From ec1576e017cf7d550a52bf8dca45a5820eb80083 Mon Sep 17 00:00:00 2001
From: Giulio De Pasquale <giulio@depasquale.eu>
Date: Mon, 15 May 2017 12:49:11 +0200
Subject: [PATCH] Assembler con MOVI, DBG flags

---
 .gitignore          |   1 +
 cpp/emulator.cpp    |  23 +++++++--
 cpp/vm.cpp          |  96 ++++++++++++++++++++++++++++---------
 cpp/vm.h            |   5 ++
 cpp/vmas.cpp        |  23 ++++-----
 python/assembler.py | 113 ++++++++++++++++++++++++++++++++++++++++++++
 6 files changed, 220 insertions(+), 41 deletions(-)
 create mode 100644 python/assembler.py

diff --git a/.gitignore b/.gitignore
index a6ac9c9..b23b836 100644
--- a/.gitignore
+++ b/.gitignore
@@ -1,2 +1,3 @@
+*.gipu
 *.out
 .vscode/
diff --git a/cpp/emulator.cpp b/cpp/emulator.cpp
index 1d1ebcf..1446b76 100644
--- a/cpp/emulator.cpp
+++ b/cpp/emulator.cpp
@@ -5,13 +5,28 @@
 #include <stdio.h>
 #include <stdlib.h>
 
-using namespace std;
+int main(int argc, char *argv[]) {
+  std::ifstream f;
+  std::streamsize fsize;
+  uint8_t * bytecode;
+
+  if (argc < 2) {
+    printf("Usage: %s <program>\n", argv[0]);
+    return 1;
+  }
+
+  /*
+  reading bytecode
+  */
+  f.open(argv[1], std::ios::binary | std::ios::ate);
+  fsize = f.tellg();
+  f.seekg(0, std::ios::beg);
+
+  bytecode = new uint8_t[fsize];
+  f.read((char*)bytecode, fsize);
 
-int main() {
-  uint8_t bytecode[] = "\x00\x00\x00";
   VM vm(bytecode, sizeof(bytecode));
   vm.run();
-  printf("\n\n");
   vm.status();
   return 0;
 }
\ No newline at end of file
diff --git a/cpp/vm.cpp b/cpp/vm.cpp
index 1964901..0448a25 100644
--- a/cpp/vm.cpp
+++ b/cpp/vm.cpp
@@ -4,32 +4,56 @@
 #include <string.h>
 
 /*
-CONSTRUCTORS
+DBG UTILS
 */
-VM::VM() {
-  DBG_INFO(("Creating VM without code.\n"));
-  as.allocate();
-  init_regs();
-}
 
-VM::VM(uint8_t *code, uint32_t codesize) {
-  DBG_INFO(("Creating VM with code.\n"));
-  if (as.allocate()) {
-    as.insCode(code, codesize);
+uint8_t *VM::reg_name(uint8_t regvalue) {
+  uint8_t *buf = new uint8_t[2];
+#ifdef DBG
+  switch (regvalue) {
+  case R0:
+    memcpy(buf, "R0", 2);
+    break;
+  case R1:
+    memcpy(buf, "R1", 2);
+    break;
+  case R2:
+    memcpy(buf, "R2", 2);
+    break;
+  case R3:
+    memcpy(buf, "R3", 2);
+    break;
+  case S0:
+    memcpy(buf, "S0", 2);
+    break;
+  case S1:
+    memcpy(buf, "S1", 2);
+    break;
+  case S2:
+    memcpy(buf, "S2", 2);
+    break;
+  case S3:
+    memcpy(buf, "S3", 2);
+    break;
+  case IP:
+    memcpy(buf, "IP", 2);
+    break;
+  case BP:
+    memcpy(buf, "BP", 2);
+    break;
+  case SP:
+    memcpy(buf, "SP", 2);
+    break;
+  default:
+    memcpy(buf, "??", 2);
+    break;
   }
-  init_regs();
-}
-
-void VM::init_regs(void) {
-  uint8_t i;
-
-  for (i = R0; i <= SP; i++) {
-    this->regs[i] = 0;
-  }
-  return;
+#endif
+  return buf;
 }
 
