From ab5333c60dad0fce905963e008a9c55660184cbd Mon Sep 17 00:00:00 2001
From: raskad <32105367+raskad@users.noreply.github.com>
Date: Sun, 7 Jan 2024 21:40:04 +0100
Subject: [PATCH] Add parsing for `DerivedNormalizationProps.txt`

---
 .../src/derived_normalization_properties.rs   | 64 +++++++++++++++++++
 ucd-parse/src/lib.rs                          |  2 +
 2 files changed, 66 insertions(+)
 create mode 100644 ucd-parse/src/derived_normalization_properties.rs

diff --git a/ucd-parse/src/derived_normalization_properties.rs b/ucd-parse/src/derived_normalization_properties.rs
new file mode 100644
index 0000000..dd7f352
--- /dev/null
+++ b/ucd-parse/src/derived_normalization_properties.rs
@@ -0,0 +1,64 @@
+use std::path::Path;
+
+use crate::{
+    common::{
+        parse_codepoint_association, CodepointIter, Codepoints, UcdFile,
+        UcdFileByCodepoint,
+    },
+    error::Error,
+};
+
+/// A single row in the `DerivedNormalizationProps.txt` file.
+#[derive(Clone, Debug, Default, Eq, PartialEq)]
+pub struct DerivedNormalizationProperty {
+    /// The codepoint or codepoint range for this entry.
+    pub codepoints: Codepoints,
+    /// The property name assigned to the codepoints in this entry.
+    pub property: String,
+}
+
+impl UcdFile for DerivedNormalizationProperty {
+    fn relative_file_path() -> &'static Path {
+        Path::new("DerivedNormalizationProps.txt")
+    }
+}
+
+impl UcdFileByCodepoint for DerivedNormalizationProperty {
+    fn codepoints(&self) -> CodepointIter {
+        self.codepoints.into_iter()
+    }
+}
+
+impl std::str::FromStr for DerivedNormalizationProperty {
+    type Err = Error;
+
+    fn from_str(line: &str) -> Result<DerivedNormalizationProperty, Error> {
+        let (codepoints, property) = parse_codepoint_association(line)?;
+        Ok(DerivedNormalizationProperty {
+            codepoints,
+            property: property.to_string(),
+        })
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::DerivedNormalizationProperty;
+
+    #[test]
+    fn parse_single() {
+        let line =
+            "00A0          ; Changes_When_NFKC_Casefolded # Zs       NO-BREAK SPACE\n";
+        let row: DerivedNormalizationProperty = line.parse().unwrap();
+        assert_eq!(row.codepoints, 0xA0);
+        assert_eq!(row.property, "Changes_When_NFKC_Casefolded");
+    }
+
+    #[test]
+    fn parse_range() {
+        let line = "0041..005A    ; Changes_When_NFKC_Casefolded # L&  [26] LATIN CAPITAL LETTER A..LATIN CAPITAL LETTER Z\n";
+        let row: DerivedNormalizationProperty = line.parse().unwrap();
+        assert_eq!(row.codepoints, (0x41, 0x5A));
+        assert_eq!(row.property, "Changes_When_NFKC_Casefolded");
+    }
+}
diff --git a/ucd-parse/src/lib.rs b/ucd-parse/src/lib.rs
index 61a33b1..fbd301d 100644
--- a/ucd-parse/src/lib.rs
+++ b/ucd-parse/src/lib.rs
@@ -19,6 +19,7 @@ pub use crate::{
     bidi_mirroring_glyph::BidiMirroring,
     case_folding::{CaseFold, CaseStatus},
     core_properties::CoreProperty,
+    derived_normalization_properties::DerivedNormalizationProperty,
     east_asian_width::EastAsianWidth,
     emoji_properties::EmojiProperty,
     grapheme_cluster_break::{GraphemeClusterBreak, GraphemeClusterBreakTest},
@@ -65,6 +66,7 @@ mod arabic_shaping;
 mod bidi_mirroring_glyph;
 mod case_folding;
 mod core_properties;
+mod derived_normalization_properties;
 mod east_asian_width;
 mod emoji_properties;
 mod grapheme_cluster_break;