[Refactor] scheduler: Drop the dead Req.decoded_text and BatchTokenIDOutput.decoded_texts

python/sglang/srt/managers/detokenizer_manager.py

@@ -317,7 +317,7 @@ def _decode_batch_token_id_output(self, recv_obj: BatchTokenIDOutput):
             rid = recv_obj.rids[i]
             if rid not in self.decode_status:
                 s = DecodeStatus(
-                    decoded_text=recv_obj.decoded_texts[i],
+                    decoded_text="",
                     decode_ids=self._clamp_decode_ids(
                         recv_obj.decode_ids[i], vocab_size
                     ),

python/sglang/srt/managers/io_struct.py

@@ -1629,7 +1629,6 @@ class BatchTokenIDOutput(BaseBatchReq, kw_only=True):
     # The finish reason
     finished_reasons: List[Optional[FinishReasonDict]]
     # For incremental decoding
-    decoded_texts: List[str]
     decode_ids: List[array]  # List[array[int]]
     read_offsets: List[int]
     # Only used when `--skip-tokenizer-init` is on

python/sglang/srt/managers/multi_tokenizer_mixin.py

@@ -191,7 +191,6 @@ def _handle_output_by_index(output, i):
             ),
             time_stats=_extract_field_by_index(output, "time_stats", i),
             finished_reasons=_extract_field_by_index(output, "finished_reasons", i),
-            decoded_texts=_extract_field_by_index(output, "decoded_texts", i),
             decode_ids=_extract_field_by_index(output, "decode_ids", i),
             read_offsets=_extract_field_by_index(output, "read_offsets", i),
             output_ids=_extract_field_by_index(output, "output_ids", i),

python/sglang/srt/managers/schedule_batch.py

@@ -1215,7 +1215,6 @@ def __init__(
         # 3: last token
         self.surr_offset = None  # Surrounding offset to defeat the cleanup algorithm
         self.read_offset = None
-        self.decoded_text = ""
 
         # For multimodal inputs
         self.multimodal_inputs: Optional[MultimodalInputs] = None
@@ -1930,13 +1929,11 @@ def _check_str_based_finish(self, new_accepted_len: int = 1):
             # Check stop strings
             if len(self.sampling_params.stop_strs) > 0:
                 for stop_str in self.sampling_params.stop_strs:
-                    stop_str_in_tail = stop_str in tail_str
-                    if stop_str_in_tail or stop_str in self.decoded_text:
+                    if stop_str in tail_str:
                         self.finished_reason = FINISH_MATCHED_STR(matched=stop_str)
-                        if stop_str_in_tail:
-                            self.finished_len = self._locate_str_stop_finished_len(
-                                new_accepted_len, stop_str=stop_str
-                            )
+                        self.finished_len = self._locate_str_stop_finished_len(
+                            new_accepted_len, stop_str=stop_str
+                        )
                         return True
 
             # Check stop regex

python/sglang/srt/managers/scheduler_components/output_streamer.py

@@ -357,7 +357,6 @@ class _GenerationStreamAccumulator:
     output_reqs: list[Req] = field(default_factory=list)
     http_worker_ipcs: list = field(default_factory=list)
     finished_reasons: list = field(default_factory=list)
-    decoded_texts: list = field(default_factory=list)
     decode_ids_list: list = field(default_factory=list)
     read_offsets: list = field(default_factory=list)
     output_ids: list = field(default_factory=list)
@@ -515,7 +514,6 @@ def accept(self, *, req: Req) -> None:
             # block. The parallel lists stay empty; the payload goes straight to
             # `push_generation`, which only indexes the fields appended above.
             self.http_worker_ipcs.append(req.http_worker_ipc)
-            self.decoded_texts.append(req.decoded_text)
             decode_ids, read_offset = req.init_incremental_detokenize()
             self.decode_ids_list.append(decode_ids[req.send_decode_id_offset :])
             req.send_decode_id_offset = len(decode_ids)
@@ -738,7 +736,6 @@ def to_payload(
             spec_cap_lens_histogram=self.spec_cap_lens_histogram,
             time_stats=self.time_stats or None,
             finished_reasons=self.finished_reasons,
-            decoded_texts=self.decoded_texts,
             decode_ids=self.decode_ids_list,
             read_offsets=self.read_offsets,
             output_ids=self.output_ids,

test/registered/unit/managers/test_flat_raw_top_logprobs.py

@@ -730,7 +730,6 @@ def _make_batch_token_id_output(**overrides) -> BatchTokenIDOutput:
     fields = dict(
         rids=["r0", "r1"],
         finished_reasons=[None] * n,
-        decoded_texts=["", ""],
         decode_ids=[array("q", [1]), array("q", [2])],
         read_offsets=[0] * n,
         output_ids=None,

test/registered/unit/managers/test_multi_tokenizer_mixin.py

@@ -120,7 +120,6 @@ def _make_batch_token_id_output() -> BatchTokenIDOutput:
         spec_num_cap_tokens=[],
         spec_cap_lens_histogram=[],
         finished_reasons=[None, None],
-        decoded_texts=["first", "second"],
         decode_ids=[[1], [2]],
         read_offsets=[0, 1],
         output_ids=[[1], [2]],
@@ -376,7 +375,7 @@ def test_router_groups_token_output_without_speculative_decoding(self):
         self.assertEqual(len(sent), 1)
         self.assertEqual(sent[0][0], "detok-a")
         self.assertEqual(sent[0][1].rids, ["token-rid-0", "token-rid-1"])
-        self.assertEqual(sent[0][1].decoded_texts, ["first", "second"])
+        self.assertEqual(sent[0][1].prompt_tokens, [10, 20])
         self.assertIsNone(sent[0][1].spec_verify_ct)
 
     def test_embedding_split_preserves_per_request_fields(self):

test/registered/unit/managers/test_output_streamer_customized_info.py

@@ -72,7 +72,6 @@ def __init__(
         self.send_token_offset = 0
         self.send_output_token_logprobs_offset = 0
         self.send_decode_id_offset = 0
-        self.decoded_text = ""
         self.origin_input_ids = []
         self.reasoning_tokens = 0
         self.cached_tokens = 0

test/registered/unit/managers/test_output_streamer_logprobs.py

@@ -31,7 +31,6 @@ def __init__(self, *, is_retracted: bool, max_new_tokens: int):
         self.send_token_offset = 0
         self.send_output_token_logprobs_offset = 0
         self.send_decode_id_offset = 0
-        self.decoded_text = ""
         self.origin_input_ids = []
         self.reasoning_tokens = 0
         self.cached_tokens = 0

test/registered/unit/managers/test_stop_str_speculative.py

@@ -145,16 +145,6 @@ def test_stop_regex_end_anchored_midchunk(self):
         self.assertEqual(req.finished_reason.matched, r"[.!?]\s*$")
         self.assertEqual(req.finished_len, 2)
 
-    # --- decoded_text-only branch ---
-    def test_stop_str_only_in_decoded_text_sets_no_finished_len(self):
-        # Documents current behavior: when the stop is only in decoded_text (not
-        # the tail), the request finishes but finished_len is left unset.
-        req = _make_req([10, 11, 12], stop=["STOP"])  # no STOP in output tokens
-        req.decoded_text = "earlier STOP text"
-        req.update_finish_state(new_accepted_len=3)
-        self.assertTrue(req.finished())
-        self.assertIsNone(req.finished_len)
-
 
 if __name__ == "__main__":
     unittest.main()