Implement implicit byte checker in C.

* Use it in `jl_load_`
* Remove the length parameter from `jl_load`, `jl_load_file_string` and
  `jl_parse_eval_all` to NOT pretend they support non-NUL-terminated strings.

base/loading.jl

@@ -378,8 +378,8 @@ end
 # remote/parallel load
 
 include_string(txt::String, fname::String) =
- ccall(:jl_load_file_string, Any, (Ptr{UInt8},Csize_t,Ptr{UInt8},Csize_t),
- txt, sizeof(txt), fname, sizeof(fname))
+ ccall(:jl_load_file_string, Any, (Ptr{UInt8},Csize_t,Cstring),
+ txt, sizeof(txt), fname)
 
 include_string(txt::AbstractString, fname::AbstractString="string") =
  include_string(String(txt), String(fname))

src/array.c

@@ -791,6 +791,46 @@ JL_DLLEXPORT jl_value_t *(jl_array_data_owner)(jl_array_t *a)
  return jl_array_data_owner(a);
 }
 
+STATIC_INLINE int jl_has_implicit_byte_owned(jl_array_t *a)
+{
+ assert(a->flags.how != 3);
+ if (!a->flags.isshared)
+ return 1;
+ return a->flags.how == 1;
+}
+
+STATIC_INLINE int jl_has_implicit_byte(jl_array_t *a)
+{
+ // * unshared:
+ // * how: 0-2
+ // We own and allocated the data.
+ // It should have the extra byte.
+ // * shared:
+ // * how: 0, 2
+ // The data might come from external source without implicit NUL byte.
+ // There could be an entra byte for a `reinterpreted` array
+ // but that should be unlikely for strings.
+ // * how: 1
+ // We allocated the data with the extra byte.
+ // * how: 3
+ // We should check the owner.
+ if (a->flags.how == 3) {
+ a = (jl_array_t*)jl_array_data_owner(a);
+ return a->elsize == 1 && jl_has_implicit_byte_owned(a);
+ }
+ return jl_has_implicit_byte_owned(a);
+}
+
+// Create an array with the same content
+JL_DLLEXPORT jl_array_t *jl_array_cconvert_cstring(jl_array_t *a)
+{
+ assert(jl_typeof(a) == jl_array_uint8_type);
+ if (!jl_has_implicit_byte(a))
+ a = jl_array_copy(a);
+ ((char*)a->data)[a->nrows] = 0;
+ return a;
+}
+
 #ifdef __cplusplus
 }
 #endif

src/ast.c

@@ -641,14 +641,15 @@ JL_DLLEXPORT jl_value_t *jl_parse_string(const char *str, size_t len,
 }
 
 // parse and eval a whole file, possibly reading from a string (`content`)
-jl_value_t *jl_parse_eval_all(const char *fname, size_t len,
+jl_value_t *jl_parse_eval_all(const char *fname,
  const char *content, size_t contentlen)
 {
  if (in_pure_callback)
  jl_error("cannot use include inside a generated function");
  jl_ast_context_t *ctx = jl_ast_ctx_enter();
  fl_context_t *fl_ctx = &ctx->fl;
  value_t f, ast;
+ size_t len = strlen(fname);
  f = cvalue_static_cstrn(fl_ctx, fname, len);
  fl_gc_handle(fl_ctx, &f);
  if (content != NULL) {
@@ -723,9 +724,9 @@ jl_value_t *jl_parse_eval_all(const char *fname, size_t len,
 }
 
 JL_DLLEXPORT jl_value_t *jl_load_file_string(const char *text, size_t len,
- char *filename, size_t namelen)
+ char *filename)
 {
- return jl_parse_eval_all(filename, namelen, text, len);
+ return jl_parse_eval_all(filename, text, len);
 }
 
 JL_DLLEXPORT int jl_parse_depwarn(int warn)

src/init.c

@@ -653,7 +653,7 @@ void _julia_init(JL_IMAGE_SEARCH rel)
  jl_all_tls_states[t]->root_task->current_module = jl_current_module;
  }
 
- jl_load("boot.jl", sizeof("boot.jl")-1);
+ jl_load("boot.jl");
  jl_get_builtin_hooks();
  jl_boot_file_loaded = 1;
  jl_init_box_caches();

src/julia.h

@@ -1315,7 +1315,7 @@ JL_DLLEXPORT jl_value_t *jl_parse_string(const char *str, size_t len,
  int pos0, int greedy);
 JL_DLLEXPORT int jl_parse_depwarn(int warn);
 JL_DLLEXPORT jl_value_t *jl_load_file_string(const char *text, size_t len,
- char *filename, size_t namelen);
+ char *filename);
 JL_DLLEXPORT jl_value_t *jl_expand(jl_value_t *expr);
 JL_DLLEXPORT jl_value_t *jl_eval_string(const char *str);
 
