cozis/Noja
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

bubblesort example in bytecode

Browse Source
This commit is contained in:
cozis
2022-08-11 22:31:26 +02:00
parent 36ffc24b9a
commit 2f87e46beb
11 changed files with 356 additions and 430 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.gitignore
+1
View File
@@ -2,3 +2,4 @@ cache
vgcore.*
libnoja.a
noja
*.gcov
build_tests.sh
+10 -11
View File
@@ -1,13 +1,12 @@
FILES="tests/assembler_test.c \
src/assembler/assembler.c \
src/common/executable.c \
src/utils/error.c \
src/utils/source.c \
src/utils/bpalloc.c \
src/utils/promise.c \
src/utils/labellist.c \
src/utils/bucketlist.c"
gcc $FILES -o build/assembler_test_cov -Wall -Wextra -g --coverage
gcc $FILES -o build/assembler_test -Wall -Wextra -g
src/noja/assembler/assembler.c \
src/noja/common/executable.c \
src/noja/utils/error.c \
src/noja/utils/source.c \
src/noja/utils/bpalloc.c \
src/noja/utils/promise.c \
src/noja/utils/labellist.c \
src/noja/utils/bucketlist.c"
gcc $FILES -o assembler_test_cov -Wall -Wextra -g --coverage
gcc $FILES -o assembler_test -Wall -Wextra -g
examples/bubble_sort.njbc
+171
View File
@@ -0,0 +1,171 @@
PUSHFUN copy, 1;
ASS "copy";
POP 1;
JUMP copy_end;
copy:
ASS "L";
POP 1;
PUSHLST 0;
ASS "L2";
POP 1;
PUSHINT 0;
ASS "i";
POP 1;
PUSHVAR "i";
PUSHVAR "L";
PUSHVAR "count";
CALL 1, 1;
LSS;
JUMPIFNOTANDPOP 31;
PUSHVAR "L";
PUSHVAR "i";
SELECT;
PUSHVAR "L2";
PUSHVAR "i";
INSERT2;
POP 1;
PUSHVAR "i";
PUSHINT 1;
ADD;
ASS "i";
POP 1;
JUMP 12;
PUSHVAR "L2";
RETURN 1;
RETURN 0;
copy_end:
PUSHFUN bubble_sort, 2;
ASS "bubble_sort";
POP 1;
JUMP bubble_sort_end;
bubble_sort:
ASS "less";
POP 1;
ASS "L";
POP 1;
PUSHVAR "less";
PUSHNNE;
EQL;
JUMPIFNOTANDPOP default_less_cb_end;
PUSHFUN 50, 2;
ASS "less";
POP 1;
JUMP default_less_cb_end;
default_less_cb:
ASS "b";
POP 1;
ASS "a";
POP 1;
PUSHVAR "a";
PUSHVAR "b";
LSS;
RETURN 1;
RETURN 0;
default_less_cb_end:
PUSHVAR "L";
PUSHVAR "copy";
CALL 1, 1;
ASS "L";
POP 1;
outer_loop_start:
PUSHFLS;
ASS "swapped";
POP 1;
PUSHINT 0;
ASS "i";
POP 1;
inner_loop_start:
PUSHVAR "i";
PUSHVAR "L";
PUSHVAR "count";
CALL 1, 1;
PUSHINT 1;
SUB;
LSS;
PUSHVAR "swapped";
NOT;
AND;
JUMPIFNOTANDPOP outer_loop_end;
PUSHVAR "L";
PUSHVAR "i";
SELECT;
PUSHVAR "L";
PUSHVAR "i";
PUSHINT 1;
ADD;
SELECT;
PUSHVAR "less";
CALL 2, 1;
JUMPIFNOTANDPOP if_branch_end;
PUSHTRU;
ASS "swapped";
POP 1;
PUSHVAR "L";
PUSHVAR "i";
PUSHINT 1;
ADD;
SELECT;
ASS "tmp";
POP 1;
PUSHVAR "L";
PUSHVAR "i";
SELECT;
PUSHVAR "L";
PUSHVAR "i";
PUSHINT 1;
ADD;
INSERT2;
POP 1;
PUSHVAR "tmp";
PUSHVAR "L";
PUSHVAR "i";
INSERT2;
POP 1;
if_branch_end:
PUSHVAR "i";
PUSHINT 1;
ADD;
ASS "i";
POP 1;
JUMP inner_loop_start;
outer_loop_end:
PUSHVAR "swapped";
