bool parser_t::detect_errors_in_argument_list(const wcstring &arg_list_src, wcstring *out,
const wchar_t *prefix) const {
bool errored = false;
parse_error_list_t errors;
// Use empty string for the prefix if it's NULL.
if (!prefix) prefix = L""; //!OCLINT(parameter reassignment)
// Parse the string as an argument list.
parse_node_tree_t tree;
if (!parse_tree_from_string(arg_list_src, parse_flag_none, &tree, &errors,
symbol_freestanding_argument_list)) {
// Failed to parse.
errored = true;
}
if (!errored) {
// Get the root argument list and extract arguments from it.
assert(!tree.empty()); //!OCLINT(multiple unary operator)
tnode_t<grammar::freestanding_argument_list> arg_list(&tree, &tree.at(0));
while (auto arg = arg_list.next_in_list<grammar::argument>()) {
const wcstring arg_src = arg.get_source(arg_list_src);
if (parse_util_detect_errors_in_argument(arg, arg_src, &errors)) {
errored = true;
}
}
}
if (!errors.empty() && out != NULL) {
out->assign(errors.at(0).describe_with_prefix(
arg_list_src, prefix, false /* not interactive */, false /* don't skip caret */));
}
return errored;
}
bool parser_t::detect_errors_in_argument_list(const wcstring &arg_list_src, wcstring *out, const wchar_t *prefix)
{
bool errored = false;
parse_error_list_t errors;
/* Use empty string for the prefix if it's NULL */
if (prefix == NULL)
{
prefix = L"";
}
/* Parse the string as an argument list */
parse_node_tree_t tree;
if (! parse_tree_from_string(arg_list_src, parse_flag_none, &tree, &errors, symbol_freestanding_argument_list))
{
/* Failed to parse. */
errored = true;
}
if (! errored)
{
/* Get the root argument list */
assert(! tree.empty());
const parse_node_t *arg_list = &tree.at(0);
assert(arg_list->type == symbol_freestanding_argument_list);
/* Extract arguments from it */
while (arg_list != NULL && ! errored)
{
const parse_node_t *arg_node = tree.next_node_in_node_list(*arg_list, symbol_argument, &arg_list);
if (arg_node != NULL)
{
const wcstring arg_src = arg_node->get_source(arg_list_src);
if (parse_util_detect_errors_in_argument(*arg_node, arg_src, &errors))
{
errored = true;
}
}
}
}
if (! errors.empty() && out != NULL)
{
out->assign(errors.at(0).describe_with_prefix(arg_list_src, prefix, false /* not interactive */, false /* don't skip caret */));
}
return errored;
}
parser_test_error_bits_t parse_util_detect_errors(const wcstring &buff_src, parse_error_list_t *out_errors, bool allow_incomplete)
{
parse_node_tree_t node_tree;
parse_error_list_t parse_errors;
parser_test_error_bits_t res = 0;
// Whether we encountered a parse error
bool errored = false;
// Whether we encountered an unclosed block
// We detect this via an 'end_command' block without source
bool has_unclosed_block = false;
// Whether there's an unclosed quote, and therefore unfinished
// This is only set if allow_incomplete is set
bool has_unclosed_quote = false;
// Parse the input string into a parse tree
// Some errors are detected here
bool parsed = parse_tree_from_string(buff_src, allow_incomplete ? parse_flag_leave_unterminated : parse_flag_none, &node_tree, &parse_errors);
if (allow_incomplete)
{
for (size_t i=0; i < parse_errors.size(); i++)
{
if (parse_errors.at(i).code == parse_error_tokenizer_unterminated_quote)
{
// Remove this error, since we don't consider it a real error
has_unclosed_quote = true;
parse_errors.erase(parse_errors.begin() + i);
i--;
}
}
}
// #1238: If the only error was unterminated quote, then consider this to have parsed successfully. A better fix would be to have parse_tree_from_string return this information directly (but it would be a shame to munge up its nice bool return).
