Atte149/book-story
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

🛠️ Fix empty lines pass .isNotBlank() & head > title is not filtered out

Browse source 
* Fixed issue, where some kind of invisible symbol passes .isNotBlank() check, resulting in empty lines
* Fixed issue with "head > title" is present and is not filtered out

Resolves: #122
This commit is contained in:
Acclorite 2024-12-16 15:02:53 +02:00
parent 9bac48c5cb
commit 85712cf566
2 changed files with 34 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

6
app/src/main/java/ua/acclorite/book_story/data/parser/DocumentParser.kt
View file

@ -3,6 +3,7 @@ package ua.acclorite.book_story.data.parser
import kotlinx.coroutines.yield import kotlinx.coroutines.yield
import org.jsoup.nodes.Document import org.jsoup.nodes.Document
import ua.acclorite.book_story.presentation.core.util.clearMarkdown import ua.acclorite.book_story.presentation.core.util.clearMarkdown
import ua.acclorite.book_story.presentation.core.util.containsVisibleText
import javax.inject.Inject import javax.inject.Inject
class DocumentParser @Inject constructor() { class DocumentParser @Inject constructor() {
@ -26,6 +27,9 @@ class DocumentParser @Inject constructor() {
element.append("\n") element.append("\n")
} }
// Remove <head>'s title
select("title").remove()
// Markdown // Markdown
select("hr").append("\n---\n") select("hr").append("\n---\n")
select("b").append("**").prepend("**") select("b").append("**").prepend("**")
@ -52,7 +56,7 @@ class DocumentParser @Inject constructor() {
Regex("""_\s*(.*?)\s*_"""), "_$1_" Regex("""_\s*(.*?)\s*_"""), "_$1_"
).trim() ).trim()
if (formattedLine.clearMarkdown().isNotBlank()) { if (formattedLine.clearMarkdown().containsVisibleText()) {
lines.add(formattedLine) lines.add(formattedLine)
} }
} }

29
app/src/main/java/ua/acclorite/book_story/presentation/core/util/PrimitiveExtensions.kt
View file

@ -30,3 +30,32 @@ fun String.clearMarkdown(): String {
fun String.clearAllMarkdown(): String { fun String.clearAllMarkdown(): String {
return replace(Regex("(_+)|(\\*+)|(#+)"), "").trim() return replace(Regex("(_+)|(\\*+)|(#+)"), "").trim()
} }
fun String.containsVisibleText(): Boolean {
return any { it.isVisibleCharacter() }
}
fun Char.isVisibleCharacter(): Boolean {
return when (this.category) {
CharCategory.UPPERCASE_LETTER,
CharCategory.LOWERCASE_LETTER,
CharCategory.TITLECASE_LETTER,
CharCategory.MODIFIER_LETTER,
CharCategory.OTHER_LETTER,
CharCategory.DECIMAL_DIGIT_NUMBER,
CharCategory.LETTER_NUMBER,
CharCategory.OTHER_NUMBER,
CharCategory.MATH_SYMBOL,
CharCategory.CURRENCY_SYMBOL,
CharCategory.OTHER_SYMBOL,
CharCategory.INITIAL_QUOTE_PUNCTUATION,
CharCategory.FINAL_QUOTE_PUNCTUATION,
CharCategory.CONNECTOR_PUNCTUATION,
CharCategory.DASH_PUNCTUATION,
CharCategory.START_PUNCTUATION,
CharCategory.END_PUNCTUATION,
CharCategory.OTHER_PUNCTUATION -> true
else -> false
}
}