@@ -150,7 +150,7 @@ def test_ast_chunking_python_preserves_top_level_blocks() -> None:
150150 chunking_strategy = "ast" ,
151151 chunk_size = 500 ,
152152 chunk_overlap = 100 ,
153- target_tokens = 60 ,
153+ target_tokens = 64 ,
154154 overlap_tokens = 0 ,
155155 ast_overlap_lines = 0 ,
156156 preserve_imports = 0 , # imports merge into the first block chunk
@@ -159,7 +159,7 @@ def test_ast_chunking_python_preserves_top_level_blocks() -> None:
159159 tok_cfg = TokenizationConfig (strategy = "whitespace" , normalize_unicode = False , lowercase = False )
160160 ch = Chunker (cfg , tok_cfg )
161161
162- body_tokens = " " .join ([f"tok{ i } " for i in range (18 )])
162+ body_tokens = " " .join ([f"tok{ i } " for i in range (45 )])
163163 content = (
164164 "import os\n import sys\n \n "
165165 "def foo():\n "
@@ -182,7 +182,7 @@ def test_ast_chunking_python_can_isolate_imports_when_enabled() -> None:
182182 chunking_strategy = "ast" ,
183183 chunk_size = 500 ,
184184 chunk_overlap = 100 ,
185- target_tokens = 20 ,
185+ target_tokens = 64 ,
186186 overlap_tokens = 0 ,
187187 ast_overlap_lines = 0 ,
188188 preserve_imports = 1 ,
@@ -194,10 +194,10 @@ def test_ast_chunking_python_can_isolate_imports_when_enabled() -> None:
194194 content = (
195195 "import os\n import sys\n \n "
196196 "def foo():\n "
197- " # a a a a a a a a a a \n "
197+ " # " + " " . join ([ "a" for _i in range ( 90 )]) + " \n "
198198 " return 1\n \n "
199199 "def bar():\n "
200- " # b b b b b b b b b b \n "
200+ " # " + " " . join ([ "b" for _i in range ( 90 )]) + " \n "
201201 " return 2\n "
202202 )
203203 chunks = ch .chunk_file ("x.py" , content )
@@ -213,7 +213,7 @@ def test_hybrid_chunking_falls_back_on_syntax_error() -> None:
213213 chunking_strategy = "hybrid" ,
214214 chunk_size = 500 ,
215215 chunk_overlap = 100 ,
216- target_tokens = 16 ,
216+ target_tokens = 64 ,
217217 overlap_tokens = 4 ,
218218 min_chunk_chars = 10 ,
219219 )
@@ -231,14 +231,14 @@ def test_ast_chunking_typescript_respects_top_level_brace_blocks() -> None:
231231 chunking_strategy = "ast" ,
232232 chunk_size = 500 ,
233233 chunk_overlap = 100 ,
234- target_tokens = 40 ,
234+ target_tokens = 64 ,
235235 overlap_tokens = 0 ,
236236 min_chunk_chars = 10 ,
237237 )
238238 tok_cfg = TokenizationConfig (strategy = "whitespace" , normalize_unicode = False , lowercase = False )
239239 ch = Chunker (cfg , tok_cfg )
240240
241- toks = " " .join ([f"w{ i } " for i in range (18 )])
241+ toks = " " .join ([f"w{ i } " for i in range (45 )])
242242 content = (
243243 "export function foo() {\n "
244244 f" // { toks } \n "
0 commit comments