// Package aep is a minimal, stdlib-only reader for After Effects project (.aep)
// files. AEP is a RIFX container (big-endian RIFF). This reader does NOT fully
// decode the project — it walks the chunk tree to extract composition names and
// (best-effort) durations, which is enough for a headless "quick scan" that
// scaffolds scenes without After Effects. Full fidelity (layers, colours, fonts)
// requires the AE-JSX scanner running on a node.
package aep

import (
	"encoding/binary"
	"errors"
	"strings"
)

// Comp is a composition discovered in the project.
type Comp struct {
	Name        string
	DurationSec float64  // 0 when it could not be derived
	Layers      []string // layer names in file order (used by FIX-mode scanning)
}

// ParseComps walks a RIFX (.aep) buffer and returns its compositions.
//
// A composition is an "Item" LIST that contains a "cdta" (composition data)
// chunk; its name is the Item's "Utf8" chunk. Folders and footage items lack
// "cdta" and are skipped. This rule is robust across AE versions because only
// comps carry cdta.
func ParseComps(data []byte) ([]Comp, error) {
	if len(data) < 12 || string(data[0:4]) != "RIFX" {
		return nil, errors.New("not a RIFX/.aep file")
	}
	// data[4:8] = file size (BE), data[8:12] = form type ("Egg!"). Body follows.
	var comps []Comp
	seen := map[string]bool{}
	walk(data[12:], &comps, seen)
	return comps, nil
}

// ParseNames returns every non-empty "Utf8" string in the project tree — layer
// names, item/footage names, comp names (and incidentally some text/expression
// strings). FIX-mode scanning filters this for the frl_/frd_ naming convention,
// which is reliable because those names are globally unique and distinctive.
//
// This is needed because FIX media placeholders are renamed *footage items*
// (e.g. frl_c1m1), not layers, so a layers-only walk misses them. Order is file
// order; duplicates are preserved (callers dedup).
func ParseNames(data []byte) ([]string, error) {
	if len(data) < 12 || string(data[0:4]) != "RIFX" {
		return nil, errors.New("not a RIFX/.aep file")
	}
	var names []string
	collectUtf8(data[12:], &names)
	return names, nil
}

func collectUtf8(buf []byte, out *[]string) {
	off := 0
	for off+8 <= len(buf) {
		id := string(buf[off : off+4])
		size := int(binary.BigEndian.Uint32(buf[off+4 : off+8]))
		ds := off + 8
		de := ds + size
		if size < 0 || de > len(buf) {
			break
		}
		if id == "Utf8" {
			if s := clean(buf[ds:de]); s != "" {
				*out = append(*out, s)
			}
		} else if id == "LIST" && size >= 4 {
			collectUtf8(buf[ds+4:de], out)
		}
		adv := 8 + size
		if size%2 == 1 {
			adv++
		}
		off += adv
	}
}

// walk iterates the chunks in buf (a LIST body or the file root), recursing into
// every LIST so nested items inside folders are found.
func walk(buf []byte, comps *[]Comp, seen map[string]bool) {
	off := 0
	for off+8 <= len(buf) {
		id := string(buf[off : off+4])
		size := int(binary.BigEndian.Uint32(buf[off+4 : off+8]))
		ds := off + 8
		de := ds + size
		if size < 0 || de > len(buf) {
			break
		}
		if id == "LIST" && size >= 4 {
			listType := string(buf[ds : ds+4])
			body := buf[ds+4 : de]
			if listType == "Item" {
				handleItem(body, comps, seen)
			}
			walk(body, comps, seen) // recurse (folders hold nested Item LISTs)
		}
		adv := 8 + size
		if size%2 == 1 {
			adv++ // chunks are word-aligned
		}
		off += adv
	}
}

