Add French implementation

Makefile.am

@@ -12,7 +12,7 @@ licensedir = $(datarootdir)/licenses/$(TRANSFORMED_PACKAGE_NAME)
 bin_PROGRAMS = decasify
 decasify_SOURCES  = src/bin/decasify.rs src/content.rs src/cli.rs src/lib.rs src/types.rs src/traits.rs
 decasify_SOURCES += src/lua.rs src/python.rs src/wasm.rs
-decasify_SOURCES += src/en.rs src/tr.rs
+decasify_SOURCES += src/en.rs src/fr.rs src/tr.rs
 EXTRA_decasify_SOURCES = tests/cli.rs tests/lib.rs
 EXTRA_DIST = pyproject.toml spec/decasify_spec.lua tests/test_all.py plugin/decasify.lua sile/decasify.lua
 dist_doc_DATA = README.md CHANGELOG.md

pyproject.toml

@@ -12,6 +12,7 @@ classifiers = [
   "Development Status :: 5 - Production/Stable",
   "Intended Audience :: Developers",
   "Natural Language :: English",
+  "Natural Language :: French",
   "Natural Language :: Turkish",
   "Programming Language :: Python :: Implementation :: CPython",
   "Programming Language :: Python :: Implementation :: PyPy",

src/fr.rs

@@ -0,0 +1,106 @@
+// SPDX-FileCopyrightText: © 2023 Caleb Maclennan <[email protected]>
+// SPDX-License-Identifier: LGPL-3.0-only
+
+use crate::content::{Chunk, Segment, Word};
+use crate::types::StyleGuide;
+
+use regex::Regex;
+use unicode_titlecase::StrTitleCase;
+
+pub use crate::en::lowercase;
+pub use crate::en::sentencecase;
+pub use crate::en::uppercase;
+
+pub fn titlecase(chunk: Chunk, style: StyleGuide) -> String {
+    match style {
+        StyleGuide::LanguageDefault => titlecase_fr(chunk),
+        _ => todo!("French implementation doesn't support this style guide."),
+    }
+}
+
+fn titlecase_fr(chunk: Chunk) -> String {
+    let mut segments: Vec<Segment> = Vec::new();
+    chunk.clone().segments.into_iter().for_each(|segment| {
+        match segment {
+            Segment::Separator(_) => segments.push(segment),
+            Segment::Word(ref word) => {
+                let mut segs = word.word.split("-").peekable();
+                while let Some(s) = segs.next() {
+                    segments.push(Segment::Word(Word { word: s.into() }));
+                    if segs.peek().is_some() {
+                        segments.push(Segment::Separator("-".into()));
+                    }
+                }
+            }
+        };
+    });
+    let mut chunk = chunk.clone();
+    let mut words = chunk
+        .segments
+        .iter_mut()
+        .filter_map(|segment| match segment {
+            Segment::Word(word) => Some(word),
+            _ => None,
+        })
+        .peekable();
+    if let Some(word) = words.next() {
+        word.word = word.to_titlecase_lower_rest();
+    }
+    while let Some(word) = words.next() {
+        word.word = match words.peek().is_none() {
+            true => word.to_titlecase_lower_rest(),
+            false => match is_reserved(word) {
+                true => word.to_lowercase(),
+                false => word.to_titlecase_lower_rest(),
+            },
+        };
+    }
+    chunk.into()
+}
+
+fn is_reserved(word: &Word) -> bool {
+    let word = word.to_lowercase();
+    let word = word.as_str();
+    // https://github.com/benoitvallon/titlecase-french/blob/83e092e91dccdd39871dfeac0d58dc06d997dabb/config.js#L22
+    let lower_case_word_list = vec![
+        "le", "la", "les", // definite articles
+        "un", "une", "des", // indefinite articles
+        "du", "de", "des", // partitive articles
+        "au", "aux", "du", "des", // contracted articles
+        "ce", "cet", "cette", "ces", // demonstrative adjectives
+        "quel", "quels", "quelle", "quelles", // exclamative adjectives
+        "mon", "ton", "son", "notre", "votre", "leur", "ma", "ta", "sa", "mes", "tes", "ses",
+        "nos", "vos", "leurs", // possessive adjectives
+        "mais", "ou", "et", "donc", "or", "ni", "car", "voire",
+        // coordinating conjunctions
+        "que", "qu", "quand", "comme", "si", "lorsque", "lorsqu", "puisque", "puisqu", "quoique",
+        "quoiqu", // subordinating conjunctions
+        "à", "chez", "dans", "entre", "jusque", "jusqu", "hors", "par", "pour", "sans", "vers",
+        "sur", "pas", "parmi", "avec", "sous", "en", // prepositions
+        "je", "tu", "il", "elle", "on", "nous", "vous", "ils", "elles", "me", "te", "se", "y",
+        // personal pronouns
+        "qui", "que", "quoi", "dont", "où", // relative pronouns
+        "ne", // others
+    ];
+    let lower_case_words = lower_case_word_list.join("|");
+    let lower_case_word = Regex::new(format!("^({lower_case_words})$").as_ref()).unwrap();
+    lower_case_word.is_match(word)
+}
+
+//capitalizedSpecials: [
+//  { input: 'À', output: 'A'},
+//  { input: 'Â', output: 'A'},
+//  { input: 'Ä', output: 'A'},
+//  { input: 'É', output: 'E'},
+//  { input: 'È', output: 'E'},
+//  { input: 'Ê', output: 'E'},
+//  { input: 'Ë', output: 'E'},
+//  { input: 'Ç', output: 'C'},
+//  { input: 'Î', output: 'I'},
+//  { input: 'Ï', output: 'I'},
+//  { input: 'Ô', output: 'O'},
+//  { input: 'Ö', output: 'O'},
+//  { input: 'Û', output: 'U'},
+//  { input: 'Ü', output: 'U'},
+//  { input: 'Ù', output: 'U'}
+//],

