package analysis

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

// This file implements a deliberately small EXIF reader. Only the fields the
// application actually uses are decoded:
//
//	Make, Model                     IFD0, tags 0x010F / 0x0110
//	DateTimeOriginal                Exif SubIFD, tag 0x9003
//	GPSLatitude / GPSLongitude      GPS SubIFD, tags 0x0002 / 0x0004
//
// EXIF is a TIFF structure embedded in a file, so parsing it means walking an
// image file directory. Every offset read from the data is treated as untrusted
// and bounds-checked, because the input is a user upload.

// errNoEXIF is returned when the payload carries no EXIF block at all. That is
// the normal case for pictures exported from messaging apps and screenshots.
var errNoEXIF = errors.New("no EXIF block found")

// exifHeaderLen is the length of the "Exif\x00\x00" identifier that introduces
// the APP1 payload.
const exifHeaderLen = 6

// searchWindow bounds how far into the file the EXIF identifier is searched.
// APP1 comes near the front of a JPEG; a small window avoids scanning a
// multi-megabyte payload.
const searchWindow = 128 << 10

// TIFF field types from the TIFF 6.0 specification.
const (
	tiffByte      = 1
	tiffASCII     = 2
	tiffShort     = 3
	tiffLong      = 4
	tiffRational  = 5
	tiffUndefined = 7
)

type tiffTag uint16

const (
	tagMake             tiffTag = 0x010F
	tagModel            tiffTag = 0x0110
	tagExifIFDPointer   tiffTag = 0x8769
	tagGPSIFDPointer    tiffTag = 0x8825
	tagDateTimeOriginal tiffTag = 0x9003
	tagGPSLatitudeRef   tiffTag = 0x0001
	tagGPSLatitude      tiffTag = 0x0002
	tagGPSLongitudeRef  tiffTag = 0x0003
	tagGPSLongitude     tiffTag = 0x0004
)

// exifData holds the decoded fields. The zero value means "not present".
type exifData struct {
	Make  string
	Model string
	// TakenAt is the capture time. EXIF records no time zone, so it is
	// interpreted as UTC.
	TakenAt *time.Time
	// Latitude and Longitude are signed decimal degrees.
	Latitude  float64
	Longitude float64
	HasGPS    bool
}

// readEXIF extracts EXIF from a JPEG or PNG payload. A payload without EXIF
// yields errNoEXIF, which callers treat as a normal outcome.
func readEXIF(data []byte) (exifData, error) {
	var out exifData

	block, err := findExifBlock(data)
	if err != nil {
		return out, err
	}
	if err := parseTIFF(block, &out); err != nil {
		return exifData{}, err
	}
	return out, nil
}

// findExifBlock locates the TIFF structure inside the payload. JPEG stores it
// in an APP1 segment and PNG in an "eXIf" chunk; both are preceded by the
// "Exif\x00\x00" identifier in practice, so scanning for that identifier keeps
// the reader format-agnostic.
func findExifBlock(data []byte) ([]byte, error) {
	window := data
	if len(window) > searchWindow {
		window = window[:searchWindow]
	}

	const marker = "Exif\x00\x00"
	for i := 0; i+exifHeaderLen <= len(window); i++ {
		if string(window[i:i+exifHeaderLen]) != marker {
			continue
		}
		start := i + exifHeaderLen
		if start >= len(data) {
			return nil, errNoEXIF
		}
		return data[start:], nil
	}
	return nil, errNoEXIF
}

