fix: stop comment stripping from breaking base64

Author: daniel-makerx

src/algokit_utils/deploy.py

@@ -325,29 +325,45 @@ def add_deploy_template_variables(
 
 
 def _find_unquoted_string(line: str, token: str, start: int = 0, end: int = -1) -> int | None:
-    """Find the first string within a line of TEAL. Only matches outside of quotes are returned.
+    """Find the first string within a line of TEAL. Only matches outside of quotes and base64 are returned.
     Returns None if not found"""
 
     if end < 0:
         end = len(line)
     idx = start
-    in_quotes = False
+    in_quotes = in_base64 = False
     while idx < end:
         current_char = line[idx]
         match current_char:
-            case "\\":
-                if in_quotes:  # skip next character
-                    idx += 1
+            # enter base64
+            case " " | "(" if not in_quotes and _last_token_base64(line, idx):
+                in_base64 = True
+            # exit base64
+            case " " | ")" if not in_quotes and in_base64:
+                in_base64 = False
+            # escaped char
+            case "\\" if in_quotes:
+                # skip next character
+                idx += 1
+            # quote boundary
             case '"':
                 in_quotes = not in_quotes
-            case _:
+            # can test for match
+            case _ if not in_quotes and not in_base64 and line.startswith(token, idx):
                 # only match if not in quotes and string matches
-                if not in_quotes and line.startswith(token, idx):
-                    return idx
+                return idx
         idx += 1
     return None
 
 
+def _last_token_base64(line: str, idx: int) -> bool:
+    try:
+        *_, last = line[:idx].split()
+    except ValueError:
+        return False
+    return last in ("base64", "b64")
+
+
 def _find_template_token(line: str, token: str, start: int = 0, end: int = -1) -> int | None:
     """Find the first template token within a line of TEAL. Only matches outside of quotes are returned.
     Only full token matches are returned, i.e. TMPL_STR will not match against TMPL_STRING

tests/test_deploy.approvals/test_comment_stripping.approved.txt

@@ -12,4 +12,19 @@ op 123
 op 123
 op ""
 op "//"
-op "//"
+op "//"
+pushbytes base64(//8=)
+pushbytes b64(//8=)
+
+pushbytes base64(//8=)
+pushbytes b64(//8=)
+pushbytes "base64(//8=)"
+pushbytes "b64(//8=)"
+
+pushbytes base64 //8=
+pushbytes b64 //8=
+
+pushbytes base64 //8=
+pushbytes b64 //8=
+pushbytes "base64 //8="
+pushbytes "b64 //8="

tests/test_deploy.py

@@ -49,6 +49,22 @@ def test_comment_stripping() -> None:
 op "" // more comments
 op "//" //op "//"
 op "//"
+pushbytes base64(//8=)
+pushbytes b64(//8=)
+
+pushbytes base64(//8=)  // pushbytes base64(//8=)
+pushbytes b64(//8=)     // pushbytes b64(//8=)
+pushbytes "base64(//8=)"  // pushbytes "base64(//8=)"
+pushbytes "b64(//8=)"     // pushbytes "b64(//8=)"
+
+pushbytes base64 //8=
+pushbytes b64 //8=
+
+pushbytes base64 //8=  // pushbytes base64 //8=
+pushbytes b64 //8=     // pushbytes b64 //8=
+pushbytes "base64 //8="  // pushbytes "base64 //8="
+pushbytes "b64 //8="     // pushbytes "b64 //8="
+
 """
     result = strip_comments(program)
     check_output_stability(result)