X-Git-Url: http://git.squeep.com/?a=blobdiff_plain;f=as-dcpu16.c;h=2d131d7a40a167c37ee5115543aba21f9f73e8f6;hb=46ffc28b939d84716b616fdd048ca0c90f374fdd;hp=4ed6050adf2ea078f219a709cae5ddfda8127c80;hpb=4706199a81dc631b6969927e1a6ad27591852b20;p=dcpu16

diff --git a/as-dcpu16.c b/as-dcpu16.c
index 4ed6050..2d131d7 100644
--- a/as-dcpu16.c
+++ b/as-dcpu16.c
@@ -19,6 +19,7 @@
  *
  *  TODO
  *    needs ability to specify location for code or data
+ *    needs ability to specify label as relative to another label
  *    short labels not correctly computed
  */
 
@@ -77,6 +78,7 @@ struct operand_ {
 
 /* keep an array of instructions as we read them in */
 struct instruction_ {
+    size_t src_line;
     char *label;  /* set if a label points here */
     char *opcode; /* tokenized instruction text */
     struct operand_ *operands;  /* list of operands */
@@ -127,6 +129,8 @@ static
 void label_addr_calculate_(struct dynamic_array *instructionps, struct dynamic_array *labels) {
     size_t i;
 
+    /* idea:  label1:label2 - calculated as offset between labels */
+
     /* for each label.. */
     for (i = 0; i < labels->entries; i++) {
         struct label_ *l;
@@ -514,143 +518,160 @@ int instruction_print_(struct instruction_ *i, unsigned int with_label) {
     return r;
 }
 
-/*  buf_tokenize_
- *  Parses the zero-terminated line of input 'buf' into a newly-allocated struct instruction_.
- *  [label] opcode [operand[,operand[,...]]]
- *  Does not yet validate if labels, opcodes, or operands are valid...
+/*  tokenize_line_
+ *  Parses a zero-terminated line of input into a newly-allocated struct instruction_.
+ *  [label] instruction [operand[,operand[,...]]]
+ *  Does no validation of contents of any of these tokens, as of yet.
  */
 static
-int buf_tokenize_(char *buf, struct instruction_ **next_instr) {
-    const char const *sep = " \t\n";
-    const char const *quot = "'`\"";
+int tokenize_line_(char *line, struct instruction_ **next_instr) {
+    const char const *whitespace = " \t\n";
+    const char const *quotes = "\"'`";
     struct instruction_ *instr = NULL;
+    char *x, *st, *qt;
+    char *label, *opcode;
     struct operand_ *operand_list = NULL;
-    char *label = NULL,
-         *opcode = NULL,
-         *operand = NULL;
-    char *x,
-         *y,
-         *st, *qt;
-    size_t instr_words_needed = 1;
+    struct operand_ **operand_tail = &operand_list;
+    size_t instr_words_needed = 0;
 
-    assert(buf != NULL);
-    assert(next_instr != NULL);
+    assert(line);
+    assert(next_instr);
 
     *next_instr = NULL;
 
-    /* kill leading whitespace */
-    buf += strspn(buf, " \t\n");
-
-    /* kill trailing whitespace */
-    for (x = buf + strlen(buf); *x && strchr(sep, *x); x--)
-        *x = '\0';
+    /* strip leading whitespace */
+    line += strspn(line, whitespace);
+    if (*line == '\0')
+        return 0;
 
-    /* split on first non-quoted ';', ignore following comment */
-    x = strqtok_r(buf, ";", '\\', quot, &qt, &st);
-    if (x == NULL)
+    /* set first bare ';' to '\0', thus isolating any comments */
+    /* here we only care about the side-effect of truncating the first separator character */
+    (void)strqtok_r(line, ";", '\\', quotes, &qt, &st);
+    /* we don't care if there was an unmatched quote at this point, let's see what happens */
+    if (*line == '\0')
         return 0;
-    if (qt) {
-        fprintf(stderr, "unmatched %c-quote\n", *qt);
-        return -1;
-    }
 
