feat(core): create first draft of rule for missing articles

harper-comments/tests/language_support_sources/laravel_app.php

@@ -112,7 +112,7 @@
     |
     | These configuration options determine the driver used to determine and
     | manage Laravel's "maintenance mode" status. The "cache" driver will
-    | allow maintenance mode to be controlled across multiple machines.
+    | allow the maintenance mode to be controlled across multiple machines.
     |
     | Supported drivers: "file", "cache"
     |

harper-core/dictionary.dict

@@ -28011,10 +28011,10 @@ highway/~NVgS
 highwayman/~Ng
 highwaymen/~9
 hijab/~NSg
-hijack/~VNSzZGgd>
+hijack/~VNSzZGgd
 hijacker/~Ng
 hijacking/~VNwg
-hike/~NVgZGd>S
+hike/~NVgZGdS
 hiker/Ng
 hiking/~VNmg
 hilarious/~JpY

harper-core/src/expr/mod.rs

@@ -150,6 +150,8 @@ where
 pub trait OwnedExprExt {
     fn or(self, other: impl Expr + 'static) -> FirstMatchOf;
     fn or_longest(self, other: impl Expr + 'static) -> LongestMatchOf;
+    fn and(self, other: impl Expr + 'static) -> All;
+    fn and_not(self, other: impl Expr + 'static) -> All;
 }
 
 impl<E> OwnedExprExt for E
@@ -167,4 +169,12 @@ where
     fn or_longest(self, other: impl Expr + 'static) -> LongestMatchOf {
         LongestMatchOf::new(vec![Box::new(self), Box::new(other)])
     }
+
+    fn and(self, other: impl Expr + 'static) -> All {
+        All::new(vec![Box::new(self), Box::new(other)])
+    }
+
+    fn and_not(self, other: impl Expr + 'static) -> All {
+        self.and(UnlessStep::new(other, |_: &Token, _: &[char]| true))
+    }
 }

harper-core/src/lib.rs

@@ -20,6 +20,7 @@ mod punctuation;
 mod render_markdown;
 mod span;
 pub mod spell;
+mod starts_with_vowel;
 mod sync;
 mod title_case;
 mod token;

harper-core/src/linting/an_a.rs

@@ -1,9 +1,7 @@
-use std::borrow::Cow;
-
 use itertools::Itertools;
 
 use crate::linting::{Lint, LintKind, Linter, Suggestion};
-use crate::{Document, TokenStringExt};
+use crate::{Document, TokenStringExt, starts_with_vowel::starts_with_vowel};
 
 #[derive(Debug, Default)]
 pub struct AnA;
@@ -78,119 +76,6 @@ impl Linter for AnA {
     }
 }
 
