From 82f2d9740a8d47af53c389d8f6edb10a6769801e Mon Sep 17 00:00:00 2001
From: Ben Kurtovic <ben.kurtovic@gmail.com>
Date: Mon, 27 Apr 2015 14:43:14 -0500
Subject: [PATCH] Implement parse_bytes(), bugfixes, improve error messages.

---
 src/assembler/errors.c     |  2 +-
 src/assembler/parse_util.c | 36 ++++++++++++++++++++++++++++++++++--
 src/assembler/tokenizer.c  | 32 +++++++++++++++++++-------------
 3 files changed, 54 insertions(+), 16 deletions(-)

diff --git a/src/assembler/errors.c b/src/assembler/errors.c
index 6d9da59..083c7d9 100644
--- a/src/assembler/errors.c
+++ b/src/assembler/errors.c
@@ -35,7 +35,7 @@ static const char *error_descs[] = {
     [ED_LYT_BLOCK0]       = "block zero cannot be mapped into a nonzero slot",
     [ED_LYT_SLOTS]        = "multiple slot declarations for block directive",
     [ED_LYT_BLOCK_CROSS]  = "instruction or data extends past block boundary",
-    [ED_LYT_OVERLAP]      = "location overlaps with instruction or data",
+    [ED_LYT_OVERLAP]      = "multiple instructions/data occupy same location",
     [ED_LYT_OVERLAP_HEAD] = "location overlaps with ROM header",
 
     [ED_SYM_DUPE_LABELS]  = "duplicate definitions for label",
diff --git a/src/assembler/parse_util.c b/src/assembler/parse_util.c
index 9e61411..38dbc79 100644
--- a/src/assembler/parse_util.c
+++ b/src/assembler/parse_util.c
@@ -7,6 +7,7 @@
 
 #include "parse_util.h"
 #include "directives.h"
+#include "../logging.h"
 #include "../util.h"
 
 #define MAX_REGION_SIZE 32
@@ -125,6 +126,8 @@ bool parse_string(char **result, size_t *length, const char *arg, ssize_t size)
 
     *length = size - 2;
     *result = malloc(sizeof(char) * (*length));
+    if (!*result)
+        OUT_OF_MEMORY()
     memcpy(*result, arg + 1, *length);
     return true;
 }
@@ -137,8 +140,37 @@ bool parse_string(char **result, size_t *length, const char *arg, ssize_t size)
 */
 bool parse_bytes(uint8_t **result, size_t *length, const char *arg, ssize_t size)
 {
-    // TODO
-    return false;
+    if (size <= 0)
+        return false;
+
+    const char *end = arg + size;
+    uint8_t *bytes = NULL;
+    size_t nbytes = 0;
+
+    while (arg != end) {
+        const char *start = arg;
+        while (arg != end && *arg != ' ')
+            arg++;
+
+        uint32_t temp;
+        if (!parse_uint32_t(&temp, start, arg - start) || temp > UINT8_MAX) {
+            free(bytes);
+            return false;
+        }
+
+        nbytes++;
+        bytes = realloc(bytes, sizeof(uint8_t) * nbytes);
+        if (!bytes)
+            OUT_OF_MEMORY()
+        bytes[nbytes - 1] = temp;
+
+        if (arg++ == end)
+            break;
+    }
+
+    *result = bytes;
+    *length = nbytes;
+    return true;
 }
 
 /*
diff --git a/src/assembler/tokenizer.c b/src/assembler/tokenizer.c
index a3bf20e..71a7a49 100644
--- a/src/assembler/tokenizer.c
+++ b/src/assembler/tokenizer.c
@@ -16,6 +16,7 @@
 typedef struct {
     size_t size;
     const ASMLine **overlap_table;
+    const ASMLine **overlap_origins;
     const ASMLine *origin;
     uint8_t bank;
     bool cross_blocks;
@@ -125,15 +126,12 @@ static ErrorInfo* handle_block_directive(
 
     if (bank >= MMU_NUM_ROM_BANKS || slot >= MMU_NUM_SLOTS)
         return error_info_create(line, ET_PREPROC, ED_PP_ARG_RANGE);
-
-    if (nargs == 2) {
-        if (bank == 0 && slot != 0)
-            return error_info_create(line, ET_LAYOUT, ED_LYT_BLOCK0);
-        if (si->slots[bank] >= 0 && si->slots[bank] != slot) {
-            ErrorInfo *ei = error_info_create(line, ET_LAYOUT, ED_LYT_SLOTS);
-            error_info_append(ei, si->lines[bank]);
-            return ei;
-        }
+    if (bank == 0 && slot != 0)
+        return error_info_create(line, ET_LAYOUT, ED_LYT_BLOCK0);
+    if (si->slots[bank] >= 0 && si->slots[bank] != slot) {
+        ErrorInfo *ei = error_info_create(line, ET_LAYOUT, ED_LYT_SLOTS);
+        error_info_append(ei, si->lines[bank]);
+        return ei;
     }
 
     *offset = bank * MMU_ROM_BANK_SIZE;
@@ -212,13 +210,14 @@ static ErrorInfo* parse_instruction(
 static ErrorInfo* check_layout(
     ASMLayoutInfo *li, const ASMLocation *loc, const ASMLine *line)
 {
-    const ASMLine *clash = NULL;
+    const ASMLine *clash = NULL, *clash_origin;
     if (loc->offset + loc->length > li->size) {
         clash = &bounds_sentinel;
     } else {
         for (size_t i = 0; i < loc->length; i++) {
             if (li->overlap_table[loc->offset + i]) {
                 clash = li->overlap_table[loc->offset + i];
+                clash_origin = li->overlap_origins[loc->offset + i];
                 break;
             }
         }
@@ -231,8 +230,11 @@ static ErrorInfo* check_layout(
 
         if (li->origin)
             error_info_append(ei, li->origin);
-        if (clash != &header_sentinel && clash != &bounds_sentinel)
+        if (clash != &header_sentinel && clash != &bounds_sentinel) {
             error_info_append(ei, clash);
+            if (clash_origin)
+                error_info_append(ei, clash_origin);
+        }
         return ei;
     }
 
@@ -244,8 +246,10 @@ static ErrorInfo* check_layout(
         return ei;
     }
 
-    for (size_t i = 0; i < loc->length; i++)
+    for (size_t i = 0; i < loc->length; i++) {
         li->overlap_table[loc->offset + i] = line;
+        li->overlap_origins[loc->offset + i] = li->origin;
+    }
     return NULL;
 }
 
@@ -263,7 +267,8 @@ ErrorInfo* tokenize(AssemblerState *state)
         .origin = NULL, .bank = 0, .cross_blocks = state->cross_blocks
     };
     li.overlap_table = calloc(li.size, sizeof(const ASMLine*));
-    if (!li.overlap_table)
+    li.overlap_origins = calloc(li.size, sizeof(const ASMLine*));
+    if (!li.overlap_table || !li.overlap_origins)
         OUT_OF_MEMORY()
 
     ErrorInfo *ei = NULL;
@@ -332,5 +337,6 @@ ErrorInfo* tokenize(AssemblerState *state)
     state->instructions = dummy_inst.next;
     state->data = dummy_data.next;
     free(li.overlap_table);
+    free(li.overlap_origins);
     return ei;
 }