Merge branch 'master' into dep-fixes

harper-cli/Cargo.toml

@@ -9,7 +9,7 @@ repository = "https://github.com/automattic/harper"
 [dependencies]
 anyhow = "1.0.95"
 ariadne = "0.4.1"
-clap = { version = "4.5.28", features = ["derive"] }
+clap = { version = "4.5.29", features = ["derive"] }
 harper-literate-haskell = { path = "../harper-literate-haskell", version = "0.21.0" }
 harper-core = { path = "../harper-core", version = "0.21.0" }
 harper-comments = { path = "../harper-comments", version = "0.21.0" }

harper-core/Cargo.toml

@@ -19,7 +19,7 @@ paste = "1.0.14"
 pulldown-cmark = "0.12.2"
 serde = { version = "1.0.214", features = ["derive"] }
 serde_json = "1.0.138"
-smallvec = { version = "1.13.2", features = ["serde"] }
+smallvec = { version = "1.14.0", features = ["serde"] }
 thiserror = "2.0.11"
 unicode-blocks = "0.1.9"
 unicode-script = "0.5.7"

harper-core/dictionary.dict

@@ -8570,7 +8570,6 @@ Ryan/2M
 Rydberg/2M
 Ryder/2M
 Ryukyu/2M
-S/152MN
 SA/5124
 SAC/12
 SALT/2M
@@ -37361,7 +37360,6 @@ pox/14MS
 pp/1
 ppm/1
 ppr
-pr/5
 practicability/1M
 practicably/
 practical/51SMY
@@ -49928,3 +49926,19 @@ backplane/SM
 cyberattack/SM
 RTX/1SM
 PDP/1SM
+Jetpack/2M
+learnt/4
+WooCommerce/2M
+Tumblr/2M
+Akismet/2M
+Simplenote/2M
+Longreads/2M
+VaultPress/2M
+Gravatar/2M
+Crowdsignal/2M
+Mullenweg/M
+journaling/SM3
+Polldaddy/2M
+P2/M
+VideoPress/2M
+stylization/SM

harper-core/src/document.rs

@@ -129,7 +129,10 @@ impl Document {
         for token in self.tokens.iter_mut() {
             if let TokenKind::Word(meta) = &mut token.kind {
                 let word_source = token.span.get_content(&self.source);
-                let found_meta = dictionary.get_word_metadata(word_source);
+                let found_meta = dictionary
+                    .get_correct_capitalization_of(word_source)
+                    .map(|canonical_caps| dictionary.get_word_metadata(canonical_caps))
+                    .unwrap_or_default();
                 *meta = meta.or(&found_meta);
             }
         }
@@ -535,6 +538,7 @@ macro_rules! create_fns_on_doc {
 
 impl TokenStringExt for Document {
     create_fns_on_doc!(word);
+    create_fns_on_doc!(hostname);
     create_fns_on_doc!(word_like);
     create_fns_on_doc!(conjunction);
     create_fns_on_doc!(space);

harper-core/src/linting/compound_nouns/general_compound_nouns.rs

@@ -58,6 +58,7 @@ impl PatternLinter for GeneralCompoundNouns {
 
     fn match_to_lint(&self, matched_tokens: &[Token], source: &[char]) -> Lint {
         let span = matched_tokens.span().unwrap();
+        let orig = span.get_content(source);
         // If the pattern matched, this will not return `None`.
         let word = self
             .split_pattern
@@ -67,7 +68,7 @@ impl PatternLinter for GeneralCompoundNouns {
         Lint {
             span,
             lint_kind: LintKind::Spelling,
-            suggestions: vec![Suggestion::ReplaceWith(word.to_vec())],
+            suggestions: vec![Suggestion::replace_with_match_case(word.to_vec(), orig)],
             message: format!(
                 "Did you mean the closed compound noun “{}”?",
                 word.to_string()

harper-core/src/linting/compound_nouns/mod.rs

@@ -226,7 +226,7 @@ mod tests {
     #[test]
     fn web_socket() {
         let test_sentence = "Real-time updates are sent via a web socket.";
-        let expected = "Real-time updates are sent via a WebSocket.";
+        let expected = "Real-time updates are sent via a websocket.";
         assert_suggestion_result(test_sentence, CompoundNouns::default(), expected);
     }
 