-    /* determine if first token is label, opcode, or we just have a blank line to ignore */
-    x = strqtok_r(x, sep, '\\', quot, &qt, &st);
-    if (x == NULL)
+    /* carve off the first token, determine if it is a label */
+    x = strqtok_r(line, whitespace, '\\', quotes, &qt, &st);
+    if (x == NULL || *x == '\0')
         return 0;
     if (qt) {
-        fprintf(stderr, "unmatched %c-quote\n", *qt);
-        return -1;
+        /* labels could contain an unmatched quote character, I guess? */
+        qt = NULL;
     }
 
-/* I want c-style labels in my asm, but example in spec uses : in prefix rather than postfix */
+    /* we have something, try to make sense of what it is */
+
 #ifdef NON_SPEC_LABELS
-    /* labels end with :, otherwise its an opcode */
-    y = x + strlen(x) - 1;
-    if (*y == ':') {
-        *y = '\0';
+    /* I want my labels like 'label:' */
+    if ( *(x + strlen(line) - 1) == ':' ) {
+        *(x + strlen(line) - 1) = '\0';
+        DEBUG_PRINTF("label: %s\n", x);
+
         label = x;
-        opcode = strqtok_r(NULL, sep, '\\', quot, &qt, &st);
+
+        opcode = strqtok_r(NULL, whitespace, '\\', quotes, &qt, &st);
     } else {
         label = NULL;
         opcode = x;
     }
-#else /* NON_SPEC_LABELS */
-    /* labels.. begin? with ':' ? okay, I guess. Whatever. */
-    /* otherwise, it's an opcode */
+#endif /* NON_SPEC_LABELS */
+
+    /* spec gives example of labels as ':label' */
     if (*x == ':') {
-        label = x + 1;
-        opcode = strqtok_r(NULL, sep, '\\', quot, &qt, &st);
+        *x = '\0';
+        x++;
+        label = x;
+        opcode = strqtok_r(NULL, whitespace, '\\', quotes, &qt, &st);
     } else {
         label = NULL;
         opcode = x;
     }
-#endif /* NON_SPEC_LABELS */
+    /* opcodes shouldn't have quotes, so we'll ignore any unmatched quotes again */
 
-    if (opcode) {
-        operand = st;
-    }
+    if (opcode && *opcode) {
+        /* if we have an opcode, we'll need at least one word to compile instruction */
+        instr_words_needed++;
 
-    /*
-        While normal instructions just have comma-separated operands,
-        DAT can be followed by comma-separated list of:
-            label, to be resolved to address
-            value, like 0xffff
-            string, "quoted", characters to be rendered into low-byte of words
-    */
+        while ( (x = strqtok_r(NULL, ",", '\\', quotes, &qt, &st)) ) {
+            struct operand_ *new_operand;
+            char *y;
+
+            DEBUG_PRINTF("considering operand '%s'\n", x);
 
-    if (operand) {
-        struct operand_ **o_next = &operand_list;
+            /* trim whitespaces */
+            x += strspn(x, whitespace);
 
-        for (x = strqtok_r(operand, ",", '\\', quot, &qt, &st);
-             x;
-             x = strqtok_r(NULL, ",", '\\', quot, &qt, &st) ) {
+            DEBUG_PRINTF("considering ftrim operand '%s'\n", x);
 
-            /* trim leading whitespace */
-            x += strspn(x, " \t\n");
+            if (*x) {
+                for (y = x + strlen(x) - 1; *y; y--) {
+                    if (strchr(whitespace, *y)) {
+                        *y = '\0';
+                    }
+                }
+            }
+            /* nothing left? */
             if (*x == '\0') {
-                fprintf(stderr, "encountered empty operand\n");
-                return -1;
+                fprintf(stderr, "ignoring null operand in line %zu\n", instr->src_line);
+                continue;
             }
 
-            /* trim trailing whitespace */
-            y = x + strlen(x) - 1;
-            while (strchr(" \t\n", *y)) {
-                *y = '\0';
-                y--;
-            }
+            DEBUG_PRINTF("found operand '%s'\n", x);
 
-            /* new operand to append to list */
-            *o_next = malloc(sizeof **o_next);
-            if (*o_next == NULL) {
-                fprintf(stderr, "%s():%s\n", "calloc", strerror(errno));
+            new_operand = malloc(sizeof *new_operand);
+            if (new_operand == NULL) {
+                fprintf(stderr, "%s():%s\n", "malloc", strerror(errno));
                 return -1;
             }
 
