// Unbuffered, inefficient implementation of readline() for raw I/O files.
STATIC mp_obj_t stream_unbuffered_readline(size_t n_args, const mp_obj_t *args) {
const mp_stream_p_t *stream_p = mp_get_stream_raise(args[0], MP_STREAM_OP_READ);
mp_int_t max_size = -1;
if (n_args > 1) {
max_size = MP_OBJ_SMALL_INT_VALUE(args[1]);
}
vstr_t vstr;
if (max_size != -1) {
vstr_init(&vstr, max_size);
} else {
vstr_init(&vstr, 16);
}
while (max_size == -1 || max_size-- != 0) {
char *p = vstr_add_len(&vstr, 1);
if (p == NULL) {
nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_MemoryError, "out of memory"));
}
int error;
mp_uint_t out_sz = stream_p->read(args[0], p, 1, &error);
if (out_sz == MP_STREAM_ERROR) {
if (mp_is_nonblocking_error(error)) {
if (vstr.len == 1) {
// We just incremented it, but otherwise we read nothing
// and immediately got EAGAIN. This case is not well
// specified in
// https://docs.python.org/3/library/io.html#io.IOBase.readline
// unlike similar case for read(). But we follow the latter's
// behavior - return None.
vstr_clear(&vstr);
return mp_const_none;
} else {
goto done;
}
}
mp_raise_OSError(error);
}
if (out_sz == 0) {
done:
// Back out previously added byte
// Consider, what's better - read a char and get OutOfMemory (so read
// char is lost), or allocate first as we do.
vstr_cut_tail_bytes(&vstr, 1);
break;
}
if (*p == '\n') {
break;
}
}
return mp_obj_new_str_from_vstr(STREAM_CONTENT_TYPE(stream_p), &vstr);
}
// Unbuffered, inefficient implementation of readline() for raw I/O files.
STATIC mp_obj_t stream_unbuffered_readline(uint n_args, const mp_obj_t *args) {
struct _mp_obj_base_t *o = (struct _mp_obj_base_t *)args[0];
if (o->type->stream_p == NULL || o->type->stream_p->read == NULL) {
// CPython: io.UnsupportedOperation, OSError subclass
nlr_raise(mp_obj_new_exception_msg(&mp_type_OSError, "Operation not supported"));
}
mp_int_t max_size = -1;
if (n_args > 1) {
max_size = MP_OBJ_SMALL_INT_VALUE(args[1]);
}
vstr_t *vstr;
if (max_size != -1) {
vstr = vstr_new_size(max_size);
} else {
vstr = vstr_new();
}
int error;
while (max_size == -1 || max_size-- != 0) {
char *p = vstr_add_len(vstr, 1);
if (p == NULL) {
nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_MemoryError, "out of memory"));
}
mp_uint_t out_sz = o->type->stream_p->read(o, p, 1, &error);
if (out_sz == MP_STREAM_ERROR) {
nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_OSError, "[Errno %d]", error));
}
if (out_sz == 0) {
// Back out previously added byte
// Consider, what's better - read a char and get OutOfMemory (so read
// char is lost), or allocate first as we do.
vstr_cut_tail_bytes(vstr, 1);
break;
}
if (*p == '\n') {
break;
}
}
// TODO need a string creation API that doesn't copy the given data
mp_obj_t ret = mp_obj_new_str_of_type(STREAM_CONTENT_TYPE(o->type->stream_p), (byte*)vstr->buf, vstr->len);
vstr_free(vstr);
return ret;
}
STATIC int compile_and_save(const char *file, const char *output_file, const char *source_file) {
mp_lexer_t *lex = mp_lexer_new_from_file(file);
if (lex == NULL) {
printf("could not open file '%s' for reading\n", file);
return 1;
}
nlr_buf_t nlr;
if (nlr_push(&nlr) == 0) {
qstr source_name;
if (source_file == NULL) {
source_name = lex->source_name;
} else {
source_name = qstr_from_str(source_file);
}
#if MICROPY_PY___FILE__
if (input_kind == MP_PARSE_FILE_INPUT) {
mp_store_global(MP_QSTR___file__, MP_OBJ_NEW_QSTR(source_name));
}
#endif
mp_parse_tree_t parse_tree = mp_parse(lex, MP_PARSE_FILE_INPUT);
mp_raw_code_t *rc = mp_compile_to_raw_code(&parse_tree, source_name, emit_opt, false);
vstr_t vstr;
vstr_init(&vstr, 16);
if (output_file == NULL) {
vstr_add_str(&vstr, file);
vstr_cut_tail_bytes(&vstr, 2);
vstr_add_str(&vstr, "mpy");
} else {
vstr_add_str(&vstr, output_file);
}
mp_raw_code_save_file(rc, vstr_null_terminated_str(&vstr));
vstr_clear(&vstr);
nlr_pop();
return 0;
} else {
// uncaught exception
mp_obj_print_exception(&mp_stderr_print, (mp_obj_t)nlr.ret_val);
return 1;
}
}
mp_obj_t mp_builtin___import__(mp_uint_t n_args, const mp_obj_t *args) {
#if DEBUG_PRINT
DEBUG_printf("__import__:\n");
for (mp_uint_t i = 0; i < n_args; i++) {
DEBUG_printf(" ");
mp_obj_print(args[i], PRINT_REPR);
DEBUG_printf("\n");
}
#endif
mp_obj_t module_name = args[0];
mp_obj_t fromtuple = mp_const_none;
mp_int_t level = 0;
if (n_args >= 4) {
fromtuple = args[3];
if (n_args >= 5) {
level = MP_OBJ_SMALL_INT_VALUE(args[4]);
}
}
mp_uint_t mod_len;
const char *mod_str = mp_obj_str_get_data(module_name, &mod_len);
if (level != 0) {
// What we want to do here is to take name of current module,
// chop <level> trailing components, and concatenate with passed-in
// module name, thus resolving relative import name into absolue.
// This even appears to be correct per
// http://legacy.python.org/dev/peps/pep-0328/#relative-imports-and-name
// "Relative imports use a module's __name__ attribute to determine that
// module's position in the package hierarchy."
level--;
mp_obj_t this_name_q = mp_obj_dict_get(mp_globals_get(), MP_OBJ_NEW_QSTR(MP_QSTR___name__));
assert(this_name_q != MP_OBJ_NULL);
#if MICROPY_CPYTHON_COMPAT
if (MP_OBJ_QSTR_VALUE(this_name_q) == MP_QSTR___main__) {
// This is a module run by -m command-line switch, get its real name from backup attribute
this_name_q = mp_obj_dict_get(mp_globals_get(), MP_OBJ_NEW_QSTR(MP_QSTR___main__));
}
#endif
mp_map_t *globals_map = mp_obj_dict_get_map(mp_globals_get());
mp_map_elem_t *elem = mp_map_lookup(globals_map, MP_OBJ_NEW_QSTR(MP_QSTR___path__), MP_MAP_LOOKUP);
bool is_pkg = (elem != NULL);
#if DEBUG_PRINT
DEBUG_printf("Current module/package: ");
mp_obj_print(this_name_q, PRINT_REPR);
DEBUG_printf(", is_package: %d", is_pkg);
DEBUG_printf("\n");
#endif
mp_uint_t this_name_l;
const char *this_name = mp_obj_str_get_data(this_name_q, &this_name_l);
const char *p = this_name + this_name_l;
if (!is_pkg) {
// We have module, but relative imports are anchored at package, so
// go there.
chop_component(this_name, &p);
}
uint dots_seen = 0;
while (level--) {
chop_component(this_name, &p);
dots_seen++;
}
if (dots_seen == 0 && level >= 1) {
// http://legacy.python.org/dev/peps/pep-0328/#relative-imports-and-name
// "If the module's name does not contain any package information
// (e.g. it is set to '__main__') then relative imports are
// resolved as if the module were a top level module, regardless
// of where the module is actually located on the file system."