if (parse_errors.empty() && has_unclosed_quote)
{
parsed = true;
}
if (! parsed)
{
errored = true;
}
// has_unclosed_quote may only be set if allow_incomplete is true
assert(! has_unclosed_quote || allow_incomplete);
// Expand all commands
// Verify 'or' and 'and' not used inside pipelines
// Verify pipes via parser_is_pipe_forbidden
// Verify return only within a function
// Verify no variable expansions
if (! errored)
{
const size_t node_tree_size = node_tree.size();
for (size_t i=0; i < node_tree_size; i++)
{
const parse_node_t &node = node_tree.at(i);
if (node.type == symbol_end_command && ! node.has_source())
{
// an 'end' without source is an unclosed block
has_unclosed_block = true;
}
else if (node.type == symbol_boolean_statement)
{
// 'or' and 'and' can be in a pipeline, as long as they're first
parse_bool_statement_type_t type = parse_node_tree_t::statement_boolean_type(node);
if ((type == parse_bool_and || type == parse_bool_or) && node_tree.statement_is_in_pipeline(node, false /* don't count first */))
{
errored = append_syntax_error(&parse_errors, node, EXEC_ERR_MSG, (type == parse_bool_and) ? L"and" : L"or");
}
}
else if (node.type == symbol_argument)
{
const wcstring arg_src = node.get_source(buff_src);
res |= parse_util_detect_errors_in_argument(node, arg_src, &parse_errors);
}
else if (node.type == symbol_job)
{
if (node_tree.job_should_be_backgrounded(node))
{
/* Disallow background in the following cases:
foo & ; and bar
foo & ; or bar
if foo & ; end
while foo & ; end
*/
const parse_node_t *job_parent = node_tree.get_parent(node);
assert(job_parent != NULL);
switch (job_parent->type)
{
case symbol_if_clause:
case symbol_while_header:
{
assert(node_tree.get_child(*job_parent, 1) == &node);
errored = append_syntax_error(&parse_errors, node, BACKGROUND_IN_CONDITIONAL_ERROR_MSG);
break;
}
case symbol_job_list:
{
// This isn't very complete, e.g. we don't catch 'foo & ; not and bar'
assert(node_tree.get_child(*job_parent, 0) == &node);
const parse_node_t *next_job_list = node_tree.get_child(*job_parent, 1, symbol_job_list);
assert(next_job_list != NULL);
const parse_node_t *next_job = node_tree.next_node_in_node_list(*next_job_list, symbol_job, NULL);
if (next_job != NULL)
{
const parse_node_t *next_statement = node_tree.get_child(*next_job, 0, symbol_statement);
if (next_statement != NULL)
{
const parse_node_t *spec_statement = node_tree.get_child(*next_statement, 0);
if (spec_statement && spec_statement->type == symbol_boolean_statement)
{
switch (parse_node_tree_t::statement_boolean_type(*spec_statement))
{
// These are not allowed
case parse_bool_and:
errored = append_syntax_error(&parse_errors, *spec_statement, BOOL_AFTER_BACKGROUND_ERROR_MSG, L"and");
break;
case parse_bool_or:
errored = append_syntax_error(&parse_errors, *spec_statement, BOOL_AFTER_BACKGROUND_ERROR_MSG, L"or");
break;
case parse_bool_not:
// This one is OK
break;
}
}
}
}
break;
}
default:
break;
}
}
}
else if (node.type == symbol_plain_statement)
{
// In a few places below, we want to know if we are in a pipeline
const bool is_in_pipeline = node_tree.statement_is_in_pipeline(node, true /* count first */);
// We need to know the decoration
const enum parse_statement_decoration_t decoration = node_tree.decoration_for_plain_statement(node);
// Check that we don't try to pipe through exec
if (is_in_pipeline && decoration == parse_statement_decoration_exec)
{
errored = append_syntax_error(&parse_errors, node, EXEC_ERR_MSG, L"exec");
}
wcstring command;
if (node_tree.command_for_plain_statement(node, buff_src, &command))
{
// Check that we can expand the command
if (! expand_one(command, EXPAND_SKIP_CMDSUBST | EXPAND_SKIP_VARIABLES | EXPAND_SKIP_JOBS, NULL))
{
// TODO: leverage the resulting errors
errored = append_syntax_error(&parse_errors, node, ILLEGAL_CMD_ERR_MSG, command.c_str());
}
// Check that pipes are sound
if (! errored && parser_is_pipe_forbidden(command) && is_in_pipeline)
{
errored = append_syntax_error(&parse_errors, node, EXEC_ERR_MSG, command.c_str());
}
// Check that we don't return from outside a function
// But we allow it if it's 'return --help'
if (! errored && command == L"return")
{
const parse_node_t *ancestor = &node;
bool found_function = false;
while (ancestor != NULL)
{
const parse_node_t *possible_function_header = node_tree.header_node_for_block_statement(*ancestor);
if (possible_function_header != NULL && possible_function_header->type == symbol_function_header)
{
found_function = true;
break;
}
ancestor = node_tree.get_parent(*ancestor);
}
if (! found_function && ! first_argument_is_help(node_tree, node, buff_src))
{
errored = append_syntax_error(&parse_errors, node, INVALID_RETURN_ERR_MSG);
}
}
// Check that we don't break or continue from outside a loop
if (! errored && (command == L"break" || command == L"continue"))
{
// Walk up until we hit a 'for' or 'while' loop. If we hit a function first, stop the search; we can't break an outer loop from inside a function.
// This is a little funny because we can't tell if it's a 'for' or 'while' loop from the ancestor alone; we need the header. That is, we hit a block_statement, and have to check its header.
bool found_loop = false, end_search = false;
const parse_node_t *ancestor = &node;
while (ancestor != NULL && ! end_search)
{
const parse_node_t *loop_or_function_header = node_tree.header_node_for_block_statement(*ancestor);
if (loop_or_function_header != NULL)
{
switch (loop_or_function_header->type)
{
case symbol_while_header:
case symbol_for_header:
// this is a loop header, so we can break or continue
found_loop = true;
end_search = true;
break;
case symbol_function_header:
// this is a function header, so we cannot break or continue. We stop our search here.
found_loop = false;
end_search = true;
break;
default:
// most likely begin / end style block, which makes no difference
break;
}
}
ancestor = node_tree.get_parent(*ancestor);
}
if (! found_loop && ! first_argument_is_help(node_tree, node, buff_src))
{
errored = append_syntax_error(&parse_errors, node, (command == L"break" ? INVALID_BREAK_ERR_MSG : INVALID_CONTINUE_ERR_MSG));
}
}
// Check that we don't do an invalid builtin (#1252)
if (! errored && decoration == parse_statement_decoration_builtin && ! builtin_exists(command))
{
errored = append_syntax_error(&parse_errors, node, UNKNOWN_BUILTIN_ERR_MSG, command.c_str());
}
}
}
}
}
if (errored)
res |= PARSER_TEST_ERROR;
if (has_unclosed_block || has_unclosed_quote)
res |= PARSER_TEST_INCOMPLETE;
if (out_errors)
{
out_errors->swap(parse_errors);
}
return res;
}
