Add markdown parser

25 files changed, 446 insertions, 0 deletions

diff --git a/src/main/kotlin/moe/nea/blog/md/Begin.kt b/src/main/kotlin/moe/nea/blog/md/Begin.kt
new file mode 100644
index 0000000..482c114
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Begin.kt
@@ -0,0 +1,8 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+class Begin : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/BlockParser.kt b/src/main/kotlin/moe/nea/blog/md/BlockParser.kt
new file mode 100644
index 0000000..2ca812f
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/BlockParser.kt
@@ -0,0 +1,7 @@
+package moe.nea.blog.md
+
+interface BlockParser {
+    fun detect(line: String): Boolean
+    fun parse(parser: MarkdownParser): MarkdownBlock
+    val prio: Int
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/BlogPost.kt b/src/main/kotlin/moe/nea/blog/md/BlogPost.kt
new file mode 100644
index 0000000..25ec13f
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/BlogPost.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import java.io.File
+
+
+fun main() {
+    val parser = MarkdownParser(File("data/test.md").readText())
+    parser.addDefaultParsers()
+    val doc = parser.readDocument()
+    doc.debugFormat(System.out)
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Bold.kt b/src/main/kotlin/moe/nea/blog/md/Bold.kt
new file mode 100644
index 0000000..aa46622
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Bold.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class Bold(val inner: MarkdownFormat) : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+        printStream.print("<b>")
+        inner.debugFormat(printStream)
+        printStream.print("</b>")
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/CodeBlock.kt b/src/main/kotlin/moe/nea/blog/md/CodeBlock.kt
new file mode 100644
index 0000000..2882b77
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/CodeBlock.kt
@@ -0,0 +1,18 @@
+package moe.nea.blog.md
+
+import moe.nea.blog.util.indent
+import java.io.PrintStream
+
+data class CodeBlock(val language: String, val lines: List<String>) : MarkdownBlock {
+    override fun debugFormat(indent: Int, printStream: PrintStream) {
+        printStream.indent(indent)
+        printStream.println("<code lang=$language>")
+        lines.forEach {
+            printStream.indent(indent)
+            printStream.println(it)
+        }
+        printStream.indent(indent)
+        printStream.println("</code>")
+    }
+}
+
diff --git a/src/main/kotlin/moe/nea/blog/md/CodeBlockParser.kt b/src/main/kotlin/moe/nea/blog/md/CodeBlockParser.kt
new file mode 100644
index 0000000..0c894fe
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/CodeBlockParser.kt
@@ -0,0 +1,23 @@
+package moe.nea.blog.md
+
+object CodeBlockParser : BlockParser {
+    override fun detect(line: String): Boolean {
+        return line.startsWith("```")
+    }
+
+    override fun parse(parser: MarkdownParser): MarkdownBlock {
+        val language = parser.consumeLine()!!.removePrefix("```")
+        val lines = mutableListOf<String>()
+        while (true) {
+            val line = parser.consumeLine() ?: error("Unfinished code block")
+            if (line == "```")
+                break
+            lines.add(line)
+        }
+        return CodeBlock(language, lines)
+    }
+
+    override val prio: Int
+        get() = 10
+
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Document.kt b/src/main/kotlin/moe/nea/blog/md/Document.kt
new file mode 100644
index 0000000..ce83ef1
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Document.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class Document(
+    val list: List<MarkdownBlock>
+) : MarkdownElement {
+    fun debugFormat(printStream: PrintStream) {
+        list.forEach { it.debugFormat(0, printStream) }
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/FormatSequence.kt b/src/main/kotlin/moe/nea/blog/md/FormatSequence.kt
new file mode 100644
index 0000000..94b55b9
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/FormatSequence.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class FormatSequence(val list: List<MarkdownFormat>) : MarkdownFormat {
+    constructor(vararg list: MarkdownFormat) : this(list.toList())
+
+    override fun debugFormat(printStream: PrintStream) {
+        list.forEach { it.debugFormat(printStream) }
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Header.kt b/src/main/kotlin/moe/nea/blog/md/Header.kt
new file mode 100644
index 0000000..fab2306
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Header.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import moe.nea.blog.util.indent
+import java.io.PrintStream
+
+data class Header(val level: Int, val text: String) : MarkdownBlock {
+    override fun debugFormat(indent: Int, printStream: PrintStream) {
+        printStream.indent(indent)
+        printStream.println("header level=$level: $text")
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/HeaderParser.kt b/src/main/kotlin/moe/nea/blog/md/HeaderParser.kt
new file mode 100644
index 0000000..928cf58
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/HeaderParser.kt
@@ -0,0 +1,17 @@
+package moe.nea.blog.md
+
+object HeaderParser : BlockParser {
+    override fun detect(line: String): Boolean {
+        return line.matches("#+ .*".toRegex())
+    }
+
+    override fun parse(parser: MarkdownParser): MarkdownBlock {
+        val line = parser.consumeLine()!!