// Supposedly this if catches this condition and resolve it properly
// TODO: But nobody knows for sure. This condition happens when
// package's __init__.py does something like "import .submod". So,
// maybe we should check for package here? But quote above doesn't
// talk about packages, it talks about dot-less module names.
DEBUG_printf("Warning: no dots in current module name and level>0\n");
p = this_name + this_name_l;
} else if (level != -1) {
nlr_raise(mp_obj_new_exception_msg(&mp_type_ImportError, "Invalid relative import"));
}
uint new_mod_l = (mod_len == 0 ? (size_t)(p - this_name) : (size_t)(p - this_name) + 1 + mod_len);
char *new_mod = alloca(new_mod_l);
memcpy(new_mod, this_name, p - this_name);
if (mod_len != 0) {
new_mod[p - this_name] = '.';
memcpy(new_mod + (p - this_name) + 1, mod_str, mod_len);
}
qstr new_mod_q = qstr_from_strn(new_mod, new_mod_l);
DEBUG_printf("Resolved base name for relative import: '%s'\n", qstr_str(new_mod_q));
if (new_mod_q == MP_QSTR_) {
// CPython raises SystemError
nlr_raise(mp_obj_new_exception_msg(&mp_type_ImportError, "cannot perform relative import"));
}
module_name = MP_OBJ_NEW_QSTR(new_mod_q);
mod_str = new_mod;
mod_len = new_mod_l;
}
// check if module already exists
qstr module_name_qstr = mp_obj_str_get_qstr(module_name);
mp_obj_t module_obj = mp_module_get(module_name_qstr);
if (module_obj != MP_OBJ_NULL) {
DEBUG_printf("Module already loaded\n");
// If it's not a package, return module right away
char *p = strchr(mod_str, '.');
if (p == NULL) {
return module_obj;
}
// If fromlist is not empty, return leaf module
if (fromtuple != mp_const_none) {
return module_obj;
}
// Otherwise, we need to return top-level package
qstr pkg_name = qstr_from_strn(mod_str, p - mod_str);
return mp_module_get(pkg_name);
}
DEBUG_printf("Module not yet loaded\n");
#if MICROPY_MODULE_FROZEN
mp_lexer_t *lex = mp_find_frozen_module(mod_str, mod_len);
if (lex != NULL) {
module_obj = mp_obj_new_module(module_name_qstr);
// if args[3] (fromtuple) has magic value False, set up
// this module for command-line "-m" option (set module's
// name to __main__ instead of real name).
// TODO: Duplicated below too.
if (fromtuple == mp_const_false) {
mp_obj_module_t *o = module_obj;
mp_obj_dict_store(o->globals, MP_OBJ_NEW_QSTR(MP_QSTR___name__), MP_OBJ_NEW_QSTR(MP_QSTR___main__));
}
do_load_from_lexer(module_obj, lex, mod_str);
return module_obj;
}
#endif
uint last = 0;
VSTR_FIXED(path, MICROPY_ALLOC_PATH_MAX)
module_obj = MP_OBJ_NULL;
mp_obj_t top_module_obj = MP_OBJ_NULL;
mp_obj_t outer_module_obj = MP_OBJ_NULL;
uint i;
for (i = 1; i <= mod_len; i++) {
if (i == mod_len || mod_str[i] == '.') {
// create a qstr for the module name up to this depth
qstr mod_name = qstr_from_strn(mod_str, i);
DEBUG_printf("Processing module: %s\n", qstr_str(mod_name));
DEBUG_printf("Previous path: =%.*s=\n", vstr_len(&path), vstr_str(&path));
// find the file corresponding to the module name
mp_import_stat_t stat;
if (vstr_len(&path) == 0) {
// first module in the dotted-name; search for a directory or file
stat = find_file(mod_str, i, &path);
} else {
// latter module in the dotted-name; append to path
vstr_add_char(&path, PATH_SEP_CHAR);
vstr_add_strn(&path, mod_str + last, i - last);
stat = stat_dir_or_file(&path);
}
DEBUG_printf("Current path: %.*s\n", vstr_len(&path), vstr_str(&path));
if (stat == MP_IMPORT_STAT_NO_EXIST) {
#if MICROPY_MODULE_WEAK_LINKS
// check if there is a weak link to this module
if (i == mod_len) {
mp_map_elem_t *el = mp_map_lookup((mp_map_t*)&mp_builtin_module_weak_links_map, MP_OBJ_NEW_QSTR(mod_name), MP_MAP_LOOKUP);
if (el == NULL) {
goto no_exist;
}
// found weak linked module
module_obj = el->value;
} else {
no_exist:
#else
{
#endif
// couldn't find the file, so fail
if (MICROPY_ERROR_REPORTING == MICROPY_ERROR_REPORTING_TERSE) {
nlr_raise(mp_obj_new_exception_msg(&mp_type_ImportError, "module not found"));
} else {
nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_ImportError,
"no module named '%q'", mod_name));
}
}
} else {
// found the file, so get the module
module_obj = mp_module_get(mod_name);
}
if (module_obj == MP_OBJ_NULL) {
// module not already loaded, so load it!
module_obj = mp_obj_new_module(mod_name);
// if args[3] (fromtuple) has magic value False, set up
// this module for command-line "-m" option (set module's
// name to __main__ instead of real name).
if (i == mod_len && fromtuple == mp_const_false) {
mp_obj_module_t *o = module_obj;
mp_obj_dict_store(o->globals, MP_OBJ_NEW_QSTR(MP_QSTR___name__), MP_OBJ_NEW_QSTR(MP_QSTR___main__));
#if MICROPY_CPYTHON_COMPAT
// Store real name in "__main__" attribute. Choosen semi-randonly, to reuse existing qstr's.
mp_obj_dict_store(o->globals, MP_OBJ_NEW_QSTR(MP_QSTR___main__), MP_OBJ_NEW_QSTR(mod_name));
#endif
}
if (stat == MP_IMPORT_STAT_DIR) {
DEBUG_printf("%.*s is dir\n", vstr_len(&path), vstr_str(&path));
// https://docs.python.org/3/reference/import.html
// "Specifically, any module that contains a __path__ attribute is considered a package."
mp_store_attr(module_obj, MP_QSTR___path__, mp_obj_new_str(vstr_str(&path), vstr_len(&path), false));
vstr_add_char(&path, PATH_SEP_CHAR);
vstr_add_str(&path, "__init__.py");
if (mp_import_stat(vstr_null_terminated_str(&path)) != MP_IMPORT_STAT_FILE) {
vstr_cut_tail_bytes(&path, sizeof("/__init__.py") - 1); // cut off /__init__.py
mp_warning("%s is imported as namespace package", vstr_str(&path));
} else {
do_load(module_obj, &path);
vstr_cut_tail_bytes(&path, sizeof("/__init__.py") - 1); // cut off /__init__.py
}
} else { // MP_IMPORT_STAT_FILE
do_load(module_obj, &path);
// TODO: We cannot just break here, at the very least, we must execute
// trailer code below. But otherwise if there're remaining components,
// that would be (??) object path within module, not modules path within FS.
