feat(census): add demographic columns to EES census tap and staging models

tap-uk-ees: EESCensusStream now declares 27 data columns (FSM %, EAL %,
ethnicity breakdowns, pupil counts) with clean Singer field names mapped
from the verbose CSV column names (e.g. '% of pupils known to be eligible
for free school meals' → fsm_pct) via a new _column_renames mechanism on
the base stream class.

stg_ees_census: materialised as table, applies safe_numeric to all
percentage/count columns, filters to numeric URNs.

int_pupil_chars_merged + fact_pupil_characteristics: pass all columns
through from staging (previously stubs with only 3 columns).

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

pipeline/plugins/extractors/tap-uk-ees/tap_uk_ees/tap.py

@@ -44,6 +44,8 @@ class EESDatasetStream(Stream):
 
     Subclasses set _target_filename to a keyword that appears in the
     target CSV path inside the ZIP (substring match, not exact).
+    Subclasses may set _column_renames to map messy CSV column names to
+    clean Singer field names before yielding records.
     """
 
     replication_key = None
@@ -51,6 +53,7 @@ class EESDatasetStream(Stream):
     _target_filename: str = ""  # keyword that appears in the CSV path
     _urn_column: str = "school_urn"  # column name for URN in the CSV
     _encoding: str = "utf-8"  # CSV file encoding (some DfE files use latin-1)
+    _column_renames: dict = {}  # CSV column name → Singer field name
 
     def get_records(self, context):
         import pandas as pd
@@ -104,6 +107,10 @@ class EESDatasetStream(Stream):
 
         for _, row in df.iterrows():
             record = row.to_dict()
+            # Apply subclass column renames (messy CSV names → clean Singer fields)
+            for old, new in self._column_renames.items():
+                if old in record:
+                    record[new] = record.pop(old)
             # Normalise URN column to 'school_urn' for consistency
             if self._urn_column in record and self._urn_column != "school_urn":
                 record["school_urn"] = record.pop(self._urn_column)
@@ -281,15 +288,75 @@ class EESCensusStream(EESDatasetStream):
     _target_filename = "spc_school_level_underlying_data"
     _urn_column = "urn"
     _encoding = "latin-1"
+
+    # Map verbose CSV column names to clean Singer field names
+    _column_renames = {
+        "headcount of pupils":                                                        "total_pupils",
+        "headcount total female":                                                     "female_pupils",
+        "headcount total male":                                                       "male_pupils",
+        "full-time pupils":                                                           "full_time_pupils",
+        "part-time pupils":                                                           "part_time_pupils",
+        "total boarders":                                                             "total_boarders",
+        "number of pupils known to be eligible for free school meals":                "fsm_eligible_n",
+        "% of pupils known to be eligible for free school meals":                    "fsm_pct",
+        "% of pupils whose first language is known or believed to be other than English": "eal_pct",
+        "% of pupils who are a young carer":                                          "young_carer_pct",
+        "% of pupils classified as white British ethnic origin":                      "ethnicity_white_british_pct",
+        "% of pupils classified as any other white background ethnic origin":         "ethnicity_white_other_pct",
+        "% of pupils classified as Gypsy/Roma ethnic origin":                        "ethnicity_gypsy_roma_pct",
+        "% of pupils classified as white and black Caribbean ethnic origin":          "ethnicity_mixed_wbc_pct",
+        "% of pupils classified as white and black African ethnic origin":            "ethnicity_mixed_wba_pct",
+        "% of pupils classified as white and Asian ethnic origin":                   "ethnicity_mixed_wa_pct",
+        "% of pupils classified as any other mixed background ethnic origin":         "ethnicity_mixed_other_pct",
+        "% of pupils classified as Indian ethnic origin":                             "ethnicity_asian_indian_pct",
+        "% of pupils classified as Pakistani ethnic origin":                          "ethnicity_asian_pakistani_pct",
+        "% of pupils classified as Bangladeshi ethnic origin":                        "ethnicity_asian_bangladeshi_pct",
+        "% of pupils classified as any other Asian background ethnic origin":         "ethnicity_asian_other_pct",
+        "% of pupils classified as Caribbean ethnic origin":                          "ethnicity_black_caribbean_pct",
+        "% of pupils classified as African ethnic origin":                            "ethnicity_black_african_pct",
+        "% of pupils classified as any other black background ethnic origin":         "ethnicity_black_other_pct",
+        "% of pupils classified as Chinese ethnic origin":                            "ethnicity_chinese_pct",
+        "% of pupils classified as any other ethnic group ethnic origin":             "ethnicity_other_pct",
+        "% of pupils unclassified":                                                   "ethnicity_unclassified_pct",
+    }
+
     schema = th.PropertiesList(
         th.Property("time_period", th.StringType, required=True),
         th.Property("school_urn", th.StringType, required=True),
         th.Property("school_name", th.StringType),
         th.Property("laestab", th.StringType),
         th.Property("phase_type_grouping", th.StringType),
-        # TODO: Add data columns (ethnicity %, FSM %, SEN %, etc.) once
-        # actual column names are verified on the container. The CSV has
-        # 269 columns — only the first 30 (metadata) have been inspected.
+        # Pupil counts
+        th.Property("total_pupils", th.StringType),
+        th.Property("female_pupils", th.StringType),
+        th.Property("male_pupils", th.StringType),
+        th.Property("full_time_pupils", th.StringType),
+        th.Property("part_time_pupils", th.StringType),
+        th.Property("total_boarders", th.StringType),
+        # FSM
+        th.Property("fsm_eligible_n", th.StringType),
+        th.Property("fsm_pct", th.StringType),
+        # EAL & young carers
+        th.Property("eal_pct", th.StringType),
+        th.Property("young_carer_pct", th.StringType),
+        # Ethnicity percentages
+        th.Property("ethnicity_white_british_pct", th.StringType),
+        th.Property("ethnicity_white_other_pct", th.StringType),
+        th.Property("ethnicity_gypsy_roma_pct", th.StringType),
+        th.Property("ethnicity_mixed_wbc_pct", th.StringType),
+        th.Property("ethnicity_mixed_wba_pct", th.StringType),
+        th.Property("ethnicity_mixed_wa_pct", th.StringType),
+        th.Property("ethnicity_mixed_other_pct", th.StringType),
+        th.Property("ethnicity_asian_indian_pct", th.StringType),
+        th.Property("ethnicity_asian_pakistani_pct", th.StringType),
+        th.Property("ethnicity_asian_bangladeshi_pct", th.StringType),
+        th.Property("ethnicity_asian_other_pct", th.StringType),
+        th.Property("ethnicity_black_caribbean_pct", th.StringType),
+        th.Property("ethnicity_black_african_pct", th.StringType),
+        th.Property("ethnicity_black_other_pct", th.StringType),
+        th.Property("ethnicity_chinese_pct", th.StringType),
+        th.Property("ethnicity_other_pct", th.StringType),
+        th.Property("ethnicity_unclassified_pct", th.StringType),
     ).to_dict()