// parseTIFF walks the directory structure starting at the beginning of block.
func parseTIFF(block []byte, out *exifData) error {
	if len(block) < 8 {
		return errors.New("EXIF: TIFF header is truncated")
	}

	var order binary.ByteOrder
	switch {
	case block[0] == 'I' && block[1] == 'I':
		order = binary.LittleEndian
	case block[0] == 'M' && block[1] == 'M':
		order = binary.BigEndian
	default:
		return errors.New("EXIF: unknown byte order marker")
	}

	if magic := order.Uint16(block[2:4]); magic != 42 {
		return errors.New("EXIF: bad TIFF magic number")
	}
	ifd0 := order.Uint32(block[4:8])

	entries, next, err := readIFD(block, order, ifd0)
	if err != nil {
		return err
	}
	for _, e := range entries {
		switch e.tag {
		case tagMake:
			if s, ok := e.asASCII(block, order); ok {
				out.Make = s
			}
		case tagModel:
			if s, ok := e.asASCII(block, order); ok {
				out.Model = s
			}
		case tagExifIFDPointer:
			if sub, _, err := readIFD(block, order, uint32(e.firstValue(order))); err == nil {
				readExifSubIFD(block, order, sub, out)
			}
		case tagGPSIFDPointer:
			if sub, _, err := readIFD(block, order, uint32(e.firstValue(order))); err == nil {
				readGPSIFD(block, order, sub, out)
			}
		}
	}
	_ = next
	return nil
}

// entry is one directory entry.
type entry struct {
	tag   tiffTag
	typ   uint16
	count uint32
	// inline is the 4-byte value field as read, used when the payload fits.
	inline [4]byte
}

func readExifSubIFD(block []byte, order binary.ByteOrder, entries []entry, out *exifData) {
	for _, e := range entries {
		if e.tag != tagDateTimeOriginal {
			continue
		}
		if s, ok := e.asASCII(block, order); ok {
			if t, err := time.Parse(exifTimeLayout, s); err == nil {
				utc := t.UTC()
				out.TakenAt = &utc
			}
		}
	}
}

func readGPSIFD(block []byte, order binary.ByteOrder, entries []entry, out *exifData) {
	var (
		lat, lon       float64
		latOK, lonOK   bool
		latRef, lonRef string
	)

	for _, e := range entries {
		switch e.tag {
		case tagGPSLatitudeRef:
			if s, ok := e.asASCII(block, order); ok {
				latRef = s
			}
		case tagGPSLongitudeRef:
			if s, ok := e.asASCII(block, order); ok {
				lonRef = s
			}
		case tagGPSLatitude:
			if v, ok := e.asRationals(block, order); ok && len(v) >= 3 {
				lat, latOK = dmsToDecimal(v[0], v[1], v[2])
			}
		case tagGPSLongitude:
			if v, ok := e.asRationals(block, order); ok && len(v) >= 3 {
				lon, lonOK = dmsToDecimal(v[0], v[1], v[2])
			}
		}
	}

	if latOK && lonOK {
		// Southern latitudes and western longitudes are stored as magnitudes
		// with a separate hemisphere reference.
		if strings.EqualFold(latRef, "S") {
			lat = -lat
		}
		if strings.EqualFold(lonRef, "W") {
			lon = -lon
		}
		if lat >= -90 && lat <= 90 && lon >= -180 && lon <= 180 {
			out.Latitude = lat
			out.Longitude = lon
			out.HasGPS = true
		}
	}
}

// readIFD reads the directory at the given offset and returns its entries plus
// the offset of the next directory, which is unused in practice.
func readIFD(block []byte, order binary.ByteOrder, offset uint32) ([]entry, uint32, error) {
	// Each entry is 12 bytes and a directory of more than a few hundred
	// entries indicates a malformed or hostile file.
	const maxEntries = 512

	if !inRange(block, offset, 2) {
		return nil, 0, errors.New("EXIF: directory offset is out of range")
	}
	count := order.Uint16(block[offset : offset+2])
	if int(count) > maxEntries {
		return nil, 0, errors.New("EXIF: implausible directory entry count")
	}

	entries := make([]entry, 0, count)
	base := int(offset) + 2
	for i := 0; i < int(count); i++ {
		at := base + i*12
		if at < 0 || !inRange(block, uint32(at), 12) {
			return nil, 0, errors.New("EXIF: directory entry is out of range")
		}
		rec := block[at : at+12]
		e := entry{
			tag:   tiffTag(order.Uint16(rec[0:2])),
			typ:   order.Uint16(rec[2:4]),
			count: order.Uint32(rec[4:8]),
		}
		copy(e.inline[:], rec[8:12])
		entries = append(entries, e)
	}

	nextAt := base + int(count)*12
	var next uint32
	if nextAt >= 0 && inRange(block, uint32(nextAt), 4) {
		next = order.Uint32(block[nextAt : nextAt+4])
	}
	return entries, next, nil
}