// break;
}
}
if (outer_module_obj != MP_OBJ_NULL) {
qstr s = qstr_from_strn(mod_str + last, i - last);
mp_store_attr(outer_module_obj, s, module_obj);
}
outer_module_obj = module_obj;
if (top_module_obj == MP_OBJ_NULL) {
top_module_obj = module_obj;
}
last = i + 1;
}
}
STATIC void mp_lexer_next_token_into(mp_lexer_t *lex, mp_token_t *tok, bool first_token) {
// skip white space and comments
bool had_physical_newline = false;
while (!is_end(lex)) {
if (is_physical_newline(lex)) {
had_physical_newline = true;
next_char(lex);
} else if (is_whitespace(lex)) {
next_char(lex);
} else if (is_char(lex, '#')) {
next_char(lex);
while (!is_end(lex) && !is_physical_newline(lex)) {
next_char(lex);
}
// had_physical_newline will be set on next loop
} else if (is_char(lex, '\\')) {
// backslash (outside string literals) must appear just before a physical newline
next_char(lex);
if (!is_physical_newline(lex)) {
// SyntaxError: unexpected character after line continuation character
tok->src_line = lex->line;
tok->src_column = lex->column;
tok->kind = MP_TOKEN_BAD_LINE_CONTINUATION;
vstr_reset(&lex->vstr);
tok->str = vstr_str(&lex->vstr);
tok->len = 0;
return;
} else {
next_char(lex);
}
} else {
break;
}
}
// set token source information
tok->src_line = lex->line;
tok->src_column = lex->column;
// start new token text
vstr_reset(&lex->vstr);
if (first_token && lex->line == 1 && lex->column != 1) {
// check that the first token is in the first column
// if first token is not on first line, we get a physical newline and
// this check is done as part of normal indent/dedent checking below
// (done to get equivalence with CPython)
tok->kind = MP_TOKEN_INDENT;
} else if (lex->emit_dent < 0) {
tok->kind = MP_TOKEN_DEDENT;
lex->emit_dent += 1;
} else if (lex->emit_dent > 0) {
tok->kind = MP_TOKEN_INDENT;
lex->emit_dent -= 1;
} else if (had_physical_newline && lex->nested_bracket_level == 0) {
tok->kind = MP_TOKEN_NEWLINE;
uint num_spaces = lex->column - 1;
lex->emit_dent = 0;
if (num_spaces == indent_top(lex)) {
} else if (num_spaces > indent_top(lex)) {
indent_push(lex, num_spaces);
lex->emit_dent += 1;
} else {
while (num_spaces < indent_top(lex)) {
indent_pop(lex);
lex->emit_dent -= 1;
}
if (num_spaces != indent_top(lex)) {
tok->kind = MP_TOKEN_DEDENT_MISMATCH;
}
}
} else if (is_end(lex)) {
if (indent_top(lex) > 0) {
tok->kind = MP_TOKEN_NEWLINE;
lex->emit_dent = 0;
while (indent_top(lex) > 0) {
indent_pop(lex);
lex->emit_dent -= 1;
}
} else {
tok->kind = MP_TOKEN_END;
}
} else if (is_char_or(lex, '\'', '\"')
|| (is_char_or3(lex, 'r', 'u', 'b') && is_char_following_or(lex, '\'', '\"'))
|| ((is_char_and(lex, 'r', 'b') || is_char_and(lex, 'b', 'r')) && is_char_following_following_or(lex, '\'', '\"'))) {
// a string or bytes literal
// parse type codes
bool is_raw = false;
bool is_bytes = false;
if (is_char(lex, 'u')) {
next_char(lex);
} else if (is_char(lex, 'b')) {
is_bytes = true;
next_char(lex);
if (is_char(lex, 'r')) {
is_raw = true;
next_char(lex);
}
} else if (is_char(lex, 'r')) {
is_raw = true;
next_char(lex);
if (is_char(lex, 'b')) {
is_bytes = true;
next_char(lex);
}
}
// set token kind
if (is_bytes) {
tok->kind = MP_TOKEN_BYTES;
} else {
tok->kind = MP_TOKEN_STRING;
}
// get first quoting character
char quote_char = '\'';
if (is_char(lex, '\"')) {
quote_char = '\"';
}
next_char(lex);
// work out if it's a single or triple quoted literal
int num_quotes;
if (is_char_and(lex, quote_char, quote_char)) {
// triple quotes
next_char(lex);
next_char(lex);
num_quotes = 3;
} else {
// single quotes
num_quotes = 1;
}
// parse the literal
int n_closing = 0;
while (!is_end(lex) && (num_quotes > 1 || !is_char(lex, '\n')) && n_closing < num_quotes) {
if (is_char(lex, quote_char)) {
n_closing += 1;
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
} else {
n_closing = 0;
if (is_char(lex, '\\')) {
next_char(lex);
unichar c = CUR_CHAR(lex);
if (is_raw) {
// raw strings allow escaping of quotes, but the backslash is also emitted
vstr_add_char(&lex->vstr, '\\');
} else {
switch (c) {
case MP_LEXER_CHAR_EOF: break; // TODO a proper error message?
case '\n': c = MP_LEXER_CHAR_EOF; break; // TODO check this works correctly (we are supposed to ignore it
case '\\': break;
case '\'': break;
case '"': break;
case 'a': c = 0x07; break;
case 'b': c = 0x08; break;
case 't': c = 0x09; break;
case 'n': c = 0x0a; break;
case 'v': c = 0x0b; break;
case 'f': c = 0x0c; break;
case 'r': c = 0x0d; break;
case 'u':
case 'U':
if (is_bytes) {
// b'\u1234' == b'\\u1234'
vstr_add_char(&lex->vstr, '\\');
break;
}
// Otherwise fall through.
