Punctuation after formula

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/checker/Checker.kt

@@ -1,18 +1,8 @@
 package com.github.darderion.mundaneassignmentpolice.checker
 
 import com.github.darderion.mundaneassignmentpolice.checker.rule.Rule
-import com.github.darderion.mundaneassignmentpolice.checker.rule.list.ListRuleBuilder
-import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.SymbolRule
-import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.SymbolRuleBuilder
-import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.and
-import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.or
-import com.github.darderion.mundaneassignmentpolice.checker.rule.tableofcontent.TableOfContentRuleBuilder
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFArea.*
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFRegion.Companion.EVERYWHERE
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFRegion.Companion.NOWHERE
 import com.github.darderion.mundaneassignmentpolice.rules.RuleSet
 import com.github.darderion.mundaneassignmentpolice.wrapper.PDFBox
-import java.util.*
 
 class Checker {
 	fun getRuleViolations(pdfName: String, ruleSet: RuleSet) = getRuleViolations(pdfName, ruleSet.rules)

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/checker/PunctuationMark.kt

@@ -0,0 +1,10 @@
+package com.github.darderion.mundaneassignmentpolice.checker
+
+enum class PunctuationMark(val value: Char) {
+    FULL_STOP('.'),
+    COMMA(',')
+}
+
+fun Char.isPunctuationMark() = PunctuationMark.values().map { it.value }.contains(this)
+
+fun String.isPunctuationMark() = this.length == 1 && this.single().isPunctuationMark()

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/checker/rule/formula/FormulaPunctuationRule.kt

@@ -0,0 +1,47 @@
+package com.github.darderion.mundaneassignmentpolice.checker.rule.formula
+
+import com.github.darderion.mundaneassignmentpolice.checker.RuleViolation
+import com.github.darderion.mundaneassignmentpolice.checker.RuleViolationType
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFDocument
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.inside
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Formula
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Line
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Word
+
+class FormulaPunctuationRule(
+    type: RuleViolationType,
+    name: String,
+    private val ignoredWords: List<Regex>,
+    private val ruleBody:
+        (formula: Formula, filteredText: List<Word>, nextFormula: Formula?) -> List<Line>
+) : FormulaRule(type, name) {
+    override fun getViolations(document: PDFDocument, formulas: List<Formula>): List<RuleViolation> {
+        val violations = mutableListOf<RuleViolation>()
+
+        formulas.forEachIndexed { index, formula ->
+            val textAfterFormula = formula.lines.last().text
+                .takeLastWhile { it != formula.text.last() }
+                .toMutableList()
+
+            textAfterFormula.addAll(
+                document.text.asSequence().drop(formula.lines.last().documentIndex + 1)
+                    .filter { it.area!! inside area && it.isNotEmpty() }
+                    .take(2) // take a line with formula reference and a line with words after the formula
+                    .map { it.text }.flatten()
+            )
+
+            val filteredText = textAfterFormula.filterNot { word -> ignoredWords.any { it.matches(word.text) } }
+
+            val nextFormula = formulas.getOrNull(index + 1)
+
+            val violationLines = ruleBody(formula, filteredText, nextFormula)
+            if (violationLines.isNotEmpty()) {
+                violations.add(
+                    RuleViolation(violationLines, name, type)
+                )
+            }
+        }
+
+        return violations
+    }
+}

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/checker/rule/formula/FormulaPunctuationRuleBuilder.kt

@@ -0,0 +1,31 @@
+package com.github.darderion.mundaneassignmentpolice.checker.rule.formula
+
+import com.github.darderion.mundaneassignmentpolice.checker.RuleViolationType
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Formula
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Line
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Word
+
+class FormulaPunctuationRuleBuilder {
+    private var type: RuleViolationType = RuleViolationType.Error
+    private var name: String = "Rule name"
+    private var ignoredWords: MutableList<Regex> = mutableListOf()
+    private var ruleBody: (formula: Formula, filteredText: List<Word>, nextFormula: Formula?) -> List<Line> =
+        { _, _, _ -> emptyList() }
+
+    infix fun called(name: String) = this.also { this.name = name }
+
+    infix fun type(type: RuleViolationType) = this.also { this.type = type }
+
+    fun ignoredWords(vararg regexes: Regex) = this.also { ignoredWords.addAll(regexes) }
+
+    infix fun rule(
+        ruleBody: (formula: Formula, filteredText: List<Word>, nextFormula: Formula?) -> List<Line>
+    ) = this.also { this.ruleBody = ruleBody }
+
+    fun getRule() = FormulaPunctuationRule(
+        type,
+        name,
+        ignoredWords,
+        ruleBody
+    ) as FormulaRule
+}

