Merge pull request #1218 from nextstrain/james/export-fixes

Fixes to augur export

Author: huddlej

CHANGES.md

@@ -4,9 +4,15 @@
 
 ### Bug fixes
 
+* export: No longer export duplicate entries in the colorings array, a bug which has been present in Augur since at least v12 [#719][]. [#1218][] (@jameshadfield)
+
+* export: In version 22.0.0, some configurations of export may have resulted in the clade coloring appearing last in the Auspice dropdown rather than first. This is now fixed. [#1218] (@jameshadfield)
+
 * export: In version 22.0.0, validation of `augur.utils.read_node_data` was changed to error when a node data JSON did not contain any actual data. This causes export to error when an empty node data JSON is passed, as for example in ncov's pathogen-ci. This is now fixed by warning instead. The bug was originally introduced in PR [#728][]. [#1214][] (@corneliusroemer)
 
+[#719]: https://github.com/nextstrain/augur/issues/719
 [#1214]: https://github.com/nextstrain/augur/pull/1214
+[#1218]: https://github.com/nextstrain/augur/pull/1218
 
 ## 22.0.0 (9 May 2023)
 

augur/export_v2.py

@@ -355,12 +355,14 @@ def _add_title_and_type(coloring):
             return False # a warning message will have been printed before `InvalidOption` is raised
         return coloring
 
-    def _create_coloring(key):
+    def _add_coloring(colorings, key):
         # handle deprecations
         if key == "authors":
             deprecated("[colorings] The 'authors' key is now called 'author'")
             key = "author"
-        return {"key": key}
+        # check if the key has already been added by another part of the color-creating logic
+        if key not in {x['key'] for x in colorings}:
+            colorings.append({"key": key})
 
     def _is_valid(coloring):
         key = coloring["key"]
@@ -389,18 +391,18 @@ def _get_colorings():
         if command_line_colorings:
             # start with auto_colorings (already validated to be included)
             for x in auto_colorings:
-                colorings.append(_create_coloring(x))
+                _add_coloring(colorings, x)
             # then add in command line colorings
             for x in command_line_colorings:
-                colorings.append(_create_coloring(x))
+                _add_coloring(colorings, x)
         else:
             # if we have a config file, start with these (extra info, such as title&type, is added in later)
             if config:
                 for x in config.keys():
-                    colorings.append(_create_coloring(x))
+                    _add_coloring(colorings, x)
             # then add in any auto-colorings already validated to include
             for x in auto_colorings:
-                colorings.append(_create_coloring(x))
+                _add_coloring(colorings, x)
 
         explicitly_defined_colorings = [x["key"] for x in colorings]
         # add in genotype as a special case if (a) not already set and (b) the data supports it
@@ -794,8 +796,9 @@ def node_data_prop_is_normal_trait(name):
         "authors",          # authors are set as a node property, not a trait property
         "author",           # see above
         "vaccine",          # vaccine info is stored as a "special" node prop
+        'clade_membership', # explicitly set as a coloring if present
         'branch_length',
-        'num_date',
+        'num_date',         # explicitly set as a coloring if present
         'raw_date',
         'numdate',
         'clock_length',

tests/functional/export_v2/dataset-with-branch-labels.json

@@ -25,13 +25,13 @@
     },
     "colorings": [
       {
-        "key": "gt",
-        "title": "Genotype",
+        "key": "clade_membership",
+        "title": "Clade",
         "type": "categorical"
       },
       {
-        "key": "clade_membership",
-        "title": "Clade",
+        "key": "gt",
+        "title": "Genotype",
         "type": "categorical"
       }
     ],

tests/functional/export_v2/dataset1.json

@@ -69,30 +69,6 @@
         ],
         "title": "Mutations per branch",
         "type": "continuous"
-      },
-      {
-        "key": "location",
-        "legend": [
-          {
-            "display": "\u03b1",
-            "value": "alpha"
-          },
-          {
-            "value": "beta"
-          }
-        ],
-        "scale": [
-          [
-            "beta",
-            "#bd0026"
-          ],
-          [
-            "gamma",
-            "#6a51a3"
-          ]
-        ],
-        "title": "Location",
-        "type": "categorical"
       }
     ],
     "filters": [

tests/functional/export_v2/dataset2.json

@@ -69,30 +69,6 @@
         ],
         "title": "Mutations per branch",
         "type": "continuous"
-      },
-      {
-        "key": "location",
-        "legend": [
-          {
-            "display": "\u03b1",
-            "value": "alpha"
-          },
-          {
-            "value": "beta"
-          }
-        ],
-        "scale": [
-          [
-            "beta",
-            "#bd0026"
-          ],
-          [
-            "gamma",
-            "#6a51a3"
-          ]
-        ],
-        "title": "Location",
-        "type": "categorical"
       }
     ],
     "filters": [

tests/functional/export_v2/dataset3.json

@@ -82,30 +82,6 @@
             "#4042C7"
           ]
         ]
-      },
-      {
-        "key": "location",
-        "title": "Location",
-        "type": "categorical",
-        "scale": [
-          [
-            "beta",
-            "#bd0026"
-          ],
-          [
-            "gamma",
-            "#6a51a3"
-          ]
-        ],
-        "legend": [
-          {
-            "value": "alpha",
-            "display": "\u03b1"
-          },
-          {
-            "value": "beta"
-          }
-        ]
       }
     ],
     "filters": [