-            /* assume an operand uses one word, unless it's a string */
-            instr_words_needed += (*x == '"') ? strlen(x) : 1;
-
-            (*o_next)->operand = strdup(x);
-            if ((*o_next)->operand == NULL) {
+            new_operand->operand = strdup(x);
+            if (new_operand->operand == NULL) {
                 fprintf(stderr, "%s():%s\n", "strdup", strerror(errno));
                 return -1;
             }
-            (*o_next)->next = NULL;
-            o_next = &((*o_next)->next);
+
+            new_operand->next = NULL;
+
+            if (strchr(quotes, x[0])) {
+                /* if this is a quoted operand, assuming we are in a DAT statement, it will take up slightly less room than it is long */
+                instr_words_needed += strlen(x) - 1;
+            }
+            instr_words_needed++;
+
+            *operand_tail = new_operand;
+            operand_tail = &(*operand_tail)->next;
         }
     }
 
-    DEBUG_PRINTF("allocating instr with room for %zu words\n", instr_words_needed);
+    DEBUG_PRINTF("allocating instruction with room for %zu bytes\n", instr_words_needed);
 
-    /* extra room for assembled words */
     instr = calloc(1, (instr_words_needed * sizeof *instr->instr_words) + sizeof *instr);
     if (instr == NULL) {
-        fprintf(stderr, "%s():%s\n", "calloc", strerror(errno));
+        fprintf(stderr, "%s():%s\n", "malloc", strerror(errno));
         return -1;
     }
 
-    instr->label = label ? strdup(label) : NULL;
-    instr->opcode = opcode ? strdup(opcode) : NULL;
+    if (label) {
+        instr->label = strdup(label);
+        if (instr->label == NULL) {
+            fprintf(stderr, "%s():%s\n", "malloc", strerror(errno));
+            return -1;
+        }
+    } else {
+        label = NULL;
+    }
+
+    if (opcode) {
+        instr->opcode = strdup(opcode);
+        if (instr->opcode == NULL) {
+            fprintf(stderr, "%s():%s\n", "malloc", strerror(errno));
+            return -1;
+        }
+    } else {
+        opcode = NULL;
+    }
+
     instr->operands = operand_list;
 
     *next_instr = instr;
@@ -659,6 +680,7 @@ int buf_tokenize_(char *buf, struct instruction_ **next_instr) {
 }
 
 /* try to generate bytecode for an instruction */
+/* returns -1 on unrecoverable error */
 static
 int instr_assemble_(struct dynamic_array *labels, struct instruction_ *i, unsigned int allow_short_labels) {
     unsigned int nwu = 0; /* number of words used */
@@ -669,7 +691,7 @@ int instr_assemble_(struct dynamic_array *labels, struct instruction_ *i, unsign
     if (opt_.verbose > 2) {
         printf("%s: assembling %p ", __func__, i);
         instruction_print_(i, 1);
-        printf("\n");
+        printf("(line :%zu)\n", i->src_line);
     }
 
     if (i->ready) {
@@ -683,7 +705,7 @@ int instr_assemble_(struct dynamic_array *labels, struct instruction_ *i, unsign
 
         i->length = 0;
 
-        while (o) {
+        for ( /* */ ; o; o = o->next) {
             size_t j, dat_len;
             char *x;
             unsigned long l;
@@ -700,14 +722,16 @@ int instr_assemble_(struct dynamic_array *labels, struct instruction_ *i, unsign
                     for (j = 0, x = o->operand + 1;
                          j < dat_len - 1;
                           j++, x++) {
-                        i->instr_words[i->length] = (DCPU16_WORD)*x;
+                        i->instr_words[i->length] = *x;
                         i->length++;
                     }
+                    /* Note that strings in DAT do not include their zero-terminators */
+                    /* specify as 'DAT "string", 0' */
                 }
-                o = o->next;
                 continue;
             }
 
