Merge pull request #7 from aynumosir/kana-release

Prepare kana feature for release

Author: neet

.github/workflows/python.yaml

@@ -24,7 +24,7 @@ jobs:
         with:
           target: x86_64
           args: --release --out dist --find-interpreter --manifest-path ./ainu-utils-python/Cargo.toml
-          # sccache: 'true'
+          sccache: 'true'
           manylinux: auto
 
       - name: pytest

ainu-utils-js/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "ainu-utils-js"
-version = "0.2.0"
+version = "0.3.0"
 edition = "2021"
 description = "A collection of utilities for the Ainu language"
 license = "MIT"

ainu-utils-python/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "ainu-utils-python"
-version = "0.2.1"
+version = "0.3.0"
 edition = "2021"
 
 [lib]

ainu-utils-python/pyproject.toml

@@ -9,7 +9,7 @@ features = ["pyo3/extension-module"]
 name = "ainu-utils"
 description = "A collection of utility for with the Ainu language"
 requires-python = ">=3.8"
-version = "0.2.1"
+version = "0.3.0"
 license = "MIT"
 
 [project.optional-dependencies]

ainu-utils-python/tests/test_tokenize.py

@@ -7,7 +7,7 @@ def test_tokenize():
 
 
 def test_tokenize_with_whitespace():
-    result = ainu_utils.segment("irankarapte. e=iwanke ya?", keep_whitespace=True)
+    result = ainu_utils.tokenize("irankarapte. e=iwanke ya?", keep_whitespace=True)
     assert result == ["irankarapte", ".", " ", "e=", "iwanke", " ", "ya", "?"]
 
 

ainu-utils/Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "ainu-utils"
-version = "0.2.0"
+version = "0.3.0"
 edition = "2021"
 description = "A collection of utilities for the Ainu language"
 license = "MIT"

ainu-utils/src/kana/sets.rs renamed to ainu-utils/src/kana/constants.rs

@@ -7,7 +7,9 @@ pub static VOWELS: Lazy<HashSet<char>> = Lazy::new(|| {
 });
 
 pub static CONSONANTS: Lazy<HashSet<char>> = Lazy::new(|| {
-    let items = ['k', 's', 't', 'c', 'n', 'h', 'p', 'm', 'y', 'r', 'w'];
+    let items = [
+        'k', 'g', 's', 'z', 't', 'd', 'c', 'j', 'n', 'h', 'p', 'b', 'f', 'm', 'y', 'r', 'w',
+    ];
     items.iter().cloned().collect()
 });
 

ainu-utils/src/kana/kana.rs

@@ -1,10 +1,9 @@
-use super::sets::{CONSONANTS, SPECIAL_CONSONANTS, SYMBOLS, VOWELS};
-use super::tables::{TABLE_1, TABLE_2, TABLE_3};
+use super::constants::{CONSONANTS, SPECIAL_CONSONANTS, SYMBOLS, VOWELS};
+use super::linking::link;
+use super::maps::{TABLE_1, TABLE_2, TABLE_3};
 use diacritics::remove_diacritics;
 
-fn normalize(input: &str) -> String {
-    let mut input = input.to_string();
-
+fn normalize(mut input: String) -> String {
     input = input.to_lowercase();
     input = input.replace("=", "");
     input = remove_diacritics(&input).to_string();
@@ -13,7 +12,11 @@ fn normalize(input: &str) -> String {
 }
 
 pub fn to_kana(input: &str) -> String {
-    let input = normalize(input);
+    let mut input = input.to_string();
+
+    input = normalize(input);
+    input = link(input);
+
     let chars: Vec<char> = input.chars().collect();
 
     let mut kana = String::new();

