[FIX] Back to sqlalchemy, SQL query needs more optimization.

gargantext/util/toolchain/main.py

@@ -9,7 +9,8 @@ from .list_stop           import do_stoplist
 from .ngram_groups        import compute_groups
 from .metric_tfidf        import compute_occs, compute_tfidf_local, compute_ti_ranking
 from .list_main           import do_mainlist
-from .ngram_coocs         import compute_coocs
+#from .ngram_coocs         import compute_coocs
+from .ngram_coocs_old_sqlalchemy_version import compute_coocs
 from .metric_specgen      import compute_specgen
 from .list_map            import do_maplist
 from .mail_notification   import notify_owner
@@ -148,7 +149,7 @@ def parse_extract_indexhyperdata(corpus):
                             on_list_id = mainlist_id,
                             groupings_id = group_id,
                             just_pass_result = True,
-                            symmetry_filter  = True,
+                           # symmetry_filter  = True,
                             diagonal_filter = False) # preserving the diagonal
                                                      # (useful for spec/gen)
     print('CORPUS #%d: [%s] computed mainlist coocs for specif rank' % (corpus.id, t()))