Assert and incorrect error message when loading source file containing invalid UTF-8

[mdboom]

Bug report

When loading a file containing invalid UTF-8, without a source encoding marker, an assert is emitted. If assert is removed, the error message is close, but not quite correct and has incorrect position information.

To reproduce, create a file with an invalid UTF-8 start byte followed by a valid continuation byte, for example, the sort of contrived and grammatically incorrect:

open("x.py", "w", encoding="latin-1").write("\n\n\n'HOLÀ¡'\n")

results in:

$ ./python x.py 
python: Parser/pegen_errors.c:335: _PyPegen_raise_error_known_location: Assertion `p->tok->fp == NULL || p->tok->fp == stdin || p->tok->done == E_EOF' failed.
[1]    16448 IOT instruction  ./python x.py

Backtrace

#0  __pthread_kill_implementation (threadid=<optimized out>, signo=signo@entry=6, no_tid=no_tid@entry=0)
    at ./nptl/pthread_kill.c:44
#1  0x00007ffff7d6389f in __pthread_kill_internal (signo=6, threadid=<optimized out>)
    at ./nptl/pthread_kill.c:78
#2  0x00007ffff7d17a52 in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#3  0x00007ffff7d02469 in __GI_abort () at ./stdlib/abort.c:79
#4  0x00007ffff7d02395 in __assert_fail_base (fmt=0x7ffff7e8fc30 "%s%s%s:%u: %s%sAssertion `%s' failed.\n%n",
    assertion=0x5555558a5a10 "p->tok->fp == NULL || p->tok->fp == stdin || p->tok->done == E_EOF",
    file=0x5555558a5898 "Parser/pegen_errors.c", line=335, function=<optimized out>) at ./assert/assert.c:92
#5  0x00007ffff7d10b02 in __GI___assert_fail (
    assertion=assertion@entry=0x5555558a5a10 "p->tok->fp == NULL || p->tok->fp == stdin || p->tok->done == E_EOF", file=file@entry=0x5555558a5898 "Parser/pegen_errors.c", line=line@entry=335,
    function=function@entry=0x5555558a5c40 <__PRETTY_FUNCTION__.1> "_PyPegen_raise_error_known_location")
    at ./assert/assert.c:101
#6  0x000055555564c7be in _PyPegen_raise_error_known_location (p=0x7ffff7a8c880,
    errtype=0x555555a1cb20 <_PyExc_SyntaxError>, lineno=4, col_offset=8, end_lineno=4, end_col_offset=8,
    errmsg=0x5555558a5946 "(%s) %U", va=0x7fffffffd058) at Parser/pegen_errors.c:335
#7  0x000055555564cc08 in _PyPegen_raise_error (p=p@entry=0x7ffff7a8c880, errtype=<optimized out>,
    errmsg=errmsg@entry=0x5555558a5946 "(%s) %U") at Parser/pegen_errors.c:235
#8  0x000055555564cedb in _Pypegen_raise_decode_error (p=p@entry=0x7ffff7a8c880) at Parser/pegen_errors.c:133
#9  0x000055555564eb1f in _PyPegen_concatenate_strings (p=p@entry=0x7ffff7a8c880,
    strings=strings@entry=0x555555c13290) at Parser/action_helpers.c:957
#10 0x00005555556598cf in strings_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:15481
#11 0x000055555566362f in atom_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:14292
#12 0x000055555567c149 in t_primary_raw (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:18126
#13 0x000055555567c4ac in t_primary_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:17916
#14 0x0000555555680978 in single_subscript_attribute_target_rule (p=p@entry=0x7ffff7a8c880)
    at Parser/parser.c:17805
#15 0x00005555556813e4 in _tmp_12_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:24220
#16 0x00005555556a5bb0 in assignment_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:2245
#17 0x00005555556a9f37 in simple_stmt_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:1652
