textmachine/backend/cmd/tmbankprobe/passes.go

package main

import (
	"fmt"
	"os"
	"sort"
	"strings"

	"textmachine/backend/internal/pipeline"
)

// passes.go: the §4.4 instrument — ONE book, its TWO purchases of the bank role, asking what actually
// differed in a row's own input where the role's decision differed.
//
// ⛔ IT READS THE STOP TABLES, and the first version of it read the signature maps instead. That version
// was WRONG and the wrongness is worth keeping written down: the map's note names the chunk COUNT of the
// top draft proposal but never its TEXT — only the other proposals are quoted — so a comparison built on
// it folds the CONSOLIDATED rendering into the evidence and reports «the evidence moved» for every single
// row whose rendering moved. It produced 0 of 12 and 0 of 14 «identical», which is not a fact about the
// runs but a fact about the reader. The stop table carries the drafts, their counts, the frequency, the
// type and the source contexts — the row's whole input, as the role was shown it.
//
// ⚠ WHAT IS STILL OUTSIDE THE ARTIFACT, and it is what the residue of this question is made of: the BATCH
// the row travelled in (its neighbours) and the canon anchor built for that batch. Two rows with identical
// blocks can therefore still have been asked in different company, and the artifacts cannot say so.

// rowInput renders everything the stop table records about a row's INPUT — never the answer. The
// consolidated rendering, the confidence and the INVENTED mark are the role's OUTPUT and are deliberately
// absent: folding an answer into the evidence is exactly how a comparison comes to say «the evidence
// changed» every time the answer did.
func rowInput(r pipeline.BankStopRow, withCtx bool) string {
	parts := make([]string, 0, len(r.Variants))
	for _, v := range r.Variants {
		parts = append(parts, v.Label())
	}
	sort.Strings(parts)
	s := fmt.Sprintf("type=%s origin=%s freq=%d drafts=[%s] evidence=[%s]",
		r.Type, r.Origin, r.Freq, strings.Join(parts, " | "), strings.Join(r.Evidence, "; "))
	if withCtx {
		ctx := append([]string(nil), r.Contexts...)
		sort.Strings(ctx)
		s += " ctx=[" + strings.Join(ctx, " ¦ ") + "]"
	}
	return s
}

// printPassComparison joins a book's two stop tables and reports the numbers the economics of «do not
// re-ask a settled row» stands on.
func printPassComparison(name, first, second string) error {
	a, err := loadSheetRows(first)
	if err != nil {
		return err
	}
	b, err := loadSheetRows(second)
	if err != nil {
		return err
	}
	idx := func(rows []pipeline.BankStopRow) map[string]pipeline.BankStopRow {
		out := make(map[string]pipeline.BankStopRow, len(rows))
		for _, r := range rows {
			out[r.Src] = r
		}
		return out
	}
	ai, bi := idx(a), idx(b)
	var shared []string
	for src := range ai {
		if _, ok := bi[src]; ok {
			shared = append(shared, src)
		}
	}
	sort.Strings(shared)
	fmt.Printf("\n== %s: ONE BOOK, TWO PURCHASES OF THE BANK ROLE\n", name)
	fmt.Printf("   first : %s (%d rows)\n   second: %s (%d rows)\n", first, len(a), second, len(b))
	fmt.Printf("   shared surfaces: %d | only in the first: %d | only in the second: %d\n",
		len(shared), len(ai)-len(shared), len(bi)-len(shared))
	// CONTROLS, printed rather than assumed: a comparison of two documents that carry no drafts and no
	// contexts would find everything identical and mean nothing by it.
	withDrafts, withCtx := 0, 0
	for _, src := range shared {
		if len(ai[src].Variants) > 0 && len(bi[src].Variants) > 0 {
			withDrafts++
		}
		if len(ai[src].Contexts) > 0 && len(bi[src].Contexts) > 0 {
			withCtx++
		}
	}
	fmt.Printf("   control: shared rows carrying draft proposals on BOTH sides: %d of %d · carrying source contexts on both: %d of %d\n",
		withDrafts, len(shared), withCtx, len(shared))

	var changed, sameBlock, sameBlockAndCtx, settled, unresolvedBoth int
	var examples []string
	for _, src := range shared {
		x, y := ai[src], bi[src]
		// ⛔ EVERY SHARED ROW IS CLASSIFIED, and the two classes below are pulled out of «unchanged»
		// deliberately. A row the bank settled was never asked in the second purchase, and a row left
		// unresolved in BOTH was never decided in either: counting them among «rows a do-not-re-ask rule
		// would have saved» would credit the rule with rows nobody re-asked and nobody answered.
		if x.SettledByBank || y.SettledByBank {
			settled++
			continue
		}
		if strings.TrimSpace(x.Dst) == "" && strings.TrimSpace(y.Dst) == "" {
			unresolvedBoth++
			continue
		}
		if x.Dst == y.Dst {
			continue
		}
		changed++
		blockSame := rowInput(x, false) == rowInput(y, false)
		ctxSame := rowInput(x, true) == rowInput(y, true)
		if blockSame {
			sameBlock++
		}
		if ctxSame {
			sameBlockAndCtx++
		}
		if len(examples) < 12 {
			verdict := "the row's own input MOVED"
			switch {
			case ctxSame:
				verdict = "input IDENTICAL, contexts included"
			case blockSame:
				verdict = "block identical, CONTEXTS moved"
			}
			examples = append(examples, fmt.Sprintf("%s: %q → %q — %s", src, x.Dst, y.Dst, verdict))
			if !blockSame {
				examples = append(examples,
					fmt.Sprintf("      before: %s", rowInput(x, false)),
					fmt.Sprintf("      after : %s", rowInput(y, false)))
			}
		}
	}
	fmt.Printf("   rows a purchase never asked about (the bank had settled them): %d — excluded below, they are not re-decisions\n", settled)
	fmt.Printf("   rows left UNRESOLVED in both purchases: %d — also excluded: nobody decided them either time\n", unresolvedBoth)
	fmt.Printf("   rows whose rendering CHANGED between the two purchases: %d of %d shared\n", changed, len(shared))
	fmt.Printf("     of those, the row's own BLOCK was identical (drafts, freq, type, evidence): %d\n", sameBlock)
	fmt.Printf("     of those, identical INCLUDING the source contexts the role was shown:       %d\n", sameBlockAndCtx)
	for _, e := range examples {
		fmt.Printf("     · %s\n", e)
	}
	// The other half of the population, and it is the one the economics is actually about.
	unchanged := len(shared) - changed - settled - unresolvedBoth
	fmt.Printf("   rows whose rendering did NOT change: %d of %d — these are the ones a «do not re-ask what is settled» rule would have saved\n", unchanged, len(shared))
	return nil
}

func loadSheetRows(path string) ([]pipeline.BankStopRow, error) {
	raw, err := os.ReadFile(path)
	if err != nil {
		return nil, err
	}
	rows, err := pipeline.ParseBankStopTable(raw)
	if err != nil {
		return nil, fmt.Errorf("%s: %w", path, err)
	}
	return rows, nil
}