+        val level = line.indexOf(" ")
+        val text = line.substring(level).trim()
+        return Header(level, text)
+    }
+
+    override val prio: Int
+        get() = 10
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/InlineParser.kt b/src/main/kotlin/moe/nea/blog/md/InlineParser.kt
new file mode 100644
index 0000000..4b60963
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/InlineParser.kt
@@ -0,0 +1,7 @@
+package moe.nea.blog.md
+
+interface InlineParser {
+    fun detect(lookback: MarkdownFormat, rest: String): Boolean
+    val specialSyntax: Set<Char>
+    fun parse(parser: MarkdownParser, text: String): Pair<MarkdownFormat, String>
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Italics.kt b/src/main/kotlin/moe/nea/blog/md/Italics.kt
new file mode 100644
index 0000000..47d8db2
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Italics.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class Italics(val inner: MarkdownFormat) : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+        printStream.print("<i>")
+        inner.debugFormat(printStream)
+        printStream.print("</i>")
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/ItalicsParser.kt b/src/main/kotlin/moe/nea/blog/md/ItalicsParser.kt
new file mode 100644
index 0000000..afecf79
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/ItalicsParser.kt
@@ -0,0 +1,91 @@
+package moe.nea.blog.md
+
+object ItalicsParser : InlineParser {
+    override fun detect(lookback: MarkdownFormat, rest: String): Boolean {
+        return lookback is Whitespace && "\\*+[^ ].*".toRegex().matches(rest)
+    }
+
+    override val specialSyntax: Set<Char>
+        get() = setOf('*')
+
+    override fun parse(parser: MarkdownParser, text: String): Pair<MarkdownFormat, String> {
+        var firstStarCount = 0
+        while (firstStarCount in text.indices) {
+            if (text[firstStarCount] != '*') break
+            firstStarCount++
+        }
+
+        if (firstStarCount < 1 || firstStarCount > 3) error("Invalid italics/bold sequence")
+
+        val isBold = firstStarCount >= 2
+        val isItalics = (firstStarCount % 2) == 1
+
+        val firstSequence = mutableListOf<MarkdownFormat>()
+        var remainingText = text.substring(firstStarCount)
+        var lastToken: MarkdownFormat = Begin() // TODO: dedicated begin token
+        while (true) {
+            val (element, next) = parser.parseInlineTextOnce(lastToken, remainingText)
+            remainingText = next
+            lastToken = element
+            firstSequence.add(element)
+            if (element !is Whitespace && next.startsWith("*")) break
+        }
+
+        var secondStarCount = 0
+        while (secondStarCount in remainingText.indices) {
+            if (remainingText[secondStarCount] != '*') break
+            secondStarCount++
+        }
+
+        if (secondStarCount > firstStarCount)
+            secondStarCount = firstStarCount
+        if (secondStarCount < 1) error("Invalid italics/bold sequence")
+
+        remainingText = remainingText.substring(secondStarCount)
+        var firstElement = parser.collapseInlineFormat(firstSequence)
+        if (secondStarCount == 2)
+            firstElement = Bold(firstElement)
+        if (secondStarCount == 1)
+            firstElement = Italics(firstElement)
+        if (secondStarCount == firstStarCount) {
+            if (isBold)
+                firstElement = Bold(firstElement)
+            if (isItalics)
+                firstElement = Italics(firstElement)
+            return Pair(firstElement, remainingText)
+        }
+
+        val secondSequence = mutableListOf<MarkdownFormat>()
+        lastToken = Begin()
+        while (true) {
+            val (element, next) = parser.parseInlineTextOnce(lastToken, remainingText)
+            remainingText = next
+            lastToken = element
+            secondSequence.add(element)
+            if (element !is Whitespace && next.startsWith("*")) break
+        }
+
+        var thirdStarCount = 0
+        while (thirdStarCount in remainingText.indices) {
+            if (remainingText[thirdStarCount] != '*') break
+            thirdStarCount++
+        }
+
+        if (thirdStarCount > firstStarCount - secondStarCount)
+            thirdStarCount = firstStarCount
+
+        remainingText = remainingText.substring(thirdStarCount)
+
+        if (thirdStarCount != firstStarCount - secondStarCount) {
+            error("Invalid italics/bold sequence")
+        }
