package pipeline

import (
	"encoding/json"
	"fmt"
	"sort"
	"strings"

	"textmachine/backend/internal/config"
)

// snapshotdiff.go: naming WHAT moved when a snapshot moves, instead of asserting why.
//
// THE DEFECT THIS CLOSES. The re-pin guard used to tell an operator «the config/prompts changed». On the
// cold run of 31.08 nothing of the sort had happened: config and prompts were byte-identical and what had
// moved was the auto-bank, which the engine had itself WARNed about one purchase earlier. The message sent
// its reader hunting for an edit that did not exist. Under the disclosure law that is §2.2 — a message may
// name only the field it actually COMPARED — and the comparison is free, because the engine stores the
// full payload of every snapshot it ever built (store.SnapshotPayload) and buildSnapshotID hands back the
// current one.
//
// ⛔ DETERMINISTIC BY CONSTRUCTION, and that is a requirement rather than a nicety. The neighbouring
// classifySnapshotMove ranges over a decoded map and returns on the FIRST key that differs: it answers a
// yes/no question, so which key it happens to see first does not matter to it. It matters here — a message
// whose text depends on Go's map iteration order would read differently on two runs over the same two
// snapshots, and an operator comparing two logs would be told two different stories. So every level sorts
// its keys before descending, and the output is a sorted, capped list of dotted paths.
//
// It reports PATHS and never values: a payload carries prompt hashes, model names and embedded-data
// versions, and the guard's job is to say which axis moved, not to print the axis.

// snapshotDiffMax bounds the paths a guard message will list. A move that touches more than a handful of
// axes is a wholesale config change, and naming the first few plus a count is as actionable as naming
// thirty — while an unbounded list would put a screenful of dotted paths inside one error string.
const snapshotDiffMax = 6

// snapshotFieldDiff returns the sorted dotted paths at which the two payloads differ, and whether the
// comparison could be made at all.
//
// ok=false means exactly one thing: the engine could not compare (a payload is missing — an id written by
// an older schema, or a row never upserted — or one of them is not the JSON object this format promises).
// The caller must then say it cannot name the cause. It must NOT fall back to a guess: an unverified cause
// is what this file exists to remove.
func snapshotFieldDiff(storedPayload, currentPayload string) (paths []string, ok bool) {
	if strings.TrimSpace(storedPayload) == "" || strings.TrimSpace(currentPayload) == "" {
		return nil, false
	}
	var was, now any
	if json.Unmarshal([]byte(storedPayload), &was) != nil || json.Unmarshal([]byte(currentPayload), &now) != nil {
		return nil, false
	}
	var out []string
	diffJSONPaths("", was, now, &out)
	sort.Strings(out)
	return out, true
}

// diffJSONPaths walks two decoded JSON values in lockstep and appends the dotted path of every leaf that
// differs. Objects descend by SORTED key and arrays by index, so the walk order — and therefore the
// output — is a function of the payloads alone.
//
// A value that changes SHAPE (an object where there was a string, an array that grew) is reported at its
// own path rather than descended into: the axis that moved is the field itself, and enumerating the
// contents of a struct that did not exist before would bury it.
func diffJSONPaths(prefix string, was, now any, out *[]string) {
	switch w := was.(type) {
	case map[string]any:
		n, same := now.(map[string]any)
		if !same {
			*out = append(*out, pathOr(prefix))
			return
		}
		keys := make([]string, 0, len(w)+len(n))
		seen := make(map[string]bool, len(w)+len(n))
		for k := range w {
			keys, seen[k] = append(keys, k), true
		}
		for k := range n {
			if !seen[k] {
				keys = append(keys, k)
			}
		}
		sort.Strings(keys)
		for _, k := range keys {
			wv, hadW := w[k]
			nv, hadN := n[k]
			if !hadW || !hadN { // a component that appeared or was dropped IS the move
				*out = append(*out, join(prefix, k))
				continue
			}
			diffJSONPaths(join(prefix, k), wv, nv, out)
		}
	case []any:
		n, same := now.([]any)
		if !same || len(w) != len(n) {
			*out = append(*out, pathOr(prefix))
			return
		}
		for i := range w {
			diffJSONPaths(fmt.Sprintf("%s[%d]", pathOr(prefix), i), w[i], n[i], out)
		}
	default:
		if !jsonScalarEqual(was, now) {
			*out = append(*out, pathOr(prefix))
		}
	}
}

func join(prefix, key string) string {
	if prefix == "" {
		return key
	}
	return prefix + "." + key
}

