|
| 1 | +"""Tests for split thinking tag handling in ModelResponsePartsManager.""" |
| 2 | + |
| 3 | +from inline_snapshot import snapshot |
| 4 | + |
| 5 | +from pydantic_ai._parts_manager import ModelResponsePartsManager |
| 6 | +from pydantic_ai.messages import ( |
| 7 | + PartDeltaEvent, |
| 8 | + PartStartEvent, |
| 9 | + TextPart, |
| 10 | + TextPartDelta, |
| 11 | + ThinkingPart, |
| 12 | + ThinkingPartDelta, |
| 13 | +) |
| 14 | + |
| 15 | + |
| 16 | +def test_handle_text_deltas_with_split_think_tags_at_chunk_start(): |
| 17 | + """Test split thinking tags when tag starts at position 0 of chunk.""" |
| 18 | + manager = ModelResponsePartsManager() |
| 19 | + thinking_tags = ('<think>', '</think>') |
| 20 | + |
| 21 | + # Chunk 1: "<thi" - starts at position 0, buffer it # codespell:ignore thi |
| 22 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='<thi', thinking_tags=thinking_tags)) |
| 23 | + assert len(events) == 0 # Buffered, no events yet |
| 24 | + assert manager.get_parts() == [] |
| 25 | + |
| 26 | + # Chunk 2: "nk>" - completes the tag |
| 27 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='nk>', thinking_tags=thinking_tags)) |
| 28 | + assert len(events) == 1 |
| 29 | + assert events[0] == snapshot( |
| 30 | + PartStartEvent(index=0, part=ThinkingPart(content='', part_kind='thinking'), event_kind='part_start') |
| 31 | + ) |
| 32 | + assert manager.get_parts() == snapshot([ThinkingPart(content='', part_kind='thinking')]) |
| 33 | + |
| 34 | + # Chunk 3: "reasoning content" |
| 35 | + events = list( |
| 36 | + manager.handle_text_delta(vendor_part_id='content', content='reasoning content', thinking_tags=thinking_tags) |
| 37 | + ) |
| 38 | + assert len(events) == 1 |
| 39 | + assert events[0] == snapshot( |
| 40 | + PartDeltaEvent( |
| 41 | + index=0, |
| 42 | + delta=ThinkingPartDelta(content_delta='reasoning content', part_delta_kind='thinking'), |
| 43 | + event_kind='part_delta', |
| 44 | + ) |
| 45 | + ) |
| 46 | + |
| 47 | + # Chunk 4: "</think>" - end tag |
| 48 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='</think>', thinking_tags=thinking_tags)) |
| 49 | + assert len(events) == 0 |
| 50 | + |
| 51 | + # Chunk 5: "after" - text after thinking |
| 52 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='after', thinking_tags=thinking_tags)) |
| 53 | + assert len(events) == 1 |
| 54 | + assert events[0] == snapshot( |
| 55 | + PartStartEvent(index=1, part=TextPart(content='after', part_kind='text'), event_kind='part_start') |
| 56 | + ) |
| 57 | + |
| 58 | + |
| 59 | +def test_handle_text_deltas_split_tags_after_text(): |
| 60 | + """Test split thinking tags at chunk position 0 after text in previous chunk.""" |
| 61 | + manager = ModelResponsePartsManager() |
| 62 | + thinking_tags = ('<think>', '</think>') |
| 63 | + |
| 64 | + # Chunk 1: "pre-" - creates TextPart |
| 65 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='pre-', thinking_tags=thinking_tags)) |
| 66 | + assert len(events) == 1 |
| 67 | + assert events[0] == snapshot( |
| 68 | + PartStartEvent(index=0, part=TextPart(content='pre-', part_kind='text'), event_kind='part_start') |
| 69 | + ) |
| 70 | + |
| 71 | + # Chunk 2: "<thi" - starts at position 0 of THIS chunk, buffer it |
| 72 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='<thi', thinking_tags=thinking_tags)) |
| 73 | + assert len(events) == 0 # Buffered |
| 74 | + assert manager.get_parts() == snapshot([TextPart(content='pre-', part_kind='text')]) |
| 75 | + |
