# ocli.aikp.ts_mediator (Batch processing)

This is an example/boilerpate AIKP where you can get the inspiration on how to implement the batch processing.

A minimal AIKP that reuses existing CREODIAS finders to fetch satellite products (Sentinel‑1/2, Landsat‑8) and stores the results in the project cache.
It prefers invoking/extending built‑ins over re‑implementing logic.

First of all you need to get a list of products (please see the `ocli/aikp/ts_mediator/cli.py` command `task template products`).
Then batch processing should consist of the 3 states: "Initialization" -> "Processing Loop" -> "Merge results".

1. "Initialization" – Setup filters and all parameters required. Get the product list. Store it in a task. Create ocli sub-environment with sum-task.
2. "Processing Loop" – Run the sub-task which performs the specific analysis (ex. 1 task run 1 time layer).
3. "Merge results" – Merge results from sub-task run into publication ready product.

## `task template products` or How to get filtered products list in one AIKP

```
+----------------------+           +-----------------------------+
| Task (ts_mediator)   |           |  Finder Dispatch            |
|  - sat_family_name   |  selects  |  finder_by_task(task)       |
|  - finder.* filters  +---------->+  (s1/s2/landsat_8)->        |
|  - ROI               |           |  FinderCreodias* class      |
+----------+-----------+           +--------------+--------------+
           |                                      |
           | inject_finder_keys()                 |
           |                                      v
           |                          +-----------------------------+
           |                          | CREODIAS Finder (RESTO API) |
           |                          |  - builds query             |
           |                          |  - requests search.json     |
           |                          |  - parses features -> GDF   |
           |                          +-----------------------------+
           |                                      |
           |                            save_meta_to_cache(.gpkg)
           v                                      v
+----------------------+              +-----------------------------+
| CLI (ts_mediator)    |              | Project .cache (GeoPackage) |
|  task template       |              |  - reused by global product |
|   products           |              |    commands                 |
+----------------------+              +-----------------------------+
                                                   |
                                                   v
                                        product list/show/load (global)
```

Key points:
- Uses `sat_family_name` to choose the correct CREODIAS finder.
- Uses ROI from the task to constrain queries.
- Stores results in project cache (`.cache/finder_*.gpkg`).
- `finder.*` keys live in the task config and override defaults.

### CLI Usage

Prerequisites:
- An active project with ROIs set up.

1) Create a task from this template
- `task create --template ocli.aikp.ts_mediator --roi <roi_name> --name ts_demo --activate`

2) Select a satellite family (optional)
- Sentinel‑2: `task set sat_family_name=s2`
- Sentinel‑1: `task set sat_family_name=s1`
- Landsat‑8: `task set sat_family_name=landsat_8`

3) Set finder filters in task config (optional)
- Date range: `task set finder.startDate=2025-01-01`
- Date range: `task set finder.completionDate=2025-02-01`
- Cloud cover: `task set finder.cloudCover="[0,20]"`

4) Fetch and cache products with the AIKP command
- `task template products --reload --max-items 200 --columns "title,completionDate,cloud_cover"`
- What happens:
  - Uses `finder.*` overrides + task ROI
  - Calls CREODIAS REST API and writes `.gpkg` cache
  - Prints a quick preview table

5) Browse results using global product commands (read from cache)
- `product list --column productId --column startDate --limit 20`
- `product show <PRODUCT_ID>`
- `product load` can also perform the fetch-and-cache step; `ts_mediator` just gives you a lighter, template‑local variant via `products`.

### Notes
- Changing `sat_family_name` switches finder class automatically.
- The cache is per project: `PROJECT/.cache/finder_*.gpkg`.
- You can refine filters at any time via `task set finder.<key>=<value>` and rerun `products`.
- Use `--no-reload` to reuse cached results.

### Troubleshooting
- Empty list: verify ROI geometry exists in the task, date range is valid, and API is reachable.
- Unknown column in `--columns`: omit it or check columns via `product list --columns`.
- Authentication isn’t required for public CREODIAS queries used here; corporate proxies/firewalls may still affect requests.

## `task template sandbox` or How to create ocli sub-environment inside of current task folder

Please see the `ocli/aikp/ts_mediator/cli.py` command `task template sandbox`.

Idea is to invoke ocli commands in nested environment (isolated from current user's environment). 
Basically we can set path to `.tsarrc` config via `--home` parameter. 

To achieve it we next to:

1. Create folders for sub-ocli (see "Sandbox path" and "Sandbox Projects Home").
2. Generate appropriate configration file (see "Sub-ocli configuration file").
3. Create/Initialize sub-task (see "Create sub-task").

Then using method like `invoke_sub_ocli` (see the `ocli/aikp/ts_mediator/cli.py`) 
in the initialized sub-ocli environment you can run any predefined sequence of commands and have access to their results.

To save the execution state - basic state machine should be implemented.