From 295f6eb70fae100f2b74a98acbea64a6f3d1209f Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@gmail.com>
Date: Tue, 28 Apr 2015 20:03:38 -0500
Subject: [PATCH] Implement space directive; allow commas in .byte.

---
 src/assembler/directives.h |  8 +++++---
 src/assembler/parse_util.c |  8 +++++---
 src/assembler/tokenizer.c  | 31 +++++++++++++++++++++++++++++++
 3 files changed, 41 insertions(+), 6 deletions(-)

diff --git a/src/assembler/directives.h b/src/assembler/directives.h
index 9ade9bb..9e51bd7 100644
--- a/src/assembler/directives.h
+++ b/src/assembler/directives.h
@@ -6,7 +6,7 @@
 #include <string.h>
 
 #define DIRECTIVE_MARKER '.'
-#define NUM_DIRECTIVES   15
+#define NUM_DIRECTIVES   16
 
 #define DIR_INCLUDE      ".include"
 
@@ -22,6 +22,7 @@
 #define DIR_ORIGIN       ".org"
 #define DIR_BLOCK        ".block"
 #define DIR_BYTE         ".byte"
+#define DIR_SPACE        ".space"
 #define DIR_ASCII        ".ascii"
 #define DIR_ASCIZ        ".asciz"
 #define DIR_ASCIIZ       ".asciiz"
@@ -35,8 +36,9 @@
 
 #define IS_LOCAL_DIRECTIVE(line)                                              \
     (IS_DIRECTIVE(line, DIR_ORIGIN) || IS_DIRECTIVE(line, DIR_BLOCK) ||       \
-     IS_DIRECTIVE(line, DIR_BYTE)   || IS_DIRECTIVE(line, DIR_ASCII) ||       \
-     IS_DIRECTIVE(line, DIR_ASCIZ)  || IS_DIRECTIVE(line, DIR_ASCIIZ))
+     IS_DIRECTIVE(line, DIR_BYTE)   || IS_DIRECTIVE(line, DIR_SPACE) ||       \
+     IS_DIRECTIVE(line, DIR_ASCII)  || IS_DIRECTIVE(line, DIR_ASCIZ) ||       \
+     IS_DIRECTIVE(line, DIR_ASCIIZ))
 
 #define DIRECTIVE_OFFSET(line, d)                                             \
     (DIRECTIVE_HAS_ARG(line, d) ? strlen(d) : 0)
diff --git a/src/assembler/parse_util.c b/src/assembler/parse_util.c
index 38dbc79..33b0419 100644
--- a/src/assembler/parse_util.c
+++ b/src/assembler/parse_util.c
@@ -147,9 +147,9 @@ bool parse_bytes(uint8_t **result, size_t *length, const char *arg, ssize_t size
     uint8_t *bytes = NULL;
     size_t nbytes = 0;
 
-    while (arg != end) {
+    while (arg < end) {
         const char *start = arg;
-        while (arg != end && *arg != ' ')
+        while (arg != end && *arg != ' ' && *arg != ',')
             arg++;
 
         uint32_t temp;
@@ -164,7 +164,9 @@ bool parse_bytes(uint8_t **result, size_t *length, const char *arg, ssize_t size
             OUT_OF_MEMORY()
         bytes[nbytes - 1] = temp;
 
-        if (arg++ == end)
+        if (arg < end - 1 && *arg == ',' && *(arg + 1) == ' ')
+            arg += 2;
+        else if (arg++ >= end)
             break;
     }
 
diff --git a/src/assembler/tokenizer.c b/src/assembler/tokenizer.c
index e21995b..1da8552 100644
--- a/src/assembler/tokenizer.c
+++ b/src/assembler/tokenizer.c
@@ -146,6 +146,31 @@ static ErrorInfo* handle_block_directive(
 }
 
 /*
+    Parse a .space directive, which fills a region with a single byte.
+*/
+static bool parse_space(
+    uint8_t **result, size_t *length, const char *arg, ssize_t size)
+{
+    uint8_t *bytes;
+    size_t nbytes;
+    if (!parse_bytes(&bytes, &nbytes, arg, size))
+        return false;
+
+    if (nbytes < 1 || nbytes > 2) {
+        free(bytes);
+        return false;
+    }
+
+    *length = bytes[0];
+    if (!(*result = malloc(sizeof(uint8_t) * (*length))))
+        OUT_OF_MEMORY()
+
+    memset(*result, nbytes == 2 ? bytes[1] : 0, *length);
+    free(bytes);
+    return true;
+}
+
+/*
     Parse data encoded in a line into an ASMData object.
 
     On success, return NULL and store the instruction in *data_ptr. On failure,
@@ -160,6 +185,9 @@ static ErrorInfo* parse_data(
     if (IS_DIRECTIVE(line, DIR_BYTE)) {
         directive = DIR_BYTE;
         parser = parse_bytes;
+    } else if (IS_DIRECTIVE(line, DIR_SPACE)) {
+        directive = DIR_SPACE;
+        parser = parse_space;
     } else if (IS_DIRECTIVE(line, DIR_ASCII)) {
         directive = DIR_ASCII;
     } else if (IS_DIRECTIVE(line, DIR_ASCIZ)) {
@@ -205,6 +233,9 @@ static ErrorInfo* parse_instruction(
     // TODO
     DEBUG("parse_instruction(): %.*s", (int) line->length, line->data)
 
+    // SYNTAX NOTES:
+    // see http://clrhome.org/table/ and http://www.z80.info/z80undoc.htm
+
     // return error_info_create(line, ET_PARSER, ED_PARSE_SYNTAX);
 
     ASMInstruction *inst = malloc(sizeof(ASMInstruction));