| 76 | + # Chunk 3: "nk>" - completes the tag |
| 77 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='nk>', thinking_tags=thinking_tags)) |
| 78 | + assert len(events) == 1 |
| 79 | + assert events[0] == snapshot( |
| 80 | + PartStartEvent(index=1, part=ThinkingPart(content='', part_kind='thinking'), event_kind='part_start') |
| 81 | + ) |
| 82 | + assert manager.get_parts() == snapshot( |
| 83 | + [TextPart(content='pre-', part_kind='text'), ThinkingPart(content='', part_kind='thinking')] |
| 84 | + ) |
| 85 | + |
| 86 | + |
| 87 | +def test_handle_text_deltas_split_tags_mid_chunk_treated_as_text(): |
| 88 | + """Test that split tags mid-chunk (after other content in same chunk) are treated as text.""" |
| 89 | + manager = ModelResponsePartsManager() |
| 90 | + thinking_tags = ('<think>', '</think>') |
| 91 | + |
| 92 | + # Chunk 1: "pre-<thi" - tag does NOT start at position 0 of chunk |
| 93 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='pre-<thi', thinking_tags=thinking_tags)) |
| 94 | + assert len(events) == 1 # Treated as text, not buffered |
| 95 | + assert events[0] == snapshot( |
| 96 | + PartStartEvent(index=0, part=TextPart(content='pre-<thi', part_kind='text'), event_kind='part_start') |
| 97 | + ) |
| 98 | + |
| 99 | + # Chunk 2: "nk>" - appends to text (not recognized as completing a tag) |
| 100 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='nk>', thinking_tags=thinking_tags)) |
| 101 | + assert len(events) == 1 |
| 102 | + assert events[0] == snapshot( |
| 103 | + PartDeltaEvent( |
| 104 | + index=0, delta=TextPartDelta(content_delta='nk>', part_delta_kind='text'), event_kind='part_delta' |
| 105 | + ) |
| 106 | + ) |
| 107 | + assert manager.get_parts() == snapshot([TextPart(content='pre-<think>', part_kind='text')]) |
| 108 | + |
| 109 | + |
| 110 | +def test_handle_text_deltas_split_tags_no_vendor_id(): |
| 111 | + """Test that split tags don't work with vendor_part_id=None (no buffering).""" |
| 112 | + manager = ModelResponsePartsManager() |
| 113 | + thinking_tags = ('<think>', '</think>') |
| 114 | + |
| 115 | + # Chunk 1: "<thi" with no vendor_part_id - can't buffer |
| 116 | + events = list(manager.handle_text_delta(vendor_part_id=None, content='<thi', thinking_tags=thinking_tags)) |
| 117 | + assert len(events) == 1 # Treated as text immediately (simple path) |
| 118 | + assert events[0] == snapshot( |
| 119 | + PartStartEvent(index=0, part=TextPart(content='<thi', part_kind='text'), event_kind='part_start') |
| 120 | + ) |
| 121 | + |
| 122 | + # Chunk 2: "nk>" - appends to text |
| 123 | + events = list(manager.handle_text_delta(vendor_part_id=None, content='nk>', thinking_tags=thinking_tags)) |
| 124 | + assert len(events) == 1 |
| 125 | + assert events[0] == snapshot( |
| 126 | + PartDeltaEvent( |
| 127 | + index=0, delta=TextPartDelta(content_delta='nk>', part_delta_kind='text'), event_kind='part_delta' |
| 128 | + ) |
| 129 | + ) |
| 130 | + assert manager.get_parts() == snapshot([TextPart(content='<think>', part_kind='text')]) |
| 131 | + |
| 132 | + |
| 133 | +def test_handle_text_deltas_false_start_then_real_tag(): |
| 134 | + """Test buffering a false start, then processing real content.""" |
| 135 | + manager = ModelResponsePartsManager() |
| 136 | + thinking_tags = ('<think>', '</think>') |
| 137 | + |
| 138 | + # Chunk 1: "<th" - could be tag start, buffer it |
| 139 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='<th', thinking_tags=thinking_tags)) |
