qwen3xml_tool_parser.py 51.6 KB
Newer Older
1
2
3
4
5
# SPDX-License-Identifier: Apache-2.0
# SPDX-FileCopyrightText: Copyright contributors to the vLLM project
import ast
import json
from collections.abc import Sequence
6
from typing import Any
7
8
9
10
from xml.parsers.expat import ParserCreate

import regex as re

11
from vllm.entrypoints.chat_utils import make_tool_call_id
12
from vllm.entrypoints.openai.chat_completion.protocol import (
13
    ChatCompletionRequest,
14
15
)
from vllm.entrypoints.openai.engine.protocol import (
16
17
18
19
20
21
22
    DeltaFunctionCall,
    DeltaMessage,
    DeltaToolCall,
    ExtractedToolCallInformation,
    FunctionCall,
    ToolCall,
)
23
from vllm.logger import init_logger
24
from vllm.tokenizers import TokenizerLike
25
from vllm.tool_parsers.abstract_tool_parser import (
26
    Tool,
27
28
    ToolParser,
)
29
from vllm.tool_parsers.utils import find_tool_properties
30
31
32
33
34
35
36
37
38
39
40
41
42
43

logger = init_logger(__name__)


class StreamingXMLToolCallParser:
    """
    Simplified streaming XML tool call parser
    Supports streaming input, parsing, and output
    """

    def __init__(self):
        self.reset_streaming_state()

        # Tool configuration information
44
        self.tools: list[Tool] | None = None
45
46
47
48
49
50
        self.tool_call_start_token: str = "<tool_call>"
        self.tool_call_end_token: str = "</tool_call>"
        self.function_start_token: str = "<function="
        self.function_end_token: str = "</function>"
        self.parameter_start_token: str = "<parameter="
        self.parameter_end_token: str = "</parameter>"
51
52
53
54
55
56
57
58
59
60
61
62
63

    def reset_streaming_state(self):
        """Reset streaming parsing state"""

        self.deltas = []
        # state for streaming
        self.tool_call_index = 0
        self.current_call_id = None
        self.last_completed_call_id = None
        self.current_function_name = None
        self.current_function_open = False
        self.parameters = {}
        self.current_param_name = None
64
65
        self.current_param_value = ""
        self.current_param_value_converted = ""
66
67
68
69
        self.current_param_is_first = False
        self.should_emit_end_newline = False
        self.start_quote_emitted = False

70
        self.streaming_buffer = ""
71
72
        self.last_processed_pos = 0

73
        self.text_content_buffer = ""
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88

        # state for preprocessing and deferred parsing
        self._pre_inside_parameter = False
        self._pre_param_buffer = ""
        self._pre_current_param_name = None
        self.defer_current_parameter = False
        self.deferred_param_raw_value = ""

        # recreate parser
        self.parser = ParserCreate()
        self.setup_parser()

    def parse_single_streaming_chunks(self, xml_chunk: str) -> DeltaMessage:
        """
        Parse single streaming XML chunk and return Delta response
89
        This is the actual streaming interface that receives chunks
90
91
92
93
94
        one by one and maintains internal state

        Args:
            xml_chunk: Single XML chunk string
        Returns:
95
            DeltaMessage: Contains delta information generated by this chunk,
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
            returns empty response if no complete elements
        """
        # Record delta count before processing
        initial_delta_count = len(self.deltas)

        self.streaming_buffer += xml_chunk

        found_elements = self._process_complete_xml_elements()

        if found_elements:
            # If complete elements found, check if end events were missed
            # some tags may not have been triggered
            try:
                new_deltas = self.deltas[initial_delta_count:]
                # If this chunk contains </function>
                # but didn't generate '}', then complete it
112
113
114
115
                if (
                    self.current_call_id is not None
                    and self.function_end_token in xml_chunk
                ):
116
117
                    # - Added '}' (non-empty parameter ending)
                    # - Added '{}' (empty parameter function)
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
                    has_function_close = any(
                        (
                            td.tool_calls
                            and any(
                                (
                                    tc.function
                                    and tc.id == self.current_call_id
                                    and isinstance(tc.function.arguments, str)
                                    and (tc.function.arguments in ("}", "{}"))
                                )
                                for tc in td.tool_calls
                            )
                        )
                        for td in new_deltas
                    )
133
134
135
                    if not has_function_close:
                        # Close potentially unclosed element
                        if self.current_param_name:
136
                            self._end_element("parameter")
137
                        if self.current_function_name:
138
                            self._end_element("function")
139
140
                # If this chunk contains </tool_call>
                # but didn't generate final empty delta, then complete it
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
                if (
                    self.current_call_id is not None
                    and self.tool_call_end_token in xml_chunk
                ):
                    has_toolcall_close = any(
                        (
                            td.tool_calls
                            and any(
                                (
                                    tc.type == "function"
                                    and tc.function
                                    and tc.function.arguments == ""
                                    and tc.id == self.current_call_id
                                )
                                for tc in td.tool_calls
                            )
                        )
                        for td in new_deltas
                    )
160
161
162
                    if not has_toolcall_close:
                        # Close potentially unclosed element
                        if self.current_param_name:
163
                            self._end_element("parameter")
164
                        if self.current_function_name:
165
166
                            self._end_element("function")
                        self._end_element("tool_call")
167
168
169
170
            except Exception as e:
                logger.warning("Error with fallback parsing: %s", e)
            # Merge newly generated deltas into single response
            result_delta = self._merge_new_deltas_to_single_response(
171
172
                initial_delta_count
            )
173
174
175
176
177
178
179
180
            return result_delta
        else:
            # No complete elements, check if there's unoutput text content
            if self.text_content_buffer and self.tool_call_index == 0:
                # Has text content but no tool_call yet, output text content
                text_delta = DeltaMessage(content=self.text_content_buffer)
                self._emit_delta(text_delta)
                # Clear buffer to avoid duplicate output
181
                self.text_content_buffer = ""
182
183
184
185
186
187
188
                return text_delta

