.py b/test/requests/link_checker.py
index f171c12a..256bf6ef 100644
--- a/test/requests/link_checker.py
+++ b/test/requests/link_checker.py
@@ -1,9 +1,63 @@
+from __future__ import print_function
+import re
 import requests
+from lxml.html import parse
+from requests.exceptions import ConnectionError
+
+def is_root_link(link):
+    pattern = re.compile("^/$")
+    return pattern.match(link)
+
+def is_mailto_link(link):
+    pattern = re.compile("^mailto:.*")
+    return pattern.match(link)
+
+def is_internal_link(link):
+    pattern = re.compile("^/.*")
+    return pattern.match(link)
+
+def get_links(doc):
+    return filter(
+        lambda x: not (
+            is_root_link(x)
+            or is_mailto_link(x))
+        , map(lambda y: y.get("href")
+              , doc.cssselect("a")))
+
+def verify_link(link):
+    try:
+        result = requests.get(link, timeout=20)
+        if result.status_code == 200:
+            print(link+" ==> OK")
+        else:
+            print("ERROR: link `"+link+"` failed with status "
+                  , result.status_code)
+    except ConnectionError as ex:
+        print("ERROR: ", link, ex)
+
+def check_page(host, start_url):
+    print("")
+    print("Checking links in page `"+start_url+"`")
+    doc = parse(start_url).getroot()
+    links = get_links(doc)
+    internal_links = filter(is_internal_link, links)
+    external_links = filter(lambda x: not is_internal_link(x), links)
+    external_links.append("http://somenon-existentsite.brr")
+    for link in internal_links:
+        verify_link(host+link)
+
+    for link in external_links:
+        verify_link(link)
 
 def check_links(args_obj, parser):
     print("")
     print("Checking links")
-    print("########################")
-    print("Not implemented yet.")
-    print("This is supposed to check all links in the system.")
-    print("########################")
+    host = args_obj.host
+
+    # Check the home page
+    check_page(host, host)
+
+    # Check traits page
+    check_page(
+        host,
+        host+"/show_trait?trait_id=1435395_s_at&dataset=HC_M2_0606_P")
-- 
cgit v1.2.3