[FEAT] bouton recalcul de corpus sur la page projet

gargantext/util/toolchain/main.py

@@ -173,5 +173,102 @@ def parse_extract_indexhyperdata(corpus):
 
 
 
+@shared_task
+def recount(corpus):
+    """
+    Recount essential metrics of the toolchain after group modifications.
+
+    ==> updates all scores in terms table
+    ==> updates tfidf relationship b/w term and doc
+
+    When groups change, the metrics need to be updated because subforms must be
+    added to their new mainform aggregate values:
+         - occurrences
+         - ndocs
+         - ti_rank
+         - coocs
+         - specificity
+         - tfidf
+
+    NB: no new extraction, no list change, just the metrics
+    """
+    # 1) we'll need the new groups and mainlist as basis
+    group_id = corpus.children("GROUPLIST").first().id
+    mainlist_id = corpus.children("MAINLIST").first().id
+
+    # 2) and we're going to overwrite the previous metric nodes
+    try:
+        old_occ_id    = corpus.children("OCCURRENCES").first().id
+    except:
+        old_occ_id    = None
+
+    try:
+        old_tirank_id = corpus.children("TIRANK-GLOBAL").first().id
+    except:
+        old_tirank_id = None
+
+    try:
+        old_spec_id   = corpus.children("SPECIFICITY").first().id
+    except:
+        old_spec_id   = None
+
+    try:
+        old_ltfidf_id = corpus.children("TFIDF-CORPUS").first().id
+    except:
+        old_ltfidf_id = None
+
+    # 3) we redo the required toolchain parts
+    # -------------------------------------------
+
+    # Instantiate status
+    corpus.status('Recounting mini-workflow', progress=1)
+    corpus.save_hyperdata()
+    session.commit()
+
+    # -> overwrite occurrences (=> NodeNodeNgram)
+    occ_id = compute_occs(corpus,
+                            groupings_id = group_id,
+                            overwrite_id=old_occ_id)
+    print('RECOUNT #%d: [%s] updated occs node #%i' % (corpus.id, t(), occ_id))
+
+    # -> write cumulated ti_ranking (tfidf ranking vector) (=> NodeNodeNgram)
+    tirank_id = compute_ti_ranking(corpus,
+                                   groupings_id = group_id,
+                                   count_scope="global",
+                                   overwrite_id=old_tirank_id)
+    print('RECOUNT #%d: [%s] updated ti ranking node #%i' % (corpus.id, t(), tirank_id))
+
+    # -> write local tfidf similarities to (=> NodeNodeNgram)
+    ltfidf_id = compute_tfidf_local(corpus,
+                                    on_list_id = mainlist_id,
+                                    groupings_id = group_id,
+                                    overwrite_id = old_ltfidf_id)
+    print('RECOUNT #%d: [%s] updated localtfidf node #%i' % (corpus.id, t(), ltfidf_id))
+    # => used for doc <=> ngram association
+
+    # ------------
+    # -> cooccurrences on mainlist: compute + write (=> NodeNgramNgram)
+    coocs = compute_coocs(corpus,
+                            on_list_id = mainlist_id,
+                            groupings_id = group_id,
+                            just_pass_result = True)
+    print('RECOUNT #%d: [%s] updated mainlist coocs for specif rank' % (corpus.id, t()))
+
+    # -> specificity: compute + write (=> NodeNgram)
+    spec_id = compute_specificity(corpus,cooc_matrix = coocs, overwrite_id = old_spec_id)
+
+
+    print('RECOUNT #%d: [%s] updated specificity node #%i' % (corpus.id, t(), spec_id))
+
+    print('RECOUNT #%d: [%s] FINISHED metric recounts' % (corpus.id, t()))
+
+    corpus.status('Recounting mini-workflow', progress=10, complete=True)
+    corpus.save_hyperdata()
+    session.commit()
+
+
+
+
+
 def t():
     return datetime.now().strftime("%Y-%m-%d_%H:%M:%S")

gargantext/views/api/metrics.py

+from gargantext.util.db_cache   import cache
+from gargantext.util.http       import ValidationException, APIView \
+                                     , HttpResponse, JsonHttpResponse
+from gargantext.util.toolchain.main import recount
+from datetime                   import datetime
+
+class CorpusMetrics(APIView):
+
+    def patch(self, request, corpusnode_id):
+        """
+        PATCH triggers recount of metrics for the specified corpus.
+
+        ex PATCH http://localhost:8000/api/metrics/14072
+                                                   -----
+                                                 corpus_id
+        """
+        print("==> update metrics request on ", corpusnode_id)
+
+        if not request.user.is_authenticated():
+            # can't use @requires_auth because of positional 'self' within class
+            return HttpResponse('Unauthorized', status=401)
+
+        try:
+            corpus = cache.Node[int(corpusnode_id)]
+        except:
+            corpus = None
+
+        if corpus is None:
+            raise ValidationException("%s is not a valid corpus node id."
+                                        % corpusnode_id)
+
+        else:
+            t_before = datetime.now()
+            # =============
+            recount(corpus)
+            # =============
+            t_after = datetime.now()
+
+            return JsonHttpResponse({
+                'corpus_id' : corpusnode_id,
+                'took': "%f s." % (t_after - t_before).total_seconds()
+            })

gargantext/views/api/urls.py

 from django.conf.urls import url
 
 from . import nodes
+from . import metrics
 from . import ngramlists
 from . import analytics
 
