[Refactor] Use shared utils in hermes tool parser (#42570)

Signed-off-by: sfeng33 <4florafeng@gmail.com>
This commit is contained in:
Flora Feng
2026-05-13 20:35:45 -04:00
committed by GitHub
parent 63cc8a55a9
commit 1087676a90
+11 -34
View File
@@ -25,33 +25,20 @@ from vllm.tool_parsers.abstract_tool_parser import (
Tool,
ToolParser,
)
from vllm.tool_parsers.utils import is_complete_json, partial_tag_overlap
from vllm.utils.mistral import is_mistral_tokenizer
logger = init_logger(__name__)
def _partial_tag_overlap(text: str, tag: str) -> int:
"""Length of the longest prefix of `tag` that matches a suffix of `text`.
E.g. text ending in "<tool_" returns 6 when tag is "<tool_call>".
Returns 0 if there is no overlap.
"""
max_check = min(len(tag) - 1, len(text))
for k in range(max_check, 0, -1):
if text.endswith(tag[:k]):
return k
return 0
def _is_valid_json(text: str) -> bool:
try:
json.loads(text)
return True
except (json.JSONDecodeError, ValueError):
return False
class Hermes2ProToolParser(ToolParser):
tool_call_start_token: str = "<tool_call>"
tool_call_end_token: str = "</tool_call>"
tool_call_regex = re.compile(
r"<tool_call>(.*?)</tool_call>|<tool_call>(.*)", re.DOTALL
)
scratch_pad_regex = re.compile(r"<scratch_pad>(.*?)</scratch_pad>", re.DOTALL)
def __init__(self, tokenizer: TokenizerLike, tools: list[Tool] | None = None):
super().__init__(tokenizer, tools)
@@ -59,16 +46,6 @@ class Hermes2ProToolParser(ToolParser):
logger.error("Detected Mistral tokenizer when using a Hermes model")
self.model_tokenizer = tokenizer.tokenizer
self.tool_call_start_token: str = "<tool_call>"
self.tool_call_end_token: str = "</tool_call>"
self.tool_call_regex = re.compile(
r"<tool_call>(.*?)</tool_call>|<tool_call>(.*)", re.DOTALL
)
self.scratch_pad_regex = re.compile(
r"<scratch_pad>(.*?)</scratch_pad>", re.DOTALL
)
if not self.model_tokenizer:
raise ValueError(
"The model tokenizer must be passed to the ToolParser "
@@ -147,7 +124,7 @@ class Hermes2ProToolParser(ToolParser):
Holds back any suffix that could be a partial <tool_call> tag.
"""
if self.tool_call_start_token not in current_text:
overlap_length = _partial_tag_overlap(
overlap_length = partial_tag_overlap(
current_text, self.tool_call_start_token
)
sendable_idx = len(current_text) - overlap_length
@@ -176,13 +153,13 @@ class Hermes2ProToolParser(ToolParser):
else:
raw = text[json_start:]
# Strip partial </tool_call> suffix if present.
overlap = _partial_tag_overlap(raw, self.tool_call_end_token)
overlap = partial_tag_overlap(raw, self.tool_call_end_token)
if overlap:
raw = raw[:-overlap]
tc_json = raw.strip()
# Valid JSON without closing tag = complete body,
# tag tokens just haven't arrived yet.
is_complete = _is_valid_json(tc_json) if tc_json else False
is_complete = is_complete_json(tc_json) if tc_json else False
results.append((tc_json, is_complete))
break
return results