Add debug symbols, improve debugger, fix bugs

4 files changed, 79 insertions, 8 deletions

diff --git a/atk16_asm/asm_pass0.py b/atk16_asm/asm_pass0.py
index 262e5c1..f6586cd 100644
--- a/atk16_asm/asm_pass0.py
+++ b/atk16_asm/asm_pass0.py
@@ -15,9 +15,12 @@ class Result0:
   lines: list[Result0Line]
 
 def pass_0(lines: list[str], file_name: str) -> Result0:
+  file_name = file_name if file_name.endswith(".atk16") else file_name + ".atk16"
   result_lines: list[Result0Line] = []
 
   for (line_num, line) in enumerate(lines):
+    line_num += 1 # line numbers are 1-based
+
     line = line.split(";")[0].strip()
     if line == "": continue
 
@@ -25,8 +28,8 @@ def pass_0(lines: list[str], file_name: str) -> Result0:
 
     match keyword:
       case "@include":
-        asm_file_name = args[0]
-        path = os.path.join(os.path.dirname(file_name), asm_file_name + ".atk16")
+        asm_file_name = args[0] if args[0].endswith(".atk16") else args[0] + ".atk16"
+        path = os.path.join(os.path.dirname(file_name), asm_file_name)
         with open(path, "r") as f:
           incl_lines = f.readlines()
 
diff --git a/atk16_asm/asm_pass3.py b/atk16_asm/asm_pass3.py
index 09f58fa..ec8b20c 100644
--- a/atk16_asm/asm_pass3.py
+++ b/atk16_asm/asm_pass3.py
@@ -12,15 +12,43 @@ class Result3Line:
   original_parts: list[str]
 
 @dataclass
+class DbgSourceInfo:
+  src_file: str
+  line_num: int
+  expanded_text: str
+  original_text: str
+
+DbgLabelTable = dict[int, list[str]]
+DbgSourceTable = dict[int, DbgSourceInfo]
+
+@dataclass
 class Result3:
   lines: list[Result3Line]
   operations: OpExpansionDict
   symbols: dict[str, int]
+  dbg_label_table: DbgLabelTable
+  dbg_source_table: DbgSourceTable
 
 def pass_3(result2: Result2) -> Result3:
   result_lines: list[Result3Line] = []
   symbols: dict[str, int] = {}
   address = 0
+  dbg_label_table: DbgLabelTable = {}
+  dbg_source_table: DbgSourceTable = {}
+
+  def save_dbg_label(addr: int, label: str):
+    if addr in dbg_label_table:
+      dbg_label_table[addr].append(label)
+    else:
+      dbg_label_table[addr] = [label]
+
+  def save_dbg_source(addr: int, src_file: str, line_num: int, expanded_text: str, original_text: str):
+    dbg_source_table[addr] = DbgSourceInfo(
+      src_file=src_file,
+      line_num=line_num,
+      expanded_text=expanded_text,
+      original_text=original_text,
+    )
 
   for line in result2.lines:
     keyword, *args = line.parts
@@ -33,11 +61,15 @@ def pass_3(result2: Result2) -> Result3:
         if label in symbols:
           raise Exception(f"When defining label {label} as {address:>04x}, symbol {label} already defined as {symbols[label]:>04x}")
         symbols[args[0]] = address
+        save_dbg_label(address, label)
         continue
       case "@let":
         symbols[args[0]] = eval_expr(symbols, args[1])
         continue
       case _:
+        dbg_original_text = " ".join(line.original_parts)
+        dbg_expanded_text = " ".join(line.parts)
+        save_dbg_source(address, line.src_file, line.line_num, dbg_expanded_text, dbg_original_text)
         result_lines.append(Result3Line(
           line_num=line.line_num,
           src_file=line.src_file,
@@ -54,10 +86,13 @@ def pass_3(result2: Result2) -> Result3:
     if n > 1:
       formatted = format_overlapping_rows(rows_with_same_addr)
       raise Exception(f"Overlapping segments: address 0x{result_line.address:>04x} has conflicting definitions:\n{formatted}")
+
   return Result3(
     operations=result2.operations,
     lines=result_lines,
     symbols=symbols,
+    dbg_label_table=dbg_label_table,
+    dbg_source_table=dbg_source_table,
   )
 