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/checker/rule/formula/FormulaRule.kt

@@ -0,0 +1,56 @@
+package com.github.darderion.mundaneassignmentpolice.checker.rule.formula
+
+import com.github.darderion.mundaneassignmentpolice.checker.RuleViolation
+import com.github.darderion.mundaneassignmentpolice.checker.RuleViolationType
+import com.github.darderion.mundaneassignmentpolice.checker.rule.Rule
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFArea
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFDocument
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.PDFRegion
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.inside
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Formula
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Line
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.PostScriptFontType
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Word
+
+abstract class FormulaRule(
+    type: RuleViolationType,
+    name: String
+) : Rule(PDFRegion.NOWHERE.except(PDFArea.SECTION), name, type) {
+    abstract fun getViolations(document: PDFDocument, formulas: List<Formula>): List<RuleViolation>
+
+    override fun process(document: PDFDocument) =
+        getViolations(document, getAllFormulas(document))
+
+    private fun getAllFormulas(document: PDFDocument): List<Formula> {
+        val text = document.text.filter { it.area!! inside area && it.isNotEmpty() }
+
+        val formulas = mutableListOf<Formula>()
+        val formulaText = mutableListOf<Word>()
+        val formulaLines = mutableSetOf<Line>()
+
+        text.forEach { line ->
+            line.text.forEach { word ->
+                if (word.font.type == PostScriptFontType.TYPE2 || word.text == " " && formulaText.isNotEmpty()) {
+                    // also captures some records in tables and code listings
+                    formulaText.add(word)
+                    formulaLines.add(line)
+                } else if (formulaText.isNotEmpty()) {
+                    formulas.add(Formula(formulaText.dropLastWhile { it.text == " " }, formulaLines.toSet()))
+                    formulaText.clear()
+                    formulaLines.clear()
+                }
+            }
+        }
+        if (formulaText.isNotEmpty())
+            formulas.add(Formula(formulaText.dropLastWhile { it.text == " " }, formulaLines.toSet()))
+
+        return filterFormulas(formulas)
+    }
+
+    private fun filterFormulas(formulas: List<Formula>) =
+        formulas.filterNot {
+            it.text.size == 1 && it.text.first().text == "∗"   // remove single special characters
+        }.filterNot {
+            it.text.size == 1 && it.text.first().text.toDoubleOrNull() != null  // remove numbers
+        }
+}

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/pdfdocument/PDFDocument.kt

@@ -2,7 +2,6 @@ package com.github.darderion.mundaneassignmentpolice.pdfdocument
 
 import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Line
 import mu.KotlinLogging
-import java.lang.Exception
 
 class PDFDocument(val name: String = "PDF",
 				  val text: List<Line>,

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/pdfdocument/list/PDFList.kt

@@ -1,9 +1,6 @@
 package com.github.darderion.mundaneassignmentpolice.pdfdocument.list
 
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Coordinate
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Font
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Line
-import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.Word
+import com.github.darderion.mundaneassignmentpolice.pdfdocument.text.*
 import java.util.*
 
 data class PDFList<T>(val value: MutableList<T> = mutableListOf(), val nodes: MutableList<PDFList<T>> = mutableListOf()) {
@@ -54,7 +51,7 @@ data class PDFList<T>(val value: MutableList<T> = mutableListOf(), val nodes: Mu
 		 */
 		fun getLists(lines: List<Line>): List<PDFList<Line>> {
 			// Adding a line to process a text that has no lines after a list
-			val lines = lines + Line(-1, -1, -1, listOf(Word("NOT A LIST ITEM", Font(0.0f), Coordinate(1000, -1))))
+			val lines = lines + Line(-1, -1, -1, listOf(Word("NOT A LIST ITEM", Font(), Coordinate(1000, -1))))
 
 			val lists: MutableList<PDFList<Line>> = mutableListOf()
 			val stack: Stack<PDFList<Line>> = Stack()