harper-core/src/linting/lint_group.rs

@@ -47,14 +47,18 @@ use super::phrase_corrections::MutePoint;
 use super::phrase_corrections::StateOfTheArt;
 use super::phrase_corrections::WantBe;
 use super::phrase_corrections::{
-    AndAlike, BadRap, BatedBreath, BeckAndCall, ChangeTack, EnMasse, HumanLife, HungerPang,
+    AndTheLike, BadRap, BatedBreath, BeckAndCall, ChangeTack, EnMasse, HumanLife, HungerPang,
     LetAlone, LoAndBehold, NeedHelp, NoLonger, OfCourse, SneakingSuspicion, SpecialAttention,
     SupposedTo, ThanOthers, ThatChallenged, TurnItOff,
 };
 use super::pique_interest::PiqueInterest;
 use super::plural_conjugate::PluralConjugate;
 use super::possessive_your::PossessiveYour;
 use super::pronoun_contraction::PronounContraction;
+use super::proper_noun_capitalization_linters::DayOneNames;
+use super::proper_noun_capitalization_linters::JetpackNames;
+use super::proper_noun_capitalization_linters::PocketCastsNames;
+use super::proper_noun_capitalization_linters::TumblrNames;
 use super::proper_noun_capitalization_linters::{
     AmazonNames, Americas, AppleNames, Australia, AzureNames, Canada, ChineseCommunistParty,
     GoogleNames, Holidays, Koreas, Malaysia, MetaNames, MicrosoftNames, UnitedOrganizations,
@@ -72,6 +76,7 @@ use super::unclosed_quotes::UnclosedQuotes;
 use super::use_genitive::UseGenitive;
 use super::was_aloud::WasAloud;
 use super::whereas::Whereas;
+use super::wordpress_dotcom::WordPressDotcom;
 use super::wrong_quotes::WrongQuotes;
 use super::{CurrencyPlacement, Lint, Linter, NoOxfordComma, OxfordComma};
 use crate::{Dictionary, Document};
@@ -195,6 +200,11 @@ macro_rules! create_lint_group_config {
 }
 
 create_lint_group_config!(
+    WordPressDotcom => true,
+    DayOneNames => true,
+    PocketCastsNames => true,
+    TumblrNames => true,
+    JetpackNames => true,
     OutOfDate => true,
     Desktop => true,
     Laptop => true,
@@ -312,7 +322,7 @@ create_lint_group_config!(
     ThatChallenged => true,
     TurnItOff => true,
     OfCourse => true,
-    AndAlike => true,
+    AndTheLike => true,
     BadRap => true,
     BatedBreath => true,
     BeckAndCall => true,

harper-core/src/linting/mod.rs

@@ -56,6 +56,7 @@ mod unclosed_quotes;
 mod use_genitive;
 mod was_aloud;
 mod whereas;
+mod wordpress_dotcom;
 mod wrong_quotes;
 
 pub use an_a::AnA;
@@ -109,14 +110,18 @@ pub use phrase_corrections::MutePoint;
 pub use phrase_corrections::StateOfTheArt;
 pub use phrase_corrections::WantBe;
 pub use phrase_corrections::{
-    AndAlike, BadRap, BatedBreath, BeckAndCall, ChangeTack, EnMasse, HumanLife, HungerPang,
+    AndTheLike, BadRap, BatedBreath, BeckAndCall, ChangeTack, EnMasse, HumanLife, HungerPang,
     LetAlone, LoAndBehold, NeedHelp, NoLonger, OfCourse, SneakingSuspicion, SpecialAttention,
     SupposedTo, ThanOthers, ThatChallenged, TurnItOff,
 };
 pub use pique_interest::PiqueInterest;
 pub use plural_conjugate::PluralConjugate;
 pub use possessive_your::PossessiveYour;
 pub use pronoun_contraction::PronounContraction;
+pub use proper_noun_capitalization_linters::DayOneNames;
+pub use proper_noun_capitalization_linters::JetpackNames;
+pub use proper_noun_capitalization_linters::PocketCastsNames;
+pub use proper_noun_capitalization_linters::TumblrNames;
 pub use proper_noun_capitalization_linters::{
     AmazonNames, Americas, AppleNames, Australia, AzureNames, Canada, ChineseCommunistParty,
     GoogleNames, Holidays, Koreas, Malaysia, MetaNames, MicrosoftNames, UnitedOrganizations,
@@ -135,6 +140,7 @@ pub use unclosed_quotes::UnclosedQuotes;
 pub use use_genitive::UseGenitive;
 pub use was_aloud::WasAloud;
 pub use whereas::Whereas;
+pub use wordpress_dotcom::WordPressDotcom;
 pub use wrong_quotes::WrongQuotes;
 
 use crate::Document;

harper-core/src/linting/phrase_corrections.rs

@@ -78,7 +78,7 @@ create_linter_for_phrase!(ThatChallenged, "that challenged", 2);
 create_linter_for_phrase!(NoLonger, "no longer", 1);
 create_linter_for_phrase!(NeedHelp, "need help", 1);
 create_linter_for_phrase!(OfCourse, "of course", 1);
-create_linter_for_phrase!(AndAlike, "and alike", 1);
+create_linter_for_phrase!(AndTheLike, "and the like", 1);
 create_linter_for_phrase!(BadRap, "bad rap", 1);
 create_linter_for_phrase!(BatedBreath, "bated breath", 1);
 create_linter_for_phrase!(BeckAndCall, "beck and call", 1);

harper-core/src/linting/possessive_your.rs

@@ -39,7 +39,7 @@ impl PatternLinter for PossessiveYour {
                 Suggestion::replace_with_match_case("your".chars().collect(), orig_chars),
                 Suggestion::replace_with_match_case("you're an".chars().collect(), orig_chars),
             ],
-            message: "The possesive version of this word is more common in this context."
+            message: "The possessive version of this word is more common in this context."
                 .to_owned(),
             ..Default::default()
         }

harper-core/src/linting/proper_noun_capitalization_linters.rs

@@ -684,6 +684,113 @@ create_linter_for!(
     "When referring to Meta products and services, make sure to treat them as proper nouns."
 );
 
