feat(remote): add jsoup TeacherScheduleParser

2026-09-19 14:47:55 +03:00
parent 458e4c4e69
commit 32c2b60afa
5 changed files with 331 additions and 7 deletions
@@ -0,0 +1,128 @@
+package agzam.vyatsu.data.remote
+
+import agzam.vyatsu.data.remote.pdf.TextPatterns
+import agzam.vyatsu.domain.model.Day
+import agzam.vyatsu.domain.model.Lesson
+import agzam.vyatsu.domain.model.LessonType
+import agzam.vyatsu.domain.model.TimeSlot
+import org.jsoup.Jsoup
+import org.jsoup.nodes.Element
+import java.time.LocalDate
+import java.time.LocalTime
+import java.time.format.DateTimeFormatter
+
+object TeacherScheduleParser {
+
+    private val DAY_REGEX = Regex("""^(понедельник|вторник|среда|четверг|пятница|суббота|воскресенье)\s+(\d{2}\.\d{2}\.\d{2})$""")
+    private val INTERVAL_REGEX = Regex("""^(\d{2}:\d{2})-(\d{2}:\d{2})$""")
+    private val GROUP_TOKEN = Regex("""(?U)\b[А-ЯЁа-яё]{1,4}-\d{2,4}-\d{2}-\d{2}\b""")
+    private val SPLIT_BY_GROUP = Regex("""(?U)(?=\b[А-ЯЁа-яё]{1,4}-\d{2,4}-\d{2}-\d{2}\b)""")
+    private val DATE_FMT = DateTimeFormatter.ofPattern("dd.MM.yy")
+
+    data class ParsedTeacherSchedule(val teachers: List<String>, val days: List<Day>)
+
+    fun parse(html: String): ParsedTeacherSchedule {
+        val doc = Jsoup.parse(html)
+        var headerFound = false
+        var teachers: List<String> = emptyList()
+        val days = mutableListOf<Day>()
+        var currentDate: LocalDate? = null
+        var currentIntervals: MutableList<TimeSlot>? = null
+
+        fun flushDay() {
+            val date = currentDate ?: return
+            val intervals = currentIntervals ?: return
+            if (intervals.isNotEmpty()) days.add(Day(date, intervals.toList()))
+        }
+
+        for (tr in doc.select("tr")) {
+            val tds = tr.children().filter { it.tagName() == "td" }
+            if (tds.size < 2) continue
+            val first = cellText(tds[0])
+            val second = cellText(tds[1])
+            if (!headerFound && first == "День" && second == "Интервал") {
+                teachers = tds.drop(2)
+                    .map { cellText(it.selectFirst("span") ?: it) }
+                    .filter { it.isNotEmpty() }
+                    .distinct()
+                headerFound = true
+                continue
+            }
+            val dayMatch = DAY_REGEX.matchEntire(first)
+            val intervalMatch = INTERVAL_REGEX.matchEntire(if (dayMatch != null) second else first)
+            if (dayMatch != null) {
+                flushDay()
+                currentDate = LocalDate.parse(dayMatch.groupValues[2], DATE_FMT)
+                currentIntervals = mutableListOf()
+            }
+            if (intervalMatch != null && currentIntervals != null) {
+                parseIntervalRow(tds, intervalMatch, if (dayMatch != null) 2 else 1, teachers, currentIntervals!!)
