diff --git a/scspell/__init__.py b/scspell/__init__.py
index 4300f4e..c7047c3 100644
--- a/scspell/__init__.py
+++ b/scspell/__init__.py
@@ -462,8 +462,8 @@ def report_failed_check(match_desc, filename, unmatched_subtokens):
 
 
 def spell_check_token(
-        match_desc, filename, fq_filename, file_id_ref,
-        dicts, ignores, report_only):
+        match_desc, fq_filename, file_id,
+        dicts, ignores):
     """Spell check a single token.
 
     :param match_desc: description of the token matching instance
@@ -485,44 +485,22 @@ def spell_check_token(
         subtokens = decompose_token(token)
         unmatched_subtokens = [
             st for st in subtokens if len(st) > LEN_THRESHOLD and
-            (not dicts.match(st, filename, file_id_ref[0])) and
+            (not dicts.match(st, fq_filename, file_id)) and
             (st not in ignores)]
-        if unmatched_subtokens:
-            unmatched_subtokens = make_unique(unmatched_subtokens)
-            if report_only:
-                return (report_failed_check(match_desc, filename,
-                                            unmatched_subtokens),
-                        True)
-            else:
-                return (
-                    handle_failed_check_interactively(
-                        match_desc, filename, fq_filename, file_id_ref,
-                        unmatched_subtokens, dicts, ignores),
-                    True)
-    return (
-        (match_desc.get_string(), match_desc.get_ofs() + len(token)),
-        False)
+        unmatched_subtokens = make_unique(unmatched_subtokens)
+        return unmatched_subtokens
+    return []
 
 
-def spell_check_file(filename, dicts, ignores, report_only, c_escapes):
-    """Spell check a single file.
+def spell_check_str(source_text, fq_filename, dicts, ignores, c_escapes):
+    """Spell check an in-memory "file".
 
-    :param filename: name of the file to check
+    :param fq_filename: fully-qualified filename
     :param dicts: dictionary set against which to perform matching
     :type  dicts: CorporaFile
     :param ignores: set of tokens to ignore for this session
 
     """
-    fq_filename = os.path.normcase(os.path.realpath(filename))
-    try:
-        with _util.open_with_encoding(fq_filename) as source_file:
-            source_text = source_file.read()
-    except IOError as e:
-        print("Error: can't read source file '{}'; "
-              'skipping (reason: {})'.format(filename, e),
-              file=sys.stderr)
-        return False
-
     # Look for a file ID
     file_id = None
     m_id = FILE_ID_REGEX.search(source_text)
@@ -535,8 +513,6 @@ def spell_check_file(filename, dicts, ignores, report_only, c_escapes):
     else:
         file_id = dicts.file_id_of_file(fq_filename)
 
-    file_id_ref = [file_id]  # allow for spell_check() creating a file_id
-
     if c_escapes:
         token_regex = C_ESCAPE_TOKEN_REGEX
     else:
@@ -545,7 +521,6 @@ def spell_check_file(filename, dicts, ignores, report_only, c_escapes):
     # Search for tokens to spell-check
     data = source_text
     pos = 0
-    okay = True
     while True:
         m = token_regex.search(data, pos)
         if m is None:
@@ -556,19 +531,61 @@ def spell_check_file(filename, dicts, ignores, report_only, c_escapes):
             # This is matching the file-id.  Skip over it.
             pos = m_id.end()
             continue
-        result = spell_check_token(MatchDescriptor(data, m),
-                                   filename, fq_filename, file_id_ref,
-                                   dicts, ignores, report_only)
-        (data, pos) = result[0]
-        error_found = result[1]
-        if error_found:
-            okay = False
+        match_desc = MatchDescriptor(data, m)
+        unmatched_subtokens = spell_check_token(match_desc,
+                                   fq_filename, file_id,
+                                   dicts, ignores)
+        pos = match_desc.get_ofs() + len(match_desc.get_token())
+        if unmatched_subtokens:
+            new_data = yield file_id, match_desc, unmatched_subtokens
+            if new_data:
+                data, pos = new_data
+
+
+def spell_check_file(filename, dicts, ignores, report_only, c_escapes):
+    """Spell check a single file.
+
+    :param filename: name of the file to check
+    :param dicts: dictionary set against which to perform matching
+    :type  dicts: CorporaFile
+    :param ignores: set of tokens to ignore for this session
+
+    """
+    fq_filename = os.path.normcase(os.path.realpath(filename))
+    try:
+        with _util.open_with_encoding(fq_filename) as source_file:
+            source_text = source_file.read()
+    except IOError as e:
+        print("Error: can't read source file '{}'; "
+              'skipping (reason: {})'.format(filename, e),
+              file=sys.stderr)
+        return False
+
+    okay = True
+    speller = spell_check_str(source_text, fq_filename, dicts, ignores, c_escapes)
+    new_pos = None
+    while True:
+        try:
+            file_id, match_desc, unmatched_subtokens = speller.send(new_pos)
+        except StopIteration:
+            break
+        okay = False
+        if report_only:
+            report_failed_check(match_desc, filename,
+                                        unmatched_subtokens)
+        else:
+            # HACK: Satisfy handle_failed_check_interactively API.  Mutation of
+            # file_id is currently not handled.
+            file_id_ref = [file_id]
+            new_pos = handle_failed_check_interactively(
+                    match_desc, filename, fq_filename, file_id_ref,
+                    unmatched_subtokens, dicts, ignores)
 
     # Write out the source file if it was modified
-    if data != source_text:
+    if new_pos and new_pos[0] != source_text:
         with _util.open_with_encoding(fq_filename, mode='w') as source_file:
             try:
-                source_file.write(data)
+                source_file.write(new_pos[0])
             except IOError as e:
                 print(str(e), file=sys.stderr)
                 return False