+            /* is this a number? */
             char *ep;
             errno = 0;
             l = strtoul(o->operand, &ep, 0);
@@ -718,20 +742,27 @@ int instr_assemble_(struct dynamic_array *labels, struct instruction_ *i, unsign
                     fprintf(stderr, "value '%lu' out of range\n", l);
                     return -1;
                 }
+                i->instr_words[i->length] = l;
+                i->length++;
+                continue;
             }
 
-            fprintf(stderr, "FIXME finish implementing DAT\n");
-            /* check if it's a parsable number */
-
-            /* otherwise assume it's a label */
-
-
+            /* otherwise assume it's a label, even if we don't know what it is */
+            if (label_addr_(labels, o->operand, &i->instr_words[i->length])) {
+                DEBUG_PRINTF("(deferred label resolution)\n");
+                incomplete = 1;
+            }
+            i->length++;
+        }
 
-            o = o->next;
+        if (incomplete) {
+            DEBUG_PRINTF("pending label address\n");
+        } else {
+            i->ready = 1;
         }
 
         return 0;
-    }
+    } /* end of DAT */
 
     /* start with opcode bits */
     bits = opcode_bits_(i->opcode);
@@ -834,13 +865,14 @@ int parse_stream_(FILE *f, const char *src, struct dynamic_array *instructionps,
             break;
         }
 
-        if (buf_tokenize_(buf, &instr)) {
+        if (tokenize_line_(buf, &instr)) {
             fprintf(stderr, "%s:%u:%s", src, line, "trouble tokenizing input\n");
             retval = -1;
             break;
         }
 
         if (instr) {
+            instr->src_line = line;
             /* add to list of instructions */
             instr_list_entry = dynarray_add(instructionps, &instr);
             if (instr_list_entry == NULL) {
@@ -867,7 +899,10 @@ int parse_stream_(FILE *f, const char *src, struct dynamic_array *instructionps,
                 label_addr_calculate_(instructionps, labels);
             }
 
-            instr_assemble_(labels, instr, allow_short_labels);
+            if (instr_assemble_(labels, instr, allow_short_labels)) {
+                fprintf(stderr, "%s:%u:%s", src, line, "could not assemble instruction\n");
+                break;
+            }
         }
     }
     if (ferror(f)) {
@@ -904,9 +939,14 @@ int assemble_check_(struct dynamic_array *instructionps, struct dynamic_array *l
     DEBUG_PRINTF(" final pass of assembler...\n");
     for (x = 0; x < instructionps->entries; x++) {
         struct instruction_ **instrp = (struct instruction_ **)DYNARRAY_ITEM(*instructionps, x);
-        retval |= instr_assemble_(labels, *instrp, allow_short_labels);
+        retval = instr_assemble_(labels, *instrp, allow_short_labels);
         if (retval) {
             fprintf(stderr, "instruction %zu failed to assemble\n", x);
+            return retval;
+        }
+        if (! (*instrp)->ready) {
+            fprintf(stderr, "instruction not resolvable\n");
+            return -1;
         }
     }
 
@@ -933,6 +973,9 @@ int assemble_check_(struct dynamic_array *instructionps, struct dynamic_array *l
     return retval;
 }
 
+/*  output_
+ *  write assembled words to named file
+ */
 static
 int output_(struct dynamic_array *instructionps, const char *filename) {
     FILE *of = NULL;
@@ -1048,13 +1091,18 @@ int main(int argc, char *argv[]) {
             }
 
             VERBOSE_PRINTF("assembling '%s'...\n", filename);
-            parse_stream_(f, filename, instructionps_, labels_, allow_short_labels);
-
+            c = parse_stream_(f, filename, instructionps_, labels_, allow_short_labels);
+            if (c)
+                break;
             fclose(f);
         }
     } else {
         VERBOSE_PRINTF("assembling '%s'...\n", "stdin");
-        parse_stream_(stdin, "-", instructionps_, labels_, allow_short_labels);
+        c = parse_stream_(stdin, "-", instructionps_, labels_, allow_short_labels);
+    }
+    if (c) {
+        fprintf(stderr, "could not parse input, aborting\n");
+        exit(EX_DATAERR);
     }
 
     if (assemble_check_(instructionps_, labels_, allow_short_labels)) {