+create_linter_for!(
+    JetpackNames,
+    SequencePattern::default()
+        .t_aco("Jetpack")
+        .then_whitespace()
+        .then(Box::new(EitherPattern::new(vec![
+            Box::new(
+                SequencePattern::default()
+                    .t_aco("VaultPress")
+                    .then_whitespace()
+                    .t_aco("Backup")
+            ),
+            Box::new(SequencePattern::default().t_aco("VaultPress")),
+            Box::new(SequencePattern::default().t_aco("Scan")),
+            Box::new(
+                SequencePattern::default()
+                    .t_aco("Akismet")
+                    .then_whitespace()
+                    .t_aco("Anti-spam")
+            ),
+            Box::new(SequencePattern::default().t_aco("Stats")),
+            Box::new(SequencePattern::default().t_aco("Social")),
+            Box::new(SequencePattern::default().t_aco("Blaze")),
+            Box::new(
+                SequencePattern::default()
+                    .t_aco("AI")
+                    .then_whitespace()
+                    .t_aco("Assistant")
+            ),
+            Box::new(
+                SequencePattern::default()
+                    .t_aco("Site")
+                    .then_whitespace()
+                    .t_aco("Search")
+            ),
+            Box::new(SequencePattern::default().t_aco("Boost")),
+            Box::new(SequencePattern::default().t_aco("VideoPress")),
+            Box::new(
+                SequencePattern::default()
+                    .t_aco("For")
+                    .then_whitespace()
+                    .t_aco("Agencies")
+            ),
+            Box::new(SequencePattern::default().t_aco("CRM")),
+        ]))),
+    "Ensure proper capitalization of Jetpack-related terms."
+);
+
+create_linter_for!(
+    TumblrNames,
+    SequencePattern::default()
+        .t_aco("Tumblr")
+        .then_whitespace()
+        .then(Box::new(EitherPattern::new(vec![
+            Box::new(SequencePattern::default().t_aco("Blaze")),
+            Box::new(SequencePattern::default().t_aco("Pro")),
+            Box::new(SequencePattern::default().t_aco("Live")),
+            Box::new(SequencePattern::default().t_aco("Ads")),
+            Box::new(SequencePattern::default().t_aco("Communities")),
+            Box::new(SequencePattern::default().t_aco("Shop")),
+            Box::new(SequencePattern::default().t_aco("Dashboard"))
+        ]))),
+    "Ensure proper capitalization of Tumblr-related terms."
+);
+
+create_linter_for!(
+    PocketCastsNames,
+    EitherPattern::new(vec![
+        Box::new(
+            SequencePattern::default()
+                .t_aco("Pocket")
+                .then_whitespace()
+                .t_aco("Casts")
+        ),
+        Box::new(
+            SequencePattern::default()
+                .t_aco("Pocket")
+                .then_whitespace()
+                .t_aco("Casts")
+                .then_whitespace()
+                .t_aco("Plus")
+        )
+    ]),
+    "Ensure proper capitalization of Pocket Casts and Pocket Casts Plus as brand names."
+);
+
+create_linter_for!(
+    DayOneNames,
+    EitherPattern::new(vec![
+        Box::new(
+            SequencePattern::default()
+                .t_aco("Day")
+                .then_whitespace()
+                .t_aco("One")
+        ),
+        Box::new(
+            SequencePattern::default()
+                .t_aco("Day")
+                .then_whitespace()
+                .t_aco("One")
+                .then_whitespace()
+                .t_aco("Premium")
+        )
+    ]),
+    "Ensure proper capitalization of Day One and Day One Premium as brand names."
+);
+
 #[cfg(test)]
 mod tests {
     use crate::linting::tests::{assert_lint_count, assert_suggestion_result};

harper-core/src/linting/spell_check.rs

@@ -5,7 +5,7 @@ use super::Suggestion;
 use super::{Lint, LintKind, Linter};
 use crate::document::Document;
 use crate::spell::suggest_correct_spelling;
-use crate::{CharString, Dictionary, TokenStringExt};
+use crate::{CharString, CharStringExt, Dictionary, TokenStringExt};
 
 pub struct SpellCheck<T>
 where
@@ -56,7 +56,9 @@ impl<T: Dictionary> Linter for SpellCheck<T> {
 
         for word in document.iter_words() {
             let word_chars = document.get_span_content(word.span);
-            if self.dictionary.contains_word(word_chars) {
+            if self.dictionary.contains_exact_word(word_chars)
+                || self.dictionary.contains_exact_word(&word_chars.to_lower())
+            {
                 continue;
             }