From bdeafcbcc6bf2996c9834dc3519b370444d7a453 Mon Sep 17 00:00:00 2001
From: sBubshait <s.bobshit@gmail.com>
Date: Sat, 15 Jun 2024 03:03:02 +0100
Subject: [PATCH] Update the assembler file structure into subfolders

---
 src/assemble.c                    |  8 ++---
 src/{ => assembler}/encode.c      | 50 +++----------------------------
 src/assembler/encode.h            | 21 +++++++++++++
 src/{ => assembler}/parser.c      |  5 ++--
 src/{ => assembler}/parser.h      |  2 +-
 src/{ => assembler}/string_util.c |  2 +-
 src/{ => assembler}/string_util.h |  0
 src/{ => assembler}/symboltable.c |  2 +-
 src/{ => assembler}/symboltable.h |  5 ++++
 src/{ => assembler}/tokeniser.c   |  0
 src/{ => assembler}/tokeniser.h   |  0
 src/util/binary_util.c            | 43 ++++++++++++++++++++++++++
 src/util/binary_util.h            | 17 +++++++++++
 src/{ => util}/fileio.c           |  2 +-
 src/{ => util}/fileio.h           |  2 +-
 15 files changed, 102 insertions(+), 57 deletions(-)
 rename src/{ => assembler}/encode.c (80%)
 create mode 100644 src/assembler/encode.h
 rename src/{ => assembler}/parser.c (99%)
 rename src/{ => assembler}/parser.h (92%)
 rename src/{ => assembler}/string_util.c (99%)
 rename src/{ => assembler}/string_util.h (100%)
 rename src/{ => assembler}/symboltable.c (99%)
 rename src/{ => assembler}/symboltable.h (97%)
 rename src/{ => assembler}/tokeniser.c (100%)
 rename src/{ => assembler}/tokeniser.h (100%)
 create mode 100644 src/util/binary_util.c
 create mode 100644 src/util/binary_util.h
 rename src/{ => util}/fileio.c (99%)
 rename src/{ => util}/fileio.h (92%)

diff --git a/src/assemble.c b/src/assemble.c
index 9b2484c..e159720 100644
--- a/src/assemble.c
+++ b/src/assemble.c
@@ -7,10 +7,10 @@
 #include <stdlib.h>
 #include <stdio.h>
 #include "a64instruction/a64instruction.h"
-#include "parser.h"
-#include "fileio.h"
-#include "parser.h"
-#include "encode.c"
+#include "assembler/parser.h"
+#include "util/fileio.h"
+#include "assembler/encode.h"
+#include "assembler/symboltable.h"
 
 static symbol_table *firstPass(a64inst_instruction *instructions, int lineCount);
 
diff --git a/src/encode.c b/src/assembler/encode.c
similarity index 80%
rename from src/encode.c
rename to src/assembler/encode.c
index bdb89f4..ef7c498 100644
--- a/src/encode.c
+++ b/src/assembler/encode.c
@@ -7,55 +7,13 @@
  *  @author Saleh Bubshait
  */
 
-#include <assert.h>
-#include "global.h"
-#include "a64instruction/a64instruction.h"
-#include "symboltable.c"
+#include "symboltable.h"
 #include <stdlib.h>
+#include "../util/binary_util.h"
+#include "encode.h"
 
 #define HALT_BINARY 2315255808
 
-// write the provided value to the bits in the range [lsb, msb) {inclusive, exclusive} to the word.
-// Does not modify any other bits in the word.
-void setBits(word* wrd, uint8_t lsb, uint8_t msb, word value) {
-    // Ensure LSB and MSB are within range of word size, and in the correct order
-    assert(lsb < msb && msb <= 32);
-
-    // Create a mask with 1s in the range [lsb, msb) and 0s elsewhere
-    word mask = 0;
-    for (uint8_t i = lsb; i < msb; i++) {
-        mask |= 1 << i;
-    }
-
-    // Clear the bits in the range [lsb, msb) in the word
-    *wrd &= ~mask;
-
-    // Set the bits in the range [lsb, msb) to the value
-    *wrd |= (value << lsb) & mask;    
-}
-
-// Sign extend a given value to a 64-bit signed integer given the number of bits
-int64_t signExtend(dword value, unsigned int n) {
-    if (n == 0 || n >= 64) {
-        // If n_bits is 0 or greater than or equal to 64, return the value as is
-        return (int64_t)value;
-    }
-    
-    uint64_t sign_bit_mask = (uint64_t)1 << (n - 1);
-    
-    // Mask to isolate the n-bit value
-    uint64_t n_bit_mask = (sign_bit_mask << 1) - 1;
-
-    // Check if the sign bit is set
-    if (value & sign_bit_mask) {
-        // Sign bit is set, extend the sign
-        return (int64_t)(value | ~n_bit_mask);
-    } else {
-        // Sign bit is not set, return the value as is
-        return (int64_t)(value & n_bit_mask);
-    }
-}
-
 static int getLabelOffset(symbol_table* table, char* label, int currentIndex, int n_bits) {
     address target = st_get(table, label);
     return signExtend((unsigned int) (target - currentIndex), n_bits);
@@ -198,7 +156,7 @@ static word encodeLoadLiteral(a64inst_instruction cI, int arrIndex, symbol_table
     return wrd;
 }
 