 void VM::status(void) {
+#ifdef DBG
   uint8_t i;
   DBG_INFO(("VM Status:\n"));
   DBG_INFO(("~~~~~~~~~~\n"));
@@ -71,6 +95,32 @@ void VM::status(void) {
     }
   }
   DBG_INFO(("~~~~~~~~~~\n"));
+#endif
+  return;
+}
+/*
+CONSTRUCTORS
+*/
+VM::VM() {
+  DBG_INFO(("Creating VM without code.\n"));
+  as.allocate();
+  init_regs();
+}
+
+VM::VM(uint8_t *code, uint32_t codesize) {
+  DBG_INFO(("Creating VM with code.\n"));
+  if (as.allocate()) {
+    as.insCode(code, codesize);
+  }
+  init_regs();
+}
+
+void VM::init_regs(void) {
+  uint8_t i;
+
+  for (i = R0; i <= SP; i++) {
+    this->regs[i] = 0;
+  }
   return;
 }
 
@@ -86,7 +136,7 @@ void VM::exec_movi(void) {
   uint16_t imm;
   dst = as.code[regs[IP] + 1];
   imm = *((uint16_t *)&as.code[regs[IP] + 2]);
-  DBG_INFO(("MOVI 0x%x 0x%x\n", dst, imm));
+  DBG_INFO(("MOVI %s, 0x%x\n", reg_name(dst), imm));
   regs[dst] = imm;
   return;
 }
@@ -100,7 +150,7 @@ void VM::exec_movr(void) {
   uint8_t dst, src;
   dst = as.code[regs[IP] + 1] >> 4;
   src = as.code[regs[IP] + 1] & 0b00001111;
-  DBG_INFO(("MOVR 0x%x 0x%x\n", dst, src));
+  DBG_INFO(("MOVR %s, %s\n", reg_name(dst), reg_name(src)));
   regs[dst] = regs[src];
   return;
 }
@@ -113,7 +163,7 @@ void VM::exec_getm(void) {
   uint16_t src;
   dst = as.code[regs[IP] + 1];
   src = *((uint16_t *)&as.code[regs[IP] + 2]);
-  DBG_INFO(("GETM 0x%x 0x%x\n", dst, src));
+  DBG_INFO(("GETM %s, 0x%x\n", reg_name(dst), src));
   regs[dst] = *((uint16_t *)&as.data[src]);
   return;
 }
diff --git a/cpp/vm.h b/cpp/vm.h
index dd93b26..921a180 100644
--- a/cpp/vm.h
+++ b/cpp/vm.h
@@ -45,6 +45,11 @@ private:
     uint8_t cf : 1;
   };
   VMAddrSpace as;
+
+  /*
+  DBG UTILS
+  */
+  uint8_t *reg_name(uint8_t);
   /*
   IMPLEMENTATIONS
   */
diff --git a/cpp/vmas.cpp b/cpp/vmas.cpp
index 1dd6cdb..14a2502 100644
--- a/cpp/vmas.cpp
+++ b/cpp/vmas.cpp
@@ -26,32 +26,27 @@ VMAddrSpace::VMAddrSpace(uint32_t ss, uint32_t cs, uint32_t ds) {
 