case 'x':
{
uint num = 0;
if (!get_hex(lex, (c == 'x' ? 2 : c == 'u' ? 4 : 8), &num)) {
// TODO error message
assert(0);
}
c = num;
break;
}
case 'N':
// Supporting '\N{LATIN SMALL LETTER A}' == 'a' would require keeping the
// entire Unicode name table in the core. As of Unicode 6.3.0, that's nearly
// 3MB of text; even gzip-compressed and with minimal structure, it'll take
// roughly half a meg of storage. This form of Unicode escape may be added
// later on, but it's definitely not a priority right now. -- CJA 20140607
assert(!"Unicode name escapes not supported");
break;
default:
if (c >= '0' && c <= '7') {
// Octal sequence, 1-3 chars
int digits = 3;
int num = c - '0';
while (is_following_odigit(lex) && --digits != 0) {
next_char(lex);
num = num * 8 + (CUR_CHAR(lex) - '0');
}
c = num;
} else {
// unrecognised escape character; CPython lets this through verbatim as '\' and then the character
vstr_add_char(&lex->vstr, '\\');
}
break;
}
}
if (c != MP_LEXER_CHAR_EOF) {
if (c < 0x110000 && !is_bytes) {
vstr_add_char(&lex->vstr, c);
} else if (c < 0x100 && is_bytes) {
vstr_add_byte(&lex->vstr, c);
} else {
assert(!"TODO: Throw an error, invalid escape code probably");
}
}
} else {
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
}
}
next_char(lex);
}
// check we got the required end quotes
if (n_closing < num_quotes) {
tok->kind = MP_TOKEN_LONELY_STRING_OPEN;
}
// cut off the end quotes from the token text
vstr_cut_tail_bytes(&lex->vstr, n_closing);
} else if (is_head_of_identifier(lex)) {
tok->kind = MP_TOKEN_NAME;
// get first char
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
next_char(lex);
// get tail chars
while (!is_end(lex) && is_tail_of_identifier(lex)) {
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
next_char(lex);
}
} else if (is_digit(lex) || (is_char(lex, '.') && is_following_digit(lex))) {
tok->kind = MP_TOKEN_NUMBER;
// get first char
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
next_char(lex);
// get tail chars
while (!is_end(lex)) {
if (is_char_or(lex, 'e', 'E')) {
vstr_add_char(&lex->vstr, 'e');
next_char(lex);
if (is_char(lex, '+') || is_char(lex, '-')) {
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
next_char(lex);
}
} else if (is_letter(lex) || is_digit(lex) || is_char_or(lex, '_', '.')) {
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
next_char(lex);
} else {
break;
}
}
} else if (is_char(lex, '.')) {
// special handling for . and ... operators, because .. is not a valid operator
// get first char
vstr_add_char(&lex->vstr, '.');
next_char(lex);
if (is_char_and(lex, '.', '.')) {
vstr_add_char(&lex->vstr, '.');
vstr_add_char(&lex->vstr, '.');
next_char(lex);
next_char(lex);
tok->kind = MP_TOKEN_ELLIPSIS;
} else {
tok->kind = MP_TOKEN_DEL_PERIOD;
}
} else {
// search for encoded delimiter or operator
const char *t = tok_enc;
uint tok_enc_index = 0;
for (; *t != 0 && !is_char(lex, *t); t += 1) {
if (*t == 'e' || *t == 'c') {
t += 1;
} else if (*t == 'E') {
tok_enc_index -= 1;
t += 1;
}
tok_enc_index += 1;
}
next_char(lex);
if (*t == 0) {
// didn't match any delimiter or operator characters
tok->kind = MP_TOKEN_INVALID;
} else {
// matched a delimiter or operator character
// get the maximum characters for a valid token
t += 1;
uint t_index = tok_enc_index;
for (;;) {
for (; *t == 'e'; t += 1) {
t += 1;
t_index += 1;
if (is_char(lex, *t)) {
next_char(lex);
tok_enc_index = t_index;
break;
}
}
if (*t == 'E') {
t += 1;
if (is_char(lex, *t)) {
next_char(lex);
tok_enc_index = t_index;
} else {
tok->kind = MP_TOKEN_INVALID;
goto tok_enc_no_match;
}
break;
}
if (*t == 'c') {
t += 1;
t_index += 1;
if (is_char(lex, *t)) {
next_char(lex);
tok_enc_index = t_index;
t += 1;
} else {
break;
}
} else {
break;
}
}
// set token kind
tok->kind = tok_enc_kind[tok_enc_index];
tok_enc_no_match:
// compute bracket level for implicit line joining
if (tok->kind == MP_TOKEN_DEL_PAREN_OPEN || tok->kind == MP_TOKEN_DEL_BRACKET_OPEN || tok->kind == MP_TOKEN_DEL_BRACE_OPEN) {
lex->nested_bracket_level += 1;
} else if (tok->kind == MP_TOKEN_DEL_PAREN_CLOSE || tok->kind == MP_TOKEN_DEL_BRACKET_CLOSE || tok->kind == MP_TOKEN_DEL_BRACE_CLOSE) {
lex->nested_bracket_level -= 1;
}
}
}
// point token text to vstr buffer
tok->str = vstr_str(&lex->vstr);
tok->len = vstr_len(&lex->vstr);
// check for keywords
if (tok->kind == MP_TOKEN_NAME) {
// We check for __debug__ here and convert it to its value. This is so
// the parser gives a syntax error on, eg, x.__debug__. Otherwise, we
// need to check for this special token in many places in the compiler.
// TODO improve speed of these string comparisons
//for (int i = 0; tok_kw[i] != NULL; i++) {
for (int i = 0; i < MP_ARRAY_SIZE(tok_kw); i++) {
if (str_strn_equal(tok_kw[i], tok->str, tok->len)) {
if (i == MP_ARRAY_SIZE(tok_kw) - 1) {
// tok_kw[MP_ARRAY_SIZE(tok_kw) - 1] == "__debug__"
tok->kind = (mp_optimise_value == 0 ? MP_TOKEN_KW_TRUE : MP_TOKEN_KW_FALSE);
} else {
tok->kind = MP_TOKEN_KW_FALSE + i;
}
break;
}
}
}
}
// function to run extra tests for things that can't be checked by scripts
STATIC mp_obj_t extra_coverage(void) {
// mp_printf (used by ports that don't have a native printf)
{
printf("# mp_printf\n");
mp_printf(&mp_plat_print, "%"); // nothing after percent
mp_printf(&mp_plat_print, "%d %+d % d\n", -123, 123, 123); // sign
mp_printf(&mp_plat_print, "%05d\n", -123); // negative number with zero padding
mp_printf(&mp_plat_print, "%ld\n", 123); // long
mp_printf(&mp_plat_print, "%X\n", 0x1abcdef); // capital hex
mp_printf(&mp_plat_print, "%.2s %.3s\n", "abc", "abc"); // fixed string precision
mp_printf(&mp_plat_print, "%.*s\n", -1, "abc"); // negative string precision
mp_printf(&mp_plat_print, "%b %b\n", 0, 1); // bools
mp_printf(&mp_plat_print, "%s\n", NULL); // null string
mp_printf(&mp_plat_print, "%t\n"); // non-format char
}
// vstr
{
printf("# vstr\n");
vstr_t *vstr = vstr_new_size(16);
vstr_hint_size(vstr, 32);
vstr_add_str(vstr, "ts");
vstr_ins_byte(vstr, 1, 'e');
vstr_ins_char(vstr, 3, 't');
vstr_ins_char(vstr, 10, 's');
printf("%.*s\n", (int)vstr->len, vstr->buf);
vstr_cut_head_bytes(vstr, 2);
printf("%.*s\n", (int)vstr->len, vstr->buf);
vstr_cut_tail_bytes(vstr, 10);
printf("%.*s\n", (int)vstr->len, vstr->buf);
vstr_printf(vstr, "t%cst", 'e');
printf("%.*s\n", (int)vstr->len, vstr->buf);
vstr_cut_out_bytes(vstr, 3, 10);
printf("%.*s\n", (int)vstr->len, vstr->buf);
VSTR_FIXED(fix, 4);
vstr_add_str(&fix, "large");
printf("%.*s\n", (int)fix.len, fix.buf);
}
// repl autocomplete
{
printf("# repl\n");
const char *str;
mp_uint_t len = mp_repl_autocomplete("__n", 3, &mp_plat_print, &str);
printf("%.*s\n", (int)len, str);
mp_store_global(MP_QSTR_sys, mp_import_name(MP_QSTR_sys, mp_const_none, MP_OBJ_NEW_SMALL_INT(0)));
mp_repl_autocomplete("sys.", 4, &mp_plat_print, &str);
len = mp_repl_autocomplete("sys.impl", 8, &mp_plat_print, &str);
printf("%.*s\n", (int)len, str);
}
// attrtuple
{
printf("# attrtuple\n");
static const qstr fields[] = {MP_QSTR_start, MP_QSTR_stop, MP_QSTR_step};
static const mp_obj_t items[] = {MP_OBJ_NEW_SMALL_INT(1), MP_OBJ_NEW_SMALL_INT(2), MP_OBJ_NEW_SMALL_INT(3)};
mp_obj_print_helper(&mp_plat_print, mp_obj_new_attrtuple(fields, 3, items), PRINT_REPR);
printf("\n");
}
return mp_const_none;
}
STATIC mp_obj_t stream_read_generic(size_t n_args, const mp_obj_t *args, byte flags) {
// What to do if sz < -1? Python docs don't specify this case.
