Add conf_requires.

Load plugins in order

conf.py

@@ -16,11 +16,11 @@ DB_ROOT = './output/'
 DISPLAY_ROOT = './output/'
 
 pre_analysis_hooks = ['page_to_hit', 'robots']
-post_analysis_hooks = ['referers', 'top_pages']
+post_analysis_hooks = ['referers', 'top_pages', 'top_downloads']
 # post_analysis_hooks = ['top_visitors', 'reverse_dns']
-display_hooks = ['top_visitors', 'all_visits', 'referers', 'top_pages']
+display_hooks = ['top_visitors', 'all_visits', 'referers', 'top_pages', 'top_downloads']
 
 reverse_dns_timeout = 0.2
-page_to_hit_conf = [r'^.+/logo/$']
+page_to_hit_conf = [r'^.+/logo[/]?$', r'^.+/category/.+$', r'^.+/tag/.+$', r'^.+/archive/.+$', r'^.+/ljdc[/]?$']
 
-count_hit_only_visitors = False
+count_hit_only_visitors = True

default_conf.py

@@ -20,7 +20,10 @@ pre_analysis_hooks = []
 post_analysis_hooks = []
 display_hooks = []
 
-pages_extensions = ['/', 'html', 'xhtml', 'py', 'pl', 'rb', 'php']
+pages_extensions = ['/', 'htm', 'html', 'xhtml', 'py', 'pl', 'rb', 'php']
 viewed_http_codes = [200, 304]
 
 count_hit_only_visitors = True
+
+multimedia_files = ['png', 'jpg', 'jpeg', 'gif', 'ico',
+                    'css', 'js']

iplugin.py

@@ -7,6 +7,7 @@ class IPlugin(object):
     def __init__(self, iwla):
         self.iwla = iwla
         self.requires = []
+        self.conf_requires = []
         self.API_VERSION = 1
         self.ANALYSIS_CLASS = 'HTTP'
 
@@ -22,6 +23,9 @@ class IPlugin(object):
     def getRequirements(self):
         return self.requires
 
+    def getConfRequirements(self):
+        return self.conf_requires
+
     def load(self):
         return True
 
@@ -33,8 +37,8 @@ def preloadPlugins(plugins, iwla):
 
     print "==> Preload plugins"
 
-    for root in plugins.keys():
+    for (root, plugins_filenames) in plugins:
-        for plugin_filename in plugins[root]:
+        for plugin_filename in plugins_filenames:
             plugin_path = root + '.' + plugin_filename
             try:
                 mod = importlib.import_module(plugin_path)
@@ -57,9 +61,19 @@ def preloadPlugins(plugins, iwla):
                 
                 #print 'Load plugin %s' % (plugin_name)
 
+                conf_requirements = plugin.getConfRequirements()
+
+                requirement_validated = True
+                for r in conf_requirements:
+                    conf_value = iwla.getConfValue(r, None)
+                    if conf_value is None:
+                        print '\'%s\' conf value required for %s' % (r, plugin_path)
+                        requirement_validated = False
+                        break
+                if not requirement_validated: continue
+
                 requirements = plugin.getRequirements()
 
-                if requirements:
                 requirement_validated = False
                 for r in requirements:
                     for (_,p) in cache_plugins.items():
@@ -67,9 +81,9 @@ def preloadPlugins(plugins, iwla):
                             requirement_validated = True
                             break
                     if not requirement_validated:
-                            print 'Missing requirements for plugin %s' % (plugin_path)
+                        print 'Missing requirements \'%s\' for plugin %s' % (r, plugin_path)
                         break
-                    if not requirement_validated: continue
+                if requirements and not requirement_validated: continue
 
                 if not plugin.load():
                     print 'Plugin %s load failed' % (plugin_path)
@@ -78,7 +92,7 @@ def preloadPlugins(plugins, iwla):
                 print '\tRegister %s' % (plugin_path)
                 cache_plugins[plugin_path] = plugin
             except Exception as e:
-                print 'Error loading \'%s\' => %s' % (plugin_path, e)
+                print 'Error loading %s => %s' % (plugin_path, e)
                 traceback.print_exc()
 
     return cache_plugins

iwla.py

@@ -35,11 +35,11 @@ class IWLA(object):
         self.http_request_extracted = re.compile(r'(?P<http_method>\S+) (?P<http_uri>\S+) (?P<http_version>\S+)')
         self.log_re = re.compile(self.log_format_extracted)
         self.uri_re = re.compile(r'(?P<extract_uri>[^\?]+)(\?(?P<extract_parameters>.+))?')
-        self.plugins = {conf.PRE_HOOK_DIRECTORY     : conf.pre_analysis_hooks,
+        self.plugins = [(conf.PRE_HOOK_DIRECTORY     , conf.pre_analysis_hooks),
-                        conf.POST_HOOK_DIRECTORY    : conf.post_analysis_hooks,
+                        (conf.POST_HOOK_DIRECTORY    , conf.post_analysis_hooks),
-                        conf.DISPLAY_HOOK_DIRECTORY : conf.display_hooks}
+                        (conf.DISPLAY_HOOK_DIRECTORY , conf.display_hooks)]
 