-static word *encode(a64inst_instruction insts[], int instCount, symbol_table* st) {
+word *encode(a64inst_instruction insts[], int instCount, symbol_table* st) {
     word *arr = (word*)malloc(sizeof(word) * instCount);
     int index = 0;
     for (int i = 0; i < instCount; i++) {
diff --git a/src/assembler/encode.h b/src/assembler/encode.h
new file mode 100644
index 0000000..1ac8a82
--- /dev/null
+++ b/src/assembler/encode.h
@@ -0,0 +1,21 @@
+/** @file encode.h
+ *  @brief A function to encode the internal representation of ARMv8
+ *  instructions, a64inst_instruction, into binary.
+ *
+ *  @author Saleh Bubshait
+ */
+
+#include "../global.h"
+#include "../a64instruction/a64instruction.h"
+#include "symboltable.h"
+
+/** @brief Encodes the internal representation of ARMv8 instructions into binary.
+ *  The symbol table is used to resolve labels in branch instructions. Assumes
+ *  that the instructions are in the same order as they appear in the source file.
+ *
+ *  @param insts An array of a64inst_instruction to encode.
+ *  @param instCount The number of instructions in the array.
+ *  @param st The symbol table to use for label resolution.
+ *  @return An array of words representing the binary encoding of the instructions.
+ */
+word *encode(a64inst_instruction insts[], int instCount, symbol_table* st);
diff --git a/src/parser.c b/src/assembler/parser.c
similarity index 99%
rename from src/parser.c
rename to src/assembler/parser.c
index 474652e..b997cf3 100644
--- a/src/parser.c
+++ b/src/assembler/parser.c
@@ -1,6 +1,7 @@
 /** @file parser.c
  *  @brief Functions to parse ARMv8 assembly lines into an array of a special
  *  internal representation of instructions, a64inst_instruction.
+ *
  *  @author Ethan Dias Alberto
  *  @author George Niedringhaus
  *  @author Saleh Bubshait
@@ -13,8 +14,8 @@
 #include <ctype.h>
 #include <stdbool.h>
 #include "parser.h"
-#include "a64instruction/a64instruction.h"
-#include "global.h"
+#include "../a64instruction/a64instruction.h"
+#include "../global.h"
 #include "tokeniser.h"
 #include "string_util.h"
 
diff --git a/src/parser.h b/src/assembler/parser.h
similarity index 92%
rename from src/parser.h
rename to src/assembler/parser.h
index 23b76c6..12d5f7b 100644
--- a/src/parser.h
+++ b/src/assembler/parser.h
@@ -6,7 +6,7 @@
  *  @author Saleh Bubshait
  */
 
-#include "a64instruction/a64instruction.h"
+#include "../a64instruction/a64instruction.h"
 
 /** @brief Parses a list of ARMv8 assembly lines into an array of a64inst_instruction.
  * 
diff --git a/src/string_util.c b/src/assembler/string_util.c
similarity index 99%
rename from src/string_util.c
rename to src/assembler/string_util.c
index 8b7aaa0..e519ef4 100644
--- a/src/string_util.c
+++ b/src/assembler/string_util.c
@@ -10,7 +10,7 @@
 #include <stdbool.h>
 #include <stdlib.h>
 #include "string_util.h"
-#include "global.h"
+#include "../global.h"
 
 /************************************
  * CONSTANTS
diff --git a/src/string_util.h b/src/assembler/string_util.h
similarity index 100%
rename from src/string_util.h
rename to src/assembler/string_util.h
diff --git a/src/symboltable.c b/src/assembler/symboltable.c
similarity index 99%
rename from src/symboltable.c
rename to src/assembler/symboltable.c
index 50db150..9ccd3d6 100644
--- a/src/symboltable.c
+++ b/src/assembler/symboltable.c
@@ -5,7 +5,7 @@
  *
  *  @author Saleh Bubshait
  */
- 
+
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
diff --git a/src/symboltable.h b/src/assembler/symboltable.h
similarity index 97%
rename from src/symboltable.h
rename to src/assembler/symboltable.h
index ca1912d..12c99b8 100644
--- a/src/symboltable.h
+++ b/src/assembler/symboltable.h
@@ -6,6 +6,9 @@
  *  @author Saleh Bubshait
  */
 