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/pdfdocument/text/Font.kt

@@ -2,9 +2,15 @@ package com.github.darderion.mundaneassignmentpolice.pdfdocument.text
 
 import com.github.darderion.mundaneassignmentpolice.utils.floatEquals
 
-class Font(val size: Float) {
+enum class PostScriptFontType {
+	TYPE0, TYPE1, TYPE2, TYPE3, NONE
+}
+
+class Font(val type: PostScriptFontType, val size: Float) {
+	constructor(): this(PostScriptFontType.NONE, 0.0f)
+
 	override fun equals(other: Any?) = this === other ||
-			(other is Font && floatEquals(size, other.size))
+			(other is Font && type == other.type && floatEquals(size, other.size))
 
-	override fun hashCode() = size.hashCode()
+	override fun hashCode() = (type to size).hashCode()
 }

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/pdfdocument/text/Formula.kt

@@ -0,0 +1,3 @@
+package com.github.darderion.mundaneassignmentpolice.pdfdocument.text
+
+data class Formula(val text: List<Word>, val lines: Set<Line>)

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/pdfdocument/text/Line.kt

@@ -20,4 +20,8 @@ data class Line(val index: Int, val page: Int, val documentIndex: Int,
 	override fun toString() = "[$documentIndex -- $index, p.$page, $area, ${position.x}] --> '$content'"
 
 	fun drop(numberOfItems: Int) = Line(index, page, documentIndex, text.drop(numberOfItems), area)
+
+	fun isEmpty() = text.isEmpty() || text.size == 1 && text.first().text == ""
+
+	fun isNotEmpty() = !isEmpty()
 }

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/pdfdocument/text/Word.kt

@@ -5,6 +5,6 @@ data class Word(val text: String, val font: Font, val position: Coordinate) {
 
 	companion object {
 		val spaceCharacter: Word
-		get() = Word(" ", Font(0.0f), Coordinate(0, 0))
+		get() = Word(" ", Font(), Coordinate(0, 0))
 	}
 }

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/rules/RuleSet.kt

@@ -3,7 +3,7 @@ package com.github.darderion.mundaneassignmentpolice.rules
 import com.github.darderion.mundaneassignmentpolice.checker.rule.Rule
 
 val RULE_SET_RU = RuleSet(
-	mutableListOf(
+	listOf(
 		RULE_LITLINK,
 		RULE_SHORT_DASH,
 		RULE_MEDIUM_DASH,
@@ -21,6 +21,7 @@ val RULE_SET_RU = RuleSet(
 	)
 			+ RULES_SPACE_AROUND_BRACKETS
 			+ RULES_SMALL_NUMBERS
+			+ RULES_FORMULA_PUNCTUATION
 )
 
-class RuleSet(val rules: List<Rule>) {}
+class RuleSet(val rules: List<Rule>)