JUMPIFANDPOP outer_loop_start;
PUSHVAR "L";
RETURN 1;
RETURN 0;
bubble_sort_end:
PUSHLST 5;
PUSHINT 0;
PUSHINT 3;
INSERT;
PUSHINT 1;
PUSHINT 2;
INSERT;
PUSHINT 2;
PUSHINT 1;
INSERT;
PUSHINT 3;
PUSHINT 6;
INSERT;
PUSHINT 4;
PUSHINT 0;
PUSHINT 2;
SUB;
INSERT;
PUSHVAR "bubble_sort";
CALL 1, 1;
PUSHVAR "print";
CALL 1, 1;
POP 1;
RETURN 0;
examples/hello-world.njbc
+5
View File
@@ -0,0 +1,5 @@
PUSHSTR "Hello, world!\n";
PUSHVAR "print";
CALL 1, 1;
POP 1;
RETURN 0;
makefile
+2 -2
View File
@@ -57,8 +57,8 @@ LFLAGS_DEBUG =
CFLAGS_RELEASE = -DNDEBUG -O3
LFLAGS_RELEASE =
CFLAGS_COVERAGE_GCC =
LFLAGS_COVERAGE_GCC = --coverage
CFLAGS_COVERAGE_GCC = -fprofile-arcs -ftest-coverage
LFLAGS_COVERAGE_GCC = -lgcov
CFLAGS_COVERAGE_CLANG = -fprofile-instr-generate -fcoverage-mapping
LFLAGS_COVERAGE_CLANG =
# NOTE: To support BUILD_MODE=COVERAGE for a new compiler,
src/cli/main.c
+27 -1
View File
@@ -39,7 +39,9 @@ static const char usage[] =
" $ noja run file.noja\n"
" $ noja run inline \"print('some noja code');\"\n"
" $ noja dis file.noja\n"
" $ noja dis inline \"print('some noja code');\"\n";
" $ noja dis inline \"print('some noja code');\"\n"
" $ noja asm file.noja\n"
" $ noja asm inline \"PUSHINT 5; PUSHFLT 1.0; ADD; POP 1;\"\n";
int main(int argc, char **argv)
{
@@ -77,6 +79,30 @@ int main(int argc, char **argv)
return r ? 0 : -1;
}
if(!strcmp(argv[1], "asm"))
{
if(argc == 2)
{
fprintf(stderr, "Error: Missing source file.\n");
return -1;
}
_Bool r;
if(!strcmp(argv[2], "inline"))
{
if(argc == 3)
{
fprintf(stderr, "Error: Missing source string.\n");
return -1;
}
r = NOJA_runAssemblyString(argv[3]);
}
else
r = NOJA_runAssemblyFile(argv[2]);
return r ? 0 : -1;
}
if(!strcmp(argv[1], "dis"))
{
if(argc == 2)
src/noja/assembler/assembler.c
-389
View File
@@ -1,389 +0,0 @@
#include <ctype.h>
#include <limits.h>
#include <string.h>
#include <assert.h>
#include "../utils/error.h"
#include "../utils/source.h"
#include "../utils/labellist.h"
#include "../common/executable.h"
typedef struct {
const char *str;
size_t len;
size_t cur;
} Context;
static void skipIdentifier(Context *ctx)
{
while(ctx->cur < ctx->len && (isalpha(ctx->str[ctx->cur]) || isdigit(ctx->str[ctx->cur]) || ctx->str[ctx->cur] == '_'))
ctx->cur += 1;
}
static void skipSpaces(Context *ctx)
{
while(ctx->cur < ctx->len && isspace(ctx->str[ctx->cur]))
ctx->cur += 1;
}
typedef struct {
size_t offset;
size_t length;
} Slice;
static bool parseLabelAndOpcode(Context *ctx, bool *no_label,
Slice *label, Slice *opcode,
Error *error)
{
assert(ctx != NULL && no_label != NULL
&& label != NULL && opcode != NULL);
// NOTE: This function must start at
// the first byte of the label or
// opcode. All whitespace must be
// consumed by the caller.
