Update cli.py

Added time feature,
added ability to specify the input file encoding and made the new default input file encoding utf-8,
removed singular and plural features,
added output to file option (mainly json),
added ability to name output file,
made help text much better, added examples, fixed bad formatting,
added new aliases for help options and for help like -h and -o, added defaults for options,
added more changes that I don't really remember.

Author: datsom1

csv_diff/cli.py

@@ -1,10 +1,13 @@
 import click
 import json as std_json
+import time as time_module
 from . import load_csv, load_json, compare, human_text
 
-
-@click.command()
-@click.version_option()
+@click.command(
+    context_settings={"help_option_names": ["-h", "--help"]},
+    help="Compare the differences between two CSV or JSON files.",
+    epilog="Example: csv-diff old.csv new.csv --key=Id --output=jsonfile --output-file=diff.json",
+)
 @click.argument(
     "previous",
     type=click.Path(exists=True, file_okay=True, dir_okay=False, allow_dash=False),
@@ -14,28 +17,13 @@
     type=click.Path(exists=True, file_okay=True, dir_okay=False, allow_dash=False),
 )
 @click.option(
-    "--key", type=str, default=None, help="Column to use as a unique ID for each row"
+    "-k", "--key", type=str, default=None, help="Column to use as a unique ID for each row (default: first column if not specified) (Example: --key=Id)"
 )
 @click.option(
     "--format",
-    type=click.Choice(["csv", "tsv", "json"]),
-    default=None,
-    help="Explicitly specify input format (csv, tsv, json) instead of auto-detecting",
-)
-@click.option(
-    "--json", type=bool, default=False, help="Output changes as JSON", is_flag=True
-)
-@click.option(
-    "--singular",
-    type=str,
+    "input_format",
     default=None,
-    help="Singular word to use, e.g. 'tree' for '1 tree'",
-)
-@click.option(
-    "--plural",
-    type=str,
-    default=None,
-    help="Plural word to use, e.g. 'trees' for '2 trees'",
+    help="Explicitly specify input format. Available (csv|tsv|json) (default: auto-detect based on file extension)",
 )
 @click.option(
     "--show-unchanged",
@@ -49,34 +37,89 @@
     multiple=True,
     help="key: format string - define extra fields to display",
 )
-def cli(previous, current, key, format, json, singular, plural, show_unchanged, extras):
-    "Diff two CSV or JSON files"
+@click.option(
+    "--encoding",
+    type=str,
+    default="utf-8",
+    help="Input File Encoding. Available: (utf-8|utf-16|utf-16le|utf-16be|latin1|cp1252|ascii|...) (default: utf-8)",
+)
+@click.option(
+    "--time",
+    "show_time",
+    is_flag=True,
+    help="Measure and display elapsed time for the diff operation",
+)
+@click.option(
+    "-o", "--output",
+    default="readable",
+    show_default=True,
+    help="Output format. Available: (readable|json|jsonfile) (default: readable)",
+)
+@click.option(
+    "--output-file",
+    type=click.Path(file_okay=True, dir_okay=False, writable=True, resolve_path=True),
+    default=None,
+    help="File to write JSON output to (only used with --output=jsonfile)",
+)
+@click.version_option()
+def cli(
+    previous, current, key, input_format,
+    show_unchanged, extras, encoding, show_time, output, output_file
+):
+    """Compare the differences between two CSV or JSON files."""
     dialect = {
         "csv": "excel",
         "tsv": "excel-tab",
     }
 
-    if extras and json:
+    if extras and output in ("json", "jsonfile"):
         raise click.UsageError(
             "Extra fields are not supported in JSON output mode",
             ctx=click.get_current_context(),
         )
 
+    if output == "jsonfile" and not output_file:
+        raise click.UsageError(
+            "--output-file must be specified when --output=jsonfile",
+            ctx=click.get_current_context(),
+        )
+
     def load(filename):
-        if format == "json":
-            return load_json(open(filename), key=key)
-        else:
-            return load_csv(
-                open(filename, newline=""), key=key, dialect=dialect.get(format)
-            )
+        try:
+            if input_format == "json":
+                with open(filename, encoding=encoding) as f:
+                    return load_json(f, key=key)
+            else:
+                with open(filename, newline="", encoding=encoding) as f:
+                    return load_csv(f, key=key, dialect=dialect.get(input_format))
+        except Exception as e:
+            raise click.ClickException(f"Failed to load '{filename}': {e}")
 
+    if show_time:
+        start = time_module.time()
+    # Load previous data first to determine key if not set
     previous_data = load(previous)
+    # If key is not set, use the first column from previous_data
+    if key is None:
+        if previous_data and isinstance(previous_data, dict):
+            first_row = next(iter(previous_data.values()))
+            key = list(first_row.keys())[0]
+        else:
+            raise click.ClickException("Could not determine key column from the previous file.")
+    # Now load current data with the determined key
     current_data = load(current)
-
     diff = compare(previous_data, current_data, show_unchanged)
-    if json:
+
+    if output == "json":
         print(std_json.dumps(diff, indent=4))
+    elif output == "jsonfile":
+        with open(output_file, "w", encoding=encoding) as f:
+            f.write(std_json.dumps(diff, indent=4))
     else:
         print(
-            human_text(diff, key, singular, plural, current=current_data, extras=extras)
+            human_text(diff, key, current=current_data, extras=extras)
         )
+
+    if show_time:
+        end = time_module.time()
+        print(f"\nElapsed time: {end - start:.3f} seconds")