From b1d0d067f9a701155cfcc5998ba71668d84f9d41 Mon Sep 17 00:00:00 2001
From: Terrtia <or1994@hotmail.fr>
Date: Wed, 24 Jun 2020 15:07:45 +0200
Subject: [PATCH] fix: [regex tracker] fix timeout

---
 bin/RegexTracker.py     | 25 +++++++------------------
 bin/lib/regex_helper.py | 27 +++++++++++++++++++++++++++
 2 files changed, 34 insertions(+), 18 deletions(-)

diff --git a/bin/RegexTracker.py b/bin/RegexTracker.py
index 2f7e5b9f..d1bdc26e 100755
--- a/bin/RegexTracker.py
+++ b/bin/RegexTracker.py
@@ -10,7 +10,6 @@ import os
 import re
 import sys
 import time
-import signal
 
 from Helper import Process
 from pubsublogger import publisher
@@ -20,18 +19,15 @@ import NotificationHelper
 from packages import Item
 from packages import Term
 
+sys.path.append(os.path.join(os.environ['AIL_BIN'], 'lib'))
+import regex_helper
+
 full_item_url = "/showsavedpaste/?paste="
 mail_body_template = "AIL Framework,\nNew occurrence for term tracked regex: {}\nitem id: {}\nurl: {}{}"
 
 dict_regex_tracked = Term.get_regex_tracked_words_dict()
 last_refresh = time.time()
 
-class TimeoutException(Exception):
-    pass
-def timeout_handler(signum, frame):
-    raise TimeoutException
-signal.signal(signal.SIGALRM, timeout_handler)
-
 def new_term_found(term, term_type, item_id, item_date):
     uuid_list = Term.get_term_uuid_list(term, 'regex')
     print('new tracked term found: {} in {}'.format(term, item_id))
@@ -56,11 +52,14 @@ if __name__ == "__main__":
     publisher.info("Script RegexTracker started")
 
     config_section = 'RegexTracker'
+    module_name = "RegexTracker"
     p = Process(config_section)
     max_execution_time = p.config.getint(config_section, "max_execution_time")
 
     ull_item_url = p.config.get("Notifications", "ail_domain") + full_item_url
 
+    redis_cache_key = regex_helper.generate_redis_cache_key(module_name)
+
     # Regex Frequency
     while True:
 
@@ -72,20 +71,10 @@ if __name__ == "__main__":
             item_content = Item.get_item_content(item_id)
 
             for regex in dict_regex_tracked:
-
-                signal.alarm(max_execution_time)
-                try:
-                    matched = dict_regex_tracked[regex].search(item_content)
-                except TimeoutException:
-                    print ("{0} processing timeout".format(item_id))
-                    continue
-                else:
-                    signal.alarm(0)
-
+                matched = regex_helper.regex_search(module_name, redis_cache_key, dict_regex_tracked[regex], item_id, item_content, max_time=max_execution_time)
                 if matched:
                     new_term_found(regex, 'regex', item_id, item_date)
 
-
         else:
             time.sleep(5)
 
diff --git a/bin/lib/regex_helper.py b/bin/lib/regex_helper.py
index 4f38c39c..67256376 100755
--- a/bin/lib/regex_helper.py
+++ b/bin/lib/regex_helper.py
@@ -73,3 +73,30 @@ def regex_findall(module_name, redis_key, regex, item_id, item_content, max_time
         print("Caught KeyboardInterrupt, terminating workers")
         proc.terminate()
         sys.exit(0)
+
+def _regex_search(redis_key, regex, item_content):
+    first_occ = regex.search(item_content)
+    if r_set:
+        r_serv_cache.set(redis_key, first_occ)
+
+def regex_search(module_name, redis_key, regex, item_id, item_content, max_time=30):
+    proc = Proc(target=_regex_search, args=(redis_key, regex, item_content, ))
+    try:
+        proc.start()
+        proc.join(max_time)
+        if proc.is_alive():
+            proc.terminate()
+            Statistics.incr_module_timeout_statistic(module_name)
+            err_mess = "{}: processing timeout: {}".format(module_name, item_id)
+            print(err_mess)
+            publisher.info(err_mess)
+            return None
+        else:
+            first_occ = r_serv_cache.get(redis_key)
+            r_serv_cache.delete(redis_key)
+            proc.terminate()
+            return first_occ
+    except KeyboardInterrupt:
+        print("Caught KeyboardInterrupt, terminating workers")
+        proc.terminate()
+        sys.exit(0)