+        var secondElement = parser.collapseInlineFormat(secondSequence)
+        var combined: MarkdownFormat = FormatSequence(firstElement, secondElement)
+        if (thirdStarCount == 1)
+            combined = Italics(combined)
+        if (thirdStarCount == 2)
+            combined = Bold(combined)
+        return Pair(combined, remainingText)
+    }
+
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Link.kt b/src/main/kotlin/moe/nea/blog/md/Link.kt
new file mode 100644
index 0000000..f8fc858
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Link.kt
@@ -0,0 +1,11 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class Link(val label: MarkdownFormat?, val target: String) : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+        printStream.print("<link target=$target>")
+        label?.debugFormat(printStream)
+        printStream.print("</link>")
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/MarkdownBlock.kt b/src/main/kotlin/moe/nea/blog/md/MarkdownBlock.kt
new file mode 100644
index 0000000..fc432ef
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/MarkdownBlock.kt
@@ -0,0 +1,7 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+interface MarkdownBlock : MarkdownElement {
+    fun debugFormat(indent: Int, printStream: PrintStream)
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/MarkdownElement.kt b/src/main/kotlin/moe/nea/blog/md/MarkdownElement.kt
new file mode 100644
index 0000000..74aac6d
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/MarkdownElement.kt
@@ -0,0 +1,3 @@
+package moe.nea.blog.md
+
+interface MarkdownElement {}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/MarkdownFormat.kt b/src/main/kotlin/moe/nea/blog/md/MarkdownFormat.kt
new file mode 100644
index 0000000..d647854
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/MarkdownFormat.kt
@@ -0,0 +1,7 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+interface MarkdownFormat : MarkdownElement {
+    fun debugFormat(printStream: PrintStream)
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/MarkdownParser.kt b/src/main/kotlin/moe/nea/blog/md/MarkdownParser.kt
new file mode 100644
index 0000000..86b92d2
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/MarkdownParser.kt
@@ -0,0 +1,96 @@
+package moe.nea.blog.md
+
+import moe.nea.blog.util.indentSize
+import java.util.*
+
+
+class MarkdownParser(source: String) {
+    private val lines = source.lines()
+    private var lineIndex = 0
+    private var blockIndents = 0
+    private val indentStack = Stack<Int>()
+
+    private val blockParsers = mutableListOf<BlockParser>()
+    private val inlineParsers = mutableListOf<InlineParser>()
+
+    fun findParserFor(line: String): BlockParser? {
+        return blockParsers.filter { it.detect(line) }
+            .maxByOrNull { it.prio }
+    }
+
+    fun readChildBlock(): MarkdownBlock? {
+        val peek = peekLine() ?: return null
+        val blockParser = findParserFor(peek) ?: ParagraphParser
+        return blockParser.parse(this)
+    }
+
+    fun pushIndent(newIndent: Int) {
+        require(newIndent > blockIndents)
+        indentStack.push(blockIndents)
+        blockIndents = newIndent
+    }
+
+    fun popIndent() {
+        blockIndents = indentStack.pop()
+    }
+
+    fun consumeLine(): String? {
+        val line = peekLine()
+        if (line != null)
+            lineIndex++
+        return line
+    }
+
+    fun peekLine(): String? {
+        if (lineIndex !in lines.indices) return null
+        val line = lines[lineIndex]
+        val indent = line.indentSize()
+        if (indent != null && indent < blockIndents)
+            return null
+        return line.substring(blockIndents)
+    }
+
+    fun parseInlineTextOnce(lookback: MarkdownFormat, text: String): Pair<MarkdownFormat, String> {
+        val parser = inlineParsers.find { it.detect(lookback, text) }
+        if (parser != null)
+            return parser.parse(this, text)
+        require(!text.isEmpty()) // TODO handle empty string
+        if (text[0] == ' ')
+            return Pair(Whitespace(), text.substring(1))
+        val nextSpecial = text.indexOfFirst { it in inlineParsers.flatMap { it.specialSyntax } || it == ' ' }
+        return Pair(Word(text.substring(0, nextSpecial)), text.substring(nextSpecial))
+    }
+
+    fun parseInlineText(text: String): MarkdownFormat {
+        val seq = mutableListOf<MarkdownFormat>()
+        var remaining = text
+        var lastToken: MarkdownFormat = Begin()
+        while (remaining.isNotEmpty()) {