#18 0x00005555556ab747 in simple_stmts_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:1547
#19 0x00005555556ac6f6 in statement_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:1370
#20 0x00005555556ac969 in _loop1_3_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:23651
#21 0x00005555556acb51 in statements_rule (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:1302
#22 0x00005555556b4ea0 in file_rule (p=0x7ffff7a8c880) at Parser/parser.c:1061
#23 0x00005555556b6953 in _PyPegen_parse (p=p@entry=0x7ffff7a8c880) at Parser/parser.c:38772
#24 0x000055555564bfc6 in _PyPegen_run_parser (p=p@entry=0x7ffff7a8c880) at Parser/pegen.c:811
#25 0x000055555564c1b2 in _PyPegen_run_parser_from_file_pointer (fp=fp@entry=0x555555be4310,
    start_rule=start_rule@entry=257, filename_ob=filename_ob@entry=0x7ffff7aa9150, enc=enc@entry=0x0,
    ps1=ps1@entry=0x0, ps2=ps2@entry=0x0, flags=0x7fffffffd8a8, errcode=0x0, arena=0x7ffff7a939a0)
    at Parser/pegen.c:884
#26 0x00005555556b8c2f in _PyParser_ASTFromFile (fp=fp@entry=0x555555be4310,
    filename_ob=filename_ob@entry=0x7ffff7aa9150, enc=enc@entry=0x0, mode=mode@entry=257, ps1=ps1@entry=0x0,
    ps2=ps2@entry=0x0, flags=0x7fffffffd8a8, errcode=0x0, arena=0x7ffff7a939a0) at Parser/peg_api.c:26
#27 0x0000555555813ca1 in pyrun_file (fp=fp@entry=0x555555be4310, filename=filename@entry=0x7ffff7aa9150,
    start=start@entry=257, globals=globals@entry=0x7ffff7a45010, locals=locals@entry=0x7ffff7a45010,
    closeit=closeit@entry=1, flags=0x7fffffffd8a8) at Python/pythonrun.c:1620
#28 0x0000555555816941 in _PyRun_SimpleFileObject (fp=fp@entry=0x555555be4310,
    filename=filename@entry=0x7ffff7aa9150, closeit=closeit@entry=1, flags=flags@entry=0x7fffffffd8a8)
    at Python/pythonrun.c:439
#29 0x0000555555816af5 in _PyRun_AnyFileObject (fp=fp@entry=0x555555be4310,
    filename=filename@entry=0x7ffff7aa9150, closeit=closeit@entry=1, flags=flags@entry=0x7fffffffd8a8)
    at Python/pythonrun.c:78
#30 0x00005555558332bf in pymain_run_file_obj (program_name=program_name@entry=0x7ffff7aa91c0,
    filename=filename@entry=0x7ffff7aa9150, skip_source_first_line=0) at Modules/main.c:360
#31 0x00005555558333c1 in pymain_run_file (config=config@entry=0x555555b29490 <_PyRuntime+69264>)
    at Modules/main.c:379
#32 0x0000555555833a99 in pymain_run_python (exitcode=exitcode@entry=0x7fffffffd9fc) at Modules/main.c:610
#33 0x0000555555833cea in Py_RunMain () at Modules/main.c:689
#34 0x0000555555833d3f in pymain_main (args=args@entry=0x7fffffffda40) at Modules/main.c:719
#35 0x0000555555833dc4 in Py_BytesMain (argc=<optimized out>, argv=<optimized out>) at Modules/main.c:743
#36 0x000055555564a742 in main (argc=<optimized out>, argv=<optimized out>) at ./Programs/python.c:15

In the above assert, p->tok->done is E_OK.

Bypassing the assert, you get an incorrect position (it should be 7, not 3):

  File "x.py", line 4
    'HOL��'
           ^
SyntaxError: (unicode error) 'utf-8' codec can't decode byte 0xc0 in position 3: invalid start byte

Your environment

Linux, Python 3.11

This bug was found because the coverage-related test added in #94856 failed on some buildbots. For them to fail, they need to both (1) build with --with-pydebug and (2) have coredumps enabled in the OS. That test should probably be fixed so it fails consistently across all buildbots, even the ones that aren't set up that way.

[lysnikolaou]

@pablogsal @gvanrossum @mdboom Can this issue be closed?