// exifTimeLayout is the non-conforming timestamp format used by EXIF.
const exifTimeLayout = "2006:01:02 15:04:05"

// typeSize returns the byte width of one element of a TIFF type.
//
// RATIONAL and SRATIONAL are two LONGs, so they are eight bytes rather than
// four. Getting this wrong makes every subsequent offset in the directory wrong,
// which is why it is worth stating explicitly.
func typeSize(typ uint16) (int, bool) {
	switch typ {
	case tiffByte, tiffASCII, tiffUndefined:
		return 1, true
	case tiffShort:
		return 2, true
	case tiffLong:
		return 4, true
	case tiffRational:
		return 8, true
	default:
		return 0, false
	}
}

// payload locates the entry's value, which is either inline in the 4-byte value
// field or at a pointer stored in that field.
func (e entry) payload(block []byte, order binary.ByteOrder) ([]byte, bool) {
	size, ok := typeSize(e.typ)
	if !ok {
		return nil, false
	}
	total := uint64(e.count) * uint64(size)
	if total == 0 {
		return nil, false
	}
	// More than a few kilobytes of "value" means a malformed file; the fields
	// this application reads are all short.
	if total > 4096 {
		return nil, false
	}

	if total <= 4 {
		return e.inline[:total], true
	}
	offset := e.firstValue(order)
	if !inRange(block, offset, uint32(total)) {
		return nil, false
	}
	return block[int(offset) : int(offset)+int(total)], true
}

func (e entry) firstValue(order binary.ByteOrder) uint32 {
	return order.Uint32(e.inline[:4])
}

func (e entry) asASCII(block []byte, order binary.ByteOrder) (string, bool) {
	if e.typ != tiffASCII {
		return "", false
	}
	raw, ok := e.payload(block, order)
	if !ok {
		return "", false
	}
	s := string(raw)
	if i := strings.IndexByte(s, 0); i >= 0 {
		s = s[:i]
	}
	return strings.TrimSpace(s), true
}

// asRationals decodes an array of unsigned rationals. Each is a numerator and
// denominator pair, and a zero denominator makes the entry unusable.
func (e entry) asRationals(block []byte, order binary.ByteOrder) ([]float64, bool) {
	if e.typ != tiffRational {
		return nil, false
	}
	raw, ok := e.payload(block, order)
	if !ok || len(raw)%8 != 0 {
		return nil, false
	}
	n := len(raw) / 8
	out := make([]float64, 0, n)
	for i := 0; i < n; i++ {
		num := order.Uint32(raw[i*8 : i*8+4])
		den := order.Uint32(raw[i*8+4 : i*8+8])
		if den == 0 {
			return nil, false
		}
		out = append(out, float64(num)/float64(den))
	}
	return out, true
}

// dmsToDecimal converts EXIF's degrees/minutes/seconds triple to decimal
// degrees. It reports false for out-of-range or non-finite values.
func dmsToDecimal(degrees, minutes, seconds float64) (float64, bool) {
	if math.IsNaN(degrees) || math.IsNaN(minutes) || math.IsNaN(seconds) {
		return 0, false
	}
	if degrees < 0 || degrees > 180 {
		return 0, false
	}
	if minutes < 0 || minutes >= 60 {
		return 0, false
	}
	if seconds < 0 || seconds >= 60 {
		return 0, false
	}
	v := degrees + minutes/60 + seconds/3600
	return v, true
}

func inRange(block []byte, offset, length uint32) bool {
	end := uint64(offset) + uint64(length)
	return end <= uint64(len(block))
}

func math_isNaN(v float64) bool { return v != v }