Skip to content

chat: new parser should not crash inference #20708

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
jpohhhh wants to merge 1 commit into from
Closed

chat: new parser should not crash inference #20708

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
21 changes: 5 additions & 16 deletions common/chat.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1758,23 +1758,12 @@ common_chat_msg common_chat_peg_parse(const common_peg_arena & src_pars
auto result = parser.parse(ctx);

if (result.fail()) {
// During partial parsing, return partial results if any AST nodes were captured
// This allows streaming to work correctly for formats like FUNC_MARKDOWN_CODE_BLOCK
if (is_partial && result.end > 0) {
// Try to extract any partial results from what was successfully parsed
common_chat_msg msg;
msg.role = "assistant";
auto mapper = common_chat_peg_mapper(msg);
mapper.from_ast(ctx.ast, result);

if (ctx.is_debug()) {
fprintf(stderr, "\nAST for partial parse (fail):\n%s\n", ctx.ast.dump().c_str());
fflush(stderr);
}
return msg;
if (result.end == 0) {
throw std::runtime_error(std::string("Failed to parse input at pos ") + std::to_string(result.end) + ": " +
effective_input.substr(result.end));
}
throw std::runtime_error(std::string("Failed to parse input at pos ") + std::to_string(result.end) + ": " +
input.substr(result.end));
LOG_WRN("%s: parser failed at pos %zu, returning partial result. Unparsed: %s\n",
__func__, result.end, effective_input.substr(result.end).c_str());
}

