Merge pull request #363 from VoSeq/dataset-fields

Dataset fields

.travis.yml

@@ -5,6 +5,7 @@ sudo: required
 python:
     - "3.4"
     - "3.5"
+    - "3.6"
 
 services:
     - elasticsearch

README.rst

@@ -14,7 +14,7 @@ VoSeq
     :alt: Join the chat at https://gitter.im/VoSeq/VoSeq
     :target: https://gitter.im/VoSeq/VoSeq?utm_source=badge&utm_medium=badge&utm_campaign=pr-badge&utm_content=badge
 
-.. |Python_versions| image:: https://img.shields.io/badge/python-3.4%203.5-blue.svg
+.. |Python_versions| image:: https://img.shields.io/badge/python-3.4%203.5%203.6-blue.svg
     :alt: Python versions
 
 

voseq/create_dataset/utils.py

@@ -145,21 +145,22 @@ def get_all_sequences(self):
         """
         seqs_dict = {}
 
-        all_seqs = Sequences.objects.all().values('code_id',
-                                                  'gene_code',
-                                                  'sequences').order_by('code_id')
+        all_seqs = Sequences.objects.filter(
+            code__in=self.voucher_codes,
+            gene_code__in=self.gene_codes,
+        ).values('code_id', 'gene_code', 'sequences').order_by('code_id')
+
         for seq in all_seqs:
             code = seq['code_id']
             gene_code = seq['gene_code']
 
-            if code in self.voucher_codes and gene_code in self.gene_codes:
-                if code not in seqs_dict:
-                    seqs_dict[code] = {gene_code: ''}
-                seqs_dict[code][gene_code] = seq
+            if code not in seqs_dict:
+                seqs_dict[code] = {gene_code: ''}
+            seqs_dict[code][gene_code] = seq
         return seqs_dict
 
     def build_seq_obj(self, code, gene_code, our_taxon_names, all_seqs):
-        """Builds a SeqRecordExpanded object. I cannot be built, returns None.
+        """Builds a SeqRecordExpanded object. If cannot be built, returns None.
 
         """
         this_voucher_seqs = self.extract_sequence_from_all_seqs_in_db(all_seqs, code, gene_code)
@@ -221,20 +222,20 @@ def get_taxon_names_for_taxa(self):
         """
         vouchers_with_taxon_names = {}
 
-        all_vouchers = Vouchers.objects.all().order_by('code').values('code', 'orden', 'superfamily',
-                                                                      'family', 'subfamily', 'tribe',
-                                                                      'subtribe', 'genus', 'species',
-                                                                      'subspecies', 'author', 'hostorg',)
+        # TODO: add gene_code. drop it for now
+        taxon_names = [i.lower() for i in self.taxon_names]
+        if "code" not in taxon_names:
+            taxon_names.append("code")
+        try:
+            taxon_names.remove("genecode")
+        except ValueError:
+            pass
+        all_vouchers = Vouchers.objects.filter(
+            code__in=self.voucher_codes,
+        ).order_by('code').values(*taxon_names)
         for voucher in all_vouchers:
             code = voucher['code']
-            if code in self.voucher_codes:
-                obj = dict()
-                for taxon_name in self.taxon_names:
-                    if taxon_name != 'GENECODE':
-                        taxon_name = taxon_name.lower()
-                        obj[taxon_name] = voucher[taxon_name]
-                vouchers_with_taxon_names[code] = obj
-
+            vouchers_with_taxon_names[code] = voucher
         return vouchers_with_taxon_names
 
     def get_gene_codes_metadata(self):

voseq/create_dataset/views.py

@@ -37,7 +37,11 @@ def results(request):
         if form.is_valid():
             dataset_format = form.cleaned_data['file_format']
             dataset_creator = CreateDataset(form.cleaned_data)
-            dataset = dataset_creator.dataset_str[0:1500] + '\n...\n\n\n' + '#######\nComplete dataset file available for download.\n#######'
+            dataset = "{}{}{}".format(
+                dataset_creator.dataset_str[0:1500],
+                '\n...\n\n\n',
+                '#######\nComplete dataset file available for download.\n#######',
+            )
             errors = dataset_creator.errors
             warnings = set(dataset_creator.warnings)
 

voseq/public_interface/models.py

@@ -147,7 +147,8 @@ class Vouchers(TimeStampedModel):
         (LOST, 'lost'),
         (UNKNOWN, 'unknown'),
     )
-    code = models.CharField(max_length=300, unique=True, primary_key=True, help_text="Voucher code.")
+    code = models.CharField(
+        max_length=300, unique=True, primary_key=True, db_index=True, help_text="Voucher code.")
     orden = models.TextField(blank=True)
     superfamily = models.TextField(blank=True)
     family = models.TextField(blank=True)
@@ -208,8 +209,8 @@ def __str__(self):
 
 
 class Sequences(models.Model):
-    code = models.ForeignKey(Vouchers, help_text='This is your voucher code.')
-    gene_code = models.CharField(max_length=100)
+    code = models.ForeignKey(Vouchers, db_index=True, help_text='This is your voucher code.')
+    gene_code = models.CharField(max_length=100, db_index=True)
     sequences = models.TextField(blank=True)
     accession = models.CharField(max_length=100, blank=True)
     lab_person = models.CharField(max_length=100, blank=True)

voseq/stats/tests.py

@@ -19,7 +19,7 @@ def test_model_stats_str(self):
         self.assertEqual("10 vouchers", str(res))
 
     def test_model_voucherspergene_str(self):
-        res = VouchersPerGene.objects.get(id=1)
+        res = VouchersPerGene.objects.get(gene_code="wingless")
         self.assertEqual("10 vouchers in gene wingless", str(res))
 
         for code in ['CP100-11', 'CP100-12', 'CP100-13', 'CP100-14', 'CP100-15',
@@ -28,7 +28,7 @@ def test_model_voucherspergene_str(self):
             s = Sequences.objects.get(code=v, gene_code="wingless")
             s.delete()
         call_command('create_stats')
-        res = VouchersPerGene.objects.get(id=1)
+        res = VouchersPerGene.objects.get(gene_code="wingless")
         self.assertEqual("1 voucher in gene wingless", str(res))
 
     def test_create_stats(self):

voseq/voseq/settings/local.py

@@ -100,3 +100,6 @@ def get_secret(setting, secrets=secrets):
 }
 HAYSTACK_DEFAULT_OPERATOR = 'AND'
 # HAYSTACK_IDENTIFIER_METHOD = 'public_interface.models.get_identifier'
+
+if DEBUG is True:
+    INTERNAL_IPS = ["127.0.0.1"]

voseq/voseq/urls.py

@@ -35,3 +35,9 @@
     urlpatterns += [
         url(r'^media/(?P<path>.*)$', views.serve)
     ]
+
+if settings.DEBUG:
+    import debug_toolbar
+    urlpatterns = [
+        url(r'^__debug__/', include(debug_toolbar.urls)),
+    ] + urlpatterns