 def format_overlapping_rows(rows: list[Result3Line]) -> str:
diff --git a/atk16_asm/asm_pass4.py b/atk16_asm/asm_pass4.py
index 7365416..52aaf68 100644
--- a/atk16_asm/asm_pass4.py
+++ b/atk16_asm/asm_pass4.py
@@ -17,6 +17,8 @@ class Result4:
   lines: list[Result4Line]
   operations: OpExpansionDict
   symbols: dict[str, int]
+  dbg_label_table: DbgLabelTable
+  dbg_source_table: DbgSourceTable
 
 def pass_4(result3: Result3) -> Result4:
   result_lines: list[Result4Line] = []
@@ -32,7 +34,10 @@ def pass_4(result3: Result3) -> Result4:
 
     if keyword in operations:
       fn = operations[keyword]
-      word = fn(meta, result3.symbols, *args)
+      try:
+        word = fn(meta, result3.symbols, *args)
+      except Exception as e:
+        raise Exception(f"Invalid assembly at {line.src_file}:{line.line_num + 1}\n\n{line.parts}\n\n{e}") from e
       result_lines.append(Result4Line(
         line_num=line.line_num,
         src_file=line.src_file,
@@ -58,4 +63,6 @@ def pass_4(result3: Result3) -> Result4:
     operations=result3.operations,
     symbols=result3.symbols,
     lines=result_lines,
+    dbg_label_table=result3.dbg_label_table,
+    dbg_source_table=result3.dbg_source_table,
   )
diff --git a/atk16_asm/assembler.py b/atk16_asm/assembler.py
index 993472e..4cd2cef 100755
--- a/atk16_asm/assembler.py
+++ b/atk16_asm/assembler.py
@@ -7,11 +7,17 @@ from .asm_eval import *
 from .asm_pass0 import pass_0
 from .asm_pass1 import pass_1
 from .asm_pass2 import pass_2
-from .asm_pass3 import pass_3
+from .asm_pass3 import pass_3, DbgLabelTable, DbgSourceTable
 from .asm_pass4 import pass_4
 from .pad_bin import pad_binary
 
-def assemble(source: str, file_name: str) -> bytearray:
+@dataclass
+class AssemblyResult:
+  program: bytearray
+  dbg_label_table: DbgLabelTable
+  dbg_source_table: DbgSourceTable
+
+def assemble(source: str, file_name: str) -> AssemblyResult:
   src_lines = source.splitlines()
 
   result0 = pass_0(src_lines, file_name)
@@ -40,7 +46,11 @@ def assemble(source: str, file_name: str) -> bytearray:
     result[2 * line.address + 0] = ((line.word >> 8) & 0xff)
     result[2 * line.address + 1] = ((line.word >> 0) & 0xff)
 
-  return result
+  return AssemblyResult(
+    program=result,
+    dbg_label_table=result4.dbg_label_table,
+    dbg_source_table=result4.dbg_source_table,
+  )
 
 def main():
   if len(sys.argv) != 3:
@@ -62,11 +72,27 @@ def main():
   result = assemble(src, infile_path)
 
   with open(outfile_path, "wb") as f:
-    f.write(result)
+    f.write(result.program)
 
   pad_binary(outfile_path, 64 * 1024)
 
-  print(f"Wrote 64 KB to {outfile_path} ({len(result)} B without padding)")
+  print(f"Wrote 64 KB to {outfile_path} ({len(result.program)} B without padding)")
+
+  dbg_outfile_path = outfile_path + ".dbg"
+
+  with open(dbg_outfile_path, "w") as f:
+    for i in range(32 * 1024): # 32K words
+      line = f"0x{i:>04x};"
+      if i not in result.dbg_source_table:
+        continue
+
+      dbg_source_info = result.dbg_source_table[i]
+      line += f"{dbg_source_info.src_file}:{dbg_source_info.line_num};{dbg_source_info.expanded_text};{dbg_source_info.original_text};"
+
+      if i in result.dbg_label_table:
+        line += f"{';'.join(result.dbg_label_table[i])}"
+
+      f.write(line + "\n")
 
 if __name__ == "__main__":
   main()