+            val (tok, next) = parseInlineTextOnce(lastToken, remaining)
+            seq.add(tok)
+            lastToken = tok
+            remaining = next
+        }
+        return collapseInlineFormat(seq)
+    }
+
+    fun collapseInlineFormat(sequence: List<MarkdownFormat>): MarkdownFormat {
+        return FormatSequence(sequence)
+    }
+
+    fun readDocument(): Document {
+        val list = mutableListOf<MarkdownBlock>()
+        while (true) {
+            val block = readChildBlock() ?: break
+            list.add(block)
+        }
+        return Document(list)
+    }
+
+    fun addDefaultParsers() {
+        blockParsers.add(CodeBlockParser)
+        blockParsers.add(HeaderParser)
+        inlineParsers.add(ItalicsParser)
+    }
+}
+
diff --git a/src/main/kotlin/moe/nea/blog/md/Paragraph.kt b/src/main/kotlin/moe/nea/blog/md/Paragraph.kt
new file mode 100644
index 0000000..54f688b
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Paragraph.kt
@@ -0,0 +1,12 @@
+package moe.nea.blog.md
+
+import moe.nea.blog.util.indent
+import java.io.PrintStream
+
+data class Paragraph(val format: MarkdownFormat) : MarkdownBlock {
+    override fun debugFormat(indent: Int, printStream: PrintStream) {
+        printStream.indent(indent)
+        format.debugFormat(printStream)
+        printStream.println()
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/ParagraphParser.kt b/src/main/kotlin/moe/nea/blog/md/ParagraphParser.kt
new file mode 100644
index 0000000..328c49b
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/ParagraphParser.kt
@@ -0,0 +1,31 @@
+package moe.nea.blog.md
+
+object ParagraphParser : BlockParser {
+    override fun detect(line: String): Boolean {
+        return true
+    }
+
+    override fun parse(parser: MarkdownParser): MarkdownBlock {
+        var text = ""
+        var emptyLineCount = 0
+        while (true) {
+            val line = parser.peekLine() ?: break
+            if (parser.findParserFor(line) != null) break
+            val trimmedLine = parser.consumeLine()!!.trim()
+            if (trimmedLine == "") {
+                emptyLineCount++
+            } else {
+                emptyLineCount = 0
+            }
+            if (emptyLineCount == 2) {
+                break
+            }
+            text += " $trimmedLine"
+        }
+        return Paragraph(parser.parseInlineText(text))
+    }
+
+    override val prio: Int
+        get() = -10
+
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Text.kt b/src/main/kotlin/moe/nea/blog/md/Text.kt
new file mode 100644
index 0000000..fb89ff3
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Text.kt
@@ -0,0 +1,9 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class Text(val text: String) : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+        printStream.print(printStream)
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Whitespace.kt b/src/main/kotlin/moe/nea/blog/md/Whitespace.kt
new file mode 100644
index 0000000..6a44152
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Whitespace.kt
@@ -0,0 +1,9 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+class Whitespace : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+        printStream.print(" ")
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/md/Word.kt b/src/main/kotlin/moe/nea/blog/md/Word.kt
new file mode 100644
index 0000000..ff89af5
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/md/Word.kt
@@ -0,0 +1,9 @@
+package moe.nea.blog.md
+
+import java.io.PrintStream
+
+data class Word(val text: String) : MarkdownFormat {
+    override fun debugFormat(printStream: PrintStream) {
+        printStream.print(text)
+    }
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/util/Indent.kt b/src/main/kotlin/moe/nea/blog/util/Indent.kt
new file mode 100644
index 0000000..5a61a48
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/util/Indent.kt
@@ -0,0 +1,5 @@
+package moe.nea.blog.util
+
+fun Appendable.indent(indent: Int) {
+    append(" ".repeat(indent))
+}
\ No newline at end of file
diff --git a/src/main/kotlin/moe/nea/blog/util/IndentSize.kt b/src/main/kotlin/moe/nea/blog/util/IndentSize.kt
new file mode 100644
index 0000000..de6a91a
--- /dev/null
+++ b/src/main/kotlin/moe/nea/blog/util/IndentSize.kt
@@ -0,0 +1,10 @@
+package moe.nea.blog.util
+
+fun String.indentSize(): Int? {
+    var i = 0
+    while (i in indices) {
+        if (this[i] != ' ') return i
+        i++
+    }
+    return null
+}
\ No newline at end of file