+#ifndef __SYMBOLTABLE__
+#define __SYMBOLTABLE__
+
 #include <stdbool.h>
 #include <stdint.h>
 #include <stdio.h>
@@ -68,3 +71,5 @@ address st_get(symbol_table *st, char *label);
  *  @param st A pointer to the target symbol table.
  */
 void st_free(symbol_table *st);
+
+#endif
diff --git a/src/tokeniser.c b/src/assembler/tokeniser.c
similarity index 100%
rename from src/tokeniser.c
rename to src/assembler/tokeniser.c
diff --git a/src/tokeniser.h b/src/assembler/tokeniser.h
similarity index 100%
rename from src/tokeniser.h
rename to src/assembler/tokeniser.h
diff --git a/src/util/binary_util.c b/src/util/binary_util.c
new file mode 100644
index 0000000..af85377
--- /dev/null
+++ b/src/util/binary_util.c
@@ -0,0 +1,43 @@
+#include <assert.h>
+#include <stdint.h>
+#include <stdbool.h>
+#include "binary_util.h"
+
+void setBits(word* wrd, uint8_t lsb, uint8_t msb, word value) {
+    // Ensure LSB and MSB are within range of word size, and in the correct order
+    assert(lsb < msb && msb <= 32);
+
+    // Create a mask with 1s in the range [lsb, msb) and 0s elsewhere
+    word mask = 0;
+    for (uint8_t i = lsb; i < msb; i++) {
+        mask |= 1 << i;
+    }
+
+    // Clear the bits in the range [lsb, msb) in the word
+    *wrd &= ~mask;
+
+    // Set the bits in the range [lsb, msb) to the value
+    *wrd |= (value << lsb) & mask;    
+}
+
+// Sign extend a given value to a 64-bit signed integer given the number of bits
+int64_t signExtend(dword value, unsigned int n) {
+    if (n == 0 || n >= 64) {
+        // If n_bits is 0 or greater than or equal to 64, return the value as is
+        return (int64_t)value;
+    }
+    
+    uint64_t sign_bit_mask = (uint64_t)1 << (n - 1);
+    
+    // Mask to isolate the n-bit value
+    uint64_t n_bit_mask = (sign_bit_mask << 1) - 1;
+
+    // Check if the sign bit is set
+    if (value & sign_bit_mask) {
+        // Sign bit is set, extend the sign
+        return (int64_t)(value | ~n_bit_mask);
+    } else {
+        // Sign bit is not set, return the value as is
+        return (int64_t)(value & n_bit_mask);
+    }
+}
diff --git a/src/util/binary_util.h b/src/util/binary_util.h
new file mode 100644
index 0000000..348dda2
--- /dev/null
+++ b/src/util/binary_util.h
@@ -0,0 +1,17 @@
+/**
+ */
+
+#include "../global.h"
+
+/** @brief Sets a range of bits of a word (32-bit unsigned integer) to a value.
+ *  The range is inclusive of the lsb and exclusive of the msb. The value should
+ *  fit within the range.
+ *  
+ *  @param wrd A pointer to the word to set bits in.
+ *  @param lsb The least significant bit of the range to set, inclusive.
+ *  @param msb The most significant bit of the range to set, exclusive.
+ *  @param value The value to set the bits to.
+ */
+void setBits(word* wrd, uint8_t lsb, uint8_t msb, word value);
+
+int64_t signExtend(dword value, unsigned int n);
diff --git a/src/fileio.c b/src/util/fileio.c
similarity index 99%
rename from src/fileio.c
rename to src/util/fileio.c
index cd4fcc6..597cbcf 100644
--- a/src/fileio.c
+++ b/src/util/fileio.c
@@ -1,5 +1,5 @@
 #include <string.h>
-#include "global.h"
+#include "../global.h"
 #include "fileio.h"
 
 #define MAX_ASM_LINE_LENGTH 300
diff --git a/src/fileio.h b/src/util/fileio.h
similarity index 92%
rename from src/fileio.h
rename to src/util/fileio.h
index 3a509ca..b3c9cbc 100644
--- a/src/fileio.h
+++ b/src/util/fileio.h
@@ -2,7 +2,7 @@
 #define __FILEIO__
 #include <stdio.h>
 #include <stdlib.h>
-#include "global.h"
+#include "../global.h"
 
 #define EXIT_FAILURE 1