Merge pull request #793 from not-my-profile/refactor-file-type-specifics

refactor(cli): Introduce file_type_specifics module

Cargo.lock

@@ -1685,6 +1685,7 @@ dependencies = [
  "serde",
  "serde_json",
  "serde_regex",
+ "snapbox",
  "thread_local",
  "toml",
  "trycmd",

crates/typos-cli/Cargo.toml

@@ -86,6 +86,7 @@ encoding_rs = "0.8.32"
 assert_fs = "1.0"
 trycmd = "0.14.16"
 criterion = "0.5"
+snapbox = "0.4.11"
 
 [[bench]]
 name = "checks"

crates/typos-cli/src/config.rs

@@ -2,6 +2,8 @@ use std::collections::HashMap;
 
 use kstring::KString;
 
+use crate::file_type_specifics;
+
 pub const SUPPORTED_FILE_NAMES: &[&str] = &["typos.toml", "_typos.toml", ".typos.toml"];
 
 #[derive(Debug, Clone, Default, PartialEq, Eq, serde::Serialize, serde::Deserialize)]
@@ -155,9 +157,11 @@ pub struct TypeEngineConfig {
 
 impl TypeEngineConfig {
     pub fn from_defaults() -> Self {
-        let patterns = [
+        let mut patterns = HashMap::new();
-            (
+
-                KString::from("lock"),
+        for no_check_type in file_type_specifics::NO_CHECK_TYPES {
+            patterns.insert(
+                KString::from(*no_check_type),
                 GlobEngineConfig {
                     extend_glob: Vec::new(),
                     engine: EngineConfig {
@@ -165,83 +169,34 @@ impl TypeEngineConfig {
                         ..Default::default()
                     },
                 },
-            ),
+            );
-            (
+        }
-                KString::from("vim"),
+
+        for (typ, dict_config) in file_type_specifics::TYPE_SPECIFIC_DICTS {
+            patterns.insert(
+                KString::from(*typ),
                 GlobEngineConfig {
                     extend_glob: Vec::new(),
                     engine: EngineConfig {
                         dict: Some(DictConfig {
-                            extend_identifiers: maplit::hashmap! {
+                            extend_identifiers: dict_config
-                                "windo".into() => "windo".into(),
+                                .ignore_idents
-                            },
+                                .iter()
+                                .map(|key| ((*key).into(), (*key).into()))
+                                .collect(),
+                            extend_words: dict_config
+                                .ignore_words
+                                .iter()
+                                .map(|key| ((*key).into(), (*key).into()))
+                                .collect(),
                             ..Default::default()
                         }),
                         ..Default::default()
                     },
                 },
-            ),
+            );
-            (
+        }
-                KString::from("vimscript"),
+
-                GlobEngineConfig {
-                    extend_glob: Vec::new(),
-                    engine: EngineConfig {
-                        dict: Some(DictConfig {
-                            extend_identifiers: maplit::hashmap! {
-                                "windo".into() => "windo".into(),
-                            },
-                            ..Default::default()
-                        }),
-                        ..Default::default()
-                    },
-                },
-            ),
-            (
-                KString::from("rust"),
-                GlobEngineConfig {
-                    extend_glob: Vec::new(),
-                    engine: EngineConfig {
-                        dict: Some(DictConfig {
-                            extend_identifiers: maplit::hashmap! {
-                                "flate2".into() => "flate2".into(),
-                            },
-                            extend_words: maplit::hashmap! {
-                                "ser".into() => "ser".into(),
-                            },
-                            ..Default::default()
-                        }),
-                        ..Default::default()
-                    },
-                },
-            ),
-            (
-                KString::from("py"),
-                GlobEngineConfig {
-                    extend_glob: Vec::new(),
-                    engine: EngineConfig {
-                        dict: Some(DictConfig {
-                            extend_identifiers: maplit::hashmap! {
-                                "NDArray".into() => "NDArray".into(),
-                            },
-                            ..Default::default()
-                        }),
-                        ..Default::default()
-                    },
-                },
-            ),
-            (
-                KString::from("cert"),
-                GlobEngineConfig {
-                    extend_glob: Vec::new(),
-                    engine: EngineConfig {
-                        check_file: Some(false),
-                        ..Default::default()
-                    },
-                },
-            ),
-        ]
-        .into_iter()
-        .collect();
         Self { patterns }
     }
 

crates/typos-cli/src/file_type_specifics.rs

@@ -0,0 +1,80 @@
+//! This module specifies [`EngineConfig`] defaults for the file types defined in [`default_types`].
+//!
+//! [`EngineConfig`]: crate::config::EngineConfig
+//! [`default_types`]: crate::default_types
+
+/// Set `check_file` to `false` for these types.
+pub const NO_CHECK_TYPES: &[&str] = &["cert", "lock"];
+
+pub const TYPE_SPECIFIC_DICTS: &[(&str, StaticDictConfig)] = &[
+    (
+        "py",
+        StaticDictConfig {
+            ignore_idents: &[
+                "NDArray", // numpy.typing.NDArray
+            ],
+            ignore_words: &[],
+        },
+    ),
+    (
+        "rust",
+        StaticDictConfig {
+            ignore_idents: &[
+                "flate2", // https://crates.io/crates/flate2
+            ],
+            ignore_words: &[
+                "ser", // serde::ser, serde_json::ser, etc.
+            ],
+        },
+    ),
+    (
+        "vim",
+        StaticDictConfig {
+            ignore_idents: &[
+                "windo", // https://vimdoc.sourceforge.net/htmldoc/windows.html#:windo
+            ],
+            ignore_words: &[],
+        },
+    ),
+    (
+        "vimscript",
+        StaticDictConfig {
+            ignore_idents: &[
+                "windo", // https://vimdoc.sourceforge.net/htmldoc/windows.html#:windo
+            ],
+            ignore_words: &[],
+        },
+    ),
+];
+
+pub struct StaticDictConfig {
+    pub ignore_idents: &'static [&'static str],
+    pub ignore_words: &'static [&'static str],
+}
+
+#[cfg(test)]
+mod tests {
+    use itertools::Itertools;
+
+    use super::TYPE_SPECIFIC_DICTS;
+
+    #[test]
+    fn test_type_specific_dicts_contains_no_duplicates() {
+        let types: Vec<_> = TYPE_SPECIFIC_DICTS.iter().map(|(typ, _)| *typ).collect();
+        let types_unique: Vec<_> = types.clone().into_iter().unique().collect();
+
+        snapbox::assert_eq(types.join("\n"), types_unique.join("\n"));
+    }
+
+    #[test]
+    fn test_type_specific_dicts_is_sorted() {
+        // The order of the entries in TYPE_SPECIFIC_DICTS actually doesn't
+        // affect the runtime behavior, we just want them ordered
+        // so that it's easier to find entries for contributors.
+
+        let types: Vec<_> = TYPE_SPECIFIC_DICTS.iter().map(|(typ, _)| *typ).collect();
+        let types_sorted: Vec<_> = types.iter().cloned().sorted().collect();
+
+        snapbox::assert_eq(types.join("\n"), types_sorted.join("\n"));
+    }
+}

crates/typos-cli/src/lib.rs

@@ -14,3 +14,4 @@ pub mod report;
 
 mod default_types;
 mod file_type;
+mod file_type_specifics;