ainu-utils/src/kana/kana_test.rs

@@ -19,8 +19,134 @@ fn test_to_kana() {
 			"アシㇼパ エキㇺネ パテㇰ キ ワ、 メノコ モンライケ エアイカㇷ゚。",
 			"ケメイキ ネ ヤ、 イテセ ネ ヤ、 メノコ モンライケ エアイカㇷ゚ メノコ アナㇰ、 アイヌ ホク コㇿ カ エアイカㇷ゚。",
 			"タネ シヌイェ クニ パハ ネ コㇿカ、 コパン。",
-			"スgイモト ニㇱパ、 タン マッカチ エトゥン ワ エンコレ!",
+			"スギモト ニㇱパ、 タン マッカチ エトゥン ワ エンコレ!",
 			"タン クミッポホ クエポタラ ワ モシㇼ クホッパ カ コヤイクㇱ。",
 		].join(" ")
 	)
 }
+
+#[test]
+fn test_dropping_h() {
+    assert_eq!(to_kana("_hine"), "イネ")
+}
+
+#[test]
+fn test_dropping_y() {
+    assert_eq!(to_kana("_ya?"), "ア?")
+}
+
+#[test]
+fn test_linking_h() {
+    assert_eq!(to_kana("hawean __hi"), "ハウェアニ")
+}
+
+#[test]
+fn test_linking_y() {
+    assert_eq!(to_kana("nankor __ya?"), "ナンコラ?")
+}
+
+#[test]
+fn test_linking_a() {
+    assert_eq!(to_kana("cis _a cis _a"), "チサ チサ")
+}
+
+#[test]
+fn test_linking_i() {
+    assert_eq!(to_kana("oar _isam"), "オアリサㇺ")
+}
+
+#[test]
+fn test_linking_u() {
+    assert_eq!(to_kana("or _un"), "オルン")
+}
+
+#[test]
+fn test_linking_e() {
+    assert_eq!(to_kana("mat _etun"), "マテトゥン")
+}
+
+#[test]
+fn test_linking_o() {
+    assert_eq!(to_kana("pet _or _un"), "ペトルン")
+}
+
+#[test]
+fn test_linking_and_dropping_a() {
+    assert_eq!(to_kana("yaypuri ekira __ani"), "ヤイプリ エキラニ")
+}
+
+#[test]
+fn test_linking_and_dropping_i() {
+    assert_eq!(to_kana("puni __i"), "プニ")
+}
+
+#[test]
+fn test_linking_and_dropping_u() {
+    assert_eq!(to_kana("a=kotanu __un"), "アコタヌン")
+}
+
+#[test]
+fn test_linking_and_dropping_e() {
+    assert_eq!(to_kana("i=samake __en anu"), "イサマケン アヌ")
+}
+
+#[test]
+fn test_linking_and_dropping_o() {
+    // 実例なし。
+    assert_eq!(to_kana("sapporo __or"), "サッポロㇿ")
+}
+
+#[test]
+fn test_linking_r_n() {
+    assert_eq!(to_kana("a=kor_ nispa"), "アコン ニㇱパ")
+}
+
+#[test]
+fn test_linking_r_r() {
+    assert_eq!(to_kana("kor_ rusuy"), "コン ルスイ")
+}
+
+#[test]
+fn test_linking_r_t() {
+    assert_eq!(to_kana("or_ ta"), "オッ タ")
+}
+
+#[test]
+fn test_linking_r_c() {
+    assert_eq!(to_kana("yar_ cise"), "ヤッ チセ")
+}
+
+#[test]
+fn test_linking_n_s() {
+    assert_eq!(to_kana("pon_ su"), "ポイ ス")
+}
+
+#[test]
+fn test_linking_n_y() {
+    assert_eq!(to_kana("pon_ yam"), "ポイ ヤㇺ")
+}
+
+#[test]
+fn test_linking_n_w() {
+    assert_eq!(to_kana("san _wa"), "サン マ")
+}
+
+#[test]
+fn test_linking_m_w() {
+    assert_eq!(to_kana("isam _wa"), "イサン マ")
+}
+
+#[test]
+fn test_linking_p_w() {
+    assert_eq!(to_kana("sap _wa"), "サッ パ")
+}
+
+#[test]
+fn test_special_mp() {
+    assert_eq!(to_kana("tampaku"), "タンパク")
+}
+
+#[test]
+fn test_special_mm() {
+    assert_eq!(to_kana("umma"), "ウンマ")
+}

ainu-utils/src/kana/linking.rs

@@ -0,0 +1,38 @@
+// https://ainugo.nam.go.jp/pages/guide.html
+static LINKING: [(&str, &str); 26] = [
+    ("_h", ""),
+    ("_y", ""),
+    (" __h", ""),
+    (" __y", ""),
+    (" _a", "a"),
+    (" _i", "i"),
+    (" _u", "u"),
+    (" _e", "e"),
+    (" _o", "o"),
+    (" __a", ""),
+    (" __i", ""),
+    (" __u", ""),
+    (" __e", ""),
+    (" __o", ""),
+    ("r_ n", "n n"),
+    ("r_ r", "n r"),
+    ("r_ t", "t t"),
+    ("r_ c", "t c"),
+    ("n_ s", "y s"),
+    ("n_ y", "y y"),
+    ("n_ w", "u w"),
+    ("n _w", "n m"),
+    // ("n _w", "n m"),
+    ("m _w", "n m"),
+    ("p _w", "t p"),
+    ("mp", "np"),
+    ("mm", "nm"),
+];
+
+pub fn link(mut input: String) -> String {
+    for (from, to) in LINKING.iter() {
+        input = input.replace(from, to);
+    }
+
+    input
+}