Fix ZCTA universe parsing and table-ACS ZCTA ingestion

Author: iandorsey00

geocompare/database/Database.py

@@ -1198,13 +1198,22 @@ def __init__(self, path, progress_callback=None):
         # 310 = Metro/Micro Area
         # 400 = Urban Area
         # 860 = ZCTA
-        rows = [
-            row
-            for row in rows
-            if row[1] in {"010", "160", "050", "040", "860", "310", "400"}
-            and len(row[3]) >= 5
-            and row[3][3:5] == "00"
-        ]
+        allowed_sumlevels = {"010", "160", "050", "040", "860", "310", "400"}
+
+        def _keep_geography_row(row):
+            sumlevel = row[1]
+            geoid = row[3]
+            if sumlevel not in allowed_sumlevels or len(geoid) < 5:
+                return False
+
+            # Table-based ACS uses non-"00" GEOID middle codes for some
+            # summary levels (notably ZCTA = 860Z200USxxxxx).
+            if self.acs_layout == "table":
+                return "US" in geoid
+
+            return geoid[3:5] == "00"
+
+        rows = [row for row in rows if _keep_geography_row(row)]
         rows = [
             [
                 row[0],  # STUSAB [lowercase]

geocompare/tools/query_syntax.py

@@ -102,10 +102,12 @@ def build_context(
     elif in_zcta:
         group = in_zcta
 
-    if universe and group:
-        return f"{universe}+{group}"
-    if universe:
-        return f"{universe}+"
+    normalized_universe = universe.lower() if universe else None
+
+    if normalized_universe and group:
+        return f"{normalized_universe}+{group}"
+    if normalized_universe:
+        return f"{normalized_universe}+"
     if group:
         return group
     return ""

geocompare/tools/summary_level_parser.py

@@ -36,7 +36,7 @@ def __init__(self):
         }
 
     def is_summary_level_keyword(self, input_str: str) -> bool:
-        return input_str in self.keyword_to_code
+        return input_str.lower() in self.keyword_to_code
 
     def is_summary_level_code(self, input_str: str) -> bool:
         return input_str in self.code_to_keyword
@@ -48,6 +48,7 @@ def parse_context(self, context: str) -> Tuple[Optional[str], Optional[str], Opt
         if context:
             if "+" in context:
                 universe_sl, group = context.split("+", 1)
+                universe_sl = universe_sl.lower()
                 if self.is_summary_level_keyword(universe_sl):
                     universe_sl = self.keyword_to_code[universe_sl]
                 elif not self.is_summary_level_code(universe_sl):