Refactor tracker update

src/bundle_combine.py

@@ -5,6 +5,7 @@ import re
 import sys
 import common_lib as mylib
 import download_tracker  # is_tracker
+import index_domains  # load
 
 
 THRESHOLD_PERCENT_OF_LOGS = 0.33  # domain appears in % recordings
@@ -110,31 +111,18 @@ def json_evaluate_inplace(obj):
     obj['avg_logs_pm'] = float3(s_c / (obj['avg_time'] or 1) * 60)
 
 
-def process(bundle_ids, where=None):
+def process(bundle_ids):
     print('writing combined json ...')
     if bundle_ids == ['*']:
         bundle_ids = list(mylib.enum_data_appids())
 
-    affected_ids = []
-    haystack = sorted([x[::-1] for x in where]) if where else None
     for bid in bundle_ids:
-        obj = json_combine(bid)
-        should_update = False
-        if not haystack:
-            should_update = True
-        else:
-            for x in obj['subdom']:
-                if mylib.bintree_lookup(haystack, x[::-1]):
-                    should_update = True
-                    break
-        if should_update:
         print('  ' + bid)
+        obj = json_combine(bid)
         mylib.json_write(fname_combined(bid), obj, pretty=False)
         json_evaluate_inplace(obj)
         mylib.json_write(fname_evaluated(bid), obj, pretty=False)
-            affected_ids.append(bid)
     print('')
-    return affected_ids
 
 
 if __name__ == '__main__':

src/index_domains.py

@@ -97,6 +97,17 @@ def enrich_with_bundle_ids(index):
             index[key][dom] = [index['bundle'][i] for i in ids]
 
 
+def all_bundles_containing(list_of_domains):
+    affected = set()
+    json = load_json_from_disk(fname_all())
+    haystack = sorted([x[::-1] for x in list_of_domains])
+    for key in ['pardom', 'subdom']:
+        for dom, ids in json[key].items():
+            if mylib.bintree_lookup(haystack, dom[::-1]):
+                affected.update(ids)
+    return [json['bundle'][i] for i in affected]
+
+
 def process(bundle_ids, deleteOnly=False):
     print('writing index: domains ...')
     fname = fname_all()

src/main.py

@@ -6,10 +6,10 @@ import common_lib as mylib
 import bundle_combine
 import download_itunes
 import download_tracker
-import html_root
-import html_index_apps
 import html_bundle
+import html_index_apps
 import html_index_domains
+import html_root
 import index_app_names
 import index_domains
 import index_meta
@@ -58,26 +58,20 @@ def del_id(bundle_ids):
         rebuild_app_index_html(inclRoot=True)
 
 
-def combine_and_update(bundle_ids, where=None):
-    def star_reset(ids):
-        # special case needed. '*' will force rebuilt index
-        return ['*'] if not where and bundle_ids == ['*'] else ids
+def combine_and_update(bundle_ids):
     # 1. download meta data from iTunes store, incl. app icons
     new_ids = download_itunes.process(bundle_ids)
-    new_ids = star_reset(new_ids)
     # 2. if new apps, update bundle name index
+    if bundle_ids == ['*']:
+        new_ids = ['*']  # special case needed to force rebuilt index
     if len(new_ids) > 0:
         index_app_names.process(new_ids)  # after download_itunes
     # 3. re-calculate combined.json and evaluated.json files
-    affected = bundle_combine.process(bundle_ids, where=where)
-    affected = star_reset(affected)
-    # 4. was any json updated? if so, make html and update domain index
-    if len(affected) > 0:
+    bundle_combine.process(bundle_ids)
+    # 4. make html and update domain index
     index_meta.process(bundle_ids)  # after bundle_combine
-        html_bundle.process(affected)  # after index_app_names
-        rebuild_domain_index(affected)  # after bundle_combine
-    else:
-        print('no bundle affected by tracker, not generating bundle html')
+    html_bundle.process(bundle_ids)  # after index_app_names
+    rebuild_domain_index(bundle_ids)  # after bundle_combine
     # 5. make all apps index
     if len(new_ids) > 0:
         rebuild_app_index_html()  # after bundle_combine
@@ -112,8 +106,11 @@ def import_update():
 
 def tracker_update():
     new_trackers = download_tracker.process()
-    if new_trackers:
-        combine_and_update(['*'], where=new_trackers)
+    affected = index_domains.all_bundles_containing(new_trackers)
+    if len(affected) > 0:
+        combine_and_update(affected)
+    else:
+        print('no bundle affected by tracker, not generating bundle html')
 
 
 try:
@@ -138,7 +135,7 @@ try:
         elif cmd == 'run':
             if len(params) == 0:
                 print_usage_and_exit()
-            combine_and_update(params)  # ['*'], where=['test.com']
+            combine_and_update(params)
         elif cmd == 'del':
             if len(params) == 0:
                 print_usage_and_exit()