// Now we expect either a label and an
// opcode, or just an opcode
//
// <label>: <opcode> ..operands..
// <opcode> ..operands..
//
// Labels and opcodes can bot contain
// digits, letters and underscores.
Slice label_or_opcode;
{
char c = ctx->str[ctx->cur];
if(!isalpha(c) && c != '_') {
// ERROR: Missing opcode
Error_Report(error, 0, "Missing opcode");
}
label_or_opcode.offset = ctx->cur;
skipIdentifier(ctx);
label_or_opcode.length = ctx->cur - label_or_opcode.offset;
assert(label_or_opcode.length > 0);
}
// Get the character after the label or opcode
// (ignoring whitespace), and if it's a ':',
// then it was a label.
skipSpaces(ctx);
if(ctx->cur < ctx->len && ctx->str[ctx->cur] == ':') {
// Skip the ':' and the whitespace after it.
ctx->cur += 1;
skipSpaces(ctx);
if(ctx->cur == ctx->len || (!isalpha(ctx->str[ctx->cur]) && ctx->str[ctx->cur] != '_')) {
Error_Report(error, 0, "Missing opcode after label");
return false;
}
// Now the opcode is expected.
opcode->offset = ctx->cur;
skipIdentifier(ctx);
opcode->length = ctx->cur - opcode->offset;
assert(opcode->length > 0);
*no_label = false;
*label = label_or_opcode;
skipSpaces(ctx);
} else {
*no_label = true;
*opcode = label_or_opcode;
}
return true;
}
static bool parseStringOperand(Context *ctx, Error *error, BPAlloc *alloc, Operand *op)
{
// Skip the first double quote.
assert(ctx->cur < ctx->len && ctx->str[ctx->cur] == '"');
ctx->cur += 1;
size_t literal_offset = ctx->cur;
// For now do a dumb copy into the buffer
// without considering special characters.
while(ctx->cur < ctx->len && ctx->str[ctx->cur] != '"')
ctx->cur += 1;
if(ctx->cur == ctx->len) {
Error_Report(error, 0, "End of source inside a string literal");
return false;
}
// Skip the ending double quotes.
assert(ctx->cur < ctx->len && ctx->str[ctx->cur] == '"');
ctx->cur += 1;
size_t literal_length = ctx->cur - literal_offset;
char *copy = BPAlloc_Malloc(alloc, literal_length+1);
if(copy == NULL) {
Error_Report(error, 1, "No memory");
return false;
}
memcpy(copy, ctx->str + literal_offset, literal_length);
copy[literal_length] = '\0';
op->type = OPTP_STRING;
op->as_string = copy;
return true;
}
static bool parseIntegerOperand(Context *ctx, Error *error, Operand *op)
{
// It's ensured by the caller that the cursor is
// pointing to a sequence of one or more digits
// NOT followed by a dot and a digit after that
// (so this is an integer for sure, not a float).
assert(ctx->cur < ctx->len && isdigit(ctx->str[ctx->cur]));
long long int buffer = 0;
do {
// Transform each digit into it's integer value.
int d = ctx->str[ctx->cur] - '0';
assert(d >= 0 && d <= 9);
// Will this overflow?
if(buffer > (LLONG_MAX - d) / 10) {
Error_Report(error, 0, "Integer literal is too big to be represented in %d bits", 8*sizeof(buffer));
return false;
}
buffer = buffer * 10 + d;
ctx->cur += 1;
} while(ctx->cur < ctx->len && isdigit(ctx->str[ctx->cur]));
// Not a float!
op->type = OPTP_INT;
op->as_int = buffer;
return true;
}
static bool parseFloatingOperand(Context *ctx, Error *error, Operand *op)
{
(void) error; // At the moment this function doesn't report
// any error. This may change when overflows
// and underflows are detected.