+            }
+        }
+        flushDay()
+        return ParsedTeacherSchedule(teachers, days)
+    }
+
+    private fun parseIntervalRow(
+        tds: List<Element>,
+        intervalMatch: MatchResult,
+        startCol: Int,
+        teachers: List<String>,
+        intervals: MutableList<TimeSlot>
+    ) {
+        val start = LocalTime.parse(intervalMatch.groupValues[1])
+        val end = LocalTime.parse(intervalMatch.groupValues[2])
+        val lessons = mutableListOf<Lesson>()
+        for (i in teachers.indices) {
+            val cell = tds.getOrNull(startCol + i) ?: continue
+            lessons += parseCell(cell, teachers[i])
+        }
+        if (lessons.isNotEmpty()) intervals.add(TimeSlot(start, end, lessons))
+    }
+
+    private fun parseCell(cell: Element, teacher: String): List<Lesson> {
+        val raw = cellText(cell)
+        if (raw.isEmpty()) return emptyList()
+        val room = TextPatterns.ROOM_REGEX.find(raw)?.value
+        val typeMatch = TextPatterns.LESSON_TYPE_REGEX.find(raw)
+        val type = LessonType.fromRaw(typeMatch?.value?.trim() ?: "")
+        val segments = raw.split(SPLIT_BY_GROUP).map { it.trim() }.filter { it.isNotEmpty() }
+        val groupSegments = segments.filter { GROUP_TOKEN.containsMatchIn(it) }
+        if (groupSegments.isEmpty()) {
+            return listOf(
+                Lesson(
+                    raw = raw,
+                    room = room,
+                    subject = subjectOf(raw, room, typeMatch),
+                    type = type,
+                    teacher = teacher
+                )
+            )
+        }
+        val subject = subjectOf(raw, room, typeMatch)
+        return groupSegments.flatMap { segment ->
+            val groups = GROUP_TOKEN.findAll(segment).map { it.value }.distinct().toList()
+            val subgroup = TextPatterns.SUBGROUP_REGEX.find(segment)?.value
+            groups.map { group ->
+                Lesson(
+                    raw = segment,
+                    room = room,
+                    group = group,
+                    subgroup = subgroup,
+                    subject = subject,
+                    type = type,
+                    teacher = teacher
+                )
+            }
+        }
+    }
+
+    private fun subjectOf(raw: String, room: String?, typeMatch: MatchResult?): String? {
+        val end = typeMatch?.range?.first ?: GROUP_TOKEN.find(raw)?.range?.first ?: raw.length
+        val prefix = raw.substring(0, end).trim()
+        val stripped = room?.let { prefix.removePrefix(it).trim() } ?: prefix
+        return stripped.ifEmpty { null }
+    }
+
+    private fun cellText(el: Element): String = el.text().replace('\u00A0', ' ').trim()
+}
@@ -26,10 +26,6 @@ object PdfScheduleParser {
    private fun timeFrom(timeString: String?): LocalTime? { return runCatching { LocalTime.parse(timeString, TIME_FORMAT) }.getOrNull() }
    private fun dateFrom(dateString: String?): LocalDate? { return runCatching { LocalDate.parse(dateString, DATE_FORMAT) }.getOrNull() }

-    private val ROOM_REGEX = Regex("""(?<=^|\s)\d{1,2}-\d{3}[_a-zA-Zа-яА-Я]?(?=\s|$)""")
-    private val LESSON_TYPE_REGEX = Regex(LessonType.entries.joinToString(separator = "|") { Regex.escape(it.label) }.let { """(?<=^|\s)($it)(?=\s|$)""" } )
-    private val SUBGROUP_REGEX = Regex("""\d{2}\sподгруппа""", RegexOption.IGNORE_CASE)
-
    fun parse(pdfBytes: ByteArray): List<Day> {
        Log.d(TAG, "Parsing ${pdfBytes.size} bytes")
        val doc = try {
@@ -145,7 +141,7 @@ object PdfScheduleParser {
                            // Поиск кабинета в строке, ибо он последний, если их несколько - то скорее всего разделение на подгруппы