// CPython does a readall, but here we silently let negatives through,
// and they will cause a MemoryError.
mp_int_t sz;
if (n_args == 1 || ((sz = mp_obj_get_int(args[1])) == -1)) {
return stream_readall(args[0]);
}
const mp_stream_p_t *stream_p = mp_get_stream(args[0]);
#if MICROPY_PY_BUILTINS_STR_UNICODE
if (stream_p->is_text) {
// We need to read sz number of unicode characters. Because we don't have any
// buffering, and because the stream API can only read bytes, we must read here
// in units of bytes and must never over read. If we want sz chars, then reading
// sz bytes will never over-read, so we follow this approach, in a loop to keep
// reading until we have exactly enough chars. This will be 1 read for text
// with ASCII-only chars, and about 2 reads for text with a couple of non-ASCII
// chars. For text with lots of non-ASCII chars, it'll be pretty inefficient
// in time and memory.
vstr_t vstr;
vstr_init(&vstr, sz);
mp_uint_t more_bytes = sz;
mp_uint_t last_buf_offset = 0;
while (more_bytes > 0) {
char *p = vstr_add_len(&vstr, more_bytes);
int error;
mp_uint_t out_sz = mp_stream_read_exactly(args[0], p, more_bytes, &error);
if (error != 0) {
vstr_cut_tail_bytes(&vstr, more_bytes);
if (mp_is_nonblocking_error(error)) {
// With non-blocking streams, we read as much as we can.
// If we read nothing, return None, just like read().
// Otherwise, return data read so far.
// TODO what if we have read only half a non-ASCII char?
if (vstr.len == 0) {
vstr_clear(&vstr);
return mp_const_none;
}
break;
}
mp_raise_OSError(error);
}
if (out_sz < more_bytes) {
// Finish reading.
// TODO what if we have read only half a non-ASCII char?
vstr_cut_tail_bytes(&vstr, more_bytes - out_sz);
if (out_sz == 0) {
break;
}
}
// count chars from bytes just read
for (mp_uint_t off = last_buf_offset;;) {
byte b = vstr.buf[off];
int n;
if (!UTF8_IS_NONASCII(b)) {
// 1-byte ASCII char
n = 1;
} else if ((b & 0xe0) == 0xc0) {
// 2-byte char
n = 2;
} else if ((b & 0xf0) == 0xe0) {
// 3-byte char
n = 3;
} else if ((b & 0xf8) == 0xf0) {
// 4-byte char
n = 4;
} else {
// TODO
n = 5;
}
if (off + n <= vstr.len) {
// got a whole char in n bytes
off += n;
sz -= 1;
last_buf_offset = off;
if (off >= vstr.len) {
more_bytes = sz;
break;
}
} else {
// didn't get a whole char, so work out how many extra bytes are needed for
// this partial char, plus bytes for additional chars that we want
more_bytes = (off + n - vstr.len) + (sz - 1);
break;
}
}
}
return mp_obj_new_str_from_vstr(&mp_type_str, &vstr);
}
#endif
vstr_t vstr;
vstr_init_len(&vstr, sz);
int error;
mp_uint_t out_sz = mp_stream_rw(args[0], vstr.buf, sz, &error, flags);
if (error != 0) {
vstr_clear(&vstr);
if (mp_is_nonblocking_error(error)) {
// https://docs.python.org/3.4/library/io.html#io.RawIOBase.read
// "If the object is in non-blocking mode and no bytes are available,
// None is returned."
// This is actually very weird, as naive truth check will treat
// this as EOF.
return mp_const_none;
}
mp_raise_OSError(error);
} else {
vstr.len = out_sz;
return mp_obj_new_str_from_vstr(STREAM_CONTENT_TYPE(stream_p), &vstr);
}
}
mp_obj_t mp_builtin___import__(uint n_args, mp_obj_t *args) {
#if DEBUG_PRINT
printf("__import__:\n");
for (int i = 0; i < n_args; i++) {
printf(" ");
mp_obj_print(args[i], PRINT_REPR);
printf("\n");
}
#endif
mp_obj_t module_name = args[0];
mp_obj_t fromtuple = mp_const_none;
int level = 0;
if (n_args >= 4) {
fromtuple = args[3];
if (n_args >= 5) {
level = MP_OBJ_SMALL_INT_VALUE(args[4]);
}
}
uint mod_len;
const char *mod_str = (const char*)mp_obj_str_get_data(module_name, &mod_len);
if (level != 0) {
// What we want to do here is to take name of current module,
// chop <level> trailing components, and concatenate with passed-in
// module name, thus resolving relative import name into absolue.
// This even appears to be correct per
// http://legacy.python.org/dev/peps/pep-0328/#relative-imports-and-name
// "Relative imports use a module's __name__ attribute to determine that
// module's position in the package hierarchy."
mp_obj_t this_name_q = mp_obj_dict_get(mp_globals_get(), MP_OBJ_NEW_QSTR(MP_QSTR___name__));
assert(this_name_q != MP_OBJ_NULL);
#if DEBUG_PRINT
printf("Current module: ");
mp_obj_print(this_name_q, PRINT_REPR);
printf("\n");
#endif
uint this_name_l;
const char *this_name = (const char*)mp_obj_str_get_data(this_name_q, &this_name_l);
uint dots_seen = 0;
const char *p = this_name + this_name_l - 1;
while (p > this_name) {
if (*p == '.') {
dots_seen++;
if (--level == 0) {
break;
}
}
p--;
}
if (dots_seen == 0 && level == 1) {
// http://legacy.python.org/dev/peps/pep-0328/#relative-imports-and-name
// "If the module's name does not contain any package information
// (e.g. it is set to '__main__') then relative imports are
// resolved as if the module were a top level module, regardless
// of where the module is actually located on the file system."