// It's ensured by the caller that the cursor is
// pointing to a sequence of one or more digits
// followed by a dot and a digit after that.
assert(ctx->cur < ctx->len && isdigit(ctx->str[ctx->cur]));
double buffer = 0;
do {
// Transform each digit into it's integer value.
int d = ctx->str[ctx->cur] - '0';
assert(d >= 0 && d <= 9);
// Should overflow be checked?
buffer = buffer * 10 + d;
ctx->cur += 1;
} while(ctx->str[ctx->cur] != '.');
assert(ctx->cur+1 < ctx->len && ctx->str[ctx->cur] == '.' && isdigit(ctx->str[ctx->cur+1]));
double q = 1;
do {
// Transform each digit into it's integer value.
int d = ctx->str[ctx->cur] - '0';
assert(d >= 0 && d <= 9);
buffer += q * d;
ctx->cur += 1;
} while(ctx->cur < ctx->len && isdigit(ctx->str[ctx->cur]));
op->type = OPTP_FLOAT;
op->as_float = buffer;
return true;
}
static bool parseOperands(Context *ctx, BPAlloc *alloc, Error *error,
LabelList *list, Operand *opv, int *opc, int opc_max)
{
// NOTE: The whitespace before the first operand must
// be consumed by the caller.
if(ctx->cur < ctx->len && ctx->str[ctx->cur] != ';')
while(1) {
Operand op;
char c = ctx->str[ctx->cur];
if(c == '"') {
if(!parseStringOperand(ctx, error, alloc, &op))
return false;
} else if(isdigit(c)) {
/* Integer or float operand */
size_t k = ctx->cur;
while(k < ctx->len && isdigit(ctx->str[k]))
k += 1;
bool ok;
if(k+1 >= ctx->len || ctx->str[k] != '.' || !isdigit(ctx->str[k+1]))
ok = parseIntegerOperand(ctx, error, &op);
else
ok = parseFloatingOperand(ctx, error, &op);
if(!ok)
return false;
} else if(isalpha(c) || c == '_') {
/* Label */
size_t offset = ctx->cur;
skipIdentifier(ctx);
size_t length = ctx->cur - offset;
Promise *promise = LabelList_GetLabel(list, ctx->str + offset, length);
if(promise == NULL) {
Error_Report(error, 1, "No memory");
return false;
}
op.type = OPTP_PROMISE;
op.as_promise = promise;
} else {
// ERROR: Unexpected character
Error_Report(error, 0, "Unexpected character '%c'", c);
return false;
}
if(*opc == opc_max) {
Error_Report(error, 0, "Too many operands");
return false;
}
opv[*opc] = op;
*opc += 1;
// Now prepare for the next operand
skipSpaces(ctx);
if(ctx->cur == ctx->len || ctx->str[ctx->cur] == ';')
break;
c = ctx->str[ctx->cur];
if(c != ',') {
Error_Report(error, 0, "Unexpected character '%c' (',' or ';' were expected)", c);
return false;
}
// Skip the comma
ctx->cur += 1;
// Skip the spaces before the next operand
skipSpaces(ctx);
}
assert(ctx->cur == ctx->len || ctx->str[ctx->cur] == ';');
return true;
}
Executable *assemble(Source *src, Error *error)
{
Executable *exe = NULL;
BPAlloc *alloc = BPAlloc_Init(-1);
if(alloc == NULL) {
Error_Report(error, 1, "No memory");
return NULL;
}
LabelList *list = LabelList_New(alloc);
if(list == NULL) {
Error_Report(error, 1, "No memory");
BPAlloc_Free(alloc);
return NULL;
}
ExeBuilder *builder = ExeBuilder_New(alloc);
if(builder == NULL) {
Error_Report(error, 1, "No memory");
LabelList_Free(list);
BPAlloc_Free(alloc);
return NULL;
}
Context ctx = {
.str = Source_GetBody(src),
.len = Source_GetSize(src),
.cur = 0,
};
while(1) {
skipSpaces(&ctx);
if(ctx.cur == ctx.len)
break;
bool no_label;
Slice label, opcode_name;
if(!parseLabelAndOpcode(&ctx, &no_label, &label, &opcode_name, error))
goto done;
// If a label was defined, add it to the list.