                            var found = 0
                            var currentStart = 0
-                            for (roomMatch in ROOM_REGEX.findAll(content)) {
+                            for (roomMatch in TextPatterns.ROOM_REGEX.findAll(content)) {
                                found++

                                val raw = content.substring(currentStart, roomMatch.range.last + 1)
@@ -155,7 +151,7 @@ object PdfScheduleParser {

                                val room = roomMatch.value

-                                val typeMatch = LESSON_TYPE_REGEX.findAll(raw).toList()
+                                val typeMatch = TextPatterns.LESSON_TYPE_REGEX.findAll(raw).toList()
                                if(typeMatch.isEmpty()) {
                                    lessons.add(Lesson(
                                        raw=raw,
@@ -170,7 +166,7 @@ object PdfScheduleParser {
                                Log.d(TAG, "$subject $group")
                                if(group != null && (subject.startsWith("$group,") || subject.startsWith("$group "))) {
                                    subject = subject.substring(group.length + 1).trim()
-                                    val subgroupMatch = SUBGROUP_REGEX.find(subject)
+                                    val subgroupMatch = TextPatterns.SUBGROUP_REGEX.find(subject)
                                    if(subgroupMatch != null) {
                                        subgroup = subgroupMatch.value.trim()
                                        subject = subject.removeRange(subgroupMatch.range).trim()
@@ -0,0 +1,12 @@
+package agzam.vyatsu.data.remote.pdf
+
+import agzam.vyatsu.domain.model.LessonType
+
+object TextPatterns {
+
+    val ROOM_REGEX = Regex("""(?<=^|\s)\d{1,2}-\d{3}[_a-zA-Zа-яА-Я]?(?=\s|$)""")
+
+    val LESSON_TYPE_REGEX = Regex(LessonType.entries.joinToString(separator = "|") { Regex.escape(it.label) }.let { """(?<=^|\s)($it)(?=\s|$)""" })
+
+    val SUBGROUP_REGEX = Regex("""\d{2}\sподгруппа""", RegexOption.IGNORE_CASE)
+}
@@ -0,0 +1,86 @@
+package agzam.vyatsu
+
+import agzam.vyatsu.data.remote.TeacherScheduleParser
+import agzam.vyatsu.domain.model.LessonType
+import java.time.LocalDate
+import java.time.LocalTime
+import org.junit.Assert.assertEquals
+import org.junit.Assert.assertNull
+import org.junit.Assert.assertTrue
+import org.junit.Test
+
+class TeacherScheduleParserTest {
+
+    private fun fixture(): String =
+        checkNotNull(javaClass.getResourceAsStream("/prepod.html"))
+            .readBytes().toString(Charsets.UTF_8)
+
+    @Test
+    fun teachersHeaderParsedInOrderAndDeduped() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        assertEquals(listOf("Лелекова Е.В.", "Иванов И.И.", "Петрова А.С."), parsed.teachers)
+    }
+
+    @Test
+    fun daysWithCorrectDates() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        assertEquals(2, parsed.days.size)
+        assertEquals(LocalDate.of(2026, 9, 14), parsed.days[0].date)
+        assertEquals(LocalDate.of(2026, 9, 15), parsed.days[1].date)
+    }
+
+    @Test
+    fun intervalTimesParsed() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        val day0 = parsed.days[0]
+        assertEquals(2, day0.intervals.size)
+        assertEquals(LocalTime.of(8, 20), day0.intervals[0].start)
+        assertEquals(LocalTime.of(9, 50), day0.intervals[0].end)
+        assertEquals(LocalTime.of(10, 0), day0.intervals[1].start)
+        assertEquals(LocalTime.of(11, 30), day0.intervals[1].end)
+    }
+
+    @Test
+    fun multiGroupCellSplitsIntoPerGroupLessons() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        val cell = parsed.days[0].intervals[0].lessons.filter { it.teacher == "Лелекова Е.В." }
