feat: better parsing & display

This commit is contained in:
2026-09-04 04:40:29 +03:00
parent 699bcd81f5
commit 686976347e
2 changed files with 67 additions and 3 deletions
@@ -2,6 +2,7 @@ package agzam.vyatsu.data.remote.pdf
import agzam.vyatsu.domain.model.Day import agzam.vyatsu.domain.model.Day
import agzam.vyatsu.domain.model.Lesson import agzam.vyatsu.domain.model.Lesson
import agzam.vyatsu.domain.model.LessonType
import agzam.vyatsu.domain.model.TimeSlot import agzam.vyatsu.domain.model.TimeSlot
import android.util.Log import android.util.Log
import com.tom_roush.pdfbox.pdmodel.PDDocument import com.tom_roush.pdfbox.pdmodel.PDDocument
@@ -25,6 +26,10 @@ object PdfScheduleParser {
private fun timeFrom(timeString: String?): LocalTime? { return runCatching { LocalTime.parse(timeString, TIME_FORMAT) }.getOrNull() } 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 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> { fun parse(pdfBytes: ByteArray): List<Day> {
Log.d(TAG, "Parsing ${pdfBytes.size} bytes") Log.d(TAG, "Parsing ${pdfBytes.size} bytes")
val doc = try { val doc = try {
@@ -35,6 +40,8 @@ object PdfScheduleParser {
} }
val days = mutableListOf<Day>() val days = mutableListOf<Day>()
var group: String? = null
try { try {
for (pageIndex in 0 until doc.numberOfPages) { for (pageIndex in 0 until doc.numberOfPages) {
Log.v(TAG, "============ Page $pageIndex ============") Log.v(TAG, "============ Page $pageIndex ============")
@@ -100,9 +107,13 @@ object PdfScheduleParser {
} }
} }
for (dayCell in daysColumn.cells) { for (dayCell in daysColumn.cells) {
val day = content(dayCell).split(' ') val day = content(dayCell).split(' ')
if(day.size == 1 && day.first() == "День") {
val groupCell = select(lessonsColumn, dayCell).firstOrNull() ?: continue
group = content(groupCell)
continue
}
if(day.size != 2) continue // XXX: добавить защиту от заголовка и выбрасывать исключения вместо пропуска? if(day.size != 2) continue // XXX: добавить защиту от заголовка и выбрасывать исключения вместо пропуска?
val date = dateFrom(day[1]) ?: continue val date = dateFrom(day[1]) ?: continue
@@ -124,7 +135,55 @@ object PdfScheduleParser {
val content = content(lessonCell) val content = content(lessonCell)
Log.v(TAG, "- $content") Log.v(TAG, "- $content")
// TODO: парсинг контента лекции // TODO: парсинг контента лекции
lessons.add(Lesson(raw=content))
// Поиск кабинета в строке, ибо он последний, если их несколько - то скорее всего разделение на подгруппы
var found = 0
var currentStart = 0
for (roomMatch in ROOM_REGEX.findAll(content)) {
found++
val raw = content.substring(currentStart, roomMatch.range.last + 1)
currentStart = roomMatch.range.last + 1
// Предполагаемый формат: [группа], [XX подгруппа] <предмет> <тип> [id + тип] <ФИО> <кабинет>
val room = roomMatch.value
val typeMatch = LESSON_TYPE_REGEX.findAll(raw).toList()
if(typeMatch.isEmpty()) {
lessons.add(Lesson(
raw=raw,
room=room,
))
continue
}
var subject = raw.substring(0, typeMatch.first().range.first).trim() // TODO: сейчас [группа], [XX подгруппа] <предмет> - можно достать подгруппу
val type = LessonType.fromRaw(typeMatch.first().value.trim())
val teacher = raw.substring(typeMatch.last().range.last+1, raw.length - room.length).trim()
var subgroup: String? = null
Log.d(TAG, subject + " " + group)
if(group != null && subject.startsWith("$group,")) {
subject = subject.substring(group.length + 1).trim()
val subgroupMatch = SUBGROUP_REGEX.find(subject)
if(subgroupMatch != null) {
subgroup = subgroupMatch.value.trim()
subject = subject.removeRange(subgroupMatch.range).trim()
}
}
lessons.add(Lesson(
raw=raw,
subject=subject,
room=room,
type=type,
teacher=teacher,
subgroup=subgroup
))
}
if(found == 0) lessons.add(Lesson(raw=content))
} }
intervals.add(TimeSlot(start = start, end = end, lessons = lessons)) intervals.add(TimeSlot(start = start, end = end, lessons = lessons))
} }
@@ -285,13 +285,18 @@ private fun LessonSlot(slotIndex: Int, slot: TimeSlot) {
} else { } else {
slot.lessons.forEach { lesson -> slot.lessons.forEach { lesson ->
Column(Modifier.fillMaxWidth().padding(vertical = 2.dp)) { Column(Modifier.fillMaxWidth().padding(vertical = 2.dp)) {
Text(
lesson.type.label,
style = MaterialTheme.typography.bodySmall,
color = MaterialTheme.colorScheme.onSurfaceVariant
)
Text( Text(
lesson.subject?.takeIf { it.isNotBlank() } ?: lesson.raw, lesson.subject?.takeIf { it.isNotBlank() } ?: lesson.raw,
style = MaterialTheme.typography.bodyLarge style = MaterialTheme.typography.bodyLarge
) )
Text( Text(
listOfNotNull( listOfNotNull(
lesson.type.label,
lesson.subgroup, lesson.subgroup,
lesson.teacher, lesson.teacher,
lesson.room lesson.room