Merge pull request #3297 from antgonza/redo-3275

Fix #3275

Author: charles-cowart

qiita_pet/handlers/admin_processing_job.py

@@ -17,6 +17,7 @@
 from qiita_db.exceptions import QiitaDBUnknownIDError
 
 from json import dumps
+from collections import Counter
 
 
 class AdminProcessingJobBaseClass(BaseHandler):
@@ -109,21 +110,43 @@ def post(self):
         # Get user-inputted qiita id and sample names
         qid = self.get_argument("qid")
         snames = self.get_argument("snames").split()
-        error, matching, missing, extra, blank = [None]*5
+        error, matching, missing, extra, blank, duplicates = [None]*6
 
         # Stripping leading qiita id from sample names
         # Example: 1.SKB1.640202 -> SKB1.640202
         try:
-            qsnames = list(Study(qid).sample_template)
+            sample_info = Study(qid).sample_template
+            qsnames = list(sample_info)
         except TypeError:
             error = f'Study {qid} seems to have no sample template'
         except QiitaDBUnknownIDError:
             error = f'Study {qid} does not exist'
 
         if error is None:
+            # if tube_id is present then this should take precedence in qsnames
+            tube_ids = dict()
+            if "tube_id" in sample_info.categories:
+                for k, v in sample_info.get_category("tube_id").items():
+                    # ignoring empty values
+                    if v in (None, 'None', ''):
+                        continue
+                    if k.startswith(qid):
+                        k = k.replace(f'{qid}.', "", 1)
+                    tube_ids[k] = v
+
             for i, qsname in enumerate(qsnames):
                 if qsname.startswith(qid):
-                    qsnames[i] = qsname.replace(f'{qid}.', "", 1)
+                    qsname = qsname.replace(f'{qid}.', "", 1)
+                if qsname in tube_ids:
+                    nname = f'{qsname}, tube_id: {tube_ids[qsname]}'
+                    snames = [s if s != tube_ids[qsname] else nname
+                              for s in snames]
+                    qsname = nname
+                qsnames[i] = qsname
+
+            # Finds duplicates in the samples
+            seen = Counter(snames)
+            duplicates = [f'{s} \u00D7 {seen[s]}' for s in seen if seen[s] > 1]
 
             # Remove blank samples from sample names
             blank = [x for x in snames if x.lower().startswith('blank')]
@@ -137,4 +160,5 @@ def post(self):
             extra = snames.difference(qsnames)
 
         self.render("sample_validation.html", input=False, matching=matching,
-                    missing=missing, extra=extra, blank=blank, error=error)
+                    missing=missing, extra=extra, blank=blank,
+                    duplicates=duplicates, error=error)

qiita_pet/templates/sample_validation.html

@@ -3,7 +3,7 @@
 <style>
     .column {
     float: left;
-    width: 25%;
+    width: 20%;
     }
 
     .row:after {
@@ -53,6 +53,14 @@ <h2>Blank</h2>
             {% end %}
         </ul>
     </div>
+    <div class="column">
+        <h2>Duplicates</h2>
+        <ul>
+            {% for sample in duplicates %}
+            <li>{{ sample }}</li>
+            {% end %}
+        </ul>
+    </div>
     <div class="column">
         <h2>Extra</h2>
         <ul>

qiita_pet/test/test_admin_processing_job_handlers.py

@@ -8,10 +8,11 @@
 
 from unittest import main
 from json import loads
-
+import pandas as pd
 from mock import Mock
 
 from qiita_db.user import User
+from qiita_db.metadata_template.sample_template import SampleTemplate as ST
 from qiita_pet.handlers.base_handlers import BaseHandler
 from qiita_pet.test.tornado_test_base import TestHandlerBase
 
@@ -63,6 +64,21 @@ def test_post(self):
         for name in snames:
             self.assertIn(name, body)
 
+        # Check success with tube_id
+        md_dict = {'SKB1.640202': {'tube_id': '12345'}}
+        md_ext = pd.DataFrame.from_dict(md_dict, orient='index', dtype=str)
+        ST(1).extend(md_ext)
+        post_args = {
+            'qid': 1,
+            'snames': '12345 SKB2.640194 BLANK.1A BLANK.1B'
+        }
+        response = self.post('/admin/sample_validation/', post_args)
+        self.assertEqual(response.code, 200)
+        snames = ['SKB2.640194', 'SKB1.640202, tube_id: 12345']
+        body = response.body.decode('ascii')
+        for name in snames:
+            self.assertIn(name, body)
+
         # Check failure: invalid qiita id
         post_args = {
             'qid': 2,