+        assertEquals(2, cell.size)
+        val lesson0 = cell[0]
+        val lesson1 = cell[1]
+        assertEquals("Бб-1101-04-00", lesson0.group)
+        assertNull(lesson0.subgroup)
+        assertEquals("01 подгруппа", lesson1.subgroup)
+        assertEquals("4-302", lesson0.room)
+        assertEquals("Биоразнообразие", lesson0.subject)
+        assertEquals(LessonType.LECTURE, lesson0.type)
+        assertEquals("Лелекова Е.В.", lesson0.teacher)
+    }
+
+    @Test
+    fun emptyCellContributesNoLesson() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        val threeSlots = (0..6).map { slot ->
+            parsed.days[0].intervals.firstOrNull { it.start.hour == 8 && it.start.minute == 20 } ?: return@map true
+        }
+        val petrovLessons = parsed.days[0].intervals[0].lessons.filter { it.teacher == "Петрова А.С." }
+        assertEquals(0, petrovLessons.size)
+        assertTrue(threeSlots.isEmpty() || threeSlots.size == 7)
+    }
+
+    @Test
+    fun noRoomCellKeepsRawAndNullRoom() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        val cell = parsed.days[0].intervals[1].lessons.filter { it.teacher == "Лелекова Е.В." }
+        assertEquals(2, cell.size)
+        assertNull(cell[0].room)
+        assertTrue(cell[0].raw.contains("Бб-1101-04-00"))
+        assertNull(cell[0].subject)
+    }
+
+    @Test
+    fun exactTotalLessonCount() {
+        val parsed = TeacherScheduleParser.parse(fixture())
+        val total = parsed.days.sumOf { day -> day.intervals.sumOf { it.lessons.size } }
+        assertEquals(11, total)
+    }
+}
@@ -0,0 +1,102 @@
+<html>
+<head>
+<meta charset="UTF-8">
+</head>
+<body>
+<TABLE>
+<TR>
+<TD><SPAN>День</SPAN></TD>
+<TD><SPAN>Интервал</SPAN></TD>
+<TD><SPAN>Лелекова Е.В.</SPAN></TD>
+<TD><SPAN>Иванов И.И.</SPAN></TD>
+<TD><SPAN>Петрова А.С.</SPAN></TD>
+</TR>
+<TR>
+<TD ROWSPAN=7><SPAN>понедельник 14.09.26</SPAN></TD>
+<TD><SPAN>08:20-09:50</SPAN></TD>
+<TD><SPAN>4-302 Биоразнообразие Лекция Бб-1101-04-00 Бб-1101-04-00, 01 подгруппа</SPAN></TD>
+<TD><SPAN>4-308 Методика познания (учитель биологии) Практическое занятие ПОДб-4701-08-00</SPAN></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>10:00-11:30</SPAN></TD>
+<TD><SPAN>Бб-1101-04-00 Бб-1101-04-00, 01 подгруппа</SPAN></TD>
+<TD><SPAN>Методическое совещание</SPAN></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>12:00-13:30</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>14:00-15:30</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>15:40-17:10</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>17:15-18:45</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>18:55-20:25</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD ROWSPAN=7><SPAN>вторник 15.09.26</SPAN></TD>
+<TD><SPAN>08:20-09:50</SPAN></TD>
+<TD><SPAN>4-507 Биоразнообразие Лабораторная работа Бб-1102-04-00 Бб-1102-04-00, 01 подгруппа</SPAN></TD>
+<TD></TD>
+<TD><SPAN>1-116 Статистическая обработка экспериментальных данных Практическое занятие БТб-2101-03-00 БТб-2101-03-00, 02 подгруппа</SPAN></TD>
+</TR>
+<TR>
+<TD><SPAN>10:00-11:30</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD><SPAN>4-402a Фиторесурсы Лекция Бм-2102-03-00</SPAN></TD>
+</TR>
+<TR>
+<TD><SPAN>12:00-13:30</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>14:00-15:30</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>15:40-17:10</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>17:15-18:45</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+<TR>
+<TD><SPAN>18:55-20:25</SPAN></TD>
+<TD></TD>
+<TD></TD>
+<TD></TD>
+</TR>
+</TABLE>
+</body>
+</html>