Generate special svg names. (#308)

datacommonsorg · May 22, 2024 · 0daedbf · 0daedbf
1 parent 14c3dd8
commit 0daedbf
Show file tree

Hide file tree

Showing 9 changed files with 243 additions and 194 deletions.
diff --git a/simple/requirements.txt b/simple/requirements.txt
@@ -9,6 +9,7 @@ idna==3.4
 importlib-metadata==6.8.0
 numpy==1.25.2
 pandas==2.1.0
+parameterized==0.9.0
 platformdirs==3.10.0
 PyMySQL==1.1.0
 python-dateutil==2.8.2

diff --git a/simple/stats/stat_var_hierarchy_generator.py b/simple/stats/stat_var_hierarchy_generator.py
@@ -120,11 +120,27 @@ def gen_svg_name(self):
       svg_name = f"{svg_name} With {pvs_str}"
     return svg_name
 
+  def gen_specialized_name(self, parent_pvs: Self) -> str:
+    parent_parts = parent_pvs._get_pv_parts()
+    child_parts = self._get_pv_parts()
+    parts = [part for part in child_parts if part not in parent_parts]
+    return ", ".join(map(lambda part: _capitalize_and_split(part), parts))
+
   # Creates and returns a new SVPropVals object with the same fields as this object
   # except for PVs which are set to the specified list.
   def with_pvs(self, pvs: list[PropVal]) -> Self:
     return replace(self, pvs=pvs)
 
+  # Returns an ordered set of PVs as a dict (since sets don't maintain order).
+  def _get_pv_parts(self) -> dict[str, bool]:
+    parts: dict[str, bool] = {}
+    for pv in self.pvs:
+      if pv.prop:
+        parts[pv.prop] = True
+      if pv.val:
+        parts[pv.val] = True
+    return parts
+
 
 class SVG:
 
@@ -136,7 +152,7 @@ def __init__(self, svg_id: str, svg_name: str) -> None:
     # Maintaining order maintains results consistency and helps with tests.
     self.sv_ids: dict[str, bool] = {}
     self.parent_svg_ids: dict[str, bool] = {}
-    self.child_svg_ids: dict[str, bool] = {}
+    self.child_svg_id_2_specialized_name: dict[str, str] = {}
     self.measured_properties: dict[str, bool] = {}
 
     self.parent_svgs_processed: bool = False
@@ -171,14 +187,19 @@ def triples(self) -> list[Triple]:
 
     return triples
 
+  def gen_specialized_name(self, parent_svg: Self) -> str:
+    if self.sample_sv and parent_svg.sample_sv:
+      return self.sample_sv.gen_specialized_name(parent_svg.sample_sv)
+    return ""
+
   # For testing.
   def json(self) -> dict:
     return {
         "svg_id": self.svg_id,
         "svg_name": self.svg_name,
         "sv_ids": list(self.sv_ids.keys()),
         "parent_svg_ids": list(self.parent_svg_ids.keys()),
-        "child_svg_ids": list(self.child_svg_ids.keys()),
+        "child_svg_specialized_names": self.child_svg_id_2_specialized_name,
         "mprops": sorted(list(self.measured_properties.keys()))
     }
 
@@ -210,7 +231,7 @@ def _attach_verticals(poptype2svg: dict[str, SVG],
     for vertical in vertical_spec.verticals:
       vertical_svg = _get_or_create_vertical_svg(vertical, vertical_svgs)
       vertical_svgs[vertical_svg.svg_id] = vertical_svg
-      vertical_svg.child_svg_ids[pop_type_svg.svg_id] = True
+      vertical_svg.child_svg_id_2_specialized_name[pop_type_svg.svg_id] = ""
       pop_type_svg.parent_svg_ids[vertical_svg.svg_id] = True
       for mprop in vertical_spec.measured_properties:
         vertical_svg.measured_properties[mprop] = True
@@ -298,7 +319,8 @@ def _create_parent_svg(parent_sv: SVPropVals, svg: SVG, svgs: dict[str, SVG],
 
   # Add parent child relationships.
   svg.parent_svg_ids[parent_svg.svg_id] = True
-  parent_svg.child_svg_ids[svg.svg_id] = True
+  parent_svg.child_svg_id_2_specialized_name[
+      svg.svg_id] = svg.gen_specialized_name(parent_svg)
 
   # Add child mprops to all parents recursively.
   _add_measured_properties_to_parent_svgs(svg.measured_properties,

diff --git a/simple/tests/stats/stat_var_hierarchy_generator_test.py b/simple/tests/stats/stat_var_hierarchy_generator_test.py
@@ -20,6 +20,7 @@
 
 from kg_util import mcf_parser
 import pandas as pd
+from parameterized import parameterized
 from stats.data import Triple
 from stats.stat_var_hierarchy_generator import *
 from stats.stat_var_hierarchy_generator import _extract_svs
@@ -179,3 +180,28 @@ def test_extract_svs(self):
     svs = _extract_svs(input_triples)
 