-fn to_lower_word(word: &[char]) -> Cow<'_, [char]> {
-    if word.iter().any(|c| c.is_uppercase()) {
-        Cow::Owned(
-            word.iter()
-                .flat_map(|c| c.to_lowercase())
-                .collect::<Vec<_>>(),
-        )
-    } else {
-        Cow::Borrowed(word)
-    }
-}
-
-/// Checks whether a provided word begins with a vowel _sound_.
-///
-/// It was produced through trial and error.
-/// Matches with 99.71% and 99.77% of vowels and non-vowels in the
-/// Carnegie-Mellon University word -> pronunciation dataset.
-fn starts_with_vowel(word: &[char]) -> bool {
-    let is_likely_initialism = word.iter().all(|c| c.is_uppercase());
-
-    if is_likely_initialism && !word.is_empty() {
-        return matches!(
-            word[0],
-            'A' | 'E' | 'F' | 'H' | 'I' | 'L' | 'M' | 'N' | 'O' | 'R' | 'S' | 'X'
-        );
-    }
-
-    let word = to_lower_word(word);
-    let word = word.as_ref();
-
-    if matches!(
-        word,
-        [] | ['u', 'k', ..]
-            | ['e', 'u', 'p', 'h', ..]
-            | ['e', 'u', 'g' | 'l' | 'c', ..]
-            | ['o', 'n', 'e']
-            | ['o', 'n', 'c', 'e']
-    ) {
-        return false;
-    }
-
-    if matches!(word, |['h', 'o', 'u', 'r', ..]| ['h', 'o', 'n', ..]
-        | ['u', 'n', 'i', 'n' | 'm', ..]
-        | ['u', 'n', 'a' | 'u', ..]
-        | ['h', 'e', 'r', 'b', ..]
-        | ['u', 'r', 'b', ..]
-        | ['i', 'n', 't', ..])
-    {
-        return true;
-    }
-
-    if matches!(word, ['u', 'n' | 's', 'i' | 'a' | 'u', ..]) {
-        return false;
-    }
-
-    if matches!(word, ['u', 'n', ..]) {
-        return true;
-    }
-
-    if matches!(word, ['u', 'r', 'g', ..]) {
-        return true;
-    }
-
-    if matches!(word, ['u', 't', 't', ..]) {
-        return true;
-    }
-
-    if matches!(
-        word,
-        ['u', 't' | 'r' | 'n', ..] | ['e', 'u', 'r', ..] | ['u', 'w', ..] | ['u', 's', 'e', ..]
-    ) {
-        return false;
-    }
-
-    if matches!(word, ['o', 'n', 'e', 'a' | 'e' | 'i' | 'u', 'l' | 'd', ..]) {
-        return true;
-    }
-
-    if matches!(word, ['o', 'n', 'e', 'a' | 'e' | 'i' | 'u' | '-' | 's', ..]) {
-        return false;
-    }
-
-    if matches!(
-        word,
-        ['s', 'o', 's']
-            | ['r', 'z', ..]
-            | ['n', 'g', ..]
-            | ['n', 'v', ..]
-            | ['x']
-            | ['x', 'b', 'o', 'x']
-            | ['h', 'e', 'i', 'r', ..]
-            | ['h', 'o', 'n', 'o', 'r', ..]
-    ) {
-        return true;
-    }
-
-    if matches!(
-        word,
-        ['j', 'u' | 'o', 'n', ..] | ['j', 'u', 'r', 'a' | 'i' | 'o', ..]
-    ) {
-        return false;
-    }
-
-    if matches!(word, ['x', '-' | '\'' | '.' | 'o' | 's', ..]) {
-        return true;
-    }
-
-    matches!(
-        word,
-        ['a', ..] | ['e', ..] | ['i', ..] | ['o', ..] | ['u', ..]
-    )
-}
-
 #[cfg(test)]
 mod tests {
     use super::AnA;

harper-core/src/linting/confident.rs

@@ -45,7 +45,7 @@ impl ExprLinter for Confident {
     }
 
     fn description(&self) -> &'static str {
-        "This linter detects instances where the noun `confidant` is incorrectly used in place of the adjective `confident`. `Confidant` refers to a trusted person, whereas `confident` describes certainty or self-assurance. The rule suggests replacing `confidant` with `confident` when used in an adjectival context."
+        "This linter detects instances where the noun `confidant` is incorrectly used in place of the adjective `confident`. `Confidant` refers to a trusted person, whereas `confident` describes some certainty or self-assurance. The rule suggests replacing `confidant` with `confident` when used in an adjectival context."
     }
 }
 

harper-core/src/linting/lint_group.rs

@@ -57,6 +57,7 @@ use super::likewise::Likewise;
 use super::long_sentences::LongSentences;
 use super::looking_forward_to::LookingForwardTo;
 use super::merge_words::MergeWords;
+use super::missing_article::MissingArticle;
 use super::missing_preposition::MissingPreposition;
 use super::modal_of::ModalOf;
 use super::most_number::MostNumber;
@@ -400,6 +401,7 @@ impl LintGroup {
         insert_struct_rule!(CapitalizePersonalPronouns, true);
         insert_expr_rule!(ChockFull, true);
         insert_expr_rule!(DoubleModal, true);
+        insert_expr_rule!(MissingArticle, true);
         insert_expr_rule!(MissingPreposition, true);
         insert_struct_rule!(DiscourseMarkers, true);
         insert_expr_rule!(WayTooAdjective, true);