// Supposedly this if catches this condition and resolve it properly
// TODO: But nobody knows for sure. This condition happens when
// package's __init__.py does something like "import .submod". So,
// maybe we should check for package here? But quote above doesn't
// talk about packages, it talks about dot-less module names.
p = this_name + this_name_l;
} else if (level != 0) {
nlr_raise(mp_obj_new_exception_msg(&mp_type_ImportError, "Invalid relative import"));
}
uint new_mod_l = (mod_len == 0 ? p - this_name : p - this_name + 1 + mod_len);
char *new_mod = alloca(new_mod_l);
memcpy(new_mod, this_name, p - this_name);
if (mod_len != 0) {
new_mod[p - this_name] = '.';
memcpy(new_mod + (p - this_name) + 1, mod_str, mod_len);
}
qstr new_mod_q = qstr_from_strn(new_mod, new_mod_l);
DEBUG_printf("Resolved relative name: %s\n", qstr_str(new_mod_q));
module_name = MP_OBJ_NEW_QSTR(new_mod_q);
mod_str = new_mod;
mod_len = new_mod_l;
}
// check if module already exists
mp_obj_t module_obj = mp_module_get(mp_obj_str_get_qstr(module_name));
if (module_obj != MP_OBJ_NULL) {
DEBUG_printf("Module already loaded\n");
// If it's not a package, return module right away
char *p = strchr(mod_str, '.');
if (p == NULL) {
return module_obj;
}
// If fromlist is not empty, return leaf module
if (fromtuple != mp_const_none) {
return module_obj;
}
// Otherwise, we need to return top-level package
qstr pkg_name = qstr_from_strn(mod_str, p - mod_str);
return mp_module_get(pkg_name);
}
DEBUG_printf("Module not yet loaded\n");
uint last = 0;
VSTR_FIXED(path, MICROPY_ALLOC_PATH_MAX)
module_obj = MP_OBJ_NULL;
mp_obj_t top_module_obj = MP_OBJ_NULL;
mp_obj_t outer_module_obj = MP_OBJ_NULL;
uint i;
for (i = 1; i <= mod_len; i++) {
if (i == mod_len || mod_str[i] == '.') {
// create a qstr for the module name up to this depth
qstr mod_name = qstr_from_strn(mod_str, i);
DEBUG_printf("Processing module: %s\n", qstr_str(mod_name));
DEBUG_printf("Previous path: %s\n", vstr_str(&path));
// find the file corresponding to the module name
mp_import_stat_t stat;
if (vstr_len(&path) == 0) {
// first module in the dotted-name; search for a directory or file
stat = find_file(mod_str, i, &path);
} else {
// latter module in the dotted-name; append to path
vstr_add_char(&path, PATH_SEP_CHAR);
vstr_add_strn(&path, mod_str + last, i - last);
stat = stat_dir_or_file(&path);
}
DEBUG_printf("Current path: %s\n", vstr_str(&path));
// fail if we couldn't find the file
if (stat == MP_IMPORT_STAT_NO_EXIST) {
nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_ImportError, "No module named '%s'", qstr_str(mod_name)));
}
module_obj = mp_module_get(mod_name);
if (module_obj == MP_OBJ_NULL) {
// module not already loaded, so load it!
module_obj = mp_obj_new_module(mod_name);
if (stat == MP_IMPORT_STAT_DIR) {
DEBUG_printf("%s is dir\n", vstr_str(&path));
// https://docs.python.org/3/reference/import.html
// "Specifically, any module that contains a __path__ attribute is considered a package."
mp_store_attr(module_obj, MP_QSTR___path__, mp_obj_new_str(vstr_str(&path), vstr_len(&path), false));
vstr_add_char(&path, PATH_SEP_CHAR);
vstr_add_str(&path, "__init__.py");
if (mp_import_stat(vstr_str(&path)) != MP_IMPORT_STAT_FILE) {
vstr_cut_tail_bytes(&path, sizeof("/__init__.py") - 1); // cut off /__init__.py
printf("Notice: %s is imported as namespace package\n", vstr_str(&path));
} else {
do_load(module_obj, &path);
vstr_cut_tail_bytes(&path, sizeof("/__init__.py") - 1); // cut off /__init__.py
}
} else { // MP_IMPORT_STAT_FILE
do_load(module_obj, &path);
// TODO: We cannot just break here, at the very least, we must execute
// trailer code below. But otherwise if there're remaining components,
// that would be (??) object path within module, not modules path within FS.
// break;
}
}
if (outer_module_obj != MP_OBJ_NULL) {
qstr s = qstr_from_strn(mod_str + last, i - last);
mp_store_attr(outer_module_obj, s, module_obj);
}
outer_module_obj = module_obj;
if (top_module_obj == MP_OBJ_NULL) {
top_module_obj = module_obj;
}
last = i + 1;
}
}
if (i < mod_len) {
// we loaded a package, now need to load objects from within that package
// TODO
assert(0);
}
// If fromlist is not empty, return leaf module
if (fromtuple != mp_const_none) {
return module_obj;
}
// Otherwise, we need to return top-level package
return top_module_obj;
}
// Unbuffered, inefficient implementation of readline() for raw I/O files.
STATIC mp_obj_t stream_unbuffered_readline(uint n_args, const mp_obj_t *args) {
struct _mp_obj_base_t *o = (struct _mp_obj_base_t *)args[0];
if (o->type->stream_p == NULL || o->type->stream_p->read == NULL) {
// CPython: io.UnsupportedOperation, OSError subclass
nlr_raise(mp_obj_new_exception_msg(&mp_type_OSError, "Operation not supported"));
}
mp_int_t max_size = -1;
if (n_args > 1) {
max_size = MP_OBJ_SMALL_INT_VALUE(args[1]);
}
vstr_t *vstr;
if (max_size != -1) {
vstr = vstr_new_size(max_size);
} else {
vstr = vstr_new();
}
int error;
while (max_size == -1 || max_size-- != 0) {
char *p = vstr_add_len(vstr, 1);
if (p == NULL) {
nlr_raise(mp_obj_new_exception_msg_varg(&mp_type_MemoryError, "out of memory"));
}
mp_uint_t out_sz = o->type->stream_p->read(o, p, 1, &error);
if (out_sz == MP_STREAM_ERROR) {
if (is_nonblocking_error(error)) {
if (vstr->len == 1) {
// We just incremented it, but otherwise we read nothing
// and immediately got EAGAIN. This is case is not well
// specified in
// https://docs.python.org/3/library/io.html#io.IOBase.readline
// unlike similar case for read(). But we follow the latter's
// behavior - return None.
vstr_free(vstr);
return mp_const_none;
} else {
goto done;
}
}
nlr_raise(mp_obj_new_exception_arg1(&mp_type_OSError, MP_OBJ_NEW_SMALL_INT(error)));
}
if (out_sz == 0) {
done:
// Back out previously added byte
// Consider, what's better - read a char and get OutOfMemory (so read
// char is lost), or allocate first as we do.