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/rules/Rules.kt

@@ -1,8 +1,10 @@
 package com.github.darderion.mundaneassignmentpolice.rules
 
+import com.github.darderion.mundaneassignmentpolice.checker.PunctuationMark
 import com.github.darderion.mundaneassignmentpolice.checker.RuleViolationType
+import com.github.darderion.mundaneassignmentpolice.checker.isPunctuationMark
+import com.github.darderion.mundaneassignmentpolice.checker.rule.formula.FormulaPunctuationRuleBuilder
 import com.github.darderion.mundaneassignmentpolice.checker.rule.list.ListRuleBuilder
-import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.SymbolRule
 import com.github.darderion.mundaneassignmentpolice.checker.rule.regex.RegexRuleBuilder
 import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.SymbolRuleBuilder
 import com.github.darderion.mundaneassignmentpolice.checker.rule.symbol.and
@@ -265,3 +267,91 @@ val RULE_ORDER_OF_REFERENCES = RegexRuleBuilder()
 			referencesInIntList != referencesInIntList.sorted()
 		}.map { it.second }
 	}.getRule()
+
+private val ignoringAfterFormula = listOf(
+	"""\s""".toRegex(),
+	"""\([0-9]+\)""".toRegex()  // ignore formula reference, e.g. "(1)"
+)
+
+val fullStopAfterFormulaRule = FormulaPunctuationRuleBuilder()
+	.called("Отсутствует точка после формулы")
+	.ignoredWords(*ignoringAfterFormula.toTypedArray())
+	.rule { formula, filteredText, nextFormula ->
+		val violationLines = listOf(formula.lines.last())
+		val lastFormulaSymbol = formula.text.last().text.last()
+
+		if (filteredText.isEmpty()) {
+			return@rule if (lastFormulaSymbol != PunctuationMark.FULL_STOP.value) violationLines else emptyList()
+		}
+
+		// full stop is not required if there is another formula after the formula
+		val (firstAfterFormula, secondAfterFormula) = filteredText.first() to filteredText.getOrNull(1)
+		if (nextFormula != null &&
+			(firstAfterFormula == nextFormula.text.first() ||
+				firstAfterFormula.text.isPunctuationMark() && secondAfterFormula == nextFormula.text.first())
+		) {
+			return@rule emptyList()
+		}
+
+		val indicator = """[A-ZА-Я].*?""".toRegex() // capitalized word that indicates the beginning of a new sentence
+		if (indicator.matches(firstAfterFormula.text)) {
+			return@rule if (lastFormulaSymbol != PunctuationMark.FULL_STOP.value) violationLines else emptyList()
+		}
+
+		// case when a punctuation mark is after the formula and not the last symbol of the formula
+		if (firstAfterFormula.text.isPunctuationMark() &&
+			secondAfterFormula != null && indicator.matches(secondAfterFormula.text)
+		) {
+			return@rule if (firstAfterFormula.text.single() != PunctuationMark.FULL_STOP.value) violationLines
+			else emptyList()
+		}
+
+		return@rule emptyList()
+	}
+	.getRule()
+
+val commaAfterFormulaRule = FormulaPunctuationRuleBuilder()
+	.called("Отсутствует запятая после формулы")
+	.ignoredWords(*ignoringAfterFormula.toTypedArray())
+	.rule { formula, filteredText, nextFormula ->
+		val violationLines = listOf(formula.lines.last())
+		val lastFormulaSymbol = formula.text.last().text.last()
+
+		if (filteredText.isEmpty()) return@rule emptyList()
+
+		// comma is required if there is another formula after the formula
+		val (firstAfterFormula, secondAfterFormula) = filteredText.first() to filteredText.getOrNull(1)
+		if (nextFormula != null) {
+			if (firstAfterFormula == nextFormula.text.first()) {
+				return@rule if (lastFormulaSymbol == PunctuationMark.COMMA.value ||
+					lastFormulaSymbol == PunctuationMark.FULL_STOP.value
+				) emptyList()
+				else violationLines
+			}
+
+			if (firstAfterFormula.text.isPunctuationMark() && secondAfterFormula == nextFormula.text.first()) {
+				return@rule if (firstAfterFormula.text.single() == PunctuationMark.COMMA.value ||
+					firstAfterFormula.text.single() == PunctuationMark.FULL_STOP.value
+				) emptyList()
+				else violationLines
+			}
+		}
+
+		val indicator = """где""".toRegex()
+		if (indicator.matches(firstAfterFormula.text)) {
+			return@rule if (lastFormulaSymbol != PunctuationMark.COMMA.value) violationLines else emptyList()
+		}
+
+		// case when a punctuation mark is after the formula and not the last symbol of the formula
+		if (firstAfterFormula.text.isPunctuationMark() &&
+			secondAfterFormula != null && indicator.matches(secondAfterFormula.text)
+		) {
+			return@rule if (firstAfterFormula.text.single() != PunctuationMark.COMMA.value) violationLines
+			else emptyList()
+		}
+
+		return@rule emptyList()
+	}
+	.getRule()
+
+val RULES_FORMULA_PUNCTUATION = listOf(fullStopAfterFormulaRule, commaAfterFormulaRule)

src/main/kotlin/com/github/darderion/mundaneassignmentpolice/wrapper/PDFBox.kt

@@ -100,8 +100,8 @@ class PDFBox {
 
 		val strippers = listOf(stripper, textStripper)
 
-		var lineIndex = 0
-		for(pageIndex in (0..document.pages.count)) {
+		var lineIndex = -1
+		for (pageIndex in (0 until document.pages.count)) {
 			// For each page
 			strippers.forEach {
 				it.startPage = pageIndex + 1
@@ -142,7 +142,7 @@ class PDFBox {
 					contentIndex += contentItem.length
 
 					if (contentItem == " ") {
-						words.add(Word(word, font?: Font(0.0f), coordinates))
+						words.add(Word(word, font?: Font(), coordinates))
 						words.add(Word.spaceCharacter)
 						font = null
 						word = ""
@@ -163,7 +163,7 @@ class PDFBox {
 						stripperIndex++
 					}
 				}
-				if (font == null && word.isEmpty()) font = Font(0.0f)
+				if (font == null && word.isEmpty()) font = Font()
 				words.add(Word(word, font!!, coordinates))
 
 				Line(line, pageIndex, lineIndex, words.toList())