@@ -1350,7 +1350,7 @@ JL_DLLEXPORT const char *jl_lookup_soname(const char *pfx, size_t n);
 // compiler
 JL_DLLEXPORT jl_value_t *jl_toplevel_eval(jl_value_t *v);
 JL_DLLEXPORT jl_value_t *jl_toplevel_eval_in(jl_module_t *m, jl_value_t *ex);
-JL_DLLEXPORT jl_value_t *jl_load(const char *fname, size_t len);
+JL_DLLEXPORT jl_value_t *jl_load(const char *fname);
 JL_DLLEXPORT jl_value_t *jl_interpret_toplevel_expr_in(jl_module_t *m, jl_value_t *e,
  jl_lambda_info_t *lam);
 JL_DLLEXPORT jl_module_t *jl_base_relative_to(jl_module_t *m);

src/julia_internal.h

@@ -196,7 +196,7 @@ jl_value_t *jl_toplevel_eval_in_warn(jl_module_t *m, jl_value_t *ex,
 
 jl_lambda_info_t *jl_wrap_expr(jl_value_t *expr);
 jl_value_t *jl_eval_global_var(jl_module_t *m, jl_sym_t *e);
-jl_value_t *jl_parse_eval_all(const char *fname, size_t len,
+jl_value_t *jl_parse_eval_all(const char *fname,
  const char *content, size_t contentlen);
 jl_value_t *jl_interpret_toplevel_thunk(jl_lambda_info_t *lam);
 jl_value_t *jl_interpret_toplevel_expr(jl_value_t *e);
@@ -662,6 +662,8 @@ STATIC_INLINE void *jl_get_frame_addr(void)
 #endif
 }
 
+JL_DLLEXPORT jl_array_t *jl_array_cconvert_cstring(jl_array_t *a);
+
 #ifdef __cplusplus
 }
 #endif

src/toplevel.c

@@ -551,28 +551,30 @@ JL_DLLEXPORT jl_value_t *jl_toplevel_eval(jl_value_t *v)
  return jl_toplevel_eval_flex(v, 1, 0);
 }
 
-JL_DLLEXPORT jl_value_t *jl_load(const char *fname, size_t len)
+JL_DLLEXPORT jl_value_t *jl_load(const char *fname)
 {
- if (NULL != memchr(fname, 0, len))
- jl_exceptionf(jl_argumenterror_type, "file name may not contain \\0");
  if (jl_current_module->istopmod) {
  jl_printf(JL_STDOUT, "%s\r\n", fname);
 #ifdef _OS_WINDOWS_
  uv_run(uv_default_loop(), (uv_run_mode)1);
 #endif
  }
- char *fpath = (char*)fname;
  uv_stat_t stbuf;
- if (jl_stat(fpath, (char*)&stbuf) != 0 || (stbuf.st_mode & S_IFMT) != S_IFREG) {
- jl_errorf("could not open file %s", fpath);
+ if (jl_stat(fname, (char*)&stbuf) != 0 || (stbuf.st_mode & S_IFMT) != S_IFREG) {
+ jl_errorf("could not open file %s", fname);
  }
- return jl_parse_eval_all(fpath, len, NULL, 0);
+ return jl_parse_eval_all(fname, NULL, 0);
 }
 
 // load from filename given as a String object
 JL_DLLEXPORT jl_value_t *jl_load_(jl_value_t *str)
 {
- return jl_load(jl_string_data(str), jl_string_len(str));
+ jl_array_t *ary =
+ jl_array_cconvert_cstring((jl_array_t*)(jl_data_ptr(str)[0]));
+ JL_GC_PUSH1(&ary);
+ jl_value_t *res = jl_load((const char*)ary->data);
+ JL_GC_POP();
+ return res;
 }
 
 // method definition ----------------------------------------------------------

test/loading.jl

@@ -4,10 +4,8 @@ using Base.Test
 
 @test @__LINE__ == 5
 
-@test_throws ArgumentError Core.include("test_sourcepath.jl\0")
-
 include("test_sourcepath.jl")
-thefname = "the fname!//\\&\0\1*"
+thefname = "the fname!//\\&\1*"
 @test include_string("include_string_test() = @__FILE__", thefname)() == Base.source_path()
 @test include_string("Base.source_path()", thefname) == Base.source_path()
 @test basename(@__FILE__) == "loading.jl"

ui/repl.c

@@ -487,7 +487,7 @@ static int exec_program(char *program)
  JL_EH_POP();
  return 1;
  }
- jl_load(program, strlen(program));
+ jl_load(program);
  }
  JL_CATCH {
  err = 1;