adding basic list support

This commit is contained in:
Adam Brown 2022-10-29 11:23:44 +01:00
parent 7016f7c4ee
commit 446d729981
2 changed files with 87 additions and 3 deletions

View File

@ -5,10 +5,11 @@ import app.dapk.st.matrix.common.UserId
private const val TAG_OPEN = '<' private const val TAG_OPEN = '<'
private const val TAG_CLOSE = '>' private const val TAG_CLOSE = '>'
private const val NO_RESULT_FOUND = -1 private const val NO_RESULT_FOUND = -1
private const val LI_VALUE_CAPTURE = "value=\""
internal class HtmlParser { internal class HtmlParser {
fun parseHtmlTags(input: String, searchIndex: Int, builder: PartBuilder) = input.findTag( fun parseHtmlTags(input: String, searchIndex: Int, builder: PartBuilder): SearchIndex = input.findTag(
fromIndex = searchIndex, fromIndex = searchIndex,
onInvalidTag = { builder.appendText(input[it].toString()) }, onInvalidTag = { builder.appendText(input[it].toString()) },
onTag = { tagOpen, tagClose -> onTag = { tagOpen, tagClose ->
@ -59,7 +60,7 @@ internal class HtmlParser {
wholeTag: String, wholeTag: String,
builder: PartBuilder, builder: PartBuilder,
tagContent: String, tagContent: String,
exitTagCloseIndex: Int exitTagCloseIndex: Int,
) = when (tagName) { ) = when (tagName) {
"a" -> { "a" -> {
val findHrefUrl = wholeTag.substringAfter("href=").replace("\"", "").removeSuffix(">") val findHrefUrl = wholeTag.substringAfter("href=").replace("\"", "").removeSuffix(">")
@ -84,6 +85,11 @@ internal class HtmlParser {
exitTagCloseIndex exitTagCloseIndex
} }
"ul", "ol" -> {
parseList(tagName, tagContent, builder)
exitTagCloseIndex
}
"h1", "h2", "h3", "h4", "h5" -> { "h1", "h2", "h3", "h4", "h5" -> {
builder.appendBold(tagContent) builder.appendBold(tagContent)
builder.appendNewline() builder.appendNewline()
@ -128,8 +134,65 @@ internal class HtmlParser {
} }
} }
private fun parseList(parentTag: String, parentContent: String, builder: PartBuilder) {
var nextIndex = 0
var index = 1
while (nextIndex != END_SEARCH) {
nextIndex = singleTagParser(parentContent, "li", nextIndex, builder) { wholeTag, tagContent ->
val content = when (parentTag) {
"ol" -> {
index = wholeTag.indexOf(LI_VALUE_CAPTURE).let {
if (it == -1) {
index
} else {
val start = it + LI_VALUE_CAPTURE.length
wholeTag.substring(start).substringBefore('\"').toInt()
}
}
"$index. $tagContent".also {
index++
}
}
else -> "- $tagContent"
}
builder.appendText(content)
builder.appendNewline()
}
}
}
private fun singleTagParser(content: String, wantedTagName: String, searchIndex: Int, builder: PartBuilder, onTag: (String, String) -> Unit): SearchIndex {
return content.findTag(
fromIndex = searchIndex,
onInvalidTag = { builder.appendText(content[it].toString()) },
onTag = { tagOpen, tagClose ->
val wholeTag = content.substring(tagOpen, tagClose + 1)
val tagName = wholeTag.substring(1, wholeTag.indexOfFirst { it == '>' || it == ' ' })
if (tagName == wantedTagName) {
val exitTag = "</$tagName>"
val exitIndex = content.indexOf(exitTag, startIndex = tagClose)
val exitTagCloseIndex = exitIndex + exitTag.length
if (exitIndex == END_SEARCH) {
builder.appendText(content[searchIndex].toString())
searchIndex.next()
} else {
val tagContent = content.substring(tagClose + 1, exitIndex)
onTag(wholeTag, tagContent)
exitTagCloseIndex
}
} else {
END_SEARCH
}
}
)
}
fun test(startingFrom: Int, intput: String): Int { fun test(startingFrom: Int, intput: String): Int {
return intput.indexOf('<', startingFrom) return intput.indexOf(TAG_OPEN, startingFrom)
} }
} }

View File

@ -74,6 +74,27 @@ class RichMessageParserTest {
expected = RichText(setOf(Normal("Hello world!\nnext line\nanother line"))) expected = RichText(setOf(Normal("Hello world!\nnext line\nanother line")))
) )
@Test
fun `parses lists`() = runParserTest(
Case(
input = "<ul><li>content in list item</li><li>another item in list</li></ul>",
expected = RichText(setOf(Normal("- content in list item\n- another item in list")))
),
Case(
input = "<ol><li>content in list item</li><li>another item in list</li></ol>",
expected = RichText(setOf(Normal("1. content in list item\n2. another item in list")))
),
Case(
input = """<ol><li value="5">content in list item</li><li>another item in list</li></ol>""",
expected = RichText(setOf(Normal("5. content in list item\n6. another item in list")))
),
Case(
input = """<ol><li value="3">content in list item</li><li>another item in list</li><li value="10">another change</li><li>without value</li></ol>""",
expected = RichText(setOf(Normal("3. content in list item\n4. another item in list\n10. another change\n11. without value")))
),
)
@Test @Test
fun `parses urls`() = runParserTest( fun `parses urls`() = runParserTest(
Case( Case(