15818567920

Committed 23 Jun 2025 07:58AM UTC coverage: 90.157% (-0.02%) from 90.178%

Build # 15818567920

Build Type

Pull #9525

github

Committed by

web-flow

Commit Message

Merge 7c097a34e into f91145964

Pull Request Pull Request #9525: refactor: Update to `StreamingChunk`, better `index` setting and change `tool_call` to `tool_calls`

Run Details

11569 of 12832 relevant lines covered (90.16%)

0.9 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

75.0

haystack/components/generators/utils.py

# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
#
# SPDX-License-Identifier: Apache-2.0

import json
from typing import Dict, List

from haystack import logging
from haystack.dataclasses import ChatMessage, StreamingChunk, ToolCall

logger = logging.getLogger(__name__)


def print_streaming_chunk(chunk: StreamingChunk) -> None:
    """
    Callback function to handle and display streaming output chunks.

    This function processes a `StreamingChunk` object by:
    - Printing tool call metadata (if any), including function names and arguments, as they arrive.
    - Printing tool call results when available.
    - Printing the main content (e.g., text tokens) of the chunk as it is received.

    The function outputs data directly to stdout and flushes output buffers to ensure immediate display during
    streaming.

    :param chunk: A chunk of streaming data containing content and optional metadata, such as tool calls and
        tool results.
    """
    if chunk.start and chunk.index and chunk.index > 0:
        # If this is the start of a new content block but not the first content block, print two new lines
        print("\n\n", flush=True, end="")

    ## Tool Call streaming
    if chunk.tool_calls:
        # Typically, if there are multiple tool calls in the chunk this means that the tool calls are fully formed and
        # not just a delta.
        for tool_call in chunk.tool_calls:
            # If chunk.start is True indicates beginning of a tool call
            # Also presence of tool_call.tool_name indicates the start of a tool call too
            if chunk.start:
                # If there is more than one tool call in the chunk, we print two new lines to separate them
                # We know there is more than one tool call if the index of the tool call is greater than the index of
                # the chunk.
                if chunk.index and tool_call.index > chunk.index:
                    print("\n\n", flush=True, end="")

                print("[TOOL CALL]\nTool: {tool_call.tool_name} \nArguments: ", flush=True, end="")

            # print the tool arguments
            if tool_call.arguments:
                print(tool_call.arguments, flush=True, end="")

    ## Tool Call Result streaming
    # Print tool call results if available (from ToolInvoker)
    if chunk.tool_call_result:
        # Tool Call Result is fully formed so delta accumulation is not needed
        print(f"[TOOL RESULT]\n{chunk.tool_call_result.result}", flush=True, end="")

    ## Normal content streaming
    # Print the main content of the chunk (from ChatGenerator)
    if chunk.content:
        if chunk.start:
            print("[ASSISTANT]\n", flush=True, end="")
        print(chunk.content, flush=True, end="")

    # End of LLM assistant message so we add two new lines
    # This ensures spacing between multiple LLM messages (e.g. Agent) or multiple Tool Call Results
    if chunk.meta.get("finish_reason") is not None:
        print("\n\n", flush=True, end="")


def _convert_streaming_chunks_to_chat_message(chunks: List[StreamingChunk]) -> ChatMessage:
    """
    Connects the streaming chunks into a single ChatMessage.

    :param chunks: The list of all `StreamingChunk` objects.

    :returns: The ChatMessage.
    """
    text = "".join([chunk.content for chunk in chunks])
    tool_calls = []

    # Process tool calls if present in any chunk
    tool_call_data: Dict[int, Dict[str, str]] = {}  # Track tool calls by index
    for chunk in chunks:
        if chunk.tool_calls:
            # We do this to make sure mypy is happy, but we enforce index is not None in the StreamingChunk dataclass if
            # tool_call is present
            assert chunk.index is not None

            for tool_call in chunk.tool_calls:
                # We use the index of the tool_call to track the tool call across chunks since the ID is not always
                # provided
                if tool_call.index not in tool_call_data:
                    tool_call_data[chunk.index] = {"id": "", "name": "", "arguments": ""}