// pathOr names the ROOT when a difference is found at the very top — a payload that is not an object at
// all. Without it such a move would be reported as an empty string.
func pathOr(prefix string) string {
	if prefix == "" {
		return "<whole payload>"
	}
	return prefix
}

// jsonScalarEqual compares two decoded JSON scalars. encoding/json decodes every number to float64, so ==
// is exact for the integers and small decimals a snapshot payload carries; nil compares equal only to nil.
func jsonScalarEqual(a, b any) bool { return a == b }

// describeSnapshotMove renders the guard's WHAT-MOVED clause from a diff.
//
// It is a separate function from the guard so the text has ONE definition and can be asserted directly by
// a test, rather than fished out of a formatted error — the D39.171 trap, where an assertion on a
// substring of a shared log buffer stays green while the message it claims to pin has changed.
func describeSnapshotMove(paths []string, ok bool) string {
	if !ok {
		// §2.2: no comparison, no cause. The reader is told what the engine does NOT know, which is
		// actionable (it says «do not go looking for a config edit on my word»), unlike a guess.
		return "the stored snapshot's payload is not available, so WHAT moved cannot be named — do not assume a config or prompt edit"
	}
	if len(paths) == 0 {
		// Two different ids over identical payloads: impossible by construction, which is exactly why it
		// must be said rather than smoothed into a plausible sentence.
		return "the ids differ but no payload field does — the snapshot format and the id are out of step; this is an engine bug, not a config edit"
	}
	shown := paths
	extra := 0
	if len(shown) > snapshotDiffMax {
		shown, extra = shown[:snapshotDiffMax], len(paths)-snapshotDiffMax
	}
	s := "what moved: " + strings.Join(shown, ", ")
	if extra > 0 {
		s += fmt.Sprintf(" and %d more field(s)", extra)
	}
	// The one hint worth carrying, because it changes what the operator DOES: a bank-only move is not an
	// edit anybody made to a file, it is the auto-bank growing between purchases, and the engine warns
	// about it a purchase earlier (mining.go). Naming it here closes the loop between the two messages.
	if len(paths) == 1 && paths[0] == memoryVersionField {
		s += " — that is the BANK, not a config or prompt edit: the auto-bank grew between purchases"
	}
	return s
}

// describeSnapshotMoveFor is the guard's own accessor: it fetches the stored payload, renders the current
// one for the stage's wave, and describes the difference.
//
// EVERY failure degrades to «cannot name it» rather than to a guess, and each degradation is logged with
// its reason. That asymmetry is the whole point: a guard that cannot compare must say so, because the
// alternative — the sentence this function replaced — is a cause the engine never verified, and it costs
// the next session a hunt through an unedited config.
func (r *Runner) describeSnapshotMoveFor(storedID string, st config.Stage) string {
	stored, err := r.Store.SnapshotPayload(storedID)
	if err != nil {
		r.Log.Warn("snapshot guard: the stored payload could not be read, so the moved field cannot be named",
			"book", r.Book.BookID, "stored", storedID, "err", err)
		return describeSnapshotMove(nil, false)
	}
	_, current, err := r.snapshotIDForWave(waveOfStage(r.Pipeline.Stages, st.Name))
	if err != nil {
		r.Log.Warn("snapshot guard: the current payload could not be rendered, so the moved field cannot be named",
			"book", r.Book.BookID, "stage", st.Name, "err", err)
		return describeSnapshotMove(nil, false)
	}
	return describeSnapshotMove(snapshotFieldDiff(stored, current))
}