feat: CLI UX improvements for agents

parallel_web_tools/cli/commands.py

@@ -1,8 +1,10 @@
 """CLI commands for Parallel."""
 
+import csv
 import json
 import logging
 import os
+import tempfile
 from typing import Any
 
 import click
@@ -39,6 +41,22 @@
 # =============================================================================
 
 
+def parse_comma_separated(values: tuple[str, ...]) -> list[str]:
+    """Parse a tuple of values that may contain comma-separated items.
+
+    Supports both repeated flags and comma-separated values:
+        --flag a,b --flag c  ->  ['a', 'b', 'c']
+        --flag a --flag b    ->  ['a', 'b']
+        --flag "a,b,c"       ->  ['a', 'b', 'c']
+    """
+    result = []
+    for value in values:
+        # Split by comma and strip whitespace
+        parts = [p.strip() for p in value.split(",")]
+        result.extend(p for p in parts if p)  # Skip empty strings
+    return result
+
+
 def write_json_output(data: dict[str, Any], output_file: str | None, output_json: bool) -> None:
     """Write output data to file and/or stdout as JSON.
 
@@ -132,6 +150,51 @@ def build_config_from_args(
     }
 
 
+def parse_inline_data(data_json: str) -> tuple[str, list[dict[str, str]]]:
+    """Parse inline JSON data and write to a temporary CSV file.
+
+    Args:
+        data_json: JSON string containing array of objects
+
+    Returns:
+        Tuple of (temp_csv_path, inferred_source_columns)
+
+    Raises:
+        click.BadParameter: If JSON is invalid or not an array of objects
+    """
+    try:
+        data = json.loads(data_json)
+    except json.JSONDecodeError as e:
+        raise click.BadParameter(f"Invalid JSON data: {e}") from e
+
+    if not isinstance(data, list):
+        raise click.BadParameter("Data must be a JSON array")
+
+    if len(data) == 0:
+        raise click.BadParameter("Data array cannot be empty")
+
+    if not isinstance(data[0], dict):
+        raise click.BadParameter("Data must be an array of objects")
+
+    # Infer columns from the first row
+    columns = list(data[0].keys())
+    if not columns:
+        raise click.BadParameter("Data objects must have at least one field")
+
+    # Create source_columns with inferred descriptions
+    source_columns = [{"name": col, "description": f"The {col} field"} for col in columns]
+
+    # Write to a temporary CSV file
+    temp_file = tempfile.NamedTemporaryFile(mode="w", suffix=".csv", delete=False, newline="")
+    writer = csv.DictWriter(temp_file, fieldnames=columns)
+    writer.writeheader()
+    for row in data:
+        writer.writerow(row)
+    temp_file.close()
+
+    return temp_file.name, source_columns
+
+
 def suggest_from_intent(
     intent: str,
     source_columns: list[dict[str, str]] | None = None,
@@ -256,8 +319,8 @@ def logout_cmd():
     "--mode", type=click.Choice(["one-shot", "agentic"]), default="one-shot", help="Search mode", show_default=True
 )
 @click.option("--max-results", type=int, default=10, help="Maximum results", show_default=True)
-@click.option("--include-domains", multiple=True, help="Only search these domains")
-@click.option("--exclude-domains", multiple=True, help="Exclude these domains")
+@click.option("--include-domains", multiple=True, help="Only search these domains (comma-separated or repeated)")
+@click.option("--exclude-domains", multiple=True, help="Exclude these domains (comma-separated or repeated)")
 @click.option("--after-date", help="Only results after this date (YYYY-MM-DD)")
 @click.option("-o", "--output", "output_file", type=click.Path(), help="Save results to file (JSON)")
 @click.option("--json", "output_json", is_flag=True, help="Output as JSON")
@@ -291,9 +354,9 @@ def search(
 
