gistfile1.txt

import re
import time

quotefix_re = re.compile(
    r'''
        ^(
            (?:
                (?:
                    (?:[^\n\\"]|(?:\\{2})*\\"|\\[^\n"])*
                    (?:\\{2})*
                    "
                ){2}
            )*
            ((?:[^\n\\"]|(?:\\{2})*\\"|\\[^\n"])*)
            "
            ((?:[^\n\\"]|(?:\\{2})*\\"|\\[^\n"])*)
        )$\n
    ''',
    flags=re.MULTILINE | re.VERBOSE
)

def sb_json_regex(raw):
    while True:
        raw, n = quotefix_re.subn(r'\1\\n', raw, count=1)
        if n == 0:
            break
    return raw

def sb_json_boring(s):
    out = []
    state = ''
    for c in s:
        if state == '':
            if c == '"':
                out.append(c)
                state = 'string'
            else:
                out.append(c)
        elif state == 'string':
            if c == '"':
                out.append(c)
                state = ''
            elif c == '\n':
                out.append('\\')
                out.append('n')
            elif c == '\\':
                out.append(c)
                state = 'escape'
            else:
                out.append(c)
        elif state == 'escape':
            out.append(c)
            state = 'string'
        else:
            assert 0, "impossible"
    return ''.join(out)

def timeit(f, name, s):
    now = time.time()
    f(s)
    t = time.time() - now
    r = s
    if len(r) > 20:
        r = r[:20] + "..."
    print("%s: s=%r len(s)=%d: %fs" % (name, r, len(s), t))

def do(s):
    timeit(sb_json_regex, "regex", s)
    timeit(sb_json_boring, "boring", s)
    assert sb_json_regex(s) == sb_json_boring(s)

do("{}")
do("""{"hello": "world"}""")
do(r"""{"\"\\ \n": ""}""")
do("{\n" + """   "this\nis an example": "of a very\nlong json file\nwith embedded newlines",\n""" * 100 + """ "the": "end" } """)

gistfile2.txt

regex: s='{}' len(s)=2: 0.000076s
boring: s='{}' len(s)=2: 0.000005s
regex: s='{"hello": "world"}' len(s)=18: 0.000034s
boring: s='{"hello": "world"}' len(s)=18: 0.000013s
regex: s='{"\\"\\\\ \\n": ""}' len(s)=15: 0.000026s
boring: s='{"\\"\\\\ \\n": ""}' len(s)=15: 0.000011s
regex: s='{\n   "this\nis an exa...' len(s)=7618: 0.855799s
boring: s='{\n   "this\nis an exa...' len(s)=7618: 0.003821s