[Stata] Importing SAS (XPT & SAS7BDAT) Files into Stata with Value Labels using Python

import sasxport5 “dataset.xpt”, clear 
save "dataset.dta", replace // save it in Stata dta format

!pip install pyreadstat
import pyreadstat

# Define file paths
xpt_file = "LLCP2021.XPT"
dta_file = "LLCP2021.dta"

# Read the XPT file 
df, meta = pyreadstat.read_xport(xpt_file, encoding="iso-8859-1")

# Rename variables that start with an underscore
df.columns = [col.lstrip("_") if col.startswith("_") else col for col in df.columns]

# Save to Stata format
pyreadstat.write_dta(df, dta_file)

print(f"File converted successfully: {dta_file}")

import re
import chardet

# Paths to your files
sas_file = "FORMAT22.sas"
stata_do_file = "stata_labels_format22.do"

# Function to ensure valid Stata variable name
def make_valid_stata_name(name):
    return name.lstrip("_")

# Detect file encoding
with open(sas_file, "rb") as f:
    raw_data = f.read()  
    detected_encoding = chardet.detect(raw_data)["encoding"]
    print(f"Detected Encoding: {detected_encoding}")

# Read file content
with open(sas_file, "r", encoding=detected_encoding) as f:
    lines = f.readlines()

# Extract all VALUE formats with restructured names
label_define_commands = []
current_format = None
current_values = []
processed_formats = {}  # Maps original format name to restructured format name

for line in lines:
    line = line.strip()
    
    # Detect format definitions (VALUE)
    match_value_start = re.match(r'^value\s+([^\s]+)', line, re.IGNORECASE)
    if match_value_start:
        if current_format and current_values:
            original_format = current_format.upper()
            digits = re.findall(r'\d+', original_format)
            alpha_part = re.sub(r'\d+', '', original_format)
            restructured_format = make_valid_stata_name(alpha_part + ''.join(digits))
            processed_formats[original_format] = restructured_format
            label_define_commands.append(f'label define {restructured_format} {" ".join(current_values)}')
        
        current_format = match_value_start.group(1)
        current_values = []
        continue
    
    # Detect mappings within VALUE formats
    code_label_pair = re.match(r'^(?!\.)(\d+|\w+)\s*=\s*"(.+?)"', line)
    if code_label_pair and current_format:
        code, label = code_label_pair.group(1), code_label_pair.group(2)
        current_values.append(f'{code} "{label}"')

# Process the last format
if current_format and current_values:
    original_format = current_format.upper()
    digits = re.findall(r'\d+', original_format)
    alpha_part = re.sub(r'\d+', '', original_format)
    restructured_format = make_valid_stata_name(alpha_part + ''.join(digits))
    processed_formats[original_format] = restructured_format
    label_define_commands.append(f'label define {restructured_format} {" ".join(current_values)}')

# Replace label define _varname with label define varname
label_define_commands = [re.sub(r'label define _(\w+)', r'label define \1', cmd) for cmd in label_define_commands]

# Write to the Stata do file
with open(stata_do_file, "w", encoding="utf-8") as f:
    f.write("// Stata do file for labels generated from SAS formats\n\n")
    
    # Write label definitions with restructured names
    f.write("// Value label definitions (restructured to match variable naming pattern)\n")
    f.write("// Original SAS format names have been modified to move numbers to the end\n")
    f.write("\n".join(label_define_commands))
    f.write("\n\n")
    
    # Write mapping information for reference
    f.write("// Format name restructuring (for reference only):\n")
    for orig, restr in processed_formats.items():
        f.write(f"// Original: {orig} → Restructured: {restr}\n")
    f.write("\n\n")

# Print the generated Stata do file content
with open(stata_do_file, "r", encoding="utf-8") as f:
    stata_code_content = f.read()
    print(stata_code_content)

print(f"\u2705 Successfully created '{stata_do_file}' with labels restructured to match variable naming pattern.")

// Auto-label assignment program
capture program drop auto_label_vars
program define auto_label_vars

    * Retrieve the list of all variables in the current dataset
    quietly ds
    local allvars = r(varlist)

    * Retrieve the list of all defined labels
    quietly label dir
    local labellist = r(names)

    * Apply labeling to each variable
    foreach var of local allvars {
        * Check if the variable exists in the defined label list
        local found : list var in labellist

        if `found' {
            label values `var' `var'
            display as result "✓ Labeled variable `var'"
        }
    }

end

* Execute the program on the currently active dataset
auto_label_vars