            # If this chunk contains end tags but wasn't triggered by parser,
            # manually complete end events
            # Only execute when still on the same call as when entered,
            # to prevent accidentally closing new calls
            # in multi <tool_call> scenarios
189
190
191
192
            if self.current_call_id is not None and (
                self.function_end_token in xml_chunk
                or self.tool_call_end_token in xml_chunk
            ):
193
194
                # Close potentially unclosed element
                if self.current_param_name:
195
196
197
                    self._end_element("parameter")
                if self.function_end_token in xml_chunk and self.current_function_name:
                    self._end_element("function")
198
                if self.tool_call_end_token in xml_chunk:
199
                    self._end_element("tool_call")
200
201
                # Return the merged delta result generated by this fallback
                result_delta = self._merge_new_deltas_to_single_response(
202
203
                    initial_delta_count
                )
204
205
206
207
208
209
210
211
212
213
214
215
216
217
                return result_delta

            # No complete elements, return empty response
            return DeltaMessage(content=None)

    def _escape_xml_special_chars(self, text: str) -> str:
        """
        Escape XML special characters
        Args:
            text: Original text
        Returns:
            Escaped text
        """
        xml_escapes = {
218
219
220
221
222
            "&": "&amp;",
            "<": "&lt;",
            ">": "&gt;",
            '"': "&quot;",
            "'": "&apos;",
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
        }

        for char, escape in xml_escapes.items():
            text = text.replace(char, escape)

        return text

    def _process_complete_xml_elements(self) -> bool:
        """
        Process complete XML elements in buffer

        Returns:
            bool: Whether complete elements were found and processed
        """
        found_any = False

        while self.last_processed_pos < len(self.streaming_buffer):
            # Find next complete xml element
241
            element, end_pos = self._find_next_complete_element(self.last_processed_pos)
242
243
244
245
246
247
248
249
250
251
252
253
254
            if element is None:
                # No complete element found, wait for more data
                break

            # Check if this element should be skipped
            if self._should_skip_element(element):
                self.last_processed_pos = end_pos
                continue

            # Found complete XML element, process it
            try:
                preprocessed_element = self._preprocess_xml_chunk(element)
                # Check if this is the first tool_call start
255
256
257
258
259
260
261
                if (
                    (
                        preprocessed_element.strip().startswith("<tool_call>")
                        or preprocessed_element.strip().startswith("<function name=")
                    )
                    and self.tool_call_index == 0
                ) and self.text_content_buffer:
262
263
264
265
266
                    # First tool_call starts,
                    # output previously collected text content first
                    text_delta = DeltaMessage(content=self.text_content_buffer)
                    self._emit_delta(text_delta)
                    # Clear buffer for potential subsequent text content
267
                    self.text_content_buffer = ""
268
269
270

                # If a new tool_call starts and
                # there are already completed tool_calls
271
272
273
274
275
                if (
                    preprocessed_element.strip().startswith("<tool_call>")
                    and self.tool_call_index > 0
                    and self.current_call_id
                ):
276
277
                    # Reset parser state but preserve generated deltas
                    if self.current_param_name:
278
                        self._end_element("parameter")
279
                    if self.current_function_open or self.current_function_name:
280
                        self._end_element("function")
281
282
283
284
                    # Output final tool_call tail delta
                    final_delta = DeltaMessage(
                        role=None,
                        content=None,
285
                        reasoning=None,
286
                        tool_calls=[
287
288
289
290
291
292
293
294
                            DeltaToolCall(
                                index=self.tool_call_index - 1,
                                id=self.current_call_id,
                                type="function",
                                function=DeltaFunctionCall(name=None, arguments=""),
                            )
                        ],
                    )
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
                    self._emit_delta(final_delta)
                    # Reset XML parser and current call state
                    self._reset_xml_parser_after_tool_call()
                # Parse preprocessed element
                self.parser.Parse(preprocessed_element, False)
                found_any = True

            except Exception as e:
                logger.warning("Error when parsing XML elements: %s", e)

            # Update processed position
            self.last_processed_pos = end_pos

        return found_any

    def _should_skip_element(self, element: str) -> bool:
        """
        Determine whether an element should be skipped

        Args:
            element: Element to evaluate

        Returns:
            bool: True means should skip, False means should process
        """

        # If it's a tool_call XML tag, don't skip
322
323
324
325
326
        if (
            element.startswith(self.tool_call_start_token)
            or element.startswith(self.function_start_token)
            or element.startswith(self.parameter_start_token)
        ):
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
            return False

        # If currently not parsing tool calls and not blank,
        # collect this text instead of skipping
        # Only process other XML elements after tool_call appears,
        # otherwise treat as plain text
        if self.current_call_id is None and element:
            # Collect text content to buffer
            self.text_content_buffer += element
            return True  # Still skip, but content has been collected

        # If currently parsing tool calls,
        # this might be parameter value, don't skip
        if self.current_call_id is not None:
            return False

