From 7f7a3906e57b2c3c8908badbf4d9578716cf1472 Mon Sep 17 00:00:00 2001
From: Yann Bertrand <5855339+yannbertrand@users.noreply.github.com>
Date: Thu, 19 Nov 2020 21:50:32 +0100
Subject: [PATCH] Faking user agent + handle "no server" and fallback error

---
 build/validate_links.py | 42 +++++++++++++++++++++++++----------------
 1 file changed, 26 insertions(+), 16 deletions(-)

diff --git a/build/validate_links.py b/build/validate_links.py
index 192e74fe..ab36b0c8 100755
--- a/build/validate_links.py
+++ b/build/validate_links.py
@@ -19,35 +19,45 @@ def parse_links(filename):
 
 def validate_links(links):
     """Checks each entry in JSON file for live link"""
-    print('Validating {} links...'.format(len(links)))
-    errors = []
+    print(f'Validating {len(links)} links...')
+    hasError = False
     for link in links:
         h = httplib2.Http(disable_ssl_certificate_validation=True, timeout=25)
         try:
-            resp = h.request(link, headers={'user-agent': 'python-httplib2/0.18.0'})
+            resp = h.request(link, headers={
+                # Faking user agent as some hosting services block not-whitelisted UA
+                'user-agent': 'Mozilla/5.0'
+            })
             code = int(resp[0]['status'])
-            # check if status code is a client or server error
-            if code >= 404:
-                errors.append('{}: {}'.format(code, link))
+            # Checking status code errors
+            if (code >= 300):
+                hasError = True
+                print(f"ERR:CLT:{code} : {link}")
         except TimeoutError:
-            errors.append("TMO: " + link)
+            hasError = True
+            print(f"ERR:TMO: {link}")
         except socket.error as socketerror:
-            errors.append("SOC: {} : {}".format(socketerror, link))
+            hasError = True
+            print(f"ERR:SOC: {socketerror} : {link}")
         except Exception as e:
+            hasError = True
             # Ignore some exceptions which are not actually errors.
             # The list below should be extended with other exceptions in the future if needed
-            if ((-1 != str(e).find("Content purported to be compressed with gzip but failed to decompress.")) and 
-                (-1 != str(e).find("[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:852)"))) :
-                errors.append("ERR: {} : {}".format(e, link))
-    return errors
+            if (-1 != str(e).find("[SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:852)")):
+                print(f"ERR:SSL: {e} : {link}")
+            elif (-1 != str(e).find("Content purported to be compressed with gzip but failed to decompress.")):
+                print(f"ERR:GZP: {e} : {link}")
+            elif (-1 != str(e).find("Unable to find the server at")):
+                print(f"ERR:SRV: {e} : {link}")
+            else:
+                print(f"ERR:UKN: {e} : {link}")
+    return hasError
 
 if __name__ == "__main__":
     num_args = len(sys.argv)
     if num_args < 2:
         print("No .md file passed")
         sys.exit(1)
-    errors = validate_links(parse_links(sys.argv[1]))
-    if len(errors) > 0:
-        for err in errors:
-            print(err)
+    hasError = validate_links(parse_links(sys.argv[1]))
+    if hasError:
         sys.exit(1)