Add TIMSS processing script

scripts/download_timss_locally.py

@@ -0,0 +1,51 @@
+#!/usr/bin/env python3
+"""
+TIMSS 2019 Processing Script
+=============================
+TIMSS data requires manual download from:
+  https://timssandpirls.bc.edu/timss2019/international-database/
+
+Steps:
+1. Go to URL above
+2. Download SPSS data files for your countries of interest
+3. Extract the zip
+4. Run: python download_timss_locally.py <path_to_extracted_data>
+
+Key files in download:
+  BSA*.sav = Student Achievement (item scores) 
+  BSG*.sav = Student Background (demographics)
+  
+Merge on IDCNTRY + IDSTUD for DIF analysis.
+ITSEX: 1=Girl, 2=Boy
+Item columns: M071001, S071001, etc.
+"""
+import os, sys, json, glob
+import pandas as pd
+import pyreadstat
+
+def process(data_dir):
+    OUT = "timss_2019_processed"
+    os.makedirs(OUT, exist_ok=True)
+    
+    savs = glob.glob(os.path.join(data_dir, "**/*.sav"), recursive=True)
+    ach = [f for f in savs if os.path.basename(f).upper().startswith(('ASA','BSA'))]
+    bg = [f for f in savs if os.path.basename(f).upper().startswith(('ASG','BSG'))]
+    
+    for f in ach:
+        df, _ = pyreadstat.read_sav(f)
+        items = [c for c in df.columns if c.startswith(('M0','S0','M1','S1'))]
+        print(f"{os.path.basename(f)}: {df.shape}, {len(items)} items")
+        df.to_parquet(os.path.join(OUT, os.path.basename(f).replace('.sav','.parquet')))
+    
+    for f in bg:
+        df, _ = pyreadstat.read_sav(f)
+        print(f"{os.path.basename(f)}: {df.shape}")
+        if 'ITSEX' in df.columns:
+            print(f"  Gender: {df['ITSEX'].value_counts().to_dict()}")
+        df.to_parquet(os.path.join(OUT, os.path.basename(f).replace('.sav','.parquet')))
+
+if __name__ == "__main__":
+    if len(sys.argv) < 2:
+        print(__doc__)
+    else:
+        process(sys.argv[1])