common_chat_msg msg;
Expand Down
154 changes: 154 additions & 0 deletions models/templates/Qwen-Qwen3.5-0.8B.jinja
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,154 @@
{%- set image_count = namespace(value=0) %}
{%- set video_count = namespace(value=0) %}
{%- macro render_content(content, do_vision_count, is_system_content=false) %}
{%- if content is string %}
{{- content }}
{%- elif content is iterable and content is not mapping %}
{%- for item in content %}
{%- if 'image' in item or 'image_url' in item or item.type == 'image' %}
{%- if is_system_content %}
{{- raise_exception('System message cannot contain images.') }}
{%- endif %}
{%- if do_vision_count %}
{%- set image_count.value = image_count.value + 1 %}
{%- endif %}
{%- if add_vision_id %}
{{- 'Picture ' ~ image_count.value ~ ': ' }}
{%- endif %}
{{- '<|vision_start|><|image_pad|><|vision_end|>' }}
{%- elif 'video' in item or item.type == 'video' %}
{%- if is_system_content %}
{{- raise_exception('System message cannot contain videos.') }}
{%- endif %}
{%- if do_vision_count %}
{%- set video_count.value = video_count.value + 1 %}
{%- endif %}
{%- if add_vision_id %}
{{- 'Video ' ~ video_count.value ~ ': ' }}
{%- endif %}
{{- '<|vision_start|><|video_pad|><|vision_end|>' }}
{%- elif 'text' in item %}
{{- item.text }}
{%- else %}
{{- raise_exception('Unexpected item type in content.') }}
{%- endif %}
{%- endfor %}
{%- elif content is none or content is undefined %}
{{- '' }}
{%- else %}
{{- raise_exception('Unexpected content type.') }}
{%- endif %}
{%- endmacro %}
{%- if not messages %}
{{- raise_exception('No messages provided.') }}
{%- endif %}
{%- if tools and tools is iterable and tools is not mapping %}
{{- '<|im_start|>system\n' }}
{{- "# Tools\n\nYou have access to the following functions:\n\n<tools>" }}
{%- for tool in tools %}
{{- "\n" }}
{{- tool | tojson }}
{%- endfor %}
{{- "\n</tools>" }}
{{- '\n\nIf you choose to call a function ONLY reply in the following format with NO suffix:\n\n<tool_call>\n<function=example_function_name>\n<parameter=example_parameter_1>\nvalue_1\n</parameter>\n<parameter=example_parameter_2>\nThis is the value for the second parameter\nthat can span\nmultiple lines\n</parameter>\n</function>\n</tool_call>\n\n<IMPORTANT>\nReminder:\n- Function calls MUST follow the specified format: an inner <function=...></function> block must be nested within <tool_call></tool_call> XML tags\n- Required parameters MUST be specified\n- You may provide optional reasoning for your function call in natural language BEFORE the function call, but NOT after\n- If there is no function call available, answer the question like normal with your current knowledge and do not tell the user about function calls\n</IMPORTANT>' }}
{%- if messages[0].role == 'system' %}
{%- set content = render_content(messages[0].content, false, true)|trim %}
{%- if content %}
{{- '\n\n' + content }}
{%- endif %}
{%- endif %}
{{- '<|im_end|>\n' }}
{%- else %}
{%- if messages[0].role == 'system' %}
{%- set content = render_content(messages[0].content, false, true)|trim %}
{{- '<|im_start|>system\n' + content + '<|im_end|>\n' }}
{%- endif %}
{%- endif %}
{%- set ns = namespace(multi_step_tool=true, last_query_index=messages|length - 1) %}
{%- for message in messages[::-1] %}
{%- set index = (messages|length - 1) - loop.index0 %}
{%- if ns.multi_step_tool and message.role == "user" %}
{%- set content = render_content(message.content, false)|trim %}
{%- if not(content.startswith('<tool_response>') and content.endswith('</tool_response>')) %}
{%- set ns.multi_step_tool = false %}
{%- set ns.last_query_index = index %}
{%- endif %}
{%- endif %}
{%- endfor %}
{%- if ns.multi_step_tool %}
{{- raise_exception('No user query found in messages.') }}
{%- endif %}
{%- for message in messages %}
{%- set content = render_content(message.content, true)|trim %}
{%- if message.role == "system" %}
{%- if not loop.first %}
{{- raise_exception('System message must be at the beginning.') }}
{%- endif %}
{%- elif message.role == "user" %}
{{- '<|im_start|>' + message.role + '\n' + content + '<|im_end|>' + '\n' }}
{%- elif message.role == "assistant" %}
{%- set reasoning_content = '' %}
{%- if message.reasoning_content is string %}
{%- set reasoning_content = message.reasoning_content %}
{%- else %}
{%- if '</think>' in content %}
{%- set reasoning_content = content.split('</think>')[0].rstrip('\n').split('<think>')[-1].lstrip('\n') %}
{%- set content = content.split('</think>')[-1].lstrip('\n') %}
{%- endif %}
{%- endif %}
{%- set reasoning_content = reasoning_content|trim %}
{%- if loop.index0 > ns.last_query_index %}
{{- '<|im_start|>' + message.role + '\n<think>\n' + reasoning_content + '\n</think>\n\n' + content }}
{%- else %}
{{- '<|im_start|>' + message.role + '\n' + content }}
{%- endif %}
{%- if message.tool_calls and message.tool_calls is iterable and message.tool_calls is not mapping %}
{%- for tool_call in message.tool_calls %}
{%- if tool_call.function is defined %}
{%- set tool_call = tool_call.function %}
{%- endif %}
{%- if loop.first %}
{%- if content|trim %}
{{- '\n\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
{%- else %}
{{- '<tool_call>\n<function=' + tool_call.name + '>\n' }}
{%- endif %}
{%- else %}
{{- '\n<tool_call>\n<function=' + tool_call.name + '>\n' }}
{%- endif %}
{%- if tool_call.arguments is defined %}
{%- for args_name, args_value in tool_call.arguments|items %}
{{- '<parameter=' + args_name + '>\n' }}
{%- set args_value = args_value | tojson | safe if args_value is mapping or (args_value is sequence and args_value is not string) else args_value | string %}
{{- args_value }}
{{- '\n</parameter>\n' }}
{%- endfor %}
{%- endif %}
{{- '</function>\n</tool_call>' }}
{%- endfor %}
{%- endif %}
{{- '<|im_end|>\n' }}
{%- elif message.role == "tool" %}
{%- if loop.previtem and loop.previtem.role != "tool" %}
{{- '<|im_start|>user' }}
{%- endif %}
{{- '\n<tool_response>\n' }}
{{- content }}
{{- '\n</tool_response>' }}
{%- if not loop.last and loop.nextitem.role != "tool" %}
{{- '<|im_end|>\n' }}
{%- elif loop.last %}
{{- '<|im_end|>\n' }}
{%- endif %}
{%- else %}
{{- raise_exception('Unexpected message role.') }}
{%- endif %}
{%- endfor %}
{%- if add_generation_prompt %}
{{- '<|im_start|>assistant\n' }}
{%- if enable_thinking is defined and enable_thinking is true %}
{{- '<think>\n' }}
{%- else %}
{{- '<think>\n\n</think>\n\n' }}
{%- endif %}
{%- endif %}
162 changes: 138 additions & 24 deletions tests/test-chat.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -806,6 +806,7 @@ struct peg_test_case {
std::string input;
common_chat_msg expect;
bool is_partial = false;
bool expect_partial_rollback = false;
};