-    def getConfValue(self, key, default):
+    def getConfValue(self, key, default=None):
         if not key in dir(conf):
             return default
         else:
@@ -77,7 +77,7 @@ class IWLA(object):
 
     def isValidForCurrentAnalysis(self, request):
         cur_time = self.meta_infos['start_analysis_time']
-        return (time.mktime(cur_time) < time.mktime(request['time_decoded']))
+        return (time.mktime(cur_time) <= time.mktime(request['time_decoded']))
 
     def _clearMeta(self):
         self.meta_infos = {
@@ -115,11 +115,14 @@ class IWLA(object):
             return pickle.load(f)
         return None
 
-    def _callPlugins(self, root, *args):
+    def _callPlugins(self, target_root, *args):
-        print '==> Call plugins (%s)' % root
+        print '==> Call plugins (%s)' % target_root
-        for p in self.plugins[root]:
+        for (root, plugins) in self.plugins:
+            if root != target_root: continue
+            for p in plugins:
+                mod = self.cache_plugins.get(root + '.' + p, None)
+                if mod:
                     print '\t%s' % (p)
-            mod = self.cache_plugins[root + '.' + p]
                     mod.hook(*args)
 
     def isPage(self, request):
@@ -143,9 +146,7 @@ class IWLA(object):
 
         request = hit['extract_request']
 
-        if 'extract_uri' in request.keys():
+        uri = request.get('extract_uri', request['http_uri'])
-            uri = request['extract_uri'] = request['http_uri']
-        uri = request['extract_uri']
 
         hit['is_page'] = self.isPage(uri)
 

plugins/post_analysis/referers.py

@@ -10,6 +10,7 @@ class IWLAPostAnalysisReferers(IPlugin):
     def __init__(self, iwla):
         super(IWLAPostAnalysisReferers, self).__init__(iwla)
         self.API_VERSION = 1
+        self.conf_requires = ['domain_name']
 
     def _getSearchEngine(self, hashid):
         for (k, e) in self.search_engines.items():
@@ -22,7 +23,7 @@ class IWLAPostAnalysisReferers(IPlugin):
         domain_name = self.iwla.getConfValue('domain_name', '')
         
         if not domain_name:
-            print 'domain_name required in conf'
+            print 'domain_name must not be empty !'
             return False
 
         self.own_domain_re = re.compile(r'.*%s.*' % (domain_name))

plugins/post_analysis/reverse_dns.py

@@ -4,12 +4,15 @@ from iwla import IWLA
 from iplugin import IPlugin
 
 class IWLAPostAnalysisReverseDNS(IPlugin):
+    DEFAULT_DNS_TIMEOUT = 0.5
+
     def __init__(self, iwla):
         super(IWLAPostAnalysisReverseDNS, self).__init__(iwla)
         self.API_VERSION = 1
 
     def load(self):
-        timeout = self.iwla.getConfValue('reverse_dns_timeout', 0.5)
+        timeout = self.iwla.getConfValue('reverse_dns_timeout',
+                                         IWLAPostAnalysisReverseDNS.DEFAULT_DNS_TIMEOUT)
         socket.setdefaulttimeout(timeout)
         return True
         

plugins/pre_analysis/page_to_hit.py

@@ -10,6 +10,7 @@ class IWLAPreAnalysisPageToHit(IPlugin):
     def __init__(self, iwla):
         super(IWLAPreAnalysisPageToHit, self).__init__(iwla)
         self.API_VERSION = 1
+        self.conf_requires = ['viewed_http_codes']
 
     def load(self):
 # Remove logo from indefero
@@ -21,7 +22,8 @@ class IWLAPreAnalysisPageToHit(IPlugin):
 
     def hook(self):
         hits = self.iwla.getCurrentVisists()
-        viewed_http_codes = self.iwla.getConfValue('viewed_http_codes', [200, 304])
+        viewed_http_codes = self.iwla.getConfValue('viewed_http_codes')
+
         for (k, super_hit) in hits.items():
             if super_hit['robot']: continue
 
@@ -31,6 +33,7 @@ class IWLAPreAnalysisPageToHit(IPlugin):
                 uri = request['extract_request']['extract_uri']
                 for regexp in self.regexps:
                     if regexp.match(uri):
+                        #print '%s is an hit' % uri 
                         request['is_page'] = False
                         super_hit['viewed_pages'] -= 1
                         super_hit['viewed_hits'] += 1