页面比较

# How to run
# python postgresql-dump-extractor.py \
#  -s barnard \
#  -i 576184071779_dump_20240923.sql\
#  -o 576184071779_dump_20240923_barnard.sql

import argparse
import re

def extract_copy_statements(input_file, output_file, schema):
    with open(input_file, 'r') as infile, open(output_file, 'w') as outfile:
        copy_block = []
        in_copy_block = False
        copy_pattern = re.compile(rf'^COPY {re.escape(schema)}\.(\w+)')

        for line in infile:
            if not in_copy_block:
                if copy_pattern.match(line):
                    in_copy_block = True
                    copy_block = [line]
            else:
                copy_block.append(line)
                if line.strip() == '\.':
                    outfile.writelines(copy_block)
                    outfile.write('\n')
                    in_copy_block = False
                    copy_block = []

def main():
    parser = argparse.ArgumentParser(description='Extract COPY statements from PostgreSQL dump file.')
    parser.add_argument('-i', '--input', required=True, help='Input dump file name')
    parser.add_argument('-o', '--output', required=True, help='Output file name')
    parser.add_argument('-s', '--schema', required=True, help='Schema name')

    args = parser.parse_args()

    extract_copy_statements(args.input, args.output, args.schema)
    print(f"Extraction complete. Output written to {args.output}")

if __name__ == "__main__":
    main()