Handle case "laa yukallifullah"

Cargo.toml

@@ -1,6 +1,6 @@
 [package]
 name = "quranize"
-version = "0.10.4"
+version = "0.10.5"
 authors = ["Alfan Nur Fauzan <[email protected]>"]
 edition = "2021"
 description = "Encoding transliterations into Quran forms."

src/lib.rs

@@ -14,7 +14,7 @@
 //!
 //! ```
 //! let q = quranize::Quranize::default();
-//! assert_eq!(q.encode("alhamdulillah").first().unwrap().0, "الحمد لله");
+//! assert_eq!(q.encode("alhamdulillah").first().unwrap().0, "الحمد للّه");
 //! ```
 //!
 //! ## Getting an aya text given surah number and ayah number
@@ -60,7 +60,7 @@ impl Default for Quranize {
     ///
     /// ```
     /// let q = quranize::Quranize::default(); // the same with `Quranize::new(usize::MAX)`
-    /// assert_eq!(q.encode("masyaallah").first().unwrap().0, "ما شاء الله");
+    /// assert_eq!(q.encode("masyaallah").first().unwrap().0, "ما شاء اللّه");
     /// ```
     fn default() -> Self {
         Self::new(usize::MAX)
@@ -77,7 +77,7 @@ impl Quranize {
     ///
     /// ```
     /// let q = quranize::Quranize::new(35);
-    /// assert_eq!(q.encode("masyaallah").first().unwrap().0, "ما شاء الله");
+    /// assert_eq!(q.encode("masyaallah").first().unwrap().0, "ما شاء اللّه");
     /// let q = quranize::Quranize::new(1);
     /// assert_eq!(q.encode("masyaallah").first(), None);
     /// ```
@@ -215,11 +215,16 @@ impl Quranize {
 
 #[cfg(test)]
 mod tests {
+    use crate::transliterations::TASYDID;
+
     use super::*;
 
     impl Quranize {
         fn e(&self, text: &str) -> Vec<String> {
-            self.encode(text).into_iter().map(|(q, _, _)| q).collect()
+            self.encode(text)
+                .into_iter()
+                .map(|(q, _, _)| q.chars().filter(|&c| c != TASYDID).collect())
+                .collect()
         }
     }
 
@@ -235,7 +240,7 @@ mod tests {
         assert_eq!(q.e("birobbinnas"), vec!["برب الناس"]);
         assert_eq!(q.e("inna anzalnahu"), vec!["إنا أنزلناه"]);
         assert_eq!(q.e("wa'tasimu"), vec!["واعتصموا"]);
-        assert_eq!(q.e("wabarro"), vec!["وبرا", "وبئر"]);
+        assert_eq!(q.e("wabarro"), vec!["وبرا"]);
         assert_eq!(q.e("idza qodho"), vec!["إذا قضى"]);
         assert_eq!(q.e("masyaallah"), vec!["ما شاء الله"]);
         assert_eq!(q.e("illa man taaba"), vec!["إلا من تاب"]);
@@ -248,6 +253,7 @@ mod tests {
         assert_eq!(q.e("undur kaifa"), vec!["انظر كيف"]);
         assert_eq!(q.e("lirrohman"), vec!["للرحمن"]);
         assert_eq!(q.e("wantum muslimun"), vec!["وأنتم مسلمون"]);
+        assert_eq!(q.e("laa yukallifullah"), vec!["لا يكلف الله"]);
 
         assert_eq!(
             q.e("bismillahirrohmanirrohiim"),
@@ -279,34 +285,33 @@ mod tests {
         assert_eq!(q.e("kaaaf haa yaa aiiin shoood"), vec!["كهيعص"]);
         assert_eq!(q.e("kaf ha ya 'ain shod"), vec!["كهيعص"]);
 
-        assert_eq!(q.locations_index.len(), 685_770);
+        assert_eq!(q.locations_index.len(), 686_059);
     }
 
     #[test]
     fn test_quranize_misc() {
         let q = Quranize::new(23);
-        assert_eq!(q.encode("bismillah")[0].1.len(), 8);
+        assert_eq!(q.encode("bismillah")[0].1.len(), 9);
         assert_eq!(q.encode("bismillah")[0].2, 3);
         assert_eq!(q.encode("arrohman").len(), 1);
-        assert_eq!(q.encode("arrohman")[0].1.len(), 6);
+        assert_eq!(q.encode("arrohman")[0].1.len(), 7);
         assert_eq!(q.encode("alhamdu")[0].1, vec!["al", "ha", "m", "du"]);
         assert_eq!(
             q.encode("arrohman")[0].1,
-            vec!["a", "", "ro", "h", "ma", "n"]
+            vec!["a", "", "r", "ro", "h", "ma", "n"]
         );
         let result = &q.encode("masyaallah")[0];
         assert_eq!(result.0.chars().count(), result.1.len());
         assert_eq!(
             result.1,
-            vec!["m", "a", "", "sy", "a", "a", "", "", "l", "la", "h"]
+            vec!["m", "a", "", "sy", "a", "a", "", "", "", "l", "la", "h"]
         );
     }
 