@@ -19,6 +20,14 @@ urlpatterns = [ url(r'^nodes$'                , nodes.NodeListResource.as_view()
               , url(r'^nodes/(\d+)/favorites$', nodes.CorpusFavorites.as_view()      )
               # in these two routes the node is supposed to be a *corpus* node
 
+
+              , url(r'^metrics/(\d+)$',         metrics.CorpusMetrics.as_view()      )
+                # update all metrics for a corpus
+                #  ex: PUT metrics/123
+                #                     \
+                #                   corpus id
+
+
               , url(r'^ngramlists/change$', ngramlists.ListChange.as_view()          )
                 # add or remove ngram from a list
                 #  ex: add <=> PUT ngramlists/change?list=42&ngrams=1,2

static/lib/gargantext/garganrest.js

@@ -69,10 +69,11 @@ var Resource = function(url_path) {
         });
     };
     // change an item
-    this.change = this.update = function(item, callback) {
+    this.change = this.update = function(id, callback) {
         $.ajax({
-            url: url_path + '/' + item.id,
+            url: url_path + '/' + id,
             type: 'PATCH',
+
             success: callback
         });
     };
@@ -84,14 +85,18 @@ var Resource = function(url_path) {
         $.ajax({
             url: url_path + '/' + id,
             type: 'DELETE',
+
             success: callback
         });
     };
     // add an item
     this.add = this.append = function(value, callback) {
         $.ajax({
+
+            // todo define id
             url: url_path + '/' + id,
             type: 'POST',
+
             success: callback
         });
     };
@@ -99,12 +104,12 @@ var Resource = function(url_path) {
 
 var GarganRest = function(base_path, path_list) {
     var that = this;
-    $.each(path_list, function(p, path){
+    $.each(path_list, function(i, path){
         that[path] = new Resource(base_path + path);
     });
 };
 
-garganrest = new GarganRest('/api/', ['nodes']);
+garganrest = new GarganRest('/api/', ['nodes', 'metrics']);
 
 
 // var log = function(result){console.log(result);};

templates/pages/projects/project.html

@@ -88,13 +88,29 @@
                                             {{corpus.name}}, {{ corpus.count }} documents {{ corpus.status_message }}
                                         </a>
                                     </div>
-                                    <div class="col-md-2 content">
-                                        <a href="/projects/{{project.id}}/corpora/{{corpus.id}}">
+                                    <div class="col-md-3 content">
+                                        <a href="/projects/{{project.id}}/corpora/{{corpus.id}}"
+                                        title="View the corpus">
                                         <button type="button" class="btn btn-default" aria-label="Left Align">
                                               <span class="glyphicon glyphicon-eye-open" aria-hidden="true"></span>
                                         </button>
 
                                         </a>
+                                        <!--  -->
+                                        <button type="button" class="btn btn-default yopla" data-container="body" data-toggle="popover" data-placement="bottom"  data-trigger="focus"
+                                            data-content="
+                                            <ul>
+                                                <li
+                                                onclick=&quot;
+                                                        garganrest.metrics.update({{corpus.id}}, function(){alert('The corpus ({{corpus.name|escapejs}}) was updated')});
+                                                        &quot;>
+                                                    <a href='#'>Recalculate ngram metrics</a> <br/> (can take a little while)
+                                                </li>
+                                            </ul>
+                                            ">
+                                            <span class="glyphicon glyphicon-dashboard" aria-hidden="true"
+                                            title='Recalculate ngram scores and similarities'></span>
+                                        </button>
                                         <button type="button" class="btn btn-default" data-container="body" data-toggle="popover" data-placement="bottom"
                                             data-content="
                                             <ul>
@@ -103,14 +119,15 @@
                                                         garganrest.nodes.delete({{corpus.id}}, function(){$('#corpus_'+{{corpus.id}}).remove()});
                                                         $(this).parent().parent().remove();
                                                     &quot;>
-                                                    <a href=&quot;#&quot;>Delete this</a>
+                                                    <a href='#'>Delete this</a>
                                                 </li>
                                             </ul>
                                             ">
-                                            <span class="glyphicon glyphicon-trash" aria-hidden="true"></span>
+                                            <span class="glyphicon glyphicon-trash" aria-hidden="true"
+                                            title='Delete this corpus'></span>
                                         </button>
                                     </div>
-                                    <div class="col-md-4 content">
+                                    <div class="col-md-3 content">
                                         {% for state in corpus.hyperdata.statuses %}
                                                 {% ifequal state.action "Workflow" %}
                                                     {% if state.complete %}