src/lib.rs

@@ -30,6 +30,7 @@ pub mod python;
 pub mod wasm;
 
 mod en;
+mod fr;
 mod tr;
 
 /// Convert a string to a specific case following typesetting conventions for a target locale
@@ -62,6 +63,7 @@ pub fn titlecase(
     let style: StyleGuide = style.into();
     match locale {
         Locale::EN => en::titlecase(chunk, style),
+        Locale::FR => fr::titlecase(chunk, style),
         Locale::TR => tr::titlecase(chunk, style),
     }
 }
@@ -72,6 +74,7 @@ pub fn lowercase(chunk: impl Into<Chunk>, locale: impl Into<Locale>) -> String {
     let locale: Locale = locale.into();
     match locale {
         Locale::EN => en::lowercase(chunk),
+        Locale::FR => fr::lowercase(chunk),
         Locale::TR => tr::lowercase(chunk),
     }
 }
@@ -82,6 +85,7 @@ pub fn uppercase(chunk: impl Into<Chunk>, locale: impl Into<Locale>) -> String {
     let locale: Locale = locale.into();
     match locale {
         Locale::EN => en::uppercase(chunk),
+        Locale::FR => fr::uppercase(chunk),
         Locale::TR => tr::uppercase(chunk),
     }
 }
@@ -92,6 +96,7 @@ pub fn sentencecase(chunk: impl Into<Chunk>, locale: impl Into<Locale>) -> Strin
     let locale: Locale = locale.into();
     match locale {
         Locale::EN => en::sentencecase(chunk),
+        Locale::FR => fr::sentencecase(chunk),
         Locale::TR => tr::sentencecase(chunk),
     }
 }

src/types.rs

@@ -43,6 +43,7 @@ pub type Result<T, E = Error> = std::result::Result<T, E>;
 pub enum Locale {
     #[default]
     EN,
+    FR,
     TR,
 }
 
@@ -85,6 +86,7 @@ impl FromStr for Locale {
     fn from_str(s: &str) -> Result<Self> {
         match s.to_ascii_lowercase().as_str() {
             "en" | "english" | "en_en" => Ok(Locale::EN),
+            "fr" | "french" | "fr_fr" | "français" => Ok(Locale::FR),
             "tr" | "turkish" | "tr_tr" | "türkçe" => Ok(Locale::TR),
             input => LocaleSnafu { input }.fail()?,
         }

tests/lib.rs

@@ -213,6 +213,38 @@ titlecase!(
     "  Serbest  Serseri\n  Boşluk  "
 );
 
+titlecase!(
+    french_def,
+    Locale::FR,
+    StyleGuide::LanguageDefault,
+    "le triangle rouge",
+    "Le Triangle Rouge"
+);
+
+titlecase!(
+    french_def2,
+    Locale::FR,
+    StyleGuide::LanguageDefault,
+    "loki, le détective mythique",
+    "Loki, le Détective Mythique"
+);
+
+titlecase!(
+    french_coordinating,
+    Locale::FR,
+    StyleGuide::LanguageDefault,
+    "il est studieux mais turbulent",
+    "Il Est Studieux mais Turbulent"
+);
+
+titlecase!(
+    french_coordinating2,
+    Locale::FR,
+    StyleGuide::LanguageDefault,
+    "mais comment font-ils?",
+    "Mais Comment Font-Ils?"
+);
+
 macro_rules! lowercase {
     ($name:ident, $locale:expr, $input:expr, $expected:expr) => {
         #[test]
@@ -225,6 +257,8 @@ macro_rules! lowercase {
 
 lowercase!(lower_en, Locale::EN, "foo BAR BaZ BIKE", "foo bar baz bike");
 
+lowercase!(lower_fr, Locale::FR, "foo BAR BaZ BIKE", "foo bar baz bike");
+
 lowercase!(
     lower_tr,
     Locale::TR,
@@ -244,6 +278,8 @@ macro_rules! uppercase {
 
 uppercase!(upper_en, Locale::EN, "foo BAR BaZ bike", "FOO BAR BAZ BIKE");
 
+uppercase!(upper_fr, Locale::FR, "foo BAR BaZ bike", "FOO BAR BAZ BIKE");
+
 uppercase!(
     upper_tr,
     Locale::TR,
@@ -268,4 +304,11 @@ sentencecase!(
     "Insert bike here"
 );
 
+sentencecase!(
+    sentence_fr,
+    Locale::FR,
+    "insert BIKE here",
+    "Insert bike here"
+);
+
 sentencecase!(sentence_tr, Locale::TR, "ilk DAVRANSIN", "İlk davransın");