        # Skip blank content
        return not element

346
    def _find_next_complete_element(self, start_pos: int) -> tuple[str | None, int]:
347
348
349
350
351
352
353
        """
        Find next complete XML element from specified position

        Args:
            start_pos: Position to start searching

        Returns:
354
            (Complete element string, element end position),
355
356
357
358
359
360
361
            returns (None, start_pos) if no complete element found
        """
        buffer = self.streaming_buffer[start_pos:]

        if not buffer:
            return None, start_pos

362
        if buffer.startswith("<"):
363
364
            # Need to ensure no new < appears,
            # find the nearest one between < and >
365
366
            tag_end = buffer.find("<", 1)
            tag_end2 = buffer.find(">", 1)
367
368
369
370
371
372
            if tag_end != -1 and tag_end2 != -1:
                # Next nearest is <
                if tag_end < tag_end2:
                    return buffer[:tag_end], start_pos + tag_end
                # Next nearest is >, means found XML element
                else:
373
                    return buffer[: tag_end2 + 1], start_pos + tag_end2 + 1
374
375
376
            elif tag_end != -1:
                return buffer[:tag_end], start_pos + tag_end
            elif tag_end2 != -1:
377
                return buffer[: tag_end2 + 1], start_pos + tag_end2 + 1
378
379
            else:
                # If currently not parsing tool calls (entering a tool_call),
380
                # check if starts with <tool_call> or <function=
381
382
                if self.current_call_id is None:
                    # Check if might be start of <tool_call>
383
                    if buffer == "<tool_call>"[: len(buffer)]:
384
385
                        # Might be start of <tool_call>, wait for more data
                        return None, start_pos
386
387
388
389
390
391
392
                    elif (
                        buffer.startswith("<function=")
                        or buffer == "<function="[: len(buffer)]
                    ):
                        # Might be start of <function=, wait for more data
                        # to get the complete function tag
                        return None, start_pos
393
                    else:
394
                        # Not start of <tool_call> or <function=, treat as text
395
396
397
398
399
400
401
                        return buffer, start_pos + len(buffer)
                else:
                    # When parsing tool calls,
                    # wait for more data to get complete tag
                    return None, start_pos
        else:
            # Find text content (until next < or buffer end)
402
            next_tag_pos = buffer.find("<")
403
404
405
406
407
408
409
410
411
412
            if next_tag_pos != -1:
                # Found text content
                text_content = buffer[:next_tag_pos]
                return text_content, start_pos + next_tag_pos
            else:
                # Buffer end is all text, process
                # (no longer wait for more data)
                remaining = buffer
                return remaining, start_pos + len(remaining)

413
    def _merge_new_deltas_to_single_response(self, initial_count: int) -> DeltaMessage:
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
        """
        Merge newly generated deltas from this processing
        into a single DeltaMessage

        Args:
            initial_count: Delta count before processing

        Returns:
            Merged DeltaMessage containing all newly generated delta information
        """
        if len(self.deltas) <= initial_count:
            return DeltaMessage(content=None)

        # Get newly generated deltas
        new_deltas = self.deltas[initial_count:]

        if len(new_deltas) == 1:
            # Only one new delta, return directly
            return new_deltas[0]

        # Merge multiple new deltas
        merged_tool_calls: list[DeltaToolCall] = []
436
        merged_content: str = ""
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453

        for delta in new_deltas:
            if delta.content:
                merged_content += delta.content
            if delta.tool_calls:
                # For tool_calls, we need to intelligently merge arguments
                for tool_call in delta.tool_calls:
                    # Find if there's already a tool_call with the same call_id
                    existing_call = None
                    for existing in merged_tool_calls:
                        if existing.id == tool_call.id:
                            existing_call = existing
                            break

                    if existing_call and existing_call.function:
                        # Merge to existing tool_call
                        if tool_call.function and tool_call.function.name:
454
455
456
457
458
                            existing_call.function.name = tool_call.function.name
                        if (
                            tool_call.function
                            and tool_call.function.arguments is not None
                        ):
459
                            if existing_call.function.arguments is None:
460
                                existing_call.function.arguments = ""
461
462
463
464
465
466
467
468
469
470
471

                            # For streaming JSON parameters,
                            # simply concatenate in order
                            new_args = tool_call.function.arguments
                            existing_call.function.arguments += new_args
                        if tool_call.type:
                            existing_call.type = tool_call.type
                    else:
                        # Add new tool_call
                        merged_tool_calls.append(tool_call)

472
473
474
475
        return DeltaMessage(
            content=merged_content if merged_content else None,
            tool_calls=merged_tool_calls,
        )
476
477
478

    def _preprocess_xml_chunk(self, chunk: str) -> str:
        """
479
        Preprocess XML chunk, handle non-standard formats,
480
481
482
483
484
485
486
487
488
489
490
491
        and escape special characters

        Args:
            chunk: Original XML chunk

        Returns:
            Processed XML chunk
        """

