[FIX] Quick'n'dirty termlists overwrite feature

gargantext/util/ngramlists_tools.py

@@ -568,7 +568,8 @@ def import_ngramlists(the_file, delimiter=DEFAULT_CSV_DELIM,
          'map':  UnweightedList(),
          'main': UnweightedList(),
          'stop': UnweightedList(),
-         'groupings' : Translations()
+         'groupings' : Translations(),
+         'new_ngram_count': n_added_ng,
          }
 
     for list_type in imported_nodes_ngrams:
@@ -664,12 +665,13 @@ def merge_ngramlists(new_lists={}, onto_corpus=None, del_originals=[]):
             for ng_id in new_lists[list_type].items:
                 collect(ng_id)
 
-    from gargantext.util.toolchain.main import t
-    print("MERGE DEBUG: starting index_new_ngrams", t())
-    n_added = index_new_ngrams(all_possibly_new_ngram_ids, onto_corpus)
-    print("MERGE DEBUG: finished index_new_ngrams", t())
+    if new_lists.get('new_ngram_count', 0) > 0:
+        from gargantext.util.toolchain.main import t
+        print("MERGE DEBUG: starting index_new_ngrams", t())
+        n_added = index_new_ngrams(all_possibly_new_ngram_ids, onto_corpus)
+        print("MERGE DEBUG: finished index_new_ngrams", t())
 
-    my_log.append("MERGE: added %i new ngram occurrences in docs" % n_added)
+        my_log.append("MERGE: added %i new ngram occurrences in docs" % n_added)
 
     # ======== Get the old lists =========
     old_lists = {}
@@ -828,7 +830,7 @@ def merge_ngramlists(new_lists={}, onto_corpus=None, del_originals=[]):
 
 
 @shared_task
-def import_and_merge_ngramlists(file_contents, onto_corpus_id):
+def import_and_merge_ngramlists(file_contents, onto_corpus_id, overwrite=False):
     """
     A single function to run import_ngramlists and merge_ngramlists together
     """
@@ -838,6 +840,7 @@ def import_and_merge_ngramlists(file_contents, onto_corpus_id):
     corpus_node = session.query(Node).filter(Node.id == onto_corpus_id).first()
 
     # merge the new_lists onto those of the target corpus
-    log_msg = merge_ngramlists(new_lists, onto_corpus=corpus_node)
+    del_originals = ['stop', 'main', 'map'] if overwrite else []
+    log_msg = merge_ngramlists(new_lists, onto_corpus=corpus_node, del_originals=del_originals)
 
     return log_msg

gargantext/views/api/ngramlists.py

@@ -90,10 +90,11 @@ class CSVLists(APIView):
         # import the csv
         # try:
         log_msg = "Async generation"
-        
+
         corpus_node_id = corpus_node.id
-        scheduled(import_and_merge_ngramlists)(csv_contents, corpus_node_id)
-        
+        scheduled(import_and_merge_ngramlists)(csv_contents, corpus_node_id,
+                                               overwrite=bool(params.get('overwrite')))
+
         return JsonHttpResponse({
             'log': log_msg,
             }, 200)
@@ -153,7 +154,8 @@ class CSVLists(APIView):
         # attempt to merge and send response
         try:
             # merge the source_lists onto those of the target corpus
-            log_msg = merge_ngramlists(source_lists, onto_corpus=corpus_node)
+            delete = todo_lists if bool(params.get('overwrite')) else []
+            log_msg = merge_ngramlists(source_lists, onto_corpus=corpus_node, del_originals=delete)
             return JsonHttpResponse({
                 'log': log_msg,
                 }, 200)

templates/pages/corpora/terms.html

@@ -250,6 +250,23 @@ em {
 
             <br/>
 
+            <div class="checkbox">
+                <label>
+                    <input type="checkbox" id="importoverwrite"> Overwrite old lists
+                    <script>
+                        function updateSubmitLabel() {
+                            $('#importsubmit').val($(this).is(':checked') ? 'Overwrite current table' : 'Import and merge with current table');
+                        }
+                        $(function() {
+                            updateSubmitLabel.call($('#importoverwrite'));
+                            $('#importoverwrite').change(updateSubmitLabel);
+                        });
+                    </script>
+                </label>
+            </div>
+
+            <br/>
+
             <input type="submit" class="btn btn-xs btn-info" id="importsubmit" value="Import and merge with current table" />
         </form>
     </div>
@@ -372,6 +389,8 @@ function listmergeUpdate(aFormData){
     // all params are added in the url like a GET
     theUrl += "&from_corpus="+sourceCorpusId
     theUrl += "&todo="+todoLists.join(',')
+    if ($('#importoverwrite').is(':checked'))
+        theUrl += "&overwrite=1"
 
     // result url looks like this : /api/ngramlists/import?onto_corpus=2&from=13308&todo=map,stop
     // console.log(theUrl)
@@ -424,7 +443,7 @@ function listmergeCsvPost(theFile){
 
         //postCorpusFile
         $.ajax({
-             url: "{{importroute | safe}}",
+             url: "{{importroute | safe}}" + ($('#importoverwrite').is(':checked') ? '&overwrite=1' : ''),
              type: 'POST',
              async: true,
              contentType: false,
@@ -436,11 +455,11 @@ function listmergeCsvPost(theFile){
              success: function(response) {
                  my_html  = '<h3 style="color:green">File upload, you will receive a notification email</h3>'
                  my_html += "<p class='note'>" + response['log'].replace(/\n/g, '<br/>') + "</p>"
-                 my_html += "<p'>(this page will reload in 3s)</p>"
+                 my_html += "<p'>(this page will reload in 30s)</p>"
                  $('#formanswer').html(my_html);
                  console.log(response) ;
                  // reload after 3s
-                 setTimeout("location.reload(true)", 3000);
+                 setTimeout("location.reload(true)", 30000);
                  },
               error: function(result, t) {
                   if (t != 'timeout') {