| 140 | + assert len(events) == 0 # Buffered |
| 141 | + |
| 142 | + # Chunk 2: "is is text" - proves it's not a tag, flush buffer |
| 143 | + events = list( |
| 144 | + manager.handle_text_delta(vendor_part_id='content', content='is is text', thinking_tags=thinking_tags) |
| 145 | + ) |
| 146 | + assert len(events) == 1 |
| 147 | + assert events[0] == snapshot( |
| 148 | + PartStartEvent(index=0, part=TextPart(content='<this is text', part_kind='text'), event_kind='part_start') |
| 149 | + ) |
| 150 | + assert manager.get_parts() == snapshot([TextPart(content='<this is text', part_kind='text')]) |
| 151 | + |
| 152 | + |
| 153 | +def test_buffered_content_exceeds_tag_length(): |
| 154 | + """Test that buffered content longer than tag is flushed (covers line 231).""" |
| 155 | + manager = ModelResponsePartsManager() |
| 156 | + thinking_tags = ('<think>', '</think>') |
| 157 | + |
| 158 | + # To hit line 231, we need: |
| 159 | + # 1. Buffer some content |
| 160 | + # 2. Next chunk starts with '<' (to pass first check) |
| 161 | + # 3. Combined length >= tag length |
| 162 | + |
| 163 | + # First chunk: exactly 6 chars |
| 164 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='<think', thinking_tags=thinking_tags)) |
| 165 | + assert len(events) == 0 # Buffered |
| 166 | + |
| 167 | + # Second chunk: starts with '<' so it checks _could_be_tag_start |
| 168 | + # Combined will be '<think<' (7 chars) which equals tag length '<think>' (7 chars) |
| 169 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='<', thinking_tags=thinking_tags)) |
| 170 | + # 7 >= 7 is True, so line 231 returns False |
| 171 | + assert len(events) == 1 |
| 172 | + assert events[0] == snapshot( |
| 173 | + PartStartEvent(index=0, part=TextPart(content='<think<', part_kind='text'), event_kind='part_start') |
| 174 | + ) |
| 175 | + assert manager.get_parts() == snapshot([TextPart(content='<think<', part_kind='text')]) |
| 176 | + |
| 177 | + |
| 178 | +def test_complete_thinking_tag_no_vendor_id(): |
| 179 | + """Test complete thinking tag with vendor_part_id=None (covers lines 161-164).""" |
| 180 | + manager = ModelResponsePartsManager() |
| 181 | + thinking_tags = ('<think>', '</think>') |
| 182 | + |
| 183 | + # Complete start tag with vendor_part_id=None goes through simple path |
| 184 | + # This covers lines 161-164 in _handle_text_delta_simple |
| 185 | + events = list(manager.handle_text_delta(vendor_part_id=None, content='<think>', thinking_tags=thinking_tags)) |
| 186 | + assert len(events) == 1 |
| 187 | + assert events[0] == snapshot( |
| 188 | + PartStartEvent(index=0, part=ThinkingPart(content='', part_kind='thinking'), event_kind='part_start') |
| 189 | + ) |
| 190 | + assert manager.get_parts() == snapshot([ThinkingPart(content='', part_kind='thinking')]) |
| 191 | + |
| 192 | + |
| 193 | +def test_exact_tag_length_boundary(): |
| 194 | + """Test when buffered content exactly equals tag length.""" |
| 195 | + manager = ModelResponsePartsManager() |
| 196 | + thinking_tags = ('<think>', '</think>') |
| 197 | + |
| 198 | + # Send content in one chunk that's exactly tag length |
| 199 | + events = list(manager.handle_text_delta(vendor_part_id='content', content='<think>', thinking_tags=thinking_tags)) |
| 200 | + # Exact match creates ThinkingPart |
| 201 | + assert len(events) == 1 |
| 202 | + assert events[0] == snapshot( |
| 203 | + PartStartEvent(index=0, part=ThinkingPart(content='', part_kind='thinking'), event_kind='part_start') |
| 204 | + ) |
0 commit comments