fix(tools): keep source lines when the guest pointer audit blanks comments and literals

Main's progress.code now drops blank lines, so the audit reported
lines from a shorter text (Export.cpp:145 for an access on line 172).
The audit blanks comments and literals itself, keeping every line,
including spliced ones.

Co-Authored-By: Claude Opus 5.5 <[email protected]>

tools/guest_pointer_audit.py

@@ -103,6 +103,35 @@ def accesses(body, name):
     return sorted(found)
 
 
+def splice_lines(text):
+    lines = []
+    pending = ""
+    joined = 0
+    for line in text.replace("\r\n", "\n").split("\n"):
+        if line.endswith("\\"):
+            pending += line[:-1]
+            joined += 1
+            continue
+        lines.append(pending + line + "\n" * joined)
+        pending = ""
+        joined = 0
+    if joined:
+        lines.append(pending + "\n" * joined)
+    return "\n".join(lines)
+
+
+def keep_lines(match):
+    if match.group("code") is not None:
+        return match.group()
+    if match.group("comment") is None:
+        return "LITERAL" + "\n" * match.group().count("\n")
+    return re.sub(r"[^\n]", " ", match.group())
+
+
+def source_code(text):
+    return progress.CPP_TOKEN.sub(keep_lines, splice_lines(text))
+
+
 def audit_text(text):
     results = []
     for match in progress.DEFINITION.finditer(text):
@@ -124,7 +153,7 @@ def audit(libraries):
         for source in sorted(library.rglob("*.cpp")):
             if "tests" in source.relative_to(library).parts:
                 continue
-            text = progress.code(source.read_text(encoding="utf-8", errors="ignore"))
+            text = source_code(source.read_text(encoding="utf-8", errors="ignore"))
             for entry in audit_text(text):
                 report.append({"library": library.name, "file": source.relative_to(ROOT).as_posix(), **entry})
     return report

tools/tests/test_guest_pointer_audit.py

@@ -13,7 +13,7 @@ import progress
 
 
 def findings(source):
-    return [(entry["function"], entry["parameter"], entry["kinds"]) for entry in guest_pointer_audit.audit_text(progress.code(source))]
+    return [(entry["function"], entry["parameter"], entry["kinds"]) for entry in guest_pointer_audit.audit_text(guest_pointer_audit.source_code(source))]
 
 
 class AccessTests(unittest.TestCase):
@@ -101,15 +101,21 @@ class AccessTests(unittest.TestCase):
         self.assertEqual(findings(source), [("sceExampleSn", "p", ["library call"]), ("sceExampleS", "p", ["library call"])])
 
     def test_line_is_the_earliest_access_of_any_kind(self):
-        entry, = guest_pointer_audit.audit_text(progress.code("int APS5_VABI sceExampleEarly(Info* p)\n{\n  p[0] = {};\n  p->field = 1;\n  return 0;\n}\n"))
+        entry, = guest_pointer_audit.audit_text(guest_pointer_audit.source_code("int APS5_VABI sceExampleEarly(Info* p)\n{\n  p[0] = {};\n  p->field = 1;\n  return 0;\n}\n"))
         self.assertEqual((entry["line"], entry["kinds"]), (3, ["index", "member"]))
 
     def test_comments_and_literals_are_ignored(self):
         source = "int APS5_VABI sceExampleText(Info* info) { // info->size = 4;\n  return Log(\"info->size *info info[0]\"); }\n"
         self.assertEqual(findings(source), [])
 
+    def test_line_counts_blank_spliced_and_commented_lines(self):
+        source = ("int APS5_VABI sceExampleBlank(int* out)\n{\n\n\n  /* one\n     two */\n  const char* s = R\"(a\nb)\";\n"
+                  "  int wide = 1 + \\\n    2;\n\n  *out = wide;\n  return 0;\n}\n")
+        entry, = guest_pointer_audit.audit_text(guest_pointer_audit.source_code(source))
+        self.assertEqual(entry["line"], 12)
+
     def test_line_points_at_the_first_access(self):
-        entry, = guest_pointer_audit.audit_text(progress.code("int APS5_VABI sceExampleLine(int* out)\n{\n  int a = 0;\n  *out = a;\n  return 0;\n}\n"))
+        entry, = guest_pointer_audit.audit_text(guest_pointer_audit.source_code("int APS5_VABI sceExampleLine(int* out)\n{\n  int a = 0;\n  *out = a;\n  return 0;\n}\n"))
         self.assertEqual(entry["line"], 4)