vstr_cut_tail_bytes(vstr, 1);
break;
}
if (*p == '\n') {
break;
}
}
// TODO need a string creation API that doesn't copy the given data
mp_obj_t ret = mp_obj_new_str_of_type(STREAM_CONTENT_TYPE(o->type->stream_p), (byte*)vstr->buf, vstr->len);
vstr_free(vstr);
return ret;
}
mp_obj_t mp_builtin___import__(uint n_args, mp_obj_t *args) {
/*
printf("import:\n");
for (int i = 0; i < n_args; i++) {
printf(" ");
mp_obj_print(args[i], PRINT_REPR);
printf("\n");
}
*/
mp_obj_t fromtuple = mp_const_none;
int level = 0;
if (n_args >= 4) {
fromtuple = args[3];
if (n_args >= 5) {
level = MP_OBJ_SMALL_INT_VALUE(args[4]);
}
}
if (level != 0) {
nlr_jump(mp_obj_new_exception_msg(&mp_type_NotImplementedError,
"Relative import is not implemented"));
}
uint mod_len;
const char *mod_str = (const char*)mp_obj_str_get_data(args[0], &mod_len);
// check if module already exists
mp_obj_t module_obj = mp_module_get(mp_obj_str_get_qstr(args[0]));
if (module_obj != MP_OBJ_NULL) {
// If it's not a package, return module right away
char *p = strchr(mod_str, '.');
if (p == NULL) {
return module_obj;
}
// If fromlist is not empty, return leaf module
if (fromtuple != mp_const_none) {
return module_obj;
}
// Otherwise, we need to return top-level package
qstr pkg_name = qstr_from_strn(mod_str, p - mod_str);
return mp_module_get(pkg_name);
}
uint last = 0;
VSTR_FIXED(path, MICROPY_PATH_MAX)
module_obj = MP_OBJ_NULL;
mp_obj_t top_module_obj = MP_OBJ_NULL;
mp_obj_t outer_module_obj = MP_OBJ_NULL;
uint i;
for (i = 1; i <= mod_len; i++) {
if (i == mod_len || mod_str[i] == '.') {
// create a qstr for the module name up to this depth
qstr mod_name = qstr_from_strn(mod_str, i);
// find the file corresponding to the module name
mp_import_stat_t stat;
if (vstr_len(&path) == 0) {
// first module in the dotted-name; search for a directory or file
stat = find_file(mod_str, i, &path);
} else {
// latter module in the dotted-name; append to path
vstr_add_char(&path, PATH_SEP_CHAR);
vstr_add_strn(&path, mod_str + last, i - last);
stat = stat_dir_or_file(&path);
}
// fail if we couldn't find the file
if (stat == MP_IMPORT_STAT_NO_EXIST) {
nlr_jump(mp_obj_new_exception_msg_varg(&mp_type_ImportError, "ImportError: No module named '%s'", qstr_str(mod_name)));
}
module_obj = mp_module_get(mod_name);
if (module_obj == MP_OBJ_NULL) {
// module not already loaded, so load it!
module_obj = mp_obj_new_module(mod_name);
if (stat == MP_IMPORT_STAT_DIR) {
vstr_add_char(&path, PATH_SEP_CHAR);
vstr_add_str(&path, "__init__.py");
if (mp_import_stat(vstr_str(&path)) != MP_IMPORT_STAT_FILE) {
vstr_cut_tail_bytes(&path, sizeof("/__init__.py") - 1); // cut off /__init__.py
nlr_jump(mp_obj_new_exception_msg_varg(&mp_type_ImportError,
"Per PEP-420 a dir without __init__.py (%s) is a namespace package; "
"namespace packages are not supported", vstr_str(&path)));
}
do_load(module_obj, &path);
vstr_cut_tail_bytes(&path, sizeof("/__init__.py") - 1); // cut off /__init__.py
} else { // MP_IMPORT_STAT_FILE
do_load(module_obj, &path);
// TODO: We cannot just break here, at the very least, we must execute
// trailer code below. But otherwise if there're remaining components,
// that would be (??) object path within module, not modules path within FS.
// break;
}
}
if (outer_module_obj != MP_OBJ_NULL) {
qstr s = qstr_from_strn(mod_str + last, i - last);
mp_store_attr(outer_module_obj, s, module_obj);
}
outer_module_obj = module_obj;
if (top_module_obj == MP_OBJ_NULL) {
top_module_obj = module_obj;
}
last = i + 1;
}
}
if (i < mod_len) {
// we loaded a package, now need to load objects from within that package
// TODO
assert(0);
}
// If fromlist is not empty, return leaf module
if (fromtuple != mp_const_none) {
return module_obj;
}
// Otherwise, we need to return top-level package
return top_module_obj;
}
// function to run extra tests for things that can't be checked by scripts
STATIC mp_obj_t extra_coverage(void) {
// mp_printf (used by ports that don't have a native printf)
{
mp_printf(&mp_plat_print, "# mp_printf\n");
mp_printf(&mp_plat_print, "%d %+d % d\n", -123, 123, 123); // sign
mp_printf(&mp_plat_print, "%05d\n", -123); // negative number with zero padding
mp_printf(&mp_plat_print, "%ld\n", 123); // long
mp_printf(&mp_plat_print, "%X\n", 0x1abcdef); // capital hex
mp_printf(&mp_plat_print, "%.2s %.3s\n", "abc", "abc"); // fixed string precision
mp_printf(&mp_plat_print, "%.*s\n", -1, "abc"); // negative string precision
mp_printf(&mp_plat_print, "%b %b\n", 0, 1); // bools
mp_printf(&mp_plat_print, "%s\n", NULL); // null string
mp_printf(&mp_plat_print, "%d\n", 0x80000000); // should print signed
mp_printf(&mp_plat_print, "%u\n", 0x80000000); // should print unsigned
mp_printf(&mp_plat_print, "%x\n", 0x80000000); // should print unsigned
mp_printf(&mp_plat_print, "%X\n", 0x80000000); // should print unsigned
}
// vstr
{
mp_printf(&mp_plat_print, "# vstr\n");
vstr_t *vstr = vstr_new(16);
vstr_hint_size(vstr, 32);
vstr_add_str(vstr, "ts");
vstr_ins_byte(vstr, 1, 'e');
vstr_ins_char(vstr, 3, 't');
vstr_ins_char(vstr, 10, 's');
mp_printf(&mp_plat_print, "%.*s\n", (int)vstr->len, vstr->buf);
vstr_cut_head_bytes(vstr, 2);
mp_printf(&mp_plat_print, "%.*s\n", (int)vstr->len, vstr->buf);
vstr_cut_tail_bytes(vstr, 10);
mp_printf(&mp_plat_print, "%.*s\n", (int)vstr->len, vstr->buf);
vstr_printf(vstr, "t%cst", 'e');
mp_printf(&mp_plat_print, "%.*s\n", (int)vstr->len, vstr->buf);
vstr_cut_out_bytes(vstr, 3, 10);
mp_printf(&mp_plat_print, "%.*s\n", (int)vstr->len, vstr->buf);
VSTR_FIXED(fix, 4);
vstr_add_str(&fix, "large");
mp_printf(&mp_plat_print, "%.*s\n", (int)fix.len, fix.buf);
}
// repl autocomplete
{
mp_printf(&mp_plat_print, "# repl\n");
const char *str;
mp_uint_t len = mp_repl_autocomplete("__n", 3, &mp_plat_print, &str);
mp_printf(&mp_plat_print, "%.*s\n", (int)len, str);
mp_store_global(MP_QSTR_sys, mp_import_name(MP_QSTR_sys, mp_const_none, MP_OBJ_NEW_SMALL_INT(0)));
mp_repl_autocomplete("sys.", 4, &mp_plat_print, &str);
len = mp_repl_autocomplete("sys.impl", 8, &mp_plat_print, &str);
mp_printf(&mp_plat_print, "%.*s\n", (int)len, str);
}
// attrtuple
{
mp_printf(&mp_plat_print, "# attrtuple\n");
