create-job

name	create-job
description	Creates a Siril job file for a target by scanning the data folder structure, reading FITS headers for temperature, and finding matching calibration files.
allowed-tools	["Bash","Read","Glob","Write"]

Create Astro Job File

Instructions

Given a target name (e.g., M45), create a job JSON file by scanning the data folders.

Step 1: Get Base Path

Read settings.json in the project root to get base_path.

Step 2: Scan Target Folders

List date folders for the target:

ls "{base_path}/{target}"

For each date folder, list filter/exposure subfolders:

ls "{base_path}/{target}/{date}"

Filter folders are named {filter}{exposure} (e.g., L180, R300, Ha180, S180).

Step 3: Read FITS Temperature

Read the sensor temperature from a FITS file header. Pick any .fit or .fits file from a light folder:

ls "{base_path}/{target}/{date}/{filter_folder}"

Then read the temperature:

uv run python -c "from astropy.io import fits; h=fits.getheader(r'{path_to_fit_file}'); print(h.get('CCD-TEMP') or h.get('SET-TEMP'))"

Round to nearest integer (e.g., -9.8 -> -10).

Step 4: Find Calibration Files

List available calibration dates:

ls "{base_path}/calibration/raw/biases"
ls "{base_path}/calibration/raw/darks"
ls "{base_path}/calibration/raw/flats"

Dark folders include temperature in the name: {YYYY_MM_DD}_{temp}C (e.g., 2025_01_23_-10C).

Calibration matching:

• Bias: use most recent date
• Darks: find folders matching the light temperature, use most recent
• Flats: use most recent date

If no darks exist at the target temperature, use the closest available and set dark_temp_override in options. Otherwise, do not set dark_temp_override.

Step 5: Determine Job Type

Based on filters found:

• Has L (with or without R,G,B, Ha, etc.) -> LRGB
• R, G, B only (no L) -> RGB
• S, H, O (narrowband only) -> SHO
• H, O only -> HOO

Step 6: Generate Job Name

Format: {target}_{season}{year}

Determine season from the most recent light date:

• Dec, Jan, Feb -> Winter
• Mar, Apr, May -> Spring
• Jun, Jul, Aug -> Summer
• Sep, Oct, Nov -> Fall

Use the year of the most recent session.

Step 7: Write Job File

Write to jobs/{target}.json.

Template

{
  "name": "{target}_{season}{year}",
  "type": "LRGB",
  "calibration": {
    "bias": "2024_08_05",
    "darks": "2025_01_23",
    "flats": "2025_01_19"
  },
  "lights": {
    "L": [
      "{target}/{date1}/L180",
      "{target}/{date2}/L180"
    ],
    "R": [
      "{target}/{date1}/R180",
      "{target}/{date2}/R180"
    ],
    "G": [
      "{target}/{date1}/G180",
      "{target}/{date2}/G180"
    ],
    "B": [
      "{target}/{date1}/B180",
      "{target}/{date2}/B180"
    ]
  },
  "output": "{target}/processed",
  "options": {}
}

Examples

LRGB job (M31):

• Filters: L, R, G, B across multiple nights
• Type: LRGB

Narrowband job (IC443):

• Filters: S, H, O
• Type: SHO
• Options include "palette": "SHO"

LRGB + Ha:

• Filters: L, R, G, B, Ha
• Type: LRGB
• Include Ha key in lights alongside L, R, G, B

Step 8: Run Job and Monitor

Run the job:

uv run python run_job.py jobs/{target}.json

Other run options:

uv run python run_job.py jobs/{target}.json --validate  # Validate only
uv run python run_job.py jobs/{target}.json --dry-run   # Dry run
uv run python run_job.py jobs/{target}.json --stage preprocess  # Run specific stage

Monitor the log file in logs/{target}_{timestamp}.log for:

• Calibration file matching issues
• Frame rejection (FWHM filtering)
• Stacking errors
• Any warnings or failures

After the job completes, review the log output and report any issues to the user.