        # Check if this is a tool_call related element
        is_tool_call = False
        if chunk.startswith(self.tool_call_start_token) or chunk.startswith(
492
493
            self.tool_call_end_token
        ):
494
495
            is_tool_call = True
        if chunk.startswith(self.function_start_token) or chunk.startswith(
496
497
            self.function_end_token
        ):
498
499
            is_tool_call = True
        if chunk.startswith(self.parameter_start_token) or chunk.startswith(
500
501
            self.parameter_end_token
        ):
502
503
            is_tool_call = True
        # Handle <function=name> format -> <function name="name">
504
        processed = re.sub(r"<function=([^>]+)>", r'<function name="\1">', chunk)
505
        # Handle <parameter=name> format -> <parameter name="name">
506
        processed = re.sub(r"<parameter=([^>]+)>", r'<parameter name="\1">', processed)
507
508
509
510
511
512

        original_chunk = chunk
        # If in parameter value accumulation mode
        if self._pre_inside_parameter:
            # Parameter end: output accumulated raw text
            # safely then return </parameter>
513
            if processed.startswith("</parameter>"):
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
                body_text = self._pre_param_buffer
                # Trigger deferred parsing mode
                # literal_eval+json output in end_element
                self.defer_current_parameter = True
                self.deferred_param_raw_value = body_text
                # Clean up state
                self._pre_inside_parameter = False
                self._pre_param_buffer = ""
                self._pre_current_param_name = None
                safe_text = self._escape_xml_special_chars(body_text)
                return f"{safe_text}</parameter>"
            else:
                # If this is the first block of content after entering parameter
                # evaluate if deferred parsing is needed;
                # If not needed, exit accumulation mode
                # and pass through directly
                if self._pre_param_buffer == "":
                    # Get current parameter type
532
533
534
535
536
                    param_type = (
                        self._get_param_type(self._pre_current_param_name)
                        if self._pre_current_param_name
                        else "string"
                    )
537
538
539
                    # Only these types need deferred parsing to
                    # handle Python literals containing single quotes
                    is_object_type = param_type in ["object"]
540
541
542
543
544
                    is_complex_type = (
                        param_type in ["array", "arr", "sequence"]
                        or param_type.startswith("dict")
                        or param_type.startswith("list")
                    )
545
546
547

                    # Only delay when contains container symbols
                    # and has single quotes and is complex type
548
549
550
551
552
                    has_container_hint = (
                        ("[" in original_chunk)
                        or ("{" in original_chunk)
                        or ("(" in original_chunk)
                    )
553
554
555
556
557
558

                    # Determine if deferred parsing is needed
                    need_defer = False
                    if is_complex_type:
                        # Complex type, always need deferred parsing
                        need_defer = True
559
560
561
562
563
                    elif (
                        is_object_type
                        and has_container_hint
                        and ("'" in original_chunk)
                    ):
564
565
566
567
568
569
570
571
572
573
574
575
576
                        # Object type with container symbols
                        # and single quotes, need deferred parsing
                        need_defer = True

                    if not need_defer:
                        # No need for deferred parsing,
                        # exit parameter mode directly
                        self._pre_inside_parameter = False
                        return self._escape_xml_special_chars(original_chunk)
                self._pre_param_buffer += original_chunk
                return ""

        # Parameter start: enable accumulation
577
        if processed.startswith("<parameter name="):
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
            m = re.match(r'<parameter name="([^"]+)">', processed)
            if m:
                self._pre_current_param_name = m.group(1)
            self._pre_inside_parameter = True
            self._pre_param_buffer = ""
            return processed

        # If processed doesn't contain special_token, escape processed
        # This is because XML parsing encounters special characters
        # and reports errors, so escaping is needed
        if not is_tool_call:
            processed = self._escape_xml_special_chars(processed)
        return processed

    def _emit_delta(self, delta: DeltaMessage):
        """Emit Delta response (streaming output)"""
        self.deltas.append(delta)

596
    def _auto_close_open_parameter_if_needed(self, incoming_tag: str | None = None):
597
598
        """Before starting to process new elements,
        if there are unclosed tags from before,
599
        automatically complete their endings to the parser.
600
        - If there are unclosed parameters,
601
        it's equivalent to feeding `</parameter>`
602
        - When about to start a new function or tool_call,
603
        if there are unclosed functions, complete `</function>`.
604
        - When about to start a new tool_call,
605
606
607
608
        if there are unclosed tool_calls, complete `</tool_call>`.
        """
        # First close unclosed parameters
        if self.current_param_name:
609
            self._end_element("parameter")
610
611
612

        # If about to start new function or tool_call,
        # and there are unclosed functions, close function first
613
614
        if incoming_tag in ("function", "tool_call") and self.current_function_name:
            self._end_element("function")
615
616
617

        # If about to start new tool_call,
        # and there are unclosed tool_calls, close tool_call first
618
619
        if incoming_tag == "tool_call" and self.current_call_id:
            self._end_element("tool_call")
620
621
622
623

    def _start_element(self, name: str, attrs: dict[str, str]):
        """Handle XML start element events"""

624
        if name == "root":
625
626
            return

627
        if name == "tool_call":
628
629
            # Before opening new tool_call,
            # automatically complete previous unclosed tags
630
            self._auto_close_open_parameter_if_needed("tool_call")
631
632

            self.parameters = {}
633
            self.current_call_id = make_tool_call_id()
634
635
            self.current_param_is_first = True
            self.tool_call_index += 1
636
        elif name.startswith("function") or (name == "function"):
637
638
            # If missing tool_call, manually complete
            if not self.current_call_id:
639
                self._start_element("tool_call", {})
640
641
            # Before opening new function,
            # automatically complete previous unclosed tags (parameter/function)
642
            self._auto_close_open_parameter_if_needed("function")
643
644
645
646
            function_name = self._extract_function_name(name, attrs)
            self.current_function_name = function_name
            self.current_function_open = True
            if function_name:
647
648
649
650
651
652
653
654
655
656
657
658
                delta = DeltaMessage(
                    tool_calls=[
                        DeltaToolCall(
                            index=self.tool_call_index - 1,
                            id=self.current_call_id,
                            type="function",
                            function=DeltaFunctionCall(
                                name=function_name, arguments=""
                            ),
                        )
                    ]
                )
659
                self._emit_delta(delta)
660
        elif name.startswith("parameter") or (name == "parameter"):
661
662
            # If previous parameter hasn't ended normally,
            # complete its end first, then start new parameter
663
            self._auto_close_open_parameter_if_needed("parameter")
664
665
            param_name = self._extract_parameter_name(name, attrs)
            self.current_param_name = param_name
666
667
            self.current_param_value = ""
            self.current_param_value_converted = ""
668
669
670
671
672
673
674
675
676
677
            self.start_quote_emitted = False  # Reset start quote flag