         source_policy: dict[str, Any] = {}
         if include_domains:
-            source_policy["include_domains"] = list(include_domains)
+            source_policy["include_domains"] = parse_comma_separated(include_domains)
         if exclude_domains:
-            source_policy["exclude_domains"] = list(exclude_domains)
+            source_policy["exclude_domains"] = parse_comma_separated(exclude_domains)
         if after_date:
             source_policy["after_date"] = after_date
         if source_policy:
@@ -450,6 +513,7 @@ def enrich():
 @click.option("--enriched-columns", help="Enriched columns as JSON")
 @click.option("--intent", help="Natural language description (AI suggests columns)")
 @click.option("--processor", type=click.Choice(AVAILABLE_PROCESSORS), help="Processor to use")
+@click.option("--data", "inline_data", help="Inline JSON data array (alternative to --source)")
 def enrich_run(
     config_file: str | None,
     source_type: str | None,
@@ -459,23 +523,60 @@ def enrich_run(
     enriched_columns: str | None,
     intent: str | None,
     processor: str | None,
+    inline_data: str | None,
 ):
-    """Run data enrichment from YAML config or CLI arguments."""
-    base_args = [source_type, source, target, source_columns]
-    has_cli_args = any(arg is not None for arg in base_args) or enriched_columns or intent
+    """Run data enrichment from YAML config or CLI arguments.
 
-    if config_file and has_cli_args:
-        console.print("[bold red]Error: Provide either a config file OR CLI arguments, not both.[/bold red]")
-        raise click.Abort()
+    You can provide data in three ways:
 
-    if not config_file and not has_cli_args:
-        console.print("[bold red]Error: Provide a config file or CLI arguments.[/bold red]")
-        raise click.Abort()
+    \b
+    1. YAML config file:
+       parallel-cli enrich run config.yaml
 
-    if has_cli_args:
-        validate_enrich_args(source_type, source, target, source_columns, enriched_columns, intent)
+    \b
+    2. CLI arguments with source file:
+       parallel-cli enrich run --source-type csv --source data.csv ...
+
+    \b
+    3. Inline JSON data (no CSV file needed):
+       parallel-cli enrich run --data '[{"company": "Google"}, {"company": "Apple"}]' \\
+           --target output.csv --intent "Find the CEO"
+    """
+    temp_csv_path: str | None = None
 
     try:
+        # Handle inline data - creates a temp CSV and infers source columns
+        if inline_data:
+            if source:
+                console.print("[bold red]Error: Use --data OR --source, not both.[/bold red]")
+                raise click.Abort()
+            if source_type and source_type != "csv":
+                console.print("[bold red]Error: --data only works with CSV output (--source-type csv).[/bold red]")
+                raise click.Abort()
+
+            temp_csv_path, inferred_cols = parse_inline_data(inline_data)
+            source = temp_csv_path
+            source_type = "csv"
+
+            # Use inferred columns if not explicitly provided
+            if not source_columns:
+                source_columns = json.dumps(inferred_cols)
+                console.print(f"[dim]Inferred {len(inferred_cols)} source column(s) from data[/dim]")
+
+        base_args = [source_type, source, target, source_columns]
+        has_cli_args = any(arg is not None for arg in base_args) or enriched_columns or intent
+
+        if config_file and has_cli_args:
+            console.print("[bold red]Error: Provide either a config file OR CLI arguments, not both.[/bold red]")
+            raise click.Abort()
+
+        if not config_file and not has_cli_args:
+            console.print("[bold red]Error: Provide a config file or CLI arguments.[/bold red]")
+            raise click.Abort()
+
+        if has_cli_args:
+            validate_enrich_args(source_type, source, target, source_columns, enriched_columns, intent)
+
         if config_file:
             console.print(f"[bold cyan]Running enrichment from {config_file}...[/bold cyan]\n")
             run_enrichment(config_file)
@@ -519,6 +620,10 @@ def enrich_run(
     except Exception as e:
         console.print(f"[bold red]Error during enrichment: {e}[/bold red]")
         raise
+    finally:
+        # Clean up temp file if we created one
+        if temp_csv_path and os.path.exists(temp_csv_path):
+            os.unlink(temp_csv_path)
 
 
 @enrich.command(name="plan")