[Experiment] Add symbol navigation commands into the editor

openhands/agenthub/codeact_agent/function_calling.py

@@ -214,6 +214,8 @@ def __init__(self):
 * The `create` command cannot be used if the specified `path` already exists as a file
 * If a `command` generates a long output, it will be truncated and marked with `<response clipped>`
 * The `undo_edit` command will revert the last edit made to the file at `path`
+* The `jump_to_definition` command will print out possible FULL definition(s) of the class or function/method in the whole codebase. Can be useful for navigating large repositories
+* The `find_references` command will print out possible references (or usages) to the class or function/method in the whole codebase. This can be useful to understand how a class or function/method is used in a repository
 
 Notes for using the `str_replace` command:
 * The `old_str` parameter should match EXACTLY one or more consecutive lines from the original file. Be mindful of whitespaces!
@@ -235,7 +237,7 @@ def __init__(self):
                     'type': 'string',
                 },
                 'path': {
-                    'description': 'Absolute path to file or directory, e.g. `/workspace/file.py` or `/workspace`.',
+                    'description': 'Absolute path to file or directory, e.g. `/workspace/file.py` or `/workspace`. Required for all commands, except `jump_to_definition` and `find_references`.',
                     'type': 'string',
                 },
                 'file_text': {
@@ -259,8 +261,11 @@ def __init__(self):
                     'items': {'type': 'integer'},
                     'type': 'array',
                 },
+                'symbol_name': {
+                    'description': 'Required parameter of `jump_to_definition` and `find_references` commands. The name of the class or function/method to search for.',
+                },
             },
-            'required': ['command', 'path'],
+            'required': ['command'],
         },
     ),
 )

openhands/llm/fn_call_converter.py

@@ -50,7 +50,7 @@
 IN_CONTEXT_LEARNING_EXAMPLE_PREFIX = """
 Here's a running example of how to perform a task with the provided tools.
 
---------------------- START OF EXAMPLE ---------------------
+--------------------- START OF EXAMPLE 1 ---------------------
 
 USER: Create a list of numbers from 1 to 10, and display them in a web page at port 5000.
 