     self.assertListEqual(svs, expected_svs)
+
+  @parameterized.expand([
+      (SVPropVals(sv_id="",
+                  population_type="",
+                  pvs=[PropVal("gender", ""),
+                       PropVal("race", "Asian")],
+                  measured_property=""),
+       SVPropVals(sv_id="",
+                  population_type="",
+                  pvs=[PropVal("gender", "Female"),
+                       PropVal("race", "Asian")],
+                  measured_property=""), "Female"),
+      (SVPropVals(sv_id="",
+                  population_type="",
+                  pvs=[PropVal("gender", "Female")],
+                  measured_property=""),
+       SVPropVals(sv_id="",
+                  population_type="",
+                  pvs=[PropVal("gender", "Female"),
+                       PropVal("race", "")],
+                  measured_property=""), "Race")
+  ])
+  def test_gen_specialized_name(self, parent: SVPropVals, child: SVPropVals,
+                                expected: str):
+    self.assertEqual(child.gen_specialized_name(parent), expected)
diff --git a/simple/tests/stats/test_data/stat_var_hierarchy_generator/expected/basic_svgs.json b/simple/tests/stats/test_data/stat_var_hierarchy_generator/expected/basic_svgs.json
@@ -6,10 +6,10 @@
   "parent_svg_ids": [
    "c/g/Root"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Race",
-   "c/g/Person_Gender"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Race": "Race",
+   "c/g/Person_Gender": "Gender"
+  },
   "mprops": []
  },
  {
@@ -19,9 +19,9 @@
   "parent_svg_ids": [
    "c/g/Person"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Gender-Female"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Gender-Female": "Female"
+  },
   "mprops": []
  },
  {
@@ -31,9 +31,9 @@
   "parent_svg_ids": [
    "c/g/Person_Gender"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Gender-Female_Race"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Gender-Female_Race": "Race"
+  },
   "mprops": []
  },
  {
@@ -43,9 +43,9 @@
   "parent_svg_ids": [
    "c/g/Person_Gender-Female"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Gender-Female_Race-Asian"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Gender-Female_Race-Asian": "Asian"
+  },
   "mprops": []
  },
  {
@@ -58,7 +58,7 @@
    "c/g/Person_Gender_Race-Asian",
    "c/g/Person_Gender-Female_Race"
   ],
-  "child_svg_ids": [],
+  "child_svg_specialized_names": {},
   "mprops": []
  },
  {
@@ -68,9 +68,9 @@
   "parent_svg_ids": [
    "c/g/Person_Race-Asian"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Gender-Female_Race-Asian"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Gender-Female_Race-Asian": "Female"
+  },
   "mprops": []
  },
  {
@@ -80,9 +80,9 @@
   "parent_svg_ids": [
    "c/g/Person"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Race-Asian"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Race-Asian": "Asian"
+  },
   "mprops": []
  },
  {
@@ -92,9 +92,9 @@
   "parent_svg_ids": [
    "c/g/Person_Race"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Gender_Race-Asian"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Gender_Race-Asian": "Gender"
+  },
   "mprops": []
  }
 ]
diff --git a/simple/tests/stats/test_data/stat_var_hierarchy_generator/expected/main_svgs.json b/simple/tests/stats/test_data/stat_var_hierarchy_generator/expected/main_svgs.json
@@ -9,10 +9,10 @@
   "parent_svg_ids": [
    "c/g/Root"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Race",
-   "c/g/Person_PovertyStatus"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Race": "Race",
+   "c/g/Person_PovertyStatus": "Poverty Status"
+  },
   "mprops": [
    "count",
    "unknownMProp"
@@ -25,10 +25,10 @@
   "parent_svg_ids": [
    "c/g/Person"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months",
-   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months": "Below Poverty Level In The Past12 Months",
+   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months": "Above Poverty Level In The Past12 Months"
+  },
   "mprops": [
    "count",
    "unknownMProp"
@@ -41,9 +41,9 @@
   "parent_svg_ids": [
    "c/g/Person_PovertyStatus"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race": "Race"
+  },
   "mprops": [
    "count"
   ]
@@ -55,9 +55,9 @@
   "parent_svg_ids": [
    "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race-WhiteAlone"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race-WhiteAlone": "White Alone"
+  },
   "mprops": [
    "count"
   ]
@@ -72,7 +72,7 @@
    "c/g/Person_PovertyStatus_Race-WhiteAlone",
    "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race"
   ],
-  "child_svg_ids": [],
+  "child_svg_specialized_names": {},
   "mprops": [
    "count"
   ]
@@ -84,9 +84,9 @@
   "parent_svg_ids": [
    "c/g/Person_PovertyStatus"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race": "Race"
+  },
   "mprops": [
    "count",
    "unknownMProp"
@@ -99,9 +99,9 @@
   "parent_svg_ids": [
    "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race-WhiteAlone"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race-WhiteAlone": "White Alone"
+  },
   "mprops": [
    "count",
    "unknownMProp"
@@ -118,7 +118,7 @@
    "c/g/Person_PovertyStatus_Race-WhiteAlone",
    "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race"
   ],
-  "child_svg_ids": [],
+  "child_svg_specialized_names": {},
   "mprops": [
    "count",
    "unknownMProp"
@@ -131,10 +131,10 @@
   "parent_svg_ids": [
    "c/g/Person_Race-WhiteAlone"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race-WhiteAlone",
-   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race-WhiteAlone"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus-BelowPovertyLevelInThePast12Months_Race-WhiteAlone": "Below Poverty Level In The Past12 Months",
+   "c/g/Person_PovertyStatus-AbovePovertyLevelInThePast12Months_Race-WhiteAlone": "Above Poverty Level In The Past12 Months"
+  },
   "mprops": [
    "count",
    "unknownMProp"
@@ -147,9 +147,9 @@
   "parent_svg_ids": [
    "c/g/Person"
   ],
-  "child_svg_ids": [
-   "c/g/Person_Race-WhiteAlone"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_Race-WhiteAlone": "White Alone"
+  },
   "mprops": [
    "count",
    "unknownMProp"
@@ -164,9 +164,9 @@
   "parent_svg_ids": [
    "c/g/Person_Race"
   ],
-  "child_svg_ids": [
-   "c/g/Person_PovertyStatus_Race-WhiteAlone"
-  ],
+  "child_svg_specialized_names": {
+   "c/g/Person_PovertyStatus_Race-WhiteAlone": "Poverty Status"
+  },
   "mprops": [
    "count",
    "unknownMProp"