Fix DFA issue (wip)

Author: StefanosChaliasos

pyproject.toml

@@ -23,7 +23,7 @@ dependencies = [
   "rstr",
   "exrex",
   "joblib",
-  "automata-lib",
+  #"automata-lib",
 ]
 
 [project.optional-dependencies]

src/zkregex_fuzzer/dfa.py

@@ -11,15 +11,22 @@
 
 from automata.fa.dfa import DFA
 from automata.fa.gnfa import GNFA
-from automata.fa.nfa import NFA
+from automata.fa.nfa import NFA, RESERVED_CHARACTERS
+
+
+from zkregex_fuzzer.utils import ASCII_CHARS
 
 
 def regex_to_dfa(regex: str) -> DFA:
     """
     Convert a regex to a DFA.
     """
+    # Symbols should include at least all ASCII characters
+    symbols = ASCII_CHARS
+    symbols = symbols - RESERVED_CHARACTERS
+
     try:
-        nfa = NFA.from_regex(regex)
+        nfa = NFA.from_regex(regex, input_symbols=symbols)
     except Exception as e:
         raise ValueError(f"Failed to parse '{regex}' into an automaton: {e}")
     try:

src/zkregex_fuzzer/utils.py

@@ -11,6 +11,19 @@
 from zkregex_fuzzer.dfa import wrapped_has_one_accepting_state_regex
 
 
+def create_range(start_char: str, end_char: str) -> set[str]:
+    """
+    Create a set of characters from start_char to end_char.
+    """
+    return {chr(i) for i in range(ord(start_char), ord(end_char) + 1)}
+
+
+LATIN_EXT_CHARS = create_range("¡", "ƿ")
+GREEK_CHARS = create_range("Ͱ", "Ͽ")
+CYRILLIC_CHARS = create_range("Ѐ", "ӿ")
+ASCII_CHARS = set(string.printable)
+
+
 def is_valid_regex(regex: str) -> bool:
     """
     Check if a regex is valid.

tests/test_dfa.py

@@ -8,7 +8,7 @@
     transform_dfa_to_regex,
     transform_dfa_to_single_accepting_state,
 )
-
+import re
 regex_with_multiple_accepting_states = [
     r"(ab|aba)",
     r"(ab|aba)*",
@@ -27,7 +27,7 @@
     r"(hello)",
     r"(ab)*",
     r"(a|b|c)*",
-    r"((a|b|c)*abc)",  # This is somewhat comples, do we want to support this?
+    r"((a|b|c)*abc)",  # This is somewhat complex, do we want to support this?
     r"[a-zA-Z]+",
     r"[0-9]+",
     r"(abc|abcd|abcde)f",
@@ -39,6 +39,17 @@
     r"abc",
     r"(hello)",
 ]
+zkemail_regexes = [
+    ">[^<>]+<.*",
+    r"to:[^\r\n]+\r\n",
+    r")subject:[^\r\n]+\r\n",
+    r"[A-Za-z0-9!#$%&'*+=?\-\^_`{|}~./]+@[A-Za-z0-9.\-@]+",
+    r"dkim-signature:([a-z]+=[^;]+; )+bh=[a-zA-Z0-9+/=]+;",
+    r"[A-Za-z0-9!#$%&'*+=?\-\^_`{|}~./@]+@[A-Za-z0-9.\-]+",
+    r"from:[^\r\n]+\r\n",
+    r"dkim-signature:([a-z]+=[^;]+; )+t=[0-9]+;",
+    r"message-id:<[A-Za-z0-9=@\.\+_-]+>\r\n",
+]
 
 
 def test_has_multiple_accepting_states_regex_without_multiple():
@@ -115,3 +126,17 @@ def test_dfa_string_matching():
                 break
         if regex not in single_solution_regexes:
             assert string != string2
+
+
+def test_dfa_string_matching_zkemail():
+    for regex in zkemail_regexes:
+        string = dfa_string_matching(regex)
+        print()
+        print("--------------------------------")
+        print(f"Testing regex: {regex}" )
+        print(f"String: {string}")
+        print("--------------------------------")
+        print()
+        assert string is not None
+        # we also need to check against re module
+        assert re.match(regex, string) is not None