            # Only output parameter name and colon,
            # don't output quotes
            # decide after parameter value type is determined
            if param_name:
                if not self.parameters:
                    # First parameter
                    # start JSON, only output parameter name and colon
                    json_start = f'{{"{param_name}": '
678
679
680
681
682
683
684
685
686
687
688
689
                    delta = DeltaMessage(
                        tool_calls=[
                            DeltaToolCall(
                                index=self.tool_call_index - 1,
                                id=self.current_call_id,
                                type="function",
                                function=DeltaFunctionCall(
                                    name=None, arguments=json_start
                                ),
                            )
                        ]
                    )
690
691
692
693
694
695
                    self._emit_delta(delta)
                    self.current_param_is_first = True
                else:
                    # Subsequent parameters
                    # add comma and parameter name, no quotes
                    json_continue = f', "{param_name}": '
696
697
698
699
700
701
702
703
704
705
706
707
                    delta = DeltaMessage(
                        tool_calls=[
                            DeltaToolCall(
                                index=self.tool_call_index - 1,
                                id=self.current_call_id,
                                type="function",
                                function=DeltaFunctionCall(
                                    name=None, arguments=json_continue
                                ),
                            )
                        ]
                    )
708
709
710
711
712
713
714
715
716
717
718
                    self._emit_delta(delta)
                    self.current_param_is_first = False

    def _char_data(self, data: str):
        """Handle XML character data events"""
        if data and self.current_param_name:
            # If preprocessing stage determines deferred parsing is needed,
            # only cache character data, no streaming output
            if self.defer_current_parameter:
                original_data = data
                if self.should_emit_end_newline:
719
                    original_data = "\n" + original_data
720
                    self.should_emit_end_newline = False
721
                if original_data.endswith("\n"):
722
723
724
725
726
727
728
729
730
                    self.should_emit_end_newline = True
                    original_data = original_data[:-1]
                self.current_param_value += original_data
                return

            param_type = self._get_param_type(self.current_param_name)

            # Check if this is the first time receiving data for this parameter
            # If this is the first packet of data and starts with \n, remove \n
731
            if not self.current_param_value and data.startswith("\n"):
732
733
734
                data = data[1:]

            # Output start quote for string type (if not already output)
735
736
737
738
739
740
741
742
743
744
745
746
747
748
            if (
                param_type in ["string", "str", "text", "varchar", "char", "enum"]
                and not self.start_quote_emitted
            ):
                quote_delta = DeltaMessage(
                    tool_calls=[
                        DeltaToolCall(
                            index=self.tool_call_index - 1,
                            id=self.current_call_id,
                            type="function",
                            function=DeltaFunctionCall(name=None, arguments='"'),
                        )
                    ]
                )
749
750
751
752
753
754
755
756
757
                self._emit_delta(quote_delta)
                self.start_quote_emitted = True

            if not data:
                return

            original_data = data
            # Delay output of trailing newline
            if self.should_emit_end_newline:
758
                original_data = "\n" + original_data
759
                self.should_emit_end_newline = False
760
            if original_data.endswith("\n"):
761
762
763
764
765
766
                self.should_emit_end_newline = True
                original_data = original_data[:-1]
            self.current_param_value += original_data

            # convert parameter value by param_type
            converted_value = self._convert_param_value(
767
768
769
                self.current_param_value, param_type
            )
            output_data = self._convert_for_json_streaming(converted_value, param_type)
770

771
            delta_data = output_data[len(self.current_param_value_converted) :]
772
773
            self.current_param_value_converted = output_data

774
775
776
777
778
779
780
781
782
783
            delta = DeltaMessage(
                tool_calls=[
                    DeltaToolCall(
                        index=self.tool_call_index - 1,
                        id=self.current_call_id,
                        type="function",
                        function=DeltaFunctionCall(name=None, arguments=delta_data),
                    )
                ]
            )
784
785
786
787
788
            self._emit_delta(delta)

    def _end_element(self, name: str):
        """Handle XML end element events"""

789
        if name == "root":
790
791
792
793
            return

        # If function or tool_call ends and there are still unclosed parameters,
        # complete parameter end first
794
795
796
        if (
            name.startswith("function") or name == "function" or name == "tool_call"
        ) and self.current_param_name:
797
798
            self._auto_close_open_parameter_if_needed()

799
800
801
        if (
            name.startswith("parameter") or name == "parameter"
        ) and self.current_param_name:
802
803
804
805
806
807
808
809
            # End current parameter
            param_name = self.current_param_name
            param_value = self.current_param_value