static const qstr fields[] = {MP_QSTR_start, MP_QSTR_stop, MP_QSTR_step};
static const mp_obj_t items[] = {MP_OBJ_NEW_SMALL_INT(1), MP_OBJ_NEW_SMALL_INT(2), MP_OBJ_NEW_SMALL_INT(3)};
mp_obj_print_helper(&mp_plat_print, mp_obj_new_attrtuple(fields, 3, items), PRINT_REPR);
mp_printf(&mp_plat_print, "\n");
}
// str
{
mp_printf(&mp_plat_print, "# str\n");
// intern string
mp_printf(&mp_plat_print, "%d\n", MP_OBJ_IS_QSTR(mp_obj_str_intern(mp_obj_new_str("intern me", 9, false))));
}
// mpz
{
mp_printf(&mp_plat_print, "# mpz\n");
mp_uint_t value;
mpz_t mpz;
mpz_init_zero(&mpz);
// mpz_as_uint_checked, with success
mpz_set_from_int(&mpz, 12345678);
mp_printf(&mp_plat_print, "%d\n", mpz_as_uint_checked(&mpz, &value));
mp_printf(&mp_plat_print, "%d\n", (int)value);
// mpz_as_uint_checked, with negative arg
mpz_set_from_int(&mpz, -1);
mp_printf(&mp_plat_print, "%d\n", mpz_as_uint_checked(&mpz, &value));
// mpz_as_uint_checked, with overflowing arg
mpz_set_from_int(&mpz, 1);
mpz_shl_inpl(&mpz, &mpz, 70);
mp_printf(&mp_plat_print, "%d\n", mpz_as_uint_checked(&mpz, &value));
}
// runtime utils
{
mp_printf(&mp_plat_print, "# runtime utils\n");
// call mp_call_function_1_protected
mp_call_function_1_protected(MP_OBJ_FROM_PTR(&mp_builtin_abs_obj), MP_OBJ_NEW_SMALL_INT(1));
// call mp_call_function_1_protected with invalid args
mp_call_function_1_protected(MP_OBJ_FROM_PTR(&mp_builtin_abs_obj), mp_obj_new_str("abc", 3, false));
// call mp_call_function_2_protected
mp_call_function_2_protected(MP_OBJ_FROM_PTR(&mp_builtin_divmod_obj), MP_OBJ_NEW_SMALL_INT(1), MP_OBJ_NEW_SMALL_INT(1));
// call mp_call_function_2_protected with invalid args
mp_call_function_2_protected(MP_OBJ_FROM_PTR(&mp_builtin_divmod_obj), mp_obj_new_str("abc", 3, false), mp_obj_new_str("abc", 3, false));
}
// warning
{
mp_emitter_warning(MP_PASS_CODE_SIZE, "test");
}
// format float
{
mp_printf(&mp_plat_print, "# format float\n");
// format with inadequate buffer size
char buf[5];
mp_format_float(1, buf, sizeof(buf), 'g', 0, '+');
mp_printf(&mp_plat_print, "%s\n", buf);
// format with just enough buffer so that precision must be
// set from 0 to 1 twice
char buf2[8];
mp_format_float(1, buf2, sizeof(buf2), 'g', 0, '+');
mp_printf(&mp_plat_print, "%s\n", buf2);
// format where precision is trimmed to avoid buffer overflow
mp_format_float(1, buf2, sizeof(buf2), 'e', 0, '+');
mp_printf(&mp_plat_print, "%s\n", buf2);
}
// return a tuple of data for testing on the Python side
mp_obj_t items[] = {(mp_obj_t)&str_no_hash_obj, (mp_obj_t)&bytes_no_hash_obj};
return mp_obj_new_tuple(MP_ARRAY_SIZE(items), items);
}
STATIC void parse_string_literal(mp_lexer_t *lex, bool is_raw) {
// get first quoting character
char quote_char = '\'';
if (is_char(lex, '\"')) {
quote_char = '\"';
}
next_char(lex);
// work out if it's a single or triple quoted literal
size_t num_quotes;
if (is_char_and(lex, quote_char, quote_char)) {
// triple quotes
next_char(lex);
next_char(lex);
num_quotes = 3;
} else {
// single quotes
num_quotes = 1;
}
size_t n_closing = 0;
while (!is_end(lex) && (num_quotes > 1 || !is_char(lex, '\n')) && n_closing < num_quotes) {
if (is_char(lex, quote_char)) {
n_closing += 1;
vstr_add_char(&lex->vstr, CUR_CHAR(lex));
} else {
n_closing = 0;
if (is_char(lex, '\\')) {
next_char(lex);
unichar c = CUR_CHAR(lex);
if (is_raw) {
// raw strings allow escaping of quotes, but the backslash is also emitted
vstr_add_char(&lex->vstr, '\\');
} else {
switch (c) {
// note: "c" can never be MP_LEXER_EOF because next_char
// always inserts a newline at the end of the input stream
case '\n': c = MP_LEXER_EOF; break; // backslash escape the newline, just ignore it
case '\\': break;
case '\'': break;
case '"': break;
case 'a': c = 0x07; break;
case 'b': c = 0x08; break;
case 't': c = 0x09; break;
case 'n': c = 0x0a; break;
case 'v': c = 0x0b; break;
case 'f': c = 0x0c; break;
case 'r': c = 0x0d; break;
case 'u':
case 'U':
if (lex->tok_kind == MP_TOKEN_BYTES) {
// b'\u1234' == b'\\u1234'
vstr_add_char(&lex->vstr, '\\');
break;
}
// Otherwise fall through.
case 'x':
{
mp_uint_t num = 0;
if (!get_hex(lex, (c == 'x' ? 2 : c == 'u' ? 4 : 8), &num)) {
// not enough hex chars for escape sequence
lex->tok_kind = MP_TOKEN_INVALID;
}
c = num;
break;
}
case 'N':
// Supporting '\N{LATIN SMALL LETTER A}' == 'a' would require keeping the
// entire Unicode name table in the core. As of Unicode 6.3.0, that's nearly
// 3MB of text; even gzip-compressed and with minimal structure, it'll take
// roughly half a meg of storage. This form of Unicode escape may be added
// later on, but it's definitely not a priority right now. -- CJA 20140607
mp_not_implemented("unicode name escapes");
break;
default:
if (c >= '0' && c <= '7') {
// Octal sequence, 1-3 chars
size_t digits = 3;
mp_uint_t num = c - '0';
while (is_following_odigit(lex) && --digits != 0) {
next_char(lex);
num = num * 8 + (CUR_CHAR(lex) - '0');
}
c = num;
} else {
// unrecognised escape character; CPython lets this through verbatim as '\' and then the character
vstr_add_char(&lex->vstr, '\\');
}
break;
}
}
if (c != MP_LEXER_EOF) {
if (MICROPY_PY_BUILTINS_STR_UNICODE_DYNAMIC) {
if (c < 0x110000 && lex->tok_kind == MP_TOKEN_STRING) {
vstr_add_char(&lex->vstr, c);
} else if (c < 0x100 && lex->tok_kind == MP_TOKEN_BYTES) {
vstr_add_byte(&lex->vstr, c);
} else {
// unicode character out of range
// this raises a generic SyntaxError; could provide more info
lex->tok_kind = MP_TOKEN_INVALID;
}
} else {
// without unicode everything is just added as an 8-bit byte
if (c < 0x100) {
vstr_add_byte(&lex->vstr, c);
} else {
// 8-bit character out of range
// this raises a generic SyntaxError; could provide more info
lex->tok_kind = MP_TOKEN_INVALID;
}
}
}
} else {
// Add the "character" as a byte so that we remain 8-bit clean.
// This way, strings are parsed correctly whether or not they contain utf-8 chars.
vstr_add_byte(&lex->vstr, CUR_CHAR(lex));
}
}
next_char(lex);
}
// check we got the required end quotes
if (n_closing < num_quotes) {
lex->tok_kind = MP_TOKEN_LONELY_STRING_OPEN;
}
// cut off the end quotes from the token text
vstr_cut_tail_bytes(&lex->vstr, n_closing);
}
