TryParseUrl¶

Overview¶

TryParseUrl is a Spark Catalyst expression that safely extracts components from URL strings without throwing exceptions on malformed URLs. It serves as a runtime-replaceable wrapper around ParseUrl with error handling, returning null when URL parsing fails instead of raising exceptions.

Syntax¶

try_parse_url(url_string, part_to_extract)
try_parse_url(url_string, part_to_extract, key)

Arguments¶

Argument	Type	Description
url_string	String	The URL string to parse
part_to_extract	String	The URL component to extract (HOST, PATH, QUERY, REF, PROTOCOL, FILE, AUTHORITY, USERINFO)
key	String	Optional. The query parameter key when part_to_extract is 'QUERY'

Return Type¶

String - Returns the extracted URL component as a string, or null if parsing fails.

Supported Data Types¶

Input: String expressions for URL and component parameters
Output: String type representing the extracted URL component

Algorithm¶

Delegates to the underlying ParseUrl expression with failOnError = false
Parses the input URL string using standard URL parsing logic
Extracts the requested component based on the part_to_extract parameter
Returns null instead of throwing exceptions when URL parsing fails
Handles query parameter extraction when a key is specified

Partitioning Behavior¶

Preserves partitioning: Yes, this is a deterministic per-row transformation
Requires shuffle: No, operates independently on each row
Can be pushed down to individual partitions without data movement

Edge Cases¶

Null handling: Returns null when any input argument is null
Empty input: Returns null for empty or whitespace-only URL strings
Malformed URLs: Returns null instead of throwing exceptions (key difference from parse_url)
Invalid part names: Returns null for unrecognized part_to_extract values
Missing query keys: Returns null when specified query parameter doesn't exist

Code Generation¶

This expression extends RuntimeReplaceable, meaning it is replaced by its underlying ParseUrl implementation during analysis phase. Code generation support depends on the underlying ParseUrl expression's codegen capabilities.

Examples¶

-- Extract hostname from URL
SELECT try_parse_url('https://spark.apache.org/docs', 'HOST');
-- Result: 'spark.apache.org'

-- Handle malformed URL gracefully
SELECT try_parse_url('not-a-valid-url', 'HOST');
-- Result: null

-- Extract query parameter
SELECT try_parse_url('https://example.com?param1=value1&param2=value2', 'QUERY', 'param1');
-- Result: 'value1'

// DataFrame API usage
import org.apache.spark.sql.functions._

df.select(expr("try_parse_url(url_column, 'HOST')").as("hostname"))

// Using with when() for null handling
df.select(
  when(expr("try_parse_url(url_column, 'HOST')").isNotNull, 
       expr("try_parse_url(url_column, 'HOST')"))
    .otherwise("unknown_host")
    .as("safe_hostname")
)