shawn/cantaloupe-downloader
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Add diff tracking — compare exports between runs

Browse Source 
- New cantaloupe/diff.py: diff engine comparing xlsx by Asset ID,
  manifest tracking at ~/.cantaloupe-exports/manifest.json
- New  subcommand: python -m cantaloupe diff [--from FILE] [--to FILE]
- Export --diff flag: auto-compare with previous export after download
- Every export auto-records to manifest for history tracking
- 44 new tests, 130/130 pass total
This commit is contained in:
Shawn
2026-05-21 18:53:11 -04:00
parent 9c3d3ef7ed
commit e408867aac
3 changed files with 1178 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
cantaloupe/cli.py
+168
View File
@@ -3,11 +3,15 @@ Click CLI for cantaloupe export tool.
Commands:
export Download machine list Excel from mycantaloupe.com
diff Compare two machine list exports and show what changed
configure Save credentials to ~/.cantaloupe.env
Usage:
python -m cantaloupe export --output ./exports/
python -m cantaloupe export --scheduled --output /data/exports/
python -m cantaloupe export --diff # auto-compare with previous export
python -m cantaloupe diff --from old.xlsx --to new.xlsx
python -m cantaloupe diff # compare two most recent exports
python -m cantaloupe configure
"""
@@ -16,6 +20,7 @@ import os
import sys
import time
from pathlib import Path
from typing import Optional
import click
@@ -83,6 +88,54 @@ def _resolve_output(output: str | None, scheduled: bool) -> Path:
return out_path
def _print_diff_result(result: dict, from_label: str, to_label: str) -> None:
"""Pretty-print a diff result to the console."""
click.echo()
click.echo("═" * 60)
click.echo(" Diff Report")
click.echo("═" * 60)
click.echo(f" From : {from_label} ({result['from_count']} rows)")
click.echo(f" To : {to_label} ({result['to_count']} rows)")
click.echo(f" Key : {result['id_column']}")
click.echo()
added = result["added_count"]
removed = result["removed_count"]
changed = result["changed_count"]
if added == 0 and removed == 0 and changed == 0:
click.echo(" ✅ No changes — exports are identical.")
else:
if added:
click.echo(f" 🟢 {added} machine(s) added")
for row in result["added"][:10]:
rid = row.get(result["id_column"], "?")
click.echo(f" + {rid}")
if added > 10:
click.echo(f" ... and {added - 10} more")
if removed:
click.echo(f" 🔴 {removed} machine(s) removed")
for row in result["removed"][:10]:
rid = row.get(result["id_column"], "?")
click.echo(f" - {rid}")
if removed > 10:
click.echo(f" ... and {removed - 10} more")
if changed:
click.echo(f" 🟡 {changed} machine(s) changed")
for entry in result["changed"][:5]:
rid = entry["id"]
field_names = list(entry["changes"].keys())
click.echo(f" ~ {rid} ({', '.join(field_names[:5])})")
if len(field_names) > 5:
click.echo(f" +{len(field_names) - 5} more fields")
if changed > 5:
click.echo(f" ... and {changed - 5} more")
click.echo("═" * 60)
# ── CLI group ───────────────────────────────────────────────────────────────
@@ -137,6 +190,12 @@ def cli(ctx: click.Context, verbose: bool) -> None:
is_flag=True,
help="Non-interactive mode for cron usage — skips credential prompts, uses defaults.",
)
@click.option(
"--diff/--no-diff",
"diff_flag",
default=False,
help="Compare with the previous export and show changes after download.",
)
@click.option(
"--email",
default=None,
@@ -155,6 +214,7 @@ def export(
output: str | None,
headless: bool,
scheduled: bool,
diff_flag: bool,
email: str | None,
password: str | None,
) -> None:
@@ -163,6 +223,8 @@ def export(
Authenticates with your Cantaloupe account and downloads the
VueMachineList ExcelExport file, reporting file size, row count,
and elapsed time on completion.
Use --diff to auto-compare with the previous export after download.
"""
# Resolve credentials (CLI flags > env vars > config file)
resolved_email = email or ctx.obj.get("email", "")
@@ -238,6 +300,32 @@ def export(
click.echo(f" Rows : {row_count:,}")
click.echo(f" Time : {elapsed:.1f}s")
# ── record to manifest ─────────────────────────────────────────────
from cantaloupe.diff import record_export
# Snapshot the previous export *before* recording this one, so
# we diff against the run-before-last, not this run.
previous_path: Path | None = None
if diff_flag:
from cantaloupe.diff import get_previous_export
previous_path = get_previous_export()
record_export(out_path, row_count, file_size)
# ── diff against previous export ────────────────────────────────────
if diff_flag:
if previous_path is None:
click.echo("\n⚠ No previous export found — skipping diff.")
elif previous_path.resolve() == out_path.resolve():
click.echo("\n⚠ Previous export is the same file — skipping diff.")
else:
try:
from cantaloupe.diff import diff_excels
result = diff_excels(previous_path, out_path)
_print_diff_result(result, previous_path.name, out_path.name)
except Exception as e:
click.echo(f"\n⚠ Diff failed: {e}", err=True)
# ── configure command ───────────────────────────────────────────────────────
@@ -268,6 +356,86 @@ def configure(email: str, password: str) -> None:
click.echo(f" You can now run: python -m cantaloupe export")
# ── diff command ─────────────────────────────────────────────────────────────
@cli.command()
@click.option(
"--from", "from_path",
default=None,
help="Path to the older xlsx file (default: second-most-recent export).",
)
@click.option(
"--to", "to_path",
default=None,
help="Path to the newer xlsx file (default: most recent export).",
)
@click.option(
"--id-column",
default=None,
help="Column name to use as unique machine identifier (auto-detected if omitted).",
)
@click.option(
"--json", "json_output",
is_flag=True,
help="Output the diff result as JSON instead of pretty-printed text.",
)
def diff(
from_path: str | None,
to_path: str | None,
id_column: str | None,
json_output: bool,
) -> None:
"""Compare two machine list Excel exports and show what changed.
If --from and --to are omitted, the two most recent exports are used
(tracked via ~/.cantaloupe-exports/manifest.json).
Reports machines added, removed, and changed — keyed on the Asset ID
column (auto-detected, or specified with --id-column).
"""
from cantaloupe.diff import diff_excels, get_latest_exports
# Resolve file paths
if from_path and to_path:
f_path = Path(os.path.expanduser(from_path)).resolve()
t_path = Path(os.path.expanduser(to_path)).resolve()
elif from_path or to_path:
raise click.UsageError(
"Specify both --from and --to, or neither (to use latest exports)."
)
else:
latest = get_latest_exports(2)
if len(latest) < 2:
click.echo(
"ERROR: Need at least two exports to diff. "
"Run an export first, or specify --from and --to explicitly.",
err=True,
)
raise SystemExit(1)
f_path = latest[1] # older
t_path = latest[0] # newer
if not f_path.exists():
click.echo(f"ERROR: File not found: {f_path}", err=True)
raise SystemExit(1)
if not t_path.exists():
click.echo(f"ERROR: File not found: {t_path}", err=True)
raise SystemExit(1)
try:
result = diff_excels(f_path, t_path, id_column=id_column)
except (ValueError, FileNotFoundError) as e:
click.echo(f"ERROR: {e}", err=True)
raise SystemExit(1)
if json_output:
import json
click.echo(json.dumps(result, indent=2, default=str))
else:
_print_diff_result(result, f_path.name, t_path.name)
# ── direct invocation ───────────────────────────────────────────────────────
if __name__ == "__main__":