if(no_label == false) {
long long int value = ExeBuilder_InstrCount(builder);
if(!LabelList_SetLabel(list, ctx.str + label.offset, label.length, value)) {
Error_Report(error, 1, "Out of memory");
goto done;
}
}
// Check that the opcode is valid (at this
// point it's just an unchecked identifier)
Opcode opcode;
const char *name = ctx.str + opcode_name.offset;
if(!Executable_GetOpcodeBinaryFromName(name, opcode_name.length, &opcode)) {
Error_Report(error, 0, "Opcode %.*s doesn't exist", (int) opcode_name.length, name);
goto done;
}
/* Parse operands */
// (whitespace was already skipped)
Operand opv[8];
int opc = 0;
if(!parseOperands(&ctx, alloc, error, list, opv, &opc, sizeof(opv)/sizeof(opv[0])))
goto done;
// The operand list ended with a ';' or the
// end of the file. If the file didn't end,
// then the ';' must be consumed.
assert(ctx.cur == ctx.len || ctx.str[ctx.cur] == ';');
if(ctx.cur < ctx.len) ctx.cur += 1;
if(!ExeBuilder_Append(builder, error, opcode, opv, opc, opcode_name.offset, opcode_name.length))
goto done;
}
size_t unresolved_count = LabelList_GetUnresolvedCount(list);
if(unresolved_count > 0) {
Error_Report(error, 0, "%d unresolved labels", unresolved_count);
goto done;
}
exe = ExeBuilder_Finalize(builder, error);
if(exe != NULL)
(void) Executable_SetSource(exe, src);
done:
LabelList_Free(list);
BPAlloc_Free(alloc);
return exe;
}
src/noja/assembler/assembler.h
-7
View File
@@ -1,7 +0,0 @@
#ifndef ASSEMBLE_H
#define ASSEMBLE_H
#include "../utils/error.h"
#include "../utils/source.h"
#include "../common/executable.h"
Executable *assemble(Source *src, Error *error);
#endif /* ASSEMBLE_H */
src/noja/noja.c
+88 -12
View File
@@ -4,6 +4,7 @@
#include "utils/source.h"
#include "compiler/parse.h"
#include "compiler/compile.h"
#include "assembler/assemble.h"
#include "builtins/basic.h"
#include "noja.h"
@@ -79,13 +80,8 @@ static Executable *build(Source *src)
return exe;
}
static _Bool interpret(Source *src)
static _Bool interpret(Executable *exe)
{
Executable *exe = build(src);
if(exe == NULL)
return 0;
Runtime *runt = Runtime_New(-1, 1024*1024, NULL, NULL);
if(runt == NULL)
@@ -95,7 +91,6 @@ static _Bool interpret(Source *src)
Error_Report(&error, 1, "Couldn't initialize runtime");
print_error(NULL, &error);
Error_Free(&error);
Executable_Free(exe);
return 0;
}
@@ -115,7 +110,6 @@ static _Bool interpret(Source *src)
assert(error.base.occurred == 1);
print_error(NULL, (Error*) &error);
RuntimeError_Free(&error);
Executable_Free(exe);
Runtime_Free(runt);
return 0;
}
@@ -143,8 +137,6 @@ static _Bool interpret(Source *src)
}
Runtime_Free(runt);
Executable_Free(exe);
return retc > -1;
}
@@ -174,8 +166,14 @@ static _Bool interpret_file(const char *file)
return 0;
}
_Bool r = interpret(src);
Executable *exe = build(src);
if(exe == NULL)
return 0;
_Bool r = interpret(exe);
Executable_Free(exe);
Source_Free(src);
return r;
}
@@ -195,12 +193,80 @@ static _Bool interpret_code(const char *code)
return 0;
}
_Bool r = interpret(src);
Executable *exe = build(src);
if(exe == NULL)