     #[test]
     fn test_quranize_empty_result() {
         let q = Quranize::new(23);
         assert!(q.encode("").is_empty());
-        assert!(q.encode("aaa").is_empty());
         assert!(q.encode("abcd").is_empty());
         assert!(q.encode("1+2=3").is_empty());
     }
@@ -327,8 +332,8 @@ mod tests {
     fn test_locate() {
         let q = Quranize::new(23);
         assert_eq!(q.get_locations("بسم").first(), Some(&(1, 1, 1)));
-        assert_eq!(q.get_locations("والناس").last(), Some(&(114, 6, 3)));
-        assert_eq!(q.get_locations("بسم الله الرحمن الرحيم").len(), 2);
+        assert_eq!(q.get_locations("والنّاس").last(), Some(&(114, 6, 3)));
+        assert_eq!(q.get_locations("بسم اللّه الرّحمن الرّحيم").len(), 2);
         assert_eq!(q.get_locations("ن").first(), Some(&(68, 1, 1)));
         assert!(q.get_locations("").is_empty());
         assert!(q.get_locations("نن").is_empty());

src/normalization.rs

@@ -1,11 +1,8 @@
 pub(crate) fn normalize(text: &str) -> String {
-    let mut chars = Vec::from_iter(text.chars().filter_map(|c| match c.to_ascii_lowercase() {
+    let chars = Vec::from_iter(text.chars().filter_map(|c| match c.to_ascii_lowercase() {
         c @ ('a'..='z' | '\'' | ' ') => Some(c),
         _ => None,
     }));
-    chars.dedup_by(|&mut a, &mut b| {
-        a == b && !(a == 'l' || a == 'a' || a == 'o' || a == 'i' || a == 'u')
-    });
     chars.into_iter().filter(|&c| c != ' ').collect()
 }
 
@@ -28,8 +25,8 @@ mod tests {
         assert_eq!(normalize("'aalimul ghoibi"), "'aalimulghoibi");
         assert_eq!(normalize("Qul A'udzu"), "qula'udzu");
         assert_eq!(
-            normalize("bismilla hirrohman nirrohiim"),
-            "bismillahirohmannirohiim"
+            normalize("bismilla hirrohma nirrohiim"),
+            "bismillahirrohmanirrohiim"
         );
     }
 

src/quran/mod.rs

@@ -8,7 +8,7 @@ use simple_plain::RAW_QURAN as SIMPLE_PLAIN;
 const SURA_COUNT: usize = 114;
 const AYA_COUNT: usize = 6236;
 
-/// Accept raw Quran string, return an iterator for each ayah in the Quran with surah number and ayah number.
+/// Returns an iterator of `(sura_number, aya_number, aya_text)` that iterates each ayah in the Quran.
 pub(crate) fn iter() -> impl Iterator<Item = (u8, u16, &'static str)> {
     iter_quran(SIMPLE_PLAIN)
 }
@@ -33,10 +33,11 @@ pub(crate) trait CleanCharsExt {
     fn clean_chars(&self) -> Filter<Chars, fn(&char) -> bool>;
 }
 
-use crate::transliterations as trans;
+use crate::transliterations::{self as trans, TASYDID};
 impl CleanCharsExt for str {
     fn clean_chars(&self) -> Filter<Chars, fn(&char) -> bool> {
-        self.chars().filter(|&c| !trans::map(c).is_empty())
+        self.chars()
+            .filter(|&c| c == TASYDID || !trans::map(c).is_empty())
     }
 }
 
@@ -112,7 +113,7 @@ mod tests {
         assert_same_basmalah(SIMPLE_PLAIN);
         assert_eq!(iter_quran(SIMPLE_PLAIN).count(), AYA_COUNT);
         assert_eq!(count_unique_simple_clean_chars(), 37);
-        assert_eq!(count_unique_simple_plain_chars(), 37);
+        assert_eq!(count_unique_simple_plain_chars(), 38);
     }
 
     fn assert_same_basmalah(raw: &str) {
@@ -160,7 +161,13 @@ mod tests {
     #[test]
     fn test_clean_chars() {
         for ((_, _, clean), (_, _, plain)) in iter_quran(SIMPLE_CLEAN).zip(iter()) {
-            assert_eq!(clean, plain.clean_chars().collect::<String>());
+            assert_eq!(
+                clean,
+                plain
+                    .clean_chars()
+                    .filter(|&c| c != TASYDID)
+                    .collect::<String>()
+            );
         }
     }
 }

src/transliterations.rs

@@ -1,3 +1,5 @@
+pub(crate) const TASYDID: char = '\u{0651}';
+
 pub(crate) fn map(c: char) -> &'static [&'static str] {
     match c {
         'ء' => &["'", "k", "a", "i", "u"],
@@ -43,7 +45,7 @@ pub(crate) fn map(c: char) -> &'static [&'static str] {
         'ف' => &["f", "fa", "fi", "fu"],
         'ق' => &["k", "ko", "ki", "ku", "q", "qo", "qi", "qu", "qa"],
         'ك' => &["k", "ka", "ki", "ku"],
-        'ل' => &["l", "ll", "lla", "la", "li", "lu"],
+        'ل' => &["l", "la", "li", "lu"],
         'م' => &["m", "ma", "mi", "mu"],
         'ن' => &["n", "na", "ni", "nu"],
         'ه' => &["h", "ha", "hi", "hu"],
@@ -57,9 +59,12 @@ pub(crate) fn map(c: char) -> &'static [&'static str] {
 
 pub(crate) fn contextual_map(prev_c: char, c: char) -> &'static [&'static str] {
     match (prev_c, c) {
-        (' ', 'ا') | ('ب', 'ا') | ('ا', 'ل') | ('و', 'ا') | ('أ', 'و') => &[""],
+        (' ', 'ا') | ('ب', 'ا') | ('ا', 'ل') | ('آ', 'ل') | ('و', 'ا') | ('أ', 'و') => {
+            &[""]
+        }
         ('\0', 'ا') => &["i", "u"],
         ('ل', 'ل') => &["i"],
+        (_, TASYDID) => map(prev_c),
         _ => &[],
     }
 }