Join VIMES and EIMES semantics

tools/extract_message_table.py

@@ -1,21 +1,23 @@
 #!/usr/bin/env python3
 """Extract an ID-dispatched SYS4 message table.
 
-ITMES and SKMES are long chains of:
+The shipped tables use two equivalent control-flow shapes:
 
     eq <temporary>, <selected id global>, <record id>
-    jcc ...
-    show-text ...
-    ...
+    jcc <next guard>
+    <message body>
     jmp <shared exit>
 
-This tool discovers the selector global from that repeated shape, reconstructs
-the displayed lines (including the surface text of furigana spans), and emits a
-reusable ID-to-message JSON table.
+or a compact guard block whose successful branches target message bodies stored
+later in the script. This tool discovers the selector global, follows either
+layout, reconstructs the displayed lines (including the surface text of
+furigana spans), and emits a reusable ID-to-message JSON table.
 
 Usage:
   py -3.11 -X utf8 tools/extract_message_table.py ITMES
   py -3.11 -X utf8 tools/extract_message_table.py SKMES [OUTNAME]
+  py -3.11 -X utf8 tools/extract_message_table.py VIMES
+  py -3.11 -X utf8 tools/extract_message_table.py EIMES
 """
 from __future__ import annotations
 
@@ -33,6 +35,11 @@ import sys4load
 T_IMM = 0
 T_INLINE_STRING = 2
 T_GLOBAL_INT = 3
+BRANCH_SENTINEL = 0xFFFFFFFF
+
+MESSAGE_LAYOUTS = {
+    "EIMES.BIN": "enemy-commentary",
+}
 
 
 def resolve(name: str) -> Path:
@@ -77,7 +84,9 @@ def _strings(ins, scr) -> list[str]:
     ]
 
 
-def _message_body(scr, start: int, stop: int) -> dict | None:
+def _message_body(
+    scr, start: int, stop: int, layout: str = "title-description"
+) -> dict | None:
     fragments: list[str] = []
     lines: list[str] = []
     furigana: list[dict] = []
@@ -102,15 +111,34 @@ def _message_body(scr, start: int, stop: int) -> dict | None:
         lines.append("".join(fragments))
     if not lines:
         return None
-    message = {
-        "title": lines[0],
-        "description": "\n".join(lines[1:]),
-    }
+    if layout == "enemy-commentary":
+        message = {
+            "summary": lines[0],
+            "strategy": "\n".join(lines[1:]),
+        }
+    else:
+        message = {
+            "title": lines[0],
+            "description": "\n".join(lines[1:]),
+        }
     if furigana:
         message["furigana"] = furigana
     return message
 
 
+def _branch_target_index(scr, guard_index: int, offsets: dict[int, int]) -> int | None:
+    """Resolve the conditional branch immediately following a dispatch guard."""
+    if guard_index + 1 >= len(scr.instructions):
+        return None
+    branch = scr.instructions[guard_index + 1]
+    if sys4load.display_label(branch.opcode) != "jcc":
+        return None
+    for arg_type, value in branch.args[1:]:
+        if arg_type == T_IMM and value != BRANCH_SENTINEL and value in offsets:
+            return offsets[value]
+    return None
+
+
 def extract_messages(scr, selector: int | None = None) -> tuple[list[dict], dict]:
     """Extract ordered message records and dispatch metadata from a script."""
     discovered_selector, comparison_count = discover_selector(scr)
@@ -122,24 +150,49 @@ def extract_messages(scr, selector: int | None = None) -> tuple[list[dict], dict
     ]
     records: list[dict] = []
     seen: set[int] = set()
+    offsets = {
+        instruction.offset: index
+        for index, instruction in enumerate(scr.instructions)
+    }
+    message_layout = MESSAGE_LAYOUTS.get(scr.path.name.upper(), "title-description")
+    dispatch_layout_counts = collections.Counter()
     for guard_index, (instruction_index, record_id) in enumerate(guards):
         stop = guards[guard_index + 1][0] if guard_index + 1 < len(guards) else len(scr.instructions)
-        message = _message_body(scr, instruction_index + 1, stop)
+        message_start = instruction_index + 2
+        message = _message_body(scr, message_start, stop, message_layout)
+        dispatch_layout = "fallthrough"
+        if message is None:
+            branch_target = _branch_target_index(scr, instruction_index, offsets)
+            if branch_target is not None:
+                message_start = branch_target
+                message = _message_body(
+                    scr, message_start, len(scr.instructions), message_layout
+                )
+                dispatch_layout = "branch-target"
         if message is None:
             continue
         if record_id in seen:
             raise ValueError(f"{scr.path.name}: duplicate message id {record_id}")
         seen.add(record_id)
+        dispatch_layout_counts[dispatch_layout] += 1
         records.append({
             "id": record_id,
             "dispatch_offset": f"0x{scr.instructions[instruction_index].offset:x}",
+            "message_offset": f"0x{scr.instructions[message_start].offset:x}",
             **message,
         })
+    dispatch_layout = (
+        next(iter(dispatch_layout_counts))
+        if len(dispatch_layout_counts) == 1
+        else "mixed"
+    )
     return records, {
         "selector_global": f"0x{selector:x}",
         "dispatch_guard_count": len(guards),
         "message_count": len(records),
         "selector_discovery_count": comparison_count,
+        "dispatch_layout": dispatch_layout,
+        "message_layout": message_layout,
     }