            # If in deferred parsing mode,
            # perform overall parsing on raw content
            # accumulated in preprocessing stage and output once
            if self.defer_current_parameter:
810
811
812
813
814
                raw_text = (
                    self.deferred_param_raw_value
                    if self.deferred_param_raw_value
                    else param_value
                )
815
816
817
818
819
820
                parsed_value = None
                output_arguments = None
                try:
                    # If previously delayed trailing newline,
                    # add it back before parsing
                    if self.should_emit_end_newline:
821
                        raw_for_parse = raw_text + "\n"
822
823
824
                    else:
                        raw_for_parse = raw_text
                    parsed_value = ast.literal_eval(raw_for_parse)
825
                    output_arguments = json.dumps(parsed_value, ensure_ascii=False)
826
827
828
829
830
                except Exception:
                    # Fallback: output as string as-is
                    output_arguments = json.dumps(raw_text, ensure_ascii=False)
                    parsed_value = raw_text

831
832
833
834
835
836
837
838
839
840
841
842
                delta = DeltaMessage(
                    tool_calls=[
                        DeltaToolCall(
                            index=self.tool_call_index - 1,
                            id=self.current_call_id,
                            type="function",
                            function=DeltaFunctionCall(
                                name=None, arguments=output_arguments
                            ),
                        )
                    ]
                )
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
                self._emit_delta(delta)

                # Clean up and store
                self.should_emit_end_newline = False
                self.parameters[param_name] = parsed_value
                self.current_param_name = None
                self.current_param_value = ""
                self.current_param_value_converted = ""
                self.start_quote_emitted = False
                self.defer_current_parameter = False
                self.deferred_param_raw_value = ""
                return

            param_type = self._get_param_type(param_name)

            # convert complete parameter value by param_type
859
            converted_value = self._convert_param_value(param_value, param_type)
860
861

            # Decide whether to add end quote based on parameter type
862
            if param_type in ["string", "str", "text", "varchar", "char", "enum"]:
863
864
865
866
                # For empty string parameters, need special handling
                if not param_value and not self.start_quote_emitted:
                    # No start quote output,
                    # directly output complete empty string
867
868
869
870
871
872
873
874
875
876
                    delta = DeltaMessage(
                        tool_calls=[
                            DeltaToolCall(
                                index=self.tool_call_index - 1,
                                id=self.current_call_id,
                                type="function",
                                function=DeltaFunctionCall(name=None, arguments='""'),
                            )
                        ]
                    )
877
878
879
                    self._emit_delta(delta)
                else:
                    # Non-empty parameter value, output end quote
880
881
882
883
884
885
886
887
888
889
                    delta = DeltaMessage(
                        tool_calls=[
                            DeltaToolCall(
                                index=self.tool_call_index - 1,
                                id=self.current_call_id,
                                type="function",
                                function=DeltaFunctionCall(name=None, arguments='"'),
                            )
                        ]
                    )
890
891
892
893
894
895
                    self._emit_delta(delta)

            self.should_emit_end_newline = False
            # Store converted value
            self.parameters[param_name] = converted_value
            self.current_param_name = None
896
897
            self.current_param_value = ""
            self.current_param_value_converted = ""
898
899
            self.start_quote_emitted = False

900
        elif name.startswith("function") or name == "function":
901
902
            # if there are parameters, close JSON object
            if self.parameters:
903
904
905
906
907
908
909
910
911
912
                delta = DeltaMessage(
                    tool_calls=[
                        DeltaToolCall(
                            index=self.tool_call_index - 1,
                            id=self.current_call_id,
                            type="function",
                            function=DeltaFunctionCall(name=None, arguments="}"),
                        )
                    ]
                )
913
914
915
                self._emit_delta(delta)
            # return empty object
            else:
916
917
918
919
920
921
922
923
924
925
                delta = DeltaMessage(
                    tool_calls=[
                        DeltaToolCall(
                            index=self.tool_call_index - 1,
                            id=self.current_call_id,
                            type="function",
                            function=DeltaFunctionCall(name=None, arguments="{}"),
                        )
                    ]
                )
926
927
928
                self._emit_delta(delta)
            self.current_function_open = False

929
        elif name == "tool_call":
930
931
932
933
934
            # Before ending tool_call,
            # ensure function is closed to complete missing right brace
            if self.current_function_open:
                # If there are still unclosed parameters, close them first
                if self.current_param_name:
935
                    self._end_element("parameter")
936
                # Close function, ensure output '}' or '{}'
937
                self._end_element("function")
938
            # Final Delta
939
940
941
942
943
944
945
946
947
948
            delta = DeltaMessage(
                tool_calls=[
                    DeltaToolCall(
                        index=self.tool_call_index - 1,
                        id=self.current_call_id,
                        type="function",
                        function=DeltaFunctionCall(name=None, arguments=""),
                    )
                ]
            )
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
            self._emit_delta(delta)

            # Check if there's text content to output (between tool_calls)
            if self.text_content_buffer.strip():
                text_delta = DeltaMessage(content=self.text_content_buffer)
                self._emit_delta(text_delta)

            self._reset_xml_parser_after_tool_call()

    def setup_parser(self):
        """Set up XML parser event handlers"""
        self.parser.buffer_text = True
        self.parser.StartElementHandler = self._start_element
        self.parser.EndElementHandler = self._end_element
        self.parser.CharacterDataHandler = self._char_data

965
    def set_tools(self, tools: list[Tool] | None):
966
967
968
        """Set tool configuration information"""
        self.tools = tools

969
    def _extract_function_name(self, name: str, attrs: dict[str, str]) -> str | None:
970
        """Extract function name from various formats"""
971
972
        if attrs and "name" in attrs:
            return attrs["name"]
973

