vyatsu-schedule/app/src/main/java/agzam/vyatsu/data/remote/TeachersParser.kt

package agzam.vyatsu.data.remote

import agzam.vyatsu.domain.model.DateRange
import agzam.vyatsu.domain.model.Department
import agzam.vyatsu.domain.model.ParsedDepartments
import agzam.vyatsu.domain.model.Season
import java.io.InputStream

/**
 * Стриминговый парсер страницы преподавателей: читает тело ответа по чанкам и
 * разбирает его на лету, не материализуя весь HTML в память.
 * Страница teacher.html весит ~2.4MB, поэтому весь документ грузить нельзя.
 */
object TeachersParser {

    private val fakNameTag = Regex("""<div[^>]*class="fak_name"""")
    private val fakIdInTag = Regex("""data-fak_id="(\d+)"""")
    private val kafPeriodTag = Regex("""<div[^>]*class="kafPeriod"""")
    private val kafIdInTag = Regex("""data-kaf_period_id="(\d+)"""")
    private val prepodLink = Regex("""href="/reports/schedule/prepod/(\d+)_([12])_(\d{8})_(\d{8})\.html"""")
    private val tagStrip = Regex("""<[^>]*>""")

    private const val CHUNK_SIZE = 8192
    private const val MAX_SLICE = 1 shl 16      // 64 KiB — защита от аномально длинных единиц
    private const val MAX_PENDING = 1 shl 20    // 1 MiB — жёсткий потолок буфера

    fun parseStream(input: InputStream): ParsedDepartments {
        val reader = input.bufferedReader(Charsets.UTF_8)
        val buf = CharArray(CHUNK_SIZE)
        val pending = StringBuilder(CHUNK_SIZE * 2)
        val departments = ArrayList<Department>()
        val ranges = HashMap<String, HashMap<Season, MutableList<DateRange>>>()
        var faculty = ""

        fun addRanges(text: String) {
            for (m in prepodLink.findAll(text)) {
                val departmentId = m.groupValues[1]
                val season = if (m.groupValues[2] == "1") Season.AUTUMN else Season.SPRING
                val range = DateRange(m.groupValues[3], m.groupValues[4])
                ranges.getOrPut(departmentId) {
                    hashMapOf(Season.AUTUMN to ArrayList(), Season.SPRING to ArrayList())
                }.getOrPut(season) { ArrayList() }.add(range)
            }
        }

        while (true) {
            val n = reader.read(buf)
            if (n < 0) break
            pending.append(buf, 0, n)

            var pos = 0
            while (pos < pending.length) {
                val divAt = pending.indexOf("<div", pos)
                val aAt = pending.indexOf("<a", pos)
                val tagStart = when {
                    divAt < 0 -> aAt
                    aAt < 0 -> divAt
                    else -> minOf(divAt, aAt)
                }
                if (tagStart < 0) {
                    pos = keepPartialTagTail(pending, pos)
                    break
                }
                if (tagStart > pos) {
                    addRanges(pending.substring(pos, tagStart))
                }
                val tagEnd = findTagEnd(pending, tagStart)
                if (tagEnd < 0) break
                val tagText = pending.substring(tagStart, tagEnd + 1)

                val isDiv = tagText.startsWith("<div")
                val isFakName = isDiv && fakNameTag.containsMatchIn(tagText)
                val isKafPeriod = isDiv && kafPeriodTag.containsMatchIn(tagText)
                if (isFakName || isKafPeriod) {
                    val closeDiv = pending.indexOf("</div>", tagEnd + 1)
                    if (closeDiv < 0) break
                    addRanges(tagText)
                    val content = pending.substring(tagEnd + 1, closeDiv)
                    if (isFakName) {
                        val stripped = tagStrip.replace(content, "").trim()
                        if (stripped.isNotEmpty()) faculty = FacultyNames.shortname(stripped)
                    } else {
                        val m = kafIdInTag.find(tagText)
                        if (m != null) {
                            departments += Department(
                                departmentId = m.groupValues[1].dropLast(1),
                                name = tagStrip.replace(content, "").trim(),
                                faculty = faculty
                            )
                        }
                    }
                    pos = closeDiv + 6
                } else {
                    addRanges(tagText)
                    pos = tagEnd + 1
                }

                if (pos > MAX_SLICE) {
                    pending.delete(0, pos)
                    pos = 0
                }
            }
            if (pos > 0 && pos <= pending.length) {
                pending.delete(0, pos)
            }
            if (pending.length > MAX_PENDING) {
                pending.delete(0, pending.length - MAX_SLICE)
            }
        }
        return ParsedDepartments(departments, ranges)
    }

    /** Ищет конец HTML-тега — символ '>' вне кавычек. */
    private fun findTagEnd(s: CharSequence, start: Int): Int {
        var i = start
        var quote = 0.toChar()
        while (i < s.length) {
            val c = s[i]
            if (quote != 0.toChar()) {
                if (c == quote) quote = 0.toChar()
            } else {
                when (c) {
                    '"', '\'' -> quote = c
                    '>' -> return i
                }
            }
            i++
        }
        return -1
    }

    /**
     * Хвост буфера может содержать неполное начало тега (<div... разорванный границей чанка).
     * Сдвиг перемотки назад, чтобы оставить такой префикс в pending до прихода следующего чанка.
     */
    private fun keepPartialTagTail(pending: CharSequence, from: Int): Int {
        var i = from
        while (i < pending.length) {
            val c = pending[i]
            if (c == '<') {
                val tail = pending.substring(i, pending.length)
                if (tail == "<" || tail == "<d" || tail == "<di" || tail == "<a") return i
            }
            i++
        }
        return pending.length
    }
}