text-match: Support match types and unicode-casemap collation

This commit is contained in:
Lennart
2025-12-31 12:24:57 +01:00
parent 578ddde36d
commit 17ba8faef2
6 changed files with 74 additions and 14 deletions

View File

@@ -1,23 +1,23 @@
use ical::property::Property;
use rustical_xml::{ValueDeserialize, XmlDeserialize};
use std::borrow::Cow;
#[derive(Clone, Debug, PartialEq, Eq, Default)]
pub enum TextCollation {
#[default]
AsciiCasemap,
UnicodeCasemap,
Octet,
}
impl TextCollation {
// Check whether a haystack contains a needle respecting the collation
#[must_use]
pub fn match_text(&self, needle: &str, haystack: &str) -> bool {
pub fn normalise<'a>(&self, value: &'a str) -> Cow<'a, str> {
match self {
// https://datatracker.ietf.org/doc/html/rfc4790#section-9.2
Self::AsciiCasemap => haystack
.to_ascii_uppercase()
.contains(&needle.to_ascii_uppercase()),
Self::Octet => haystack.contains(needle),
Self::AsciiCasemap => Cow::from(value.to_ascii_uppercase()),
Self::UnicodeCasemap => Cow::from(value.to_uppercase()),
Self::Octet => Cow::from(value),
}
}
}
@@ -26,6 +26,7 @@ impl AsRef<str> for TextCollation {
fn as_ref(&self) -> &str {
match self {
Self::AsciiCasemap => "i;ascii-casemap",
Self::UnicodeCasemap => "i;unicode-casemap",
Self::Octet => "i;octet",
}
}
@@ -35,6 +36,7 @@ impl ValueDeserialize for TextCollation {
fn deserialize(val: &str) -> Result<Self, rustical_xml::XmlError> {
match val {
"i;ascii-casemap" => Ok(Self::AsciiCasemap),
"i;unicode-casemap" => Ok(Self::UnicodeCasemap),
"i;octet" => Ok(Self::Octet),
_ => Err(rustical_xml::XmlError::InvalidVariant(format!(
"Invalid collation: {val}"
@@ -58,6 +60,45 @@ impl ValueDeserialize for NegateCondition {
}
}
#[derive(Clone, Debug, PartialEq, Eq, Default)]
pub enum MatchType {
Equals,
#[default]
Contains,
StartsWith,
EndsWith,
}
impl MatchType {
pub fn match_text(&self, collation: &TextCollation, needle: &str, haystack: &str) -> bool {
let haystack = collation.normalise(haystack);
let needle = collation.normalise(needle);
match &self {
Self::Equals => haystack == needle,
Self::Contains => haystack.contains(needle.as_ref()),
Self::StartsWith => haystack.starts_with(needle.as_ref()),
Self::EndsWith => haystack.ends_with(needle.as_ref()),
}
}
}
impl ValueDeserialize for MatchType {
fn deserialize(val: &str) -> Result<Self, rustical_xml::XmlError> {
Ok(match val {
"equals" => Self::Equals,
"contains" => Self::Contains,
"starts-with" => Self::StartsWith,
"ends-with" => Self::EndsWith,
_ => {
return Err(rustical_xml::XmlError::InvalidVariant(format!(
"Invalid match-type parameter: {val}"
)));
}
})
}
}
#[derive(XmlDeserialize, Clone, Debug, PartialEq, Eq)]
#[allow(dead_code)]
pub struct TextMatchElement {
@@ -65,6 +106,8 @@ pub struct TextMatchElement {
pub collation: TextCollation,
#[xml(ty = "attr", default = "Default::default")]
pub negate_condition: NegateCondition,
#[xml(ty = "attr", default = "Default::default")]
pub match_type: MatchType,
#[xml(ty = "text")]
pub needle: String,
}
@@ -76,12 +119,13 @@ impl TextMatchElement {
collation,
negate_condition,
needle,
match_type,
} = self;
let matches = property
.value
.as_ref()
.is_some_and(|haystack| collation.match_text(needle, haystack));
.is_some_and(|haystack| match_type.match_text(collation, needle, haystack));
// XOR
negate_condition.0 ^ matches
@@ -90,14 +134,20 @@ impl TextMatchElement {
#[cfg(test)]
mod tests {
use crate::xml::MatchType;
use super::TextCollation;
#[test]
fn test_collation() {
assert!(TextCollation::AsciiCasemap.match_text("GrüN", "grün"));
assert!(!TextCollation::AsciiCasemap.match_text("GrÜN", "grün"));
assert!(!TextCollation::Octet.match_text("GrÜN", "grün"));
assert!(TextCollation::Octet.match_text("hallo", "hallo"));
assert!(TextCollation::AsciiCasemap.match_text("HaLlo", "hAllo"));
assert!(!MatchType::Contains.match_text(&TextCollation::AsciiCasemap, "GrÜN", "grünsd"));
assert!(MatchType::Contains.match_text(&TextCollation::AsciiCasemap, "GrüN", "grün"));
assert!(!MatchType::Contains.match_text(&TextCollation::Octet, "GrüN", "grün"));
assert!(MatchType::Contains.match_text(&TextCollation::UnicodeCasemap, "GrÜN", "grün"));
assert!(MatchType::Contains.match_text(&TextCollation::AsciiCasemap, "GrüN", "grün"));
assert!(MatchType::Contains.match_text(&TextCollation::AsciiCasemap, "GrüN", "grün"));
assert!(MatchType::StartsWith.match_text(&TextCollation::Octet, "hello", "hello you"));
assert!(MatchType::EndsWith.match_text(&TextCollation::Octet, "joe", "joe mama"));
assert!(MatchType::Equals.match_text(&TextCollation::UnicodeCasemap, "GrÜN", "grün"));
}
}