Add pandoc conversion

This commit is contained in:
Alexander Nozik 2023-06-19 19:54:50 +03:00
parent 3c1a1bd99d
commit 09da54f4f0
3 changed files with 40 additions and 16 deletions

View File

@ -3,6 +3,7 @@ import org.jetbrains.kotlin.gradle.tasks.KotlinCompile
plugins { plugins {
kotlin("jvm") version "1.8.22" kotlin("jvm") version "1.8.22"
application application
id("com.github.johnrengelman.shadow") version "8.1.1"
} }
group = "ru.mipt.npm" group = "ru.mipt.npm"

View File

@ -2,13 +2,11 @@ package ru.mipt.npm.space.documentextractor
import java.nio.file.Path import java.nio.file.Path
import java.nio.file.StandardOpenOption import java.nio.file.StandardOpenOption
import java.nio.file.attribute.FileAttribute
import kotlin.io.path.* import kotlin.io.path.*
@OptIn(ExperimentalPathApi::class) internal fun prepareScripts(inputPath: Path): Path {
fun generateHtml(inputPath: Path, outputPath: Path) {
val scriptPath = inputPath.resolveSibling("scripts").resolve("links-to-html.lua") val scriptPath = inputPath.resolveSibling("scripts").resolve("links-to-html.lua")
if(!scriptPath.exists()) { if (!scriptPath.exists()) {
scriptPath.parent.createDirectories() scriptPath.parent.createDirectories()
scriptPath.writeText( scriptPath.writeText(
{}.javaClass.getResource("/links-to-html.lua")!!.readText(), {}.javaClass.getResource("/links-to-html.lua")!!.readText(),
@ -16,9 +14,31 @@ fun generateHtml(inputPath: Path, outputPath: Path) {
StandardOpenOption.CREATE StandardOpenOption.CREATE
) )
} }
return scriptPath
}
@OptIn(ExperimentalPathApi::class)
fun convert(inputPath: Path, outputPath: Path) {
val scriptPath = prepareScripts(inputPath)
inputPath.copyToRecursively(outputPath, followLinks = false) { source: Path, target: Path -> inputPath.copyToRecursively(outputPath, followLinks = false) { source: Path, target: Path ->
if (source.isRegularFile() && source.extension == "md") { if (source.isRegularFile() && source.extension == "md") {
val targetPath = outputPath.resolve(source.fileName.nameWithoutExtension + ".html")
val docxPath = target.parent.resolve(source.fileName.nameWithoutExtension + ".docx")
ProcessBuilder(
"pandoc",
"--from=markdown",
"--to=docx",
"--lua-filter=${scriptPath.absolute()}",
"--output=${docxPath.absolute()}",
"${source.absolute()}",
).also {
logger.info("Running pandoc: ${it.command().joinToString(separator = " ")}")
}.inheritIO().start().waitFor()
val htmlPath = target.parent.resolve(source.fileName.nameWithoutExtension + ".html")
ProcessBuilder( ProcessBuilder(
"pandoc", "pandoc",
@ -28,17 +48,21 @@ fun generateHtml(inputPath: Path, outputPath: Path) {
"--from=markdown", "--from=markdown",
"--to=html5", "--to=html5",
"--lua-filter=${scriptPath.absolute()}", "--lua-filter=${scriptPath.absolute()}",
"--output=${targetPath.absolute()}", "--output=${htmlPath.absolute()}",
"${source.absolute()}", "${source.absolute()}",
).also { ).also {
logger.info("Running pandoc: ${it.command().joinToString(separator = " ")}") logger.info("Running pandoc: ${it.command().joinToString(separator = " ")}")
}.inheritIO().start().waitFor() }.inheritIO().start().waitFor()
CopyActionResult.CONTINUE CopyActionResult.CONTINUE
} else { } else {
source.copyToIgnoringExistingDirectory(target, false) source.copyToIgnoringExistingDirectory(target, false)
} }
} }
}
// ZipOutputStream(zipFileName.outputStream().buffered()).use { zipStream -> // ZipOutputStream(zipFileName.outputStream().buffered()).use { zipStream ->
// outputPath.walk().forEach { file -> // outputPath.walk().forEach { file ->
@ -50,4 +74,3 @@ fun generateHtml(inputPath: Path, outputPath: Path) {
// } // }
// } // }
// } // }
}

View File

@ -40,15 +40,15 @@ suspend fun main(args: Array<String>) {
description = "FolderId for the folder to export. By default uses project root." description = "FolderId for the folder to export. By default uses project root."
) )
val generateHtml by parser.option( val convert by parser.option(
ArgType.Boolean, ArgType.Boolean,
description = "If defined, generate HTML directory and zip" description = "If defined, convert result to HTML and DOCX on download"
).default(false) ).default(false)
val htmlOutputPath by parser.option( val convertOutputPath by parser.option(
ArgType.String, ArgType.String,
description = "Path for html output directory sibling to 'output' directory" description = "Path for html and docx output directory sibling to 'output' directory"
).default("html") ).default("converted")
val clientId by parser.option( val clientId by parser.option(
ArgType.String, ArgType.String,
@ -83,8 +83,8 @@ suspend fun main(args: Array<String>) {
ProjectIdentifier.Key(project), ProjectIdentifier.Key(project),
folderId?.let { FolderIdentifier.Id(it) } ?: FolderIdentifier.Root folderId?.let { FolderIdentifier.Id(it) } ?: FolderIdentifier.Root
) )
if (generateHtml) { if (convert) {
generateHtml(target, target.resolveSibling(htmlOutputPath)) convert(target, target.resolveSibling(convertOutputPath))
} }
} }
} }