struct make_peg_parser {
Expand Down Expand Up @@ -889,41 +890,45 @@ static void test_peg_parser(common_chat_templates * tmpls,
std::string prefix = tc.input.substr(0, safe_len);
common_chat_msg msg_current = parser.parse(prefix, is_partial);

for (const auto & diff : common_chat_msg_diff::compute_diffs(msg_prev, msg_current)) {
if (!diff.reasoning_content_delta.empty()) {
msg_accum.reasoning_content += diff.reasoning_content_delta;
}
if (!diff.content_delta.empty()) {
msg_accum.content += diff.content_delta;
}
if (diff.tool_call_index != std::string::npos) {
// During partial parsing, a new tool call may appear with empty name initially
// The name gets filled in as more input is parsed
while (msg_accum.tool_calls.size() <= diff.tool_call_index) {
msg_accum.tool_calls.push_back({ "", "", "" });
}
// Always update name and id from diff (may change during incremental parsing), but only if the delta
// actually contains them
if (!diff.tool_call_delta.name.empty()) {
msg_accum.tool_calls[diff.tool_call_index].name = diff.tool_call_delta.name;
try {
for (const auto & diff : common_chat_msg_diff::compute_diffs(msg_prev, msg_current)) {
if (!diff.reasoning_content_delta.empty()) {
msg_accum.reasoning_content += diff.reasoning_content_delta;
}
if (!diff.tool_call_delta.id.empty()) {
msg_accum.tool_calls[diff.tool_call_index].id = diff.tool_call_delta.id;
if (!diff.content_delta.empty()) {
msg_accum.content += diff.content_delta;
}
if (!diff.tool_call_delta.arguments.empty()) {
msg_accum.tool_calls[diff.tool_call_index].arguments += diff.tool_call_delta.arguments;
if (diff.tool_call_index != std::string::npos) {
while (msg_accum.tool_calls.size() <= diff.tool_call_index) {
msg_accum.tool_calls.push_back({ "", "", "" });
}
if (!diff.tool_call_delta.name.empty()) {
msg_accum.tool_calls[diff.tool_call_index].name = diff.tool_call_delta.name;
}
if (!diff.tool_call_delta.id.empty()) {
msg_accum.tool_calls[diff.tool_call_index].id = diff.tool_call_delta.id;
}
if (!diff.tool_call_delta.arguments.empty()) {
msg_accum.tool_calls[diff.tool_call_index].arguments += diff.tool_call_delta.arguments;
}
}
}
}
try {
assert_msg_equals(msg_current, msg_accum, true);
} catch (std::exception & e) {
throw std::runtime_error((std::string("Error comparing accumulated message to current: ") + e.what()).c_str());
if (tc.expect_partial_rollback) {
return;
}
throw std::runtime_error((std::string("Error in incremental parse: ") + e.what()).c_str());
}

msg_prev = msg_current;
}

if (tc.expect_partial_rollback) {
throw std::runtime_error("Partial rollback was expected but did not occur. "
"If grammar coverage improved, update this test to expect the extracted content instead.");
}

if (!tc.is_partial) {
assert_msg_equals(tc.expect, parser.parse(tc.input, false), true);
}
Expand Down Expand Up @@ -1098,6 +1103,10 @@ class peg_test_builder {
tc_.is_partial = val;
return *this;
}
peg_test_builder & expect_partial_rollback(bool val = true) {
tc_.expect_partial_rollback = val;
return *this;
}

// Expect setters
peg_test_builder & expect(const common_chat_msg & msg) {
Expand Down Expand Up @@ -1807,6 +1816,78 @@ static void test_template_output_peg_parsers(bool detailed_debug) {
})
.run();
}

