Add the ability to override the default download extensions.

2 files changed, 11 insertions, 1 deletions

diff --git a/misc/log-analytics/import_logs.py b/misc/log-analytics/import_logs.py
index a339605acc..09d8b4e969 100755
--- a/misc/log-analytics/import_logs.py
+++ b/misc/log-analytics/import_logs.py
@@ -481,6 +481,10 @@ class Configuration(object):
             '--enable-testmode', dest='enable_testmode', default=False, action='store_true',
             help="If set, it will try to get the token_auth from the piwik_tests directory"
         )
+        option_parser.add_option(
+            '--download-extensions', dest='download_extensions', default=None,
+            help="If set (format: pdf,doc,...), this list will be used to determine what file extensions should be tracked as a download, otherwise the default list will be used."
+        )
         return option_parser
 
     def _parse_args(self, option_parser):
@@ -550,6 +554,11 @@ class Configuration(object):
         if self.options.recorders < 1:
             self.options.recorders = 1
 
+        if self.options.download_extensions:
+            self.options.download_extensions = set(self.options.download_extensions.split(','))
+        else:
+            self.options.download_extensions = DOWNLOAD_EXTENSIONS
+
     def __init__(self):
         self._parse_args(self._create_parser())
 
@@ -1373,7 +1382,7 @@ class Parser(object):
         return True
 
     def check_download(self, hit):
-        if hit.extension in DOWNLOAD_EXTENSIONS:
+        if hit.extension in config.options.download_extensions:
             stats.count_lines_downloads.increment()
             hit.is_download = True
         return True
diff --git a/misc/log-analytics/tests/tests.py b/misc/log-analytics/tests/tests.py
index 163c100d3a..37af5eee8f 100644
--- a/misc/log-analytics/tests/tests.py
+++ b/misc/log-analytics/tests/tests.py
@@ -63,6 +63,7 @@ class Options(object):
     force_lowercase_path = False
     included_paths = []
     enable_http_errors = False
+    download_extensions = 'doc,pdf'
 
 class Config(object):
     """Mock configuration."""