finish analyzer

6 files changed, 125 insertions, 10 deletions

diff --git a/src/as_analyzer.c b/src/as_analyzer.c
index dbc1bd2..a3e7ea4 100644
--- a/src/as_analyzer.c
+++ b/src/as_analyzer.c
@@ -3,11 +3,6 @@
 #include <stddef.h>
 #include <string.h>
 
-struct sym_table_entry {
-    const char * name;
-    size_t offset;
-};
-
 const char * compose_section_label(allocator_t alct, const char * section, const char * name) {
     size_t section_len = strlen(section);
     size_t name_len = strlen(name);
@@ -36,18 +31,43 @@ void process_section_label(allocator_t alct, prog_t prog) {
 }
 
 size_t instr_size(instr_t instr) {
-    // TODO
-    return 0;
+    return op_size(instr->op);
+}
+
+struct sym_table new_sym_table(allocator_t alct) {
+    struct sym_table tbl;
+    tbl.cap = 16;
+    tbl.size = 0;
+    tbl.buf = allocate(alct, sizeof(struct sym_table_entry) * 16);
+    return tbl;
+}
+
+void sym_table_add(allocator_t alct, struct sym_table* tbl, const char* name, int pos) {
+    if (tbl->cap == tbl->size) {
+        void *old_buf = tbl->buf;
+        tbl->buf = allocate(alct, sizeof(struct sym_table_entry) * tbl->cap * 2);
+        memcpy(tbl->buf, old_buf, sizeof(struct sym_table_entry) * tbl->cap);
+        tbl->cap = tbl->cap * 2;
+    }
+    tbl->buf[tbl->size] = (struct sym_table_entry){.name = name, .offset = pos,};
+    tbl->size += 1;
 }
 
 struct sym_table analyze_prog(allocator_t alct, prog_t prog) {
     process_section_label(alct, prog);
     stmt_t * stmts = prog->stmts->stmts;
+    struct sym_table tbl = new_sym_table(alct);
+    size_t cur_pos = 0;
     for (int i = 0; ; i++) {
         if (stmts[i] == NULL) break;
+        stmt_t stmt = stmts[i];
+        if (stmt->label) {
+            sym_table_add(alct, &tbl, stmt->label->name, cur_pos);
+        }
+        if (stmt->instr) {
+            cur_pos += instr_size(stmt->instr);
+        }
     }
-    struct sym_table tbl;
-    // TODO
     return tbl;
 }
 
diff --git a/src/as_analyzer.h b/src/as_analyzer.h
index 793d14a..988cca0 100644
--- a/src/as_analyzer.h
+++ b/src/as_analyzer.h
@@ -4,13 +4,20 @@
 #include "as_parser.h"
 #include "utils.h"
 
-struct sym_table_entry;
+struct sym_table_entry {
+    const char * name;
+    size_t offset;
+};
+
 struct sym_table {
     int size;
     int cap;
     struct sym_table_entry *buf;
 };
 
+struct sym_table new_sym_table(allocator_t alct);
+void sym_table_add(allocator_t alct, struct sym_table* tbl, const char* name, int pos);
+
 struct sym_table analyze_prog(allocator_t alct, prog_t prog);
 
 #endif // FVM_AS_ANALYZER_H_
diff --git a/src/as_op.c b/src/as_op.c
index 8b5d887..67c91ff 100644
--- a/src/as_op.c
+++ b/src/as_op.c
@@ -92,6 +92,46 @@ struct op_table_entry op_table [] = {
     {OP_END, NULL},
 };
 
+int op_size(enum op op) {
+    if (op == OP_IMM) {
+        return 8 + 1;
+    }
+    if (op == OP_CALL
+            || op == OP_JMP
+            || op == OP_JNZ
+            || op == OP_JZ
+            || op == OP_SYSCALL) {
+        return 8 + 1 + 1;
+    }
+    if (op == OP_LDARG || op == OP_STARG) {
+        return 9 + 9 + 1 + 9 + 4;
+    }
+    if (op == OP_LDVAR || op == OP_STVAR) {
+        return 9 + 9 + 2 + 9 + 4;
+    }
+    return 1;
+}
+
+int is_ld_op(enum op op) {
+    if (op == OP_LD
+            || op == OP_LD8
+            || op == OP_LD16
+            || op == OP_LD32) {
+        return 1;
+    }
+    return 0;
+}
+
+int is_st_op(enum op op) {
+    if (op == OP_ST
+            || op == OP_ST8
+            || op == OP_ST16
+            || op == OP_ST32) {
+        return 1;
+    }
+    return 0;
+}
+
 enum op str2op(const char* str) {
     for (int i = 0; op_table[i].name != NULL; i++) {
         if (strcmp(op_table[i].name, str) == 0) {
diff --git a/src/as_op.h b/src/as_op.h
index bded9e4..d02c29f 100644
--- a/src/as_op.h
+++ b/src/as_op.h
@@ -23,6 +23,9 @@ enum op {
 
 enum op str2op(const char *str);
 int isOp(const char *str);
+int op_size(enum op op);
+int is_ld_op(enum op);
+int is_st_op(enum op);
 
 #endif
 
diff --git a/src/as_tokenizer.c b/src/as_tokenizer.c
index 54e0969..216ee8a 100644
--- a/src/as_tokenizer.c
+++ b/src/as_tokenizer.c
@@ -55,6 +55,9 @@ int is_start_of_identifier(int c) {
     if (c == '_') {
         return 1;
     }
+    if (c == '.') {
+        return 1;
+    }
     return 0;
 }
 
diff --git a/tests/test_as_analyzer.c b/tests/test_as_analyzer.c
new file mode 100644
index 0000000..4c56131
--- /dev/null
+++ b/tests/test_as_analyzer.c
@@ -0,0 +1,42 @@
+#include <stdio.h>
+#include <string.h> 
+#include <assert.h>
+
+#include "as_tokenizer.h"
+#include "as_parser.h"
+#include "as_analyzer.h"
+#include "utils.h"
+
+char *input_buffer = 
+    "start:\n"
+    "    imm 1\n"
+    "mid: add\n"
+    "    call start\n"
+    "    sub\n"
+    ".insec: add\n"
+    "    div\n"
+    "end:    eq\n";
+
+int main(int argc, char** argv) {
+    printf("[TEST] assembler analyzer\n");
+    // make a memory buffer to FILE*
+    FILE *fp = fmemopen(input_buffer, strlen(input_buffer), "r");
+    allocator_t alct = new_allocator();
+    token_stream_t ts = new_token_stream(alct, fp);
+    prog_t prog = parse_prog(alct, ts);
+    struct sym_table sym_table = analyze_prog(alct, prog);    
+
+    assert(strcmp(sym_table.buf[0].name, "start") == 0);
+    assert(strcmp(sym_table.buf[1].name, "mid") == 0);
+    assert(strcmp(sym_table.buf[2].name, "mid.insec") == 0);
+    assert(strcmp(sym_table.buf[3].name, "end") == 0);
+
+    assert(sym_table.buf[0].offset == 0);
+    assert(sym_table.buf[1].offset == 9);
+    assert(sym_table.buf[2].offset == 21);
+    assert(sym_table.buf[3].offset == 23);
+
+    printf("[PASS] assembler analyzer\n");
+    delete_allocator(alct);
+    return 0;
+}