                # Save the ID if present
                if tool_call.id is not None:
                    tool_call_data[chunk.index]["id"] = tool_call.id

                if tool_call.tool_name is not None:
                    tool_call_data[chunk.index]["name"] += tool_call.tool_name
                if tool_call.arguments is not None:
                    tool_call_data[chunk.index]["arguments"] += tool_call.arguments

    # Convert accumulated tool call data into ToolCall objects
    sorted_keys = sorted(tool_call_data.keys())
    for key in sorted_keys:
        tool_call_dict = tool_call_data[key]
        try:
            arguments = json.loads(tool_call_dict["arguments"])
            tool_calls.append(ToolCall(id=tool_call_dict["id"], tool_name=tool_call_dict["name"], arguments=arguments))
        except json.JSONDecodeError:
            logger.warning(
                "OpenAI returned a malformed JSON string for tool call arguments. This tool call "
                "will be skipped. To always generate a valid JSON, set `tools_strict` to `True`. "
                "Tool call ID: {_id}, Tool name: {_name}, Arguments: {_arguments}",
                _id=tool_call_dict["id"],
                _name=tool_call_dict["name"],
                _arguments=tool_call_dict["arguments"],
            )

    # finish_reason can appear in different places so we look for the last one
    finish_reasons = [
        chunk.meta.get("finish_reason") for chunk in chunks if chunk.meta.get("finish_reason") is not None
    ]
    finish_reason = finish_reasons[-1] if finish_reasons else None

    meta = {
        "model": chunks[-1].meta.get("model"),
        "index": 0,
        "finish_reason": finish_reason,
        "completion_start_time": chunks[0].meta.get("received_at"),  # first chunk received
        "usage": chunks[-1].meta.get("usage"),  # last chunk has the final usage data if available
    }

    return ChatMessage.from_assistant(text=text or None, tool_calls=tool_calls, meta=meta)

