> ## Documentation Index
> Fetch the complete documentation index at: https://mintlify.com/pranavkrishnasuresh/chemAgent/llms.txt
> Use this file to discover all available pages before exploring further.

# pubchem_fetcher

> PubChem API integration for chemical data retrieval

## Overview

The `pubchem_fetcher.py` module handles all interactions with the PubChem REST API, fetching chemical compound information using multiple search strategies (CID, formula, and name-based lookups).

## Constants

### PUBCHEM\_API\_BASE

```python theme={null}
PUBCHEM_API_BASE = "https://pubchem.ncbi.nlm.nih.gov/rest/pug"
```

Base URL for the PubChem PUG REST API.

## Functions

### fetch\_pubchem\_data

Retrieves chemical compound descriptions from PubChem using multiple API endpoints.

<ParamField path="terms" type="list[str]" required>
  List of chemistry terms extracted from the user query
</ParamField>

<ResponseField name="context" type="str">
  Formatted string containing all retrieved PubChem descriptions, or empty string if no data found
</ResponseField>

#### Search Strategy

For each term, the function attempts three different PubChem endpoints:

1. **CID (Compound ID)**: `/compound/cid/{term}/description/JSON`
2. **Formula**: `/compound/formula/{term}/description/JSON`
3. **Name**: `/compound/name/{term}/description/JSON`

This multi-endpoint approach maximizes the chances of finding relevant data regardless of how the user phrases their query.

#### Error Handling

* `requests.exceptions.RequestException`: Silently skips failed requests
* `ValueError`: Handles JSON parsing errors gracefully
* Timeouts set to 10 seconds per request

### extract\_pubchem\_descriptions

Parses PubChem API responses to extract description text.

<ParamField path="data" type="dict" required>
  JSON response from PubChem API
</ParamField>

<ResponseField name="descriptions" type="str | None">
  Concatenated description strings, or None if no descriptions found
</ResponseField>

## Example Usage

```python theme={null}
from plan_execute_agent.pubchem_rag.pubchem_fetcher import fetch_pubchem_data

# Fetch data for multiple chemistry terms
terms = ["aspirin", "C9H8O4", "acetylsalicylic acid"]
context = fetch_pubchem_data(terms)
print(context)
```

## Response Format

The function returns formatted text blocks for each successful query:

```
**Term**: aspirin
**Endpoint**: /compound/name/aspirin/description/JSON
Aspirin is a salicylate drug that inhibits cyclooxygenase...

**Term**: C9H8O4
**Endpoint**: /compound/formula/C9H8O4/description/JSON
Acetylsalicylic acid is an organic compound...
```

## PubChem API Structure

The function expects responses in this format:

```json theme={null}
{
  "InformationList": {
    "Information": [
      {
        "Description": "Chemical description text..."
      }
    ]
  }
}
```

## Integration Notes

* Called by `query_chemistry_related()` in the RAG pipeline
* Provides the context that feeds into the LLM response generator
* Designed to be fault-tolerant, continuing even if some requests fail
* No API key required (uses public PubChem endpoints)

## Source Location

* `fetch_pubchem_data`: `plan_execute_agent/pubchem_rag/pubchem_fetcher.py:6`
* `extract_pubchem_descriptions`: `plan_execute_agent/pubchem_rag/pubchem_fetcher.py:40`