Script to clean encoding for https://github.com/druedin/swisscantonsmod/issues/2

shp_fix.py

import shapefile

def replace_escape(s):
    return s.replace(b'\xc3\x82\xc2',b'\xc2').replace(b'\xc3\x83\xc2',b'\xc3')

sf = shapefile.Reader("swisscantonsmod/ch-cantons.dbf")
print(f"records: {len(sf.records())}")
print(f"shapes: {len(sf.shapes())}")
out = shapefile.Writer("swisscantonsmod/ch-cantons_fixed.dbf")
for f in sf.fields:
    out.field(*f)

def clean_rec(rec):
    return tuple(replace_escape(r.encode()).decode() if isinstance(r, str) else r for r in rec)

for rec in sf.records():
    out.record(*clean_rec(rec))

for shape in sf.shapes():
    out.shape(shape)