 bool VMAddrSpace::allocate(void) {
   DBG_INFO(("Allocating sections...\n"));
-  if (code == NULL) {
-    DBG_INFO(("\tcode...\n"));
-    code = (uint8_t *)malloc(codesize);
-  }
-  if (data == NULL) {
-    DBG_INFO(("\tdata...\n"));
-    data = (uint8_t *)malloc(datasize);
-  }
-  if (stack == NULL) {
-    DBG_INFO(("\tstack...\n"));
-    stack = (uint8_t *)malloc(stacksize);
-  }
+
+  DBG_INFO(("\tcode...\n"));
+  code = new uint8_t[codesize];
+  DBG_INFO(("\tdata...\n"));
+  data = new uint8_t[datasize];
+  DBG_INFO(("\tstack...\n"));
+  stack = new uint8_t[stacksize];
+
   if (code == NULL) {
     DBG_ERROR(("Couldn't allocate code section.\n"));
     return false;
   }
-  data = (uint8_t *)malloc(datasize);
   if (data == NULL) {
     DBG_ERROR(("Couldn't allocate data section.\n"));
     return false;
   }
-  stack = (uint8_t *)malloc(stacksize);
   if (stack == NULL) {
     DBG_ERROR(("Couldn't allocate stack section.\n"));
     return false;
   }
+  
   memset(code, 0xff,
          stacksize); // auto halt in case the assembly is not correct
   memset(stack, 0x0, stacksize);
diff --git a/python/assembler.py b/python/assembler.py
new file mode 100644
index 0000000..36db40d
--- /dev/null
+++ b/python/assembler.py
@@ -0,0 +1,113 @@
+import sys
+import re
+import struct
+
+op_names = ["MOVI",
+            "MOVR",
+            "GETM",
+            "PUTM",
+            "ADDI",
+            "ADDR",
+            "SUBI",
+            "SUBR",
+            "XORI",
+            "XORR",
+            "NOTI",
+            "NOTR",
+            "MULI",
+            "MULR",
+            "DIVI",
+            "DIVR",
+            "PUSH",
+            "POOP",
+            "CALL",
+            "HALT",
+            "NOPE"]
+
+reg_names = ["R0", "R1", "R2", "R3", "S0", "S1", "S2", "S3", "IP", "BP", "SP"]
+section_names = ["DATA:", "CODE:", "STACK:"]
+section_flags = {s: i + 1 for i, s in enumerate(section_names)}
+ops = {s: i for i, s in enumerate(op_names)}
+regs = {s: i for i, s in enumerate(reg_names)}
+
+assembled = bytearray()
+
+
+def to_uint8(data):
+    alphanum = re.compile("^[0-9]+$")
+    if isinstance(data, int):
+        return struct.pack("<b", data)
+    elif data.startswith("0x"):
+        return struct.pack("<b", int(data, 16))
+    elif alphanum.match(data):  # only numbers
+        return struct.pack("<b", int(data))
+    else:
+        return None
+
+def to_uint16(data):
+    print(data)
+    alphanum = re.compile("^[0-9]+$")
+    if isinstance(data, int):
+        return struct.pack("<h", data)
+    elif data.startswith("0x"):
+        return struct.pack("<h", int(data, 16))
+    elif alphanum.match(data):  # only numbers
+        return struct.pack("<h", int(data))
+    else:
+        return None
+
+
+def is_reg(data):
+    if data not in reg_names:
+        return False
+    return True
+
+
+def assemble_code(line):
+    global assembled
+    sys.stdout.write("CODE:")
+    instruction = [x for x in re.split('\W', line) if x]
+    op_name = instruction[0]
+
+    if op_name not in op_names:
+        sys.stderr.write(
+            "ERROR WHILE ASSEMBLING UNKNOWN OPERATION: {}\n".format(op_name))
+        return False
+    if op_name == "MOVI":
+        op = ops[op_name]
+        dst = instruction[1]
+        if (is_reg(dst)):
+            dst = to_uint8(regs[dst])
+            src = to_uint16(instruction[2])
+            print("SRC: {} | DST: {}".format(src, dst))
+        else:
+            sys.stderr.write(
+                "ERROR WHILE ASSEMBLING UNKNOWN REGISTER: {}\n".format(dst))
+            return False
+        assembled += to_uint8(op) + dst + src
+    return True
+
+
+def assemble_data(line):
+    sys.stdout.write("DATA:\t")
+    sys.stdout.write(line.strip(",") + "\n")
+
+
+def main():
+    global assembled
+    with open("./test.gipu", 'r') as f:
+        gen = (line.strip("\n") for line in f if line != "\n")
+
+        for line in gen:
+            if line.startswith(tuple(section_names)):
+                flag = section_flags[line]
+                continue
+            if flag == section_flags["DATA:"]:
+                assemble_data(line)
+            elif flag == section_flags["CODE:"]:
+                assemble_code(line)
+    with open("./out.gipu", 'wb') as f:
+        f.write(assembled)
+
+if __name__ == '__main__':
+    main()