974
975
976
        if "=" in name:
            parts = name.split("=", 1)
            if len(parts) == 2 and parts[0] == "function":
977
978
979
980
                return parts[1]

        return None

981
    def _extract_parameter_name(self, name: str, attrs: dict[str, str]) -> str | None:
982
        """Extract parameter name from various formats"""
983
984
        if attrs and "name" in attrs:
            return attrs["name"]
985

986
987
988
        if "=" in name:
            parts = name.split("=", 1)
            if len(parts) == 2 and parts[0] == "parameter":
989
990
991
992
993
994
995
996
997
998
999
1000
1001
                return parts[1]

        return None

    def _get_param_type(self, param_name: str) -> str:
        """Get parameter type based on tool configuration, defaults to string
        Args:
            param_name: Parameter name

        Returns:
            Parameter type
        """
        if not self.tools or not self.current_function_name:
1002
            return "string"
1003

1004
1005
1006
1007
1008
        properties = find_tool_properties(self.tools, self.current_function_name)
        if param_name in properties and isinstance(properties[param_name], dict):
            return self.repair_param_type(
                str(properties[param_name].get("type", "string"))
            )
1009
        return "string"
1010
1011
1012
1013
1014
1015
1016
1017
1018

    def repair_param_type(self, param_type: str) -> str:
        """Repair unknown parameter types by treating them as string
        Args:
            param_type: Parameter type

        Returns:
            Repaired parameter type
        """
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
        if (
            param_type in ["string", "str", "text", "varchar", "char", "enum"]
            or param_type.startswith("int")
            or param_type.startswith("uint")
            or param_type.startswith("long")
            or param_type.startswith("short")
            or param_type.startswith("unsigned")
            or param_type.startswith("num")
            or param_type.startswith("float")
            or param_type in ["boolean", "bool", "binary"]
            or (
                param_type in ["object", "array", "arr", "sequence"]
                or param_type.startswith("dict")
                or param_type.startswith("list")
            )
        ):
1035
1036
            return param_type
        else:
1037
            return "string"
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047

    def _convert_param_value(self, param_value: str, param_type: str) -> Any:
        """Convert value based on parameter type
        Args:
            param_value: Parameter value
            param_type: Parameter type

        Returns:
            Converted value
        """
1048
        if param_value.lower() == "null":
1049
1050
1051
            return None

        param_type = param_type.strip().lower()
1052
        if param_type in ["string", "str", "text", "varchar", "char", "enum"]:
1053
            return param_value
1054
1055
1056
1057
1058
1059
1060
        elif (
            param_type.startswith("int")
            or param_type.startswith("uint")
            or param_type.startswith("long")
            or param_type.startswith("short")
            or param_type.startswith("unsigned")
        ):
1061
1062
1063
1064
1065
            try:
                return int(param_value)
            except (ValueError, TypeError):
                logger.warning(
                    "Parsed value '%s' of parameter '%s' is not an integer "
1066
1067
1068
                    "in tool '%s', degenerating to string.",
                    param_value,
                )
1069
            return param_value
1070
        elif param_type.startswith("num") or param_type.startswith("float"):
1071
1072
            try:
                float_param_value: float = float(param_value)
1073
1074
1075
1076
1077
                return (
                    float_param_value
                    if float_param_value - int(float_param_value) != 0
                    else int(float_param_value)
                )
1078
1079
1080
            except (ValueError, TypeError):
                logger.warning(
                    "Parsed value '%s' of parameter '%s' is not a float "
1081
1082
1083
                    "in tool '%s', degenerating to string.",
                    param_value,
                )
1084
            return param_value
1085
        elif param_type in ["boolean", "bool", "binary"]:
1086
            param_value = param_value.lower()
1087
            return param_value == "true"
1088
1089
1090
        else:
            return param_value

1091
1092
    def _convert_for_json_streaming(self, converted_value: Any, param_type: str) -> str:
        """Convert converted_value based on
1093
1094
1095
1096
1097
1098
1099
1100
1101
        whether it's empty and if type is string
        Args:
            converted_value: Converted value
            param_type: Parameter type

        Returns:
            Converted string for streaming output
        """
        # Check if value is empty, but exclude numeric 0
1102
1103
        if converted_value is None or converted_value == "":
            return ""
1104

1105
        if param_type in ["string", "str", "text", "varchar", "char", "enum"]:
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
            # String type, remove double quotes
            return json.dumps(converted_value, ensure_ascii=False)[1:-1]
        else:
            # Non-string type, return complete JSON string
            if not isinstance(converted_value, str):
                return json.dumps(converted_value, ensure_ascii=False)
            else:
                return converted_value

    def _reset_xml_parser_after_tool_call(self):
        """
1117
        Each tool_call is treated as a separate XML document,
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
        so we need to reset the parser after each tool_call.
        """

        # recreate XML parser
        self.parser = ParserCreate()
        self.setup_parser()

        # Reset current tool_call state
        if self.current_call_id:
            self.last_completed_call_id = self.current_call_id
        self.current_call_id = None
        self.current_function_name = None
        self.current_function_open = False
        self.parameters = {}
        self.current_param_name = None
1133
1134
        self.current_param_value = ""
        self.current_param_value_converted = ""
1135
1136
1137
        self.current_param_is_first = False
        self.should_emit_end_newline = False
        self.start_quote_emitted = False
1138
        self.text_content_buffer = ""
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148