1	# SPDX-FileCopyrightText: 2022-present deepset GmbH <info@deepset.ai>
2	#
3	# SPDX-License-Identifier: Apache-2.0
4
5	import json	1✔
6	from typing import Dict, List	1✔
7
8	from haystack import logging	1✔
9	from haystack.dataclasses import ChatMessage, StreamingChunk, ToolCall	1✔
10
11	logger = logging.getLogger(__name__)	1✔
12
13
14	def print_streaming_chunk(chunk: StreamingChunk) -> None:	1✔
15	"""
16	Callback function to handle and display streaming output chunks.
17
18	This function processes a `StreamingChunk` object by:
19	- Printing tool call metadata (if any), including function names and arguments, as they arrive.
20	- Printing tool call results when available.
21	- Printing the main content (e.g., text tokens) of the chunk as it is received.
22
23	The function outputs data directly to stdout and flushes output buffers to ensure immediate display during
24	streaming.
25
26	:param chunk: A chunk of streaming data containing content and optional metadata, such as tool calls and
27	tool results.
28	"""
29	if chunk.start and chunk.index and chunk.index > 0:	1✔
30	# If this is the start of a new content block but not the first content block, print two new lines
31	print("\n\n", flush=True, end="")	×
32
33	## Tool Call streaming
34	if chunk.tool_calls:	1✔
35	# Typically, if there are multiple tool calls in the chunk this means that the tool calls are fully formed and
36	# not just a delta.
37	for tool_call in chunk.tool_calls:	×
38	# If chunk.start is True indicates beginning of a tool call
39	# Also presence of tool_call.tool_name indicates the start of a tool call too
40	if chunk.start:	×
41	# If there is more than one tool call in the chunk, we print two new lines to separate them
42	# We know there is more than one tool call if the index of the tool call is greater than the index of
43	# the chunk.
44	if chunk.index and tool_call.index > chunk.index:	×
45	print("\n\n", flush=True, end="")	×
46
47	print("[TOOL CALL]\nTool: {tool_call.tool_name} \nArguments: ", flush=True, end="")	×
48
49	# print the tool arguments
50	if tool_call.arguments:	×
51	print(tool_call.arguments, flush=True, end="")	×
52
53	## Tool Call Result streaming
54	# Print tool call results if available (from ToolInvoker)
55	if chunk.tool_call_result:	1✔
56	# Tool Call Result is fully formed so delta accumulation is not needed
57	print(f"[TOOL RESULT]\n{chunk.tool_call_result.result}", flush=True, end="")	1✔
58
59	## Normal content streaming
60	# Print the main content of the chunk (from ChatGenerator)
61	if chunk.content:	1✔
62	if chunk.start:	×
63	print("[ASSISTANT]\n", flush=True, end="")	×
64	print(chunk.content, flush=True, end="")	×
65
66	# End of LLM assistant message so we add two new lines
67	# This ensures spacing between multiple LLM messages (e.g. Agent) or multiple Tool Call Results
68	if chunk.meta.get("finish_reason") is not None:	1✔
69	print("\n\n", flush=True, end="")	1✔
70
71
72	def _convert_streaming_chunks_to_chat_message(chunks: List[StreamingChunk]) -> ChatMessage:	1✔
73	"""
74	Connects the streaming chunks into a single ChatMessage.
75
76	:param chunks: The list of all `StreamingChunk` objects.
77
78	:returns: The ChatMessage.
79	"""
80	text = "".join([chunk.content for chunk in chunks])	1✔
81	tool_calls = []	1✔
82
83	# Process tool calls if present in any chunk
84	tool_call_data: Dict[int, Dict[str, str]] = {} # Track tool calls by index	1✔
85	for chunk in chunks:	1✔
86	if chunk.tool_calls:	1✔
87	# We do this to make sure mypy is happy, but we enforce index is not None in the StreamingChunk dataclass if
88	# tool_call is present
89	assert chunk.index is not None	1✔
90
91	for tool_call in chunk.tool_calls:	1✔
92	# We use the index of the tool_call to track the tool call across chunks since the ID is not always
93	# provided
94	if tool_call.index not in tool_call_data:	1✔
95	tool_call_data[chunk.index] = {"id": "", "name": "", "arguments": ""}	1✔
96
97	# Save the ID if present
98	if tool_call.id is not None:	1✔
99	tool_call_data[chunk.index]["id"] = tool_call.id	1✔
100
101	if tool_call.tool_name is not None:	1✔
102	tool_call_data[chunk.index]["name"] += tool_call.tool_name	1✔
103	if tool_call.arguments is not None:	1✔
104	tool_call_data[chunk.index]["arguments"] += tool_call.arguments	1✔
105
106	# Convert accumulated tool call data into ToolCall objects
107	sorted_keys = sorted(tool_call_data.keys())	1✔
108	for key in sorted_keys:	1✔
109	tool_call_dict = tool_call_data[key]	1✔
110	try:	1✔
111	arguments = json.loads(tool_call_dict["arguments"])	1✔
112	tool_calls.append(ToolCall(id=tool_call_dict["id"], tool_name=tool_call_dict["name"], arguments=arguments))	1✔
113	except json.JSONDecodeError:	×
114	logger.warning(	×
115	"OpenAI returned a malformed JSON string for tool call arguments. This tool call "
116	"will be skipped. To always generate a valid JSON, set `tools_strict` to `True`. "
117	"Tool call ID: {_id}, Tool name: {_name}, Arguments: {_arguments}",
118	_id=tool_call_dict["id"],
119	_name=tool_call_dict["name"],
120	_arguments=tool_call_dict["arguments"],
121	)
122
123	# finish_reason can appear in different places so we look for the last one
124	finish_reasons = [	1✔
125	chunk.meta.get("finish_reason") for chunk in chunks if chunk.meta.get("finish_reason") is not None
126	]
127	finish_reason = finish_reasons[-1] if finish_reasons else None	1✔
128
129	meta = {	1✔
130	"model": chunks[-1].meta.get("model"),
131	"index": 0,
132	"finish_reason": finish_reason,
133	"completion_start_time": chunks[0].meta.get("received_at"), # first chunk received
134	"usage": chunks[-1].meta.get("usage"), # last chunk has the final usage data if available
135	}
136
137	return ChatMessage.from_assistant(text=text or None, tool_calls=tool_calls, meta=meta)	1✔

deepset-ai / haystack / 15818567920

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous