# Optional advanced Snakemake skeleton for a lab analysis.
#
# Why this exists: it is the dependency graph between the agent and your outputs.
# Start with a documented script; use this when you need explicit dependencies.
# See pages/08-reproducibility.md. This is not runnable as supplied: add config.yaml,
# the referenced scripts and approved inputs. Declare script/config dependencies
# for each rule before using it to track changes.
#
#   snakemake -n                 dry run: what would rebuild, and why
#   snakemake --cores 4 all      reproduce everything
#   snakemake --dag | dot -Tsvg > dag.svg
#
# A rule can describe how a figure is produced. Snakemake is optional, not lab policy.

import os

configfile: "config.yaml"

SUBJECTS = config["subjects"]          # e.g. ["01", "02", "03"]
SEED     = config["seed"]              # one seed, one place
PY       = config["python"]            # full interpreter path


rule all:
    input:
        "results/figures/fig1_thresholds.pdf",
        "results/figures/fig2_metacognition.pdf",
        "results/tables/table1_descriptives.csv",


# ---------------------------------------------------------------------------
# Per-subject preprocessing. raw -> processed. Never writes to data/raw/.
# ---------------------------------------------------------------------------
rule preprocess:
    input:
        raw="data/raw/sub-{subject}/behaviour.csv",
    output:
        clean="data/processed/sub-{subject}_clean.csv",
    log:
        "logs/preprocess_{subject}.log",
    shell:
        "{PY} scripts/preprocess.py "
        "--input {input.raw} --output {output.clean} 2> {log}"


# ---------------------------------------------------------------------------
# Per-subject model fit. Jobs can run in parallel within allocated resources.
# ---------------------------------------------------------------------------
rule fit_subject:
    input:
        clean="data/processed/sub-{subject}_clean.csv",
    output:
        fit="data/processed/sub-{subject}_fit.nc",
    params:
        seed=SEED,
    log:
        "logs/fit_{subject}.log",
    shell:
        "{PY} scripts/fit_subject.py "
        "--input {input.clean} --output {output.fit} "
        "--seed {params.seed} 2> {log}"


rule aggregate:
    input:
        fits=expand("data/processed/sub-{subject}_fit.nc", subject=SUBJECTS),
    output:
        group="data/processed/group_estimates.csv",
    log:
        "logs/aggregate.log",
    shell:
        "{PY} scripts/aggregate.py --inputs {input.fits} --output {output.group} 2> {log}"


# ---------------------------------------------------------------------------
# Outputs. Every figure and table in the paper is a rule.
# ---------------------------------------------------------------------------
rule figures:
    input:
        group="data/processed/group_estimates.csv",
    output:
        fig1="results/figures/fig1_thresholds.pdf",
        fig2="results/figures/fig2_metacognition.pdf",
    log:
        "logs/figures.log",
    shell:
        "{PY} scripts/make_figures.py --input {input.group} "
        "--fig1 {output.fig1} --fig2 {output.fig2} 2> {log}"


rule tables:
    input:
        group="data/processed/group_estimates.csv",
    output:
        t1="results/tables/table1_descriptives.csv",
    log:
        "logs/tables.log",
    shell:
        "{PY} scripts/make_tables.py --input {input.group} --output {output.t1} 2> {log}"


# ---------------------------------------------------------------------------
# Illustrative provenance stamp. Not included in rule all.
# It does not automatically refresh or record input hashes; adapt it per run.
# Preserve the run record in approved storage alongside results.
# ---------------------------------------------------------------------------
rule provenance:
    output:
        "results/provenance.txt",
    shell:
        """
        {{
          echo "commit: $(git rev-parse HEAD)"
          echo "dirty:  $(git status --porcelain | wc -l) uncommitted files"
          echo "date:   $(date -Iseconds)"
          echo "seed:   {SEED}"
          echo "--- packages ---"
          uv pip freeze
        }} > {output}
        """