        # Reset preprocessing and deferred parsing state
        self._pre_inside_parameter = False
        self._pre_param_buffer = ""
        self._pre_current_param_name = None
        self.defer_current_parameter = False
        self.deferred_param_raw_value = ""


class Qwen3XMLToolParser(ToolParser):
1149
1150
    def __init__(self, tokenizer: TokenizerLike, tools: list[Tool] | None = None):
        super().__init__(tokenizer, tools)
1151
1152
        self.parser = StreamingXMLToolCallParser()

1153
1154
1155
1156
        # Add missing attributes for compatibility with serving_chat.py
        self.prev_tool_call_arr: list[dict] = []
        self.streamed_args_for_tool: list[str] = []

1157
1158
1159
        logger.info(
            "vLLM Successfully import tool parser %s !", self.__class__.__name__
        )
1160
1161
1162
1163
1164
1165
1166

    def extract_tool_calls(
        self,
        model_output: str,
        request: ChatCompletionRequest,
    ) -> ExtractedToolCallInformation:
        self.parser.reset_streaming_state()
1167
1168
1169
        # Reset tool call tracking arrays for new extraction
        self.prev_tool_call_arr = []
        self.streamed_args_for_tool = []
1170
        self.parser.set_tools(self.tools)
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
        result = self.parser.parse_single_streaming_chunks(model_output)
        if not result.tool_calls:
            return ExtractedToolCallInformation(
                tool_calls=[],
                tools_called=False,
                content=result.content,
            )
        else:
            tool_calls = []
            for tool_call in result.tool_calls:
                if tool_call.function and tool_call.function.name:
                    tool_calls.append(
                        ToolCall(
                            id=tool_call.id,
                            type=tool_call.type,
                            function=FunctionCall(
                                name=tool_call.function.name,
                                arguments=tool_call.function.arguments,
                            ),
1190
1191
                        )
                    )
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219

                    # Update tool call tracking arrays for compatibility
                    tool_index = (
                        tool_call.index
                        if tool_call.index is not None
                        else len(self.prev_tool_call_arr) - 1
                    )

                    # Ensure we have enough entries in our tracking arrays
                    while len(self.prev_tool_call_arr) <= tool_index:
                        self.prev_tool_call_arr.append({"name": "", "arguments": ""})
                    while len(self.streamed_args_for_tool) <= tool_index:
                        self.streamed_args_for_tool.append("")

                    # Update tool call information
                    self.prev_tool_call_arr[tool_index]["name"] = (
                        tool_call.function.name
                    )
                    self.prev_tool_call_arr[tool_index]["arguments"] = (
                        tool_call.function.arguments
                    )

                    # Update streamed arguments
                    if tool_call.function.arguments:
                        self.streamed_args_for_tool[tool_index] = (
                            tool_call.function.arguments
                        )

1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
            return ExtractedToolCallInformation(
                tool_calls=tool_calls,
                tools_called=len(tool_calls) > 0,
                content=result.content,
            )

    def extract_tool_calls_streaming(
        self,
        previous_text: str,
        current_text: str,
        delta_text: str,
        previous_token_ids: Sequence[int],
        current_token_ids: Sequence[int],
        delta_token_ids: Sequence[int],
        request: ChatCompletionRequest,
1235
    ) -> DeltaMessage | None:
1236
1237
        if not previous_text:
            self.parser.reset_streaming_state()
1238
1239
1240
            # Reset tool call tracking arrays for new streaming session
            self.prev_tool_call_arr = []
            self.streamed_args_for_tool = []
1241
            self.parser.set_tools(self.tools)
1242
1243
1244
1245
1246
1247
1248

        # Model sometimes outputs separately causing delta_text to be empty.
        # If there were tool_calls before and all current tool_calls have ended,
        # return an empty tool_call for outer streaming output
        # to correctly output tool_call field
        if not delta_text and delta_token_ids:
            open_calls = current_text.count(
1249
1250
                self.parser.tool_call_start_token
            ) - current_text.count(self.parser.tool_call_end_token)
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
            if (
                open_calls == 0
                and self.parser.tool_call_index > 0
                or not self.parser.tool_call_index
                and current_text
            ):
                return DeltaMessage(content="")
            return None

        # Parse the delta text and get the result
        result = self.parser.parse_single_streaming_chunks(delta_text)

        # Update tool call tracking arrays based on incremental parsing results
        if result and result.tool_calls:
            for tool_call in result.tool_calls:
                if tool_call.function:
                    tool_index = (
                        tool_call.index
                        if tool_call.index is not None
                        else len(self.prev_tool_call_arr) - 1
                    )

                    # Ensure we have enough entries in our tracking arrays
                    while len(self.prev_tool_call_arr) <= tool_index:
                        self.prev_tool_call_arr.append({"name": "", "arguments": ""})
                    while len(self.streamed_args_for_tool) <= tool_index:
                        self.streamed_args_for_tool.append("")

                    # Update tool name if provided
                    if tool_call.function.name:
                        self.prev_tool_call_arr[tool_index]["name"] = (
                            tool_call.function.name
1283
                        )
1284

1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
                    # Update arguments incrementally
                    if tool_call.function.arguments is not None:
                        # Concatenate the incremental arguments
                        # to the existing streamed arguments
                        self.prev_tool_call_arr[tool_index]["arguments"] += (
                            tool_call.function.arguments
                        )
                        self.streamed_args_for_tool[tool_index] += (
                            tool_call.function.arguments
                        )
        return result