Do not produce duplicate coloring entries

The logic for creating colorings involves many steps, and when adding a coloring we had neglected to check if it had already been added. For instance, colorings explicitly defined in the auspice-config which were provided via node-data files (rather than the metadata tsv) would be added twice. In our current nCoV build there are 7 such duplicates! There were also a few duplicates that were present in our tests. Whoops! These weren't a problem for auspice (the first was used), but added unnecessary complexity and bytes to the JSON. Closes #719
nextstrain · May 16, 2023 · 5daf6e8 · 5daf6e8
1 parent 497a5ea
commit 5daf6e8
Show file tree

Hide file tree

Showing 4 changed files with 10 additions and 78 deletions.
diff --git a/augur/export_v2.py b/augur/export_v2.py
@@ -355,12 +355,16 @@ 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 in {x['key'] for x in colorings}:
+            return False
+        colorings.append({"key": key})
+        return True
 
     def _is_valid(coloring):
         key = coloring["key"]
@@ -389,18 +393,18 @@ 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

diff --git a/tests/functional/export_v2/dataset1.json b/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": [

diff --git a/tests/functional/export_v2/dataset2.json b/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": [

diff --git a/tests/functional/export_v2/dataset3.json b/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": [