Fetching HiRISE RDR `_RED` products with `plp fetch`

Cheat sheet for the most common ways to pull the RED color-band JP2 RDR products. The dotted key is always mro.hirise.rdr. The product ID is the HiRISE observation ID with the _RED suffix appended (e.g. PSP_003092_0985_RED).

1. A single obsid on the command line

The simplest case — pass the full product ID directly:

plp fetch mro.hirise.rdr PSP_003092_0985_RED

The file lands under {storage_root}/mro/hirise/rdr/PSP_003092_0985_RED/. Add --here to drop it into the current directory instead.

2. A handful of obsids on the command line

Variadic positional — pass each fully-qualified product ID. The batch path engages automatically once there’s more than one PID:

plp fetch mro.hirise.rdr PSP_003092_0985_RED ESP_089803_2650_RED
plp fetch mro.hirise.rdr PSP_003092_0985_RED ESP_089803_2650_RED ESP_089671_2650_RED

Downloads run in parallel (default 4 workers; tune with --workers N). The per-PID outcome shows up as a Fetch summary: … line on stderr, with any failures formatted as multi-line blocks (PID on its own line, indented error below).

Note: --pid-suffix _RED is not applied to positional arguments — that flag only activates with --pids-from. With positional PIDs you need to spell out the full _RED suffix on each obsid yourself.

3. A whole CSV file

Your CSV column is named observation_id (lowercase). The auto-detector in pid_column only knows the uppercase canonical names (PRODUCT_ID, FILE_NAME, IMAGE_ID, OBSERVATION_ID), so it can’t pick lowercase observation_id automatically — you have to name the column with --pid-key. Once the column is known, --pid-suffix _RED appends the band to every obsid before the catalog lookup:

plp fetch mro.hirise.rdr \
  --pids-from Kolhar_summer_furrows_HiRISE-20260528053713.csv \
  --pid-key observation_id \
  --pid-suffix _RED

For CSVs where the PID column IS named with one of the canonical uppercase names, you can drop --pid-key:

plp fetch mro.hirise.rdr --pids-from targets_with_PRODUCT_ID.csv --pid-suffix _RED

Useful options:

--workers 8 — increase parallel download threads (default 4).
--report full — show one OK … or FAIL … block per PID on stdout, not just the errors on stderr.
--report jsonl — emit one JSON object per PID; pipe into jq or another downstream tool.

4. Just the first few rows of a file (head pipe)

The head idiom for testing your command on a small slice before committing to the full batch:

head -n 5 Kolhar_summer_furrows_HiRISE-20260528053713.csv \
  | plp fetch mro.hirise.rdr \
      --pids-from - \
      --pid-key observation_id \
      --pid-suffix _RED

--pids-from - reads from stdin. The first-line comma sniff auto-detects this as CSV (since the header line contains commas); --pid-key observation_id is still needed because of the lowercase column name. --pid-suffix _RED applies as in case 3.

The same idiom works with other shell filters:

# First 10 obsids after sorting by date column:
sort -t, -k7 Kolhar_summer_furrows_HiRISE-20260528053713.csv \
  | head -n 10 \
  | plp fetch mro.hirise.rdr --pids-from - --pid-key observation_id --pid-suffix _RED

# Only the COMPLETE-status rows:
grep COMPLETE Kolhar_summer_furrows_HiRISE-20260528053713.csv \
  | plp fetch mro.hirise.rdr --pids-from - --pid-key observation_id --pid-suffix _RED

Recap — when do you need which flag?

Situation	`--pids-from`	`--pid-key`	`--pid-suffix`
1 obsid on the command line	—	—	— (write `_RED` into the PID)
N obsids on the command line	—	—	— (write `_RED` into each PID)
CSV file, column named `PRODUCT_ID` (or other canonical)	path to file	—	`_RED`
CSV file, column named anything else (e.g. `observation_id`)	path to file	column name	`_RED`
Pipe a slice of a CSV via `head` / `grep` / etc.	`-`	same as the file case	same as the file case