Handle strings with interior null pointer (#222)

Author: lucic71

cpp2rust/converter/converter.cpp

@@ -1991,8 +1991,15 @@ bool Converter::VisitStringLiteral(clang::StringLiteral *expr) {
                        GetEscapedStringLiteral(expr, 1)));
     return false;
   }
-  assert(!expr->getString().contains('\0') &&
-         "interior null byte in string literal");
+  if (expr->getString().contains('\0')) {
+    std::string out = "(&[";
+    for (unsigned char c : expr->getString()) {
+      out += getTypedLiteral(std::to_string(c).c_str(), CharRustType()) + ", ";
+    }
+    out += getTypedLiteral("0", CharRustType()) + "])";
+    StrCat(out);
+    return false;
+  }
   StrCat(std::format("c{}", GetEscapedStringLiteral(expr, 0)));
   return false;
 }
@@ -2927,7 +2934,8 @@ bool Converter::VisitCompoundLiteralExpr(clang::CompoundLiteralExpr *expr) {
 
 bool Converter::VisitArraySubscriptExpr(clang::ArraySubscriptExpr *expr) {
   auto *base = expr->getBase();
-  if (base->IgnoreCasts()->getType()->isPointerType()) {
+  if (base->IgnoreCasts()->getType()->isPointerType() ||
+      clang::isa<clang::StringLiteral>(base->IgnoreCasts())) {
     ConvertPointerSubscript(expr);
   } else {
     ConvertArraySubscript(base, expr->getIdx(), expr->getType());

tests/unit/out/refcount/string_literal_interior_null.rs

@@ -0,0 +1,44 @@
+extern crate libcc2rs;
+use libcc2rs::*;
+use std::cell::RefCell;
+use std::collections::BTreeMap;
+use std::io::prelude::*;
+use std::io::{Read, Seek, Write};
+use std::os::fd::AsFd;
+use std::rc::{Rc, Weak};
+pub fn sum_bytes_0(buf: Ptr<u8>, len: u32) -> i32 {
+    let buf: Value<Ptr<u8>> = Rc::new(RefCell::new(buf));
+    let len: Value<u32> = Rc::new(RefCell::new(len));
+    let sum: Value<i32> = Rc::new(RefCell::new(0));
+    let i: Value<u32> = Rc::new(RefCell::new(0_u32));
+    'loop_: while ((*i.borrow()) < (*len.borrow())) {
+        (*sum.borrow_mut()) += (((*buf.borrow()).offset((*i.borrow()) as isize).read()) as i32);
+        (*i.borrow_mut()).postfix_inc();
+    }
+    return (*sum.borrow());
+}
+thread_local!(
+    pub static g_packet_1: Value<Ptr<u8>> =
+        Rc::new(RefCell::new(Ptr::from_string_literal(b"\x01\0")));
+);
+pub fn main() {
+    std::process::exit(main_0());
+}
+fn main_0() -> i32 {
+    let a: Value<i32> = Rc::new(RefCell::new(
+        ({ sum_bytes_0(Ptr::from_string_literal(b"\x01\0"), 2_u32) }),
+    ));
+    let b: Value<i32> = Rc::new(RefCell::new(
+        ({ sum_bytes_0((*g_packet_1.with(Value::clone).borrow()).clone(), 2_u32) }),
+    ));
+    assert!(((*a.borrow()) == (*b.borrow())));
+    assert!(((*a.borrow()) == 1));
+    let c: Value<i32> = Rc::new(RefCell::new(
+        ((b"\r\n.\r\n"[(0) as usize] as i32) + (b"\r\n.\r\n"[(3) as usize] as i32)),
+    ));
+    assert!(((*c.borrow()) == ((('\r' as u8) as i32) + (('\r' as u8) as i32))));
+    let idx: Value<i32> = Rc::new(RefCell::new(1));
+    let d: Value<i32> = Rc::new(RefCell::new((b"abcd"[(*idx.borrow()) as usize] as i32)));
+    assert!(((*d.borrow()) == (('b' as u8) as i32)));
+    return 0;
+}

tests/unit/out/unsafe/string_literal_interior_null.rs

@@ -0,0 +1,53 @@
+extern crate libc;
+use libc::*;
+extern crate libcc2rs;
+use libcc2rs::*;
+use std::collections::BTreeMap;
+use std::io::{Read, Seek, Write};
+use std::os::fd::{AsFd, FromRawFd, IntoRawFd};
+use std::rc::Rc;
+pub unsafe fn sum_bytes_0(mut buf: *const libc::c_char, mut len: u32) -> i32 {
+    let mut sum: i32 = 0;
+    let mut i: u32 = 0_u32;
+    'loop_: while ((i) < (len)) {
+        sum += (((*buf.offset((i) as isize)) as u8) as i32);
+        i.postfix_inc();
+    }
+    return sum;
+}
+pub static mut g_packet_1: *const libc::c_char = unsafe {
+    (&[
+        (1 as libc::c_char),
+        (0 as libc::c_char),
+        (0 as libc::c_char),
+    ])
+        .as_ptr()
+};
+pub fn main() {
+    unsafe {
+        std::process::exit(main_0() as i32);
+    }
+}
+unsafe fn main_0() -> i32 {
+    let mut a: i32 = (unsafe {
+        sum_bytes_0(
+            (&[
+                (1 as libc::c_char),
+                (0 as libc::c_char),
+                (0 as libc::c_char),
+            ])
+                .as_ptr(),
+            2_u32,
+        )
+    });
+    let mut b: i32 = (unsafe { sum_bytes_0(g_packet_1, 2_u32) });
+    assert!(((a) == (b)));
+    assert!(((a) == (1)));
+    let mut c: i32 = (((*c"\r\n.\r\n".as_ptr().offset((0) as isize)) as i32)
+        + ((*c"\r\n.\r\n".as_ptr().offset((3) as isize)) as i32));
+    assert!(((c) == ((('\r' as libc::c_char) as i32) + (('\r' as libc::c_char) as i32))));
+    let mut idx: i32 = 1;
+    let mut d: i32 = ((*c"abcd".as_ptr().offset((idx) as isize)) as i32);
+    assert!(((d) == (('b' as libc::c_char) as i32)));
+    return 0;
+}

tests/unit/string_literal_interior_null.cpp

@@ -0,0 +1,29 @@
+#include <assert.h>
+#include <cstdio>
+
+static int sum_bytes(const char *buf, unsigned len) {
+  int sum = 0;
+  for (unsigned i = 0; i < len; i++) {
+    sum += (unsigned char)buf[i];
+  }
+  return sum;
+}
+
+static const char *g_packet = "\x01\x00";
+
+int main() {
+  int a = sum_bytes("\x01\x00", 2);
+  int b = sum_bytes(g_packet, 2);
+
+  assert(a == b);
+  assert(a == 1);
+
+  int c = "\r\n.\r\n"[0] + "\r\n.\r\n"[3];
+  assert(c == '\r' + '\r');
+
+  int idx = 1;
+  int d = "abcd"[idx];
+  assert(d == 'b');
+
+  return 0;
+}