// Qwen3.5-0.8B: template + basic tool calling + malformed JSON regression
{
auto tst = peg_tester("models/templates/Qwen-Qwen3.5-0.8B.jinja", /*detailed_debug=*/false);

static common_chat_tool flashcards_tool{
"flashcards", "Flashcards for studying",
R"({"type":"object","properties":{"flashcards":{"type":"array","items":{"type":"object","properties":{"query":{"type":"string"},"recall":{"type":"string"}},"required":["recall","query"]}}},"required":["flashcards"]})",
};

tst.test("</think>Hello, world!\nWhat's up?")
.enable_thinking(true)
.reasoning_format(COMMON_REASONING_FORMAT_DEEPSEEK)
.expect(message_assist)
.run();

tst.test(
"</think>"
"<tool_call>\n"
"<function=special_function>\n"
"<parameter=arg1>\n"
"1\n"
"</parameter>\n"
"</function>\n"
"</tool_call>")
.enable_thinking(true)
.reasoning_format(COMMON_REASONING_FORMAT_DEEPSEEK)
.tools({ special_function_tool })
.expect(message_assist_call)
.run();

tst.test(
"The user wants flashcards about California.\n"
"</think>\n"
"\n"
"<tool_call>\n"
"<function=special_function>\n"
"<parameter=arg1>\n"
"1\n"
"</parameter>\n"
"</function>\n"
"</tool_call>")
.tools({ special_function_tool })
.enable_thinking(true)
.reasoning_format(COMMON_REASONING_FORMAT_AUTO)
.expect_reasoning("The user wants flashcards about California.\n")
.expect_tool_calls({
{ "special_function", R"({"arg1": 1})", {} },
})
.run();

// Well-formed JSON: no regression
tst.test(
"Making cards.\n"
"</think>\n\n"
"<tool_call>\n"
"<function=flashcards>\n"
"<parameter=flashcards>\n"
"[{\"query\": \"CA\", \"recall\": \"LA\"}]\n"
"</parameter>\n"
"</function>\n"
"</tool_call>")
.tools({ flashcards_tool })
.enable_thinking(true)
.reasoning_format(COMMON_REASONING_FORMAT_AUTO)
.expect_reasoning("Making cards.\n")
.expect_tool_calls({
{ "flashcards", R"({"flashcards": [{"query": "CA", "recall": "LA"}]})", {} },
})
.run();
}

{
auto tst = peg_tester("models/templates/deepseek-ai-DeepSeek-V3.1.jinja", detailed_debug);
tst.test(
Expand Down Expand Up @@ -1954,6 +2035,39 @@ static void test_template_output_peg_parsers(bool detailed_debug) {
}
}

// Qwen3.5 TAG_WITH_TAGGED: trailing newline after </tool_call>.
//
// This is not canonical inference output. It was observed during inference, but
// grammar constraints and/or parallel_tool_calls settings now prevent the model
// from generating this exact output at runtime. It exists solely as a regression
// test: the trailing \n causes result.fail() with result.end > 0 on the full
// parse. On unpatched code this throws std::runtime_error and discards all parsed
// content. With the fix, reasoning and tool calls are recovered.
{
auto tst = peg_tester("models/templates/Qwen-Qwen3.5-0.8B.jinja", /*detailed_debug=*/false);

static common_chat_tool flashcards_tool{
"flashcards", "Flashcards for studying",
R"({"type":"object","properties":{"flashcards":{"type":"array","items":{"type":"object","properties":{"query":{"type":"string"},"recall":{"type":"string"}},"required":["recall","query"]}}},"required":["flashcards"]})",
};

tst.test(
"Making cards.\n"
"</think>\n\n"
"<tool_call>\n"
"<function=flashcards>\n"
"<parameter=flashcards>\n"
"[{\"query\": \"CA\", \"recall\": \"LA\"}]\n"
"</parameter>\n"
"</function>\n"
"</tool_call>\n")
.tools({ flashcards_tool })
.enable_thinking(true)
.reasoning_format(COMMON_REASONING_FORMAT_AUTO)
.expect_partial_rollback()
.run();
}

// Kimi-K2-Thinking tests - custom parser
// Unique feature: tool call ID embeds function name as functions.<name>:<counter>
{
Expand Down