// handleItem inspects the DIRECT children of an "Item" LIST: a "cdta" marks it as
// a composition, the first "Utf8" is its name.
func handleItem(body []byte, comps *[]Comp, seen map[string]bool) {
	var name string
	var cdta []byte
	hasCdta := false
	var layers []string

	off := 0
	for off+8 <= len(body) {
		id := string(body[off : off+4])
		size := int(binary.BigEndian.Uint32(body[off+4 : off+8]))
		ds := off + 8
		de := ds + size
		if size < 0 || de > len(body) {
			break
		}
		switch {
		case id == "cdta":
			hasCdta = true
			cdta = body[ds:de]
		case id == "Utf8":
			if name == "" {
				name = clean(body[ds:de])
			}
		case id == "LIST" && size >= 4 && string(body[ds:ds+4]) == "Layr":
			// A composition's layers are nested "Layr" LISTs; the layer's display
			// name is the first "Utf8" inside it (FIX templates encode scenes in
			// these names, e.g. frl_c1t1). Empty/unnamed layers are skipped.
			if ln := layerName(body[ds+4 : de]); ln != "" {
				layers = append(layers, ln)
			}
		}
		adv := 8 + size
		if size%2 == 1 {
			adv++
		}
		off += adv
	}

	if hasCdta && name != "" && !seen[name] {
		seen[name] = true
		*comps = append(*comps, Comp{Name: name, DurationSec: durationFromCdta(cdta), Layers: layers})
	}
}

// layerName returns the first "Utf8" string among the DIRECT children of a "Layr"
// LIST body — that is the layer's display name. It does not recurse, so nested
// property/effect names (also stored as Utf8) cannot be mistaken for the name.
func layerName(body []byte) string {
	off := 0
	for off+8 <= len(body) {
		id := string(body[off : off+4])
		size := int(binary.BigEndian.Uint32(body[off+4 : off+8]))
		ds := off + 8
		de := ds + size
		if size < 0 || de > len(body) {
			break
		}
		if id == "Utf8" {
			return clean(body[ds:de])
		}
		adv := 8 + size
		if size%2 == 1 {
			adv++
		}
		off += adv
	}
	return ""
}

func clean(b []byte) string {
	return strings.TrimSpace(strings.TrimRight(string(b), "\x00"))
}

// durationFromCdta reads the comp duration (in seconds) from the cdta chunk.
//
// cdta layout (verified empirically against AE 2024/2026 project files):
//
//	off  4  u32  framerate divisor
//	off  8  u32  framerate dividend == comp time base (ticks/sec)
//	off 44  u32  duration numerator (in time-base ticks)
//	off 48  u32  duration divisor   (== time base)
//
// duration_seconds = numerator / time_base. The duration's byte offset shifts by
// AE version (44 on recent builds, 40 on older ones documented by the boltframe
// reference parser), so rather than hard-code one offset we accept whichever
// places the time base immediately *after* the numerator — that self-selects the
// correct field and rejects garbage. Returns 0 (unknown) on any doubt; the
// importer treats 0 as "leave duration unset".
func durationFromCdta(cdta []byte) float64 {
	if len(cdta) < 52 {
		return 0
	}
	frDivisor := binary.BigEndian.Uint32(cdta[4:8])
	timeBase := binary.BigEndian.Uint32(cdta[8:12]) // framerate dividend
	if frDivisor == 0 || timeBase == 0 {
		return 0
	}
	if fps := float64(timeBase) / float64(frDivisor); fps < 1 || fps > 240 {
		return 0 // not a comp cdta we recognise
	}
	for _, off := range []int{44, 40} { // recent layout first, then older
		num := binary.BigEndian.Uint32(cdta[off : off+4])
		div := binary.BigEndian.Uint32(cdta[off+4 : off+8])
		if num == 0 || div != timeBase {
			continue // divisor must equal the time base to be the duration field
		}
		sec := float64(num) / float64(timeBase)
		if sec <= 0 || sec > 36000 { // > 10h is nonsense
			continue
		}
		return float64(int(sec*100+0.5)) / 100 // round to 2 dp
	}
	return 0
}