gccrs: Add Unicode check for crate_name attributes

gcc/rust/ChangeLog:

	* lex/rust-codepoint.h: Add comment
	* lex/rust-lex.h: New method to get decoded characters
	* rust-session-manager.cc (validate_crate_name): Modify unicode check
	(rust_crate_name_validation_test): Add testcases
	* util/rust-unicode.h (RUST_UNICODE_H): New class Utf8String.
	(class Utf8String): New class.
	* util/rust-unicode.cc (binary_search_sorted_array): Add comment.
	(recursive_decomp_cano): Add comment.
	(recomp): Remove dead code.
	(dump_string): Removed.

gcc/testsuite/ChangeLog:

	* rust/compile/bad-crate-name.rs: Moved to...
	* rust/compile/bad-crate-name1.rs: ...here.
	* rust/compile/bad-crate-name2.rs: New test.

Signed-off-by: Raiki Tamura <tamaron1203@gmail.com>
This commit is contained in:
Raiki Tamura 2023-07-14 14:45:34 +09:00 committed by Arthur Cohen
parent 884dec3a42
commit f7b2e17682
7 changed files with 59 additions and 29 deletions

View file

@ -22,6 +22,8 @@
#include "rust-system.h"
namespace Rust {
// FIXME: move this to rust-unicode.h?
struct Codepoint
{
uint32_t value;

View file

@ -334,6 +334,14 @@ public:
return c;
}
}
tl::optional<std::vector<Codepoint>> get_chars ()
{
if (is_valid ())
return {chars};
else
return tl::nullopt;
}
};
class FileInputSource : public InputSource

View file

@ -42,6 +42,7 @@
#include "rust-early-name-resolver.h"
#include "rust-cfg-strip.h"
#include "rust-expand-visitor.h"
#include "rust-unicode.h"
#include "diagnostic.h"
#include "input.h"
@ -107,30 +108,39 @@ infer_crate_name (const std::string &filename)
return crate;
}
/* Validate the crate name using the ASCII rules
TODO: Support Unicode version of the rules */
/* Validate the crate name using the ASCII rules */
static bool
validate_crate_name (const std::string &crate_name, Error &error)
{
if (crate_name.empty ())
Utf8String utf8_name = {crate_name};
tl::optional<std::vector<Codepoint>> uchars_opt = utf8_name.get_chars ();
if (!uchars_opt.has_value ())
{
error = Error (UNDEF_LOCATION, "crate name is not a valid UTF-8 string");
return false;
}
std::vector<Codepoint> uchars = uchars_opt.value ();
if (uchars.empty ())
{
error = Error (UNDEF_LOCATION, "crate name cannot be empty");
return false;
}
if (crate_name.length () > kMaxNameLength)
if (uchars.size () > kMaxNameLength)
{
error = Error (UNDEF_LOCATION, "crate name cannot exceed %lu characters",
(unsigned long) kMaxNameLength);
return false;
}
for (auto &c : crate_name)
for (Codepoint &c : uchars)
{
if (!(ISALNUM (c) || c == '_'))
if (!(is_alphabetic (c.value) || is_numeric (c.value) || c.value == '_'))
{
error = Error (UNDEF_LOCATION,
"invalid character %<%c%> in crate name: %<%s%>", c,
crate_name.c_str ());
"invalid character %<%s%> in crate name: %<%s%>",
c.as_string ().c_str (), crate_name.c_str ());
return false;
}
}
@ -1273,13 +1283,17 @@ rust_crate_name_validation_test (void)
ASSERT_TRUE (Rust::validate_crate_name ("example", error));
ASSERT_TRUE (Rust::validate_crate_name ("abcdefg_1234", error));
ASSERT_TRUE (Rust::validate_crate_name ("1", error));
// FIXME: The next test does not pass as of current implementation
// ASSERT_TRUE (Rust::CompileOptions::validate_crate_name ("惊吓"));
ASSERT_TRUE (Rust::validate_crate_name ("クレート", error));
ASSERT_TRUE (Rust::validate_crate_name ("Sōkrátēs", error));
ASSERT_TRUE (Rust::validate_crate_name ("惊吓", error));
// NOTE: - is not allowed in the crate name ...
ASSERT_FALSE (Rust::validate_crate_name ("abcdefg-1234", error));
ASSERT_FALSE (Rust::validate_crate_name ("a+b", error));
ASSERT_FALSE (Rust::validate_crate_name ("/a+b/", error));
ASSERT_FALSE (Rust::validate_crate_name ("😸++", error));
ASSERT_FALSE (Rust::validate_crate_name ("", error));
/* Tests for crate name inference */
ASSERT_EQ (Rust::infer_crate_name ("c.rs"), "c");

View file

@ -12,6 +12,7 @@ typedef std::vector<codepoint_t> string_t;
template <std::size_t SIZE>
int64_t
binary_search_ranges (
// FIXME: use binray search function from <algorithm>
const std::array<std::pair<uint32_t, uint32_t>, SIZE> &ranges,
uint32_t target_cp)
{
@ -49,6 +50,7 @@ int64_t
binary_search_sorted_array (const std::array<uint32_t, SIZE> &array,
uint32_t target)
{
// FIXME: use binray search function from <algorithm>
if (SIZE == 0)
return -1;
@ -104,9 +106,7 @@ recursive_decomp_cano (codepoint_t c, string_t &buf)
{
string_t decomped = it->second;
for (codepoint_t cp : decomped)
{
recursive_decomp_cano (cp, buf);
}
recursive_decomp_cano (cp, buf);
}
else
buf.push_back (c);
@ -152,8 +152,7 @@ recomp (string_t s)
if (s.size () > 0)
{
int last_class = -1;
// int starter_pos = 0; // Assume the first character is Starter. Correct?
// int target_pos = 1;
// Assume the first character is Starter.
codepoint_t starter_ch = s[0];
for (unsigned int src_pos = 1; src_pos < s.size (); src_pos++)
{
@ -189,20 +188,6 @@ recomp (string_t s)
return buf;
}
// TODO: remove
/*
void
dump_string (std::vector<uint32_t> s)
{
std::cout << "dump=";
for (auto c : s)
{
std::cout << std::hex << c << ", ";
}
std::cout << std::endl;
}
*/
string_t
nfc_normalize (string_t s)
{

View file

@ -19,10 +19,29 @@
#ifndef RUST_UNICODE_H
#define RUST_UNICODE_H
#include "optional.h"
#include "rust-system.h"
#include "rust-lex.h"
namespace Rust {
class Utf8String
{
private:
tl::optional<std::vector<Codepoint>> chars;
public:
Utf8String (const std::string &maybe_utf8)
{
Lexer::BufferInputSource input_source = {maybe_utf8, 0};
chars = input_source.get_chars ();
}
// Returns UTF codepoints when string is valid as UTF-8, returns nullopt
// otherwise.
tl::optional<std::vector<Codepoint>> get_chars () const { return chars; }
};
// TODO: add function nfc_normalize
bool

View file

@ -0,0 +1,2 @@
#![crate_name = "😅"] // { dg-error "invalid character ...." "" }
fn main() {}