@@ -213,7 +213,148 @@ def index():
 <function=finish>
 </function>
 
---------------------- END OF EXAMPLE ---------------------
+--------------------- END OF EXAMPLE 1 ---------------------
+
+
+--------------------- START OF EXAMPLE 2 ---------------------
+
+USER: I'm working on an issue about HTTP timeouts in our data ingestion pipeline. The logs show
+TimeoutError is being raised in fetch_customer_data() but I'm not sure if we need to increase the
+timeout or if there's a deeper issue. Can you help me trace this function?
+
+ASSISTANT: I'll help track down the timeout issue in fetch_customer_data. Let me see its definition first:
+
+<function=str_replace_editor>
+<parameter=command>jump_to_definition</parameter>
+<parameter=symbol_name>fetch_customer_data</parameter>
+</function>
+
+USER: EXECUTION RESULT of [str_replace_editor]:
+Definition(s) of `fetch_customer_data`:
+/workspace/services/data_ingestion.py:
+...⋮...
+252│async def fetch_customer_data(customer_id: str) -> Dict[str, Any]:
+253│    '''Fetches customer data from multiple endpoints and aggregates results.
+254│
+255│    Args:
+256│        customer_id: Unique identifier for customer
+257│
+258│    Returns:
+259│        Combined customer data from all sources
+260│    '''
+261│    async with CustomerClient(timeout=DEFAULT_TIMEOUT) as client:
+262│        # Core profile data has 30s timeout
+263│        profile = await client.get_profile(customer_id)
+264│
+265│        # Transaction history needs more time
+266│        transactions = await _fetch_transactions(
+267│            customer_id,
+268│            timeout=TRANSACTION_TIMEOUT
+269│        )
+270│
+271│        return {
+272│            "profile": profile,
+273│            "transactions": transactions,
+274│            "last_updated": datetime.now()
+275│        }
+
+ASSISTANT: Interesting - it looks like fetch_customer_data relies on _fetch_transactions which has a separate timeout. Let me check the definition of that internal function:
+
+<function=str_replace_editor>
+<parameter=command>jump_to_definition</parameter>
+<parameter=symbol_name>_fetch_transactions</parameter>
+</function>
+
+USER: EXECUTION RESULT of [str_replace_editor]:
+Definition(s) of `_fetch_transactions`:
+/workspace/services/data_ingestion.py:
+...⋮...
+67|async def _fetch_transactions(customer_id: str, timeout: int) -> List[Dict]:
+68|    '''Fetches detailed transaction history for customer.
+69|    Handles pagination and retries on network errors.
+70|    '''
+71|    results = []
+72|    page = 1
+73|
+74|    while True:
+75|        try:
+76|            async with TransactionClient(timeout=timeout) as client:
+77|                batch = await client.get_transactions(
+78|                    customer_id,
+79|                    page=page,
+80|                    batch_size=BATCH_SIZE
+81|                )
+82|                if not batch:
+83|                    break
+84|
+85|                results.extend(batch)
+86|                page += 1
+87|
+88|        except TimeoutError:
+89|            logger.error(
+90|                f"Timeout fetching transactions for {customer_id} "
+91|                f"on page {page}"
+92|            )
+93|            raise
+94|
+95|    return results
+
+ASSISTANT: Now, let me see where fetch_customer_data is being called to understand the usage patterns:
+
+<function=str_replace_editor>
+<parameter=command>find_references</parameter>
+<parameter=symbol_name>fetch_customer_data</parameter>
+</function>
+
+USER: EXECUTION RESULT of [str_replace_editor]:
+References to `fetch_customer_data`:
+/workspace/api/customers.py:
+...⋮...
+21|    @router.get("/customers/{customer_id}/full")
+22|    async def get_full_customer_data(customer_id: str):
+23|        try:
+24|            data = await fetch_customer_data(customer_id)
+25|            return data
+26|        except TimeoutError:
+27|            raise HTTPException(status_code=504, detail="Data fetch timed out")
+...⋮...
+/workspace/jobs/daily_report.py:
+...⋮...
+314|    async def generate_customer_report(customer_ids: List[str]):
+315|        results = []
+316|        for id in customer_ids:
+317|            try:
+318|                data = await fetch_customer_data(id)
+319|                results.append(summarize_customer_data(data))
+320|            except TimeoutError:
+321|                logger.warning(f"Skipping customer {id} due to timeout")
+322|                continue
+323|        return results
+...⋮...
+/workspace/tests/test_services.py:
+...⋮...
+60|    @pytest.mark.asyncio
+61|    async def test_fetch_customer_data_large_history():
+62|        # Test with customer having 50+ pages of transactions
+63|        data = await fetch_customer_data("customer_with_large_history")
+64|        assert len(data["transactions"]) > 500
+...⋮...
+
+ASSISTANT:
+I see the issue now - the TimeoutError is likely happening because _fetch_transactions has to paginate through transaction history, and while each individual request uses TRANSACTION_TIMEOUT, we're not limiting the total time for all pages.
+
+Some customers might have hundreds of transaction pages, causing the overall fetch_customer_data call to take much longer than expected. This explains why we see timeouts in both the API endpoint and daily report job.
+
+We should probably:
+1. Add a max_pages parameter to _fetch_transactions to limit total pagination
+2. Add a total_timeout parameter separate from the per-request timeout
+3. Consider implementing parallel page fetching for large transaction histories
+
+Would you like me to implement any of these solutions for the issue?
+
+... (continue the conversation)
+
+--------------------- END OF EXAMPLE 2 ---------------------
 
 Do NOT assume the environment is the same as in the example above.
 

pyproject.toml

@@ -63,7 +63,7 @@ opentelemetry-exporter-otlp-proto-grpc = "1.25.0"
 modal = "^0.64.145"
 runloop-api-client = "0.7.0"
 pygithub = "^2.5.0"
-openhands-aci = "^0.1.1"
+openhands-aci = {git = "https://github.com/All-Hands-AI/openhands-aci.git", rev = "ht/jump-commands"}
 
 [tool.poetry.group.llama-index.dependencies]
 llama-index = "*"
@@ -95,6 +95,7 @@ reportlab = "*"
 [tool.coverage.run]
 concurrency = ["gevent"]
 
+
 [tool.poetry.group.runtime.dependencies]
 jupyterlab = "*"
 notebook = "*"
@@ -125,6 +126,7 @@ ignore = ["D1"]
 [tool.ruff.lint.pydocstyle]
 convention = "google"
 
+
 [tool.poetry.group.evaluation.dependencies]
 streamlit = "*"
 whatthepatch = "*"