return 0;
_Bool r = interpret(exe);
Executable_Free(exe);
Source_Free(src);
return r;
}
static _Bool interpret_asm_file(const char *file)
{
Error error;
Error_Init(&error);
Source *src = Source_FromFile(file, &error);
if(src == NULL)
{
assert(error.occurred == 1);
print_error(NULL, &error);
Error_Free(&error);
return 0;
}
Executable *exe = assemble(src, &error);
if(exe == NULL) {
print_error("Assemblation", &error);
Source_Free(src);
Error_Free(&error);
return 0;
}
_Bool r = interpret(exe);
Executable_Free(exe);
Source_Free(src);
Error_Free(&error);
return r;
}
static _Bool interpret_asm_code(const char *code)
{
Error error;
Error_Init(&error);
Source *src = Source_FromString(NULL, code, -1, &error);
if(src == NULL)
{
assert(error.occurred);
print_error(NULL, &error);
Error_Free(&error);
return 0;
}
Executable *exe = assemble(src, &error);
if(exe == NULL) {
print_error("Assemblation", &error);
Source_Free(src);
Error_Free(&error);
return 0;
}
_Bool r = interpret(exe);
Executable_Free(exe);
Source_Free(src);
Error_Free(&error);
return r;
}
static _Bool disassemble_file(const char *file)
{
Error error;
@@ -262,3 +328,13 @@ _Bool NOJA_dumpStringBytecode(const char *str)
{
return disassemble_code(str);
}
_Bool NOJA_runAssemblyFile(const char *file)
{
return interpret_asm_file(file);
}
_Bool NOJA_runAssemblyString(const char *str)
{
return interpret_asm_code(str);
}
src/noja/noja.h
+3 -1
View File
@@ -1,7 +1,9 @@
#ifndef NOJA_H
#define NOJA_H
_Bool NOJA_runString(const char *str);
_Bool NOJA_runFile (const char *file);
_Bool NOJA_runString(const char *str);
_Bool NOJA_runAssemblyFile(const char *file);
_Bool NOJA_runAssemblyString(const char *str);
_Bool NOJA_dumpFileBytecode(const char *file);
_Bool NOJA_dumpStringBytecode(const char *str);
#endif /* NOJA_H */
tests/assembler_test.c
+44 -2
View File
@@ -1,7 +1,7 @@
#include <stdlib.h>
#include <stdbool.h>
#include "../src/utils/bpalloc.h"
#include "../src/assembler/assembler.h"
#include "../src/noja/utils/bpalloc.h"
#include "../src/noja/assembler/assembler.h"
static void stopTesting_(const char *file, int line)
{
@@ -243,6 +243,45 @@ int main()
},
},
},
// Instruction with the wrong number of arguments.
{
.line = __LINE__,
.text = "POS 1;",
.fail = true,
},
{
.line = __LINE__,
.text = "PUSHINT 1;"
"PUSHINT 2;"
"ADD;",
.fail = false,
.expected = (StaticExecutable) {
.instr_count = 3,
.instr_list = (StaticInstruction[]) {
{
.opcode = OPCODE_PUSHINT,
.operands = (Operand[]) {
{ OPTP_INT, .as_int = 1, }
},
.operand_count = 1,
},
{
.opcode = OPCODE_PUSHINT,
.operands = (Operand[]) {
{ OPTP_INT, .as_int = 2, }
},
.operand_count = 1,
},
{
.opcode = OPCODE_ADD,
.operands = NULL,
.operand_count = 0,
},
},
},
},
};
Error error;
@@ -268,6 +307,9 @@ int main()
} else {
Executable *exp_exe = buildExpectedExecutable(alloc, &tests[i].expected);
if(got_exe == NULL)
fprintf(stderr, "Build error :: %s.\n", error.message);
testCase(tests[i].line, got_exe != NULL, "Failed to assemble a valid source");
/*
fprintf(stderr, "== Generated ==\n");