Merge pull request #1197 from octodns/chunked-text-escape-handling

Fix handling of chunked values (TXT) when escapes land on boundaries

CHANGELOG.md

@@ -3,6 +3,8 @@
 * MetaProcessor.include_extra to add support for arbitrary extra values to be
   set on the meta record.
 * Correctly handled quoted svcparams when parsing SVCB/HTTPS rdata text
+* Fix handling of chunked values (TXT, SPF) when escaped characters land at the
+  split boundaries, don't split escapes from their following chars
 
 ## v1.9.1 - 2024-06-21 - What's in a name
 

octodns/record/chunked.py

@@ -13,10 +13,20 @@ class _ChunkedValuesMixin(ValuesMixin):
 
     def chunked_value(self, value):
         value = value.replace('"', '\\"')
-        vs = [
-            value[i : i + self.CHUNK_SIZE]
-            for i in range(0, len(value), self.CHUNK_SIZE)
-        ]
+        vs = []
+        i = 0
+        n = len(value)
+        # until we've processed the whole string
+        while i < n:
+            # start with a full chunk size
+            c = min(self.CHUNK_SIZE, n - i)
+            # make sure that we don't break on escape chars
+            while value[i + c - 1] == '\\':
+                c -= 1
+            # we have our chunk now
+            vs.append(value[i : i + c])
+            # and can step over if
+            i += c
         vs = '" "'.join(vs)
         return self._value_type(f'"{vs}"')
 

tests/test_octodns_record_chunked.py

@@ -4,8 +4,9 @@
 
 from unittest import TestCase
 
-from octodns.record.chunked import _ChunkedValue
+from octodns.record.chunked import _ChunkedValue, _ChunkedValuesMixin
 from octodns.record.spf import SpfRecord
+from octodns.record.txt import TxtValue
 from octodns.zone import Zone
 
 
@@ -67,3 +68,50 @@ class TestChunkedValue(TestCase):
             ['non ASCII character in "Déjà vu"'],
             _ChunkedValue.validate('Déjà vu', 'TXT'),
         )
+
+    zone = Zone('unit.tests.', [])
+
+    # some hacks to let us work with smaller sizes
+    class Base:
+        def __init__(self, *args, **kwargs):
+            pass
+
+    class SmallerChunkedMixin(_ChunkedValuesMixin, Base):
+        CHUNK_SIZE = 8
+        _type = 'TXT'
+        _value_type = TxtValue
+
+        def __init__(self, values):
+            super().__init__(None, None, {'values': values})
+
+    def test_splitting(self):
+
+        for value, expected in (
+            # shorter
+            ('0123', '"0123"'),
+            # exact
+            ('01234567', '"01234567"'),
+            # simple
+            ('0123456789', '"01234567" "89"'),
+            # 1 extra
+            ('012345678', '"01234567" "8"'),
+            # escape in the middle
+            ('01234\\;56789', '"01234\\;5" "6789"'),
+            # escape before the boundary
+            ('012345\\;6789', '"012345\\;" "6789"'),
+            # escape after the boundary
+            ('01234567\\;89', '"01234567" "\\;89"'),
+            # escape spanning the boundary
+            ('0123456\\;789', '"0123456" "\\;789"'),
+            # multiple escapes at the boundary
+            ('012345\\\\;6789', '"012345" "\\\\;6789"'),
+            # exact size escape
+            ('012345\\;', '"012345\\;"'),
+            # spanning ending
+            ('0123456\\;', '"0123456" "\\;"'),
+        ):
+            sc = self.SmallerChunkedMixin(value)
+            self.assertEqual([expected], sc.chunked_values)
+
+        sc = self.SmallerChunkedMixin(['0123456789'])
+        self.assertEqual(['"01234567" "89"'], sc.chunked_values)