sqruff_lib_core/
dialects.rs

1pub mod common;
2pub mod init;
3pub mod syntax;
4
5use std::borrow::Cow;
6use std::collections::hash_map::Entry;
7use std::fmt::Debug;
8
9use ahash::{AHashMap, AHashSet};
10
11use crate::dialects::init::DialectKind;
12use crate::dialects::syntax::SyntaxKind;
13use crate::helpers::ToMatchable;
14use crate::parser::lexer::{Lexer, Matcher};
15use crate::parser::matchable::Matchable;
16use crate::parser::parsers::StringParser;
17use crate::parser::types::DialectElementType;
18
19#[derive(Debug, Clone, Default)]
20pub struct Dialect {
21    pub name: DialectKind,
22    lexer_matchers: Option<Vec<Matcher>>,
23    library: AHashMap<Cow<'static, str>, DialectElementType>,
24    sets: AHashMap<&'static str, AHashSet<&'static str>>,
25    pub bracket_collections: AHashMap<&'static str, AHashSet<BracketPair>>,
26    lexer: Option<Lexer>,
27}
28
29impl PartialEq for Dialect {
30    fn eq(&self, other: &Self) -> bool {
31        self.name == other.name
32    }
33}
34
35impl Dialect {
36    pub fn new() -> Self {
37        Dialect {
38            name: DialectKind::Ansi,
39            ..Default::default()
40        }
41    }
42
43    pub fn name(&self) -> DialectKind {
44        self.name
45    }
46
47    pub fn add(&mut self, iter: impl IntoIterator<Item = (Cow<'static, str>, DialectElementType)>) {
48        self.library.extend(iter);
49    }
50
51    pub fn grammar(&self, name: &str) -> Matchable {
52        match self
53            .library
54            .get(name)
55            .unwrap_or_else(|| panic!("not found {name}"))
56        {
57            DialectElementType::Matchable(matchable) => matchable.clone(),
58            DialectElementType::SegmentGenerator(_) => {
59                unreachable!("Attempted to fetch non grammar [{name}] with `Dialect::grammar`.")
60            }
61        }
62    }
63
64    #[track_caller]
65    pub fn replace_grammar(&mut self, name: &'static str, match_grammar: Matchable) {
66        match self.library.entry(Cow::Borrowed(name)) {
67            Entry::Occupied(entry) => {
68                let target = entry.into_mut();
69                match target {
70                    DialectElementType::Matchable(matchable) => {
71                        if let Some(node_matcher) = matchable.as_node_matcher() {
72                            node_matcher.replace(match_grammar);
73                        } else {
74                            *target = DialectElementType::Matchable(match_grammar);
75                        }
76                    }
77                    DialectElementType::SegmentGenerator(_) => {
78                        *target = DialectElementType::Matchable(match_grammar);
79                    }
80                }
81            }
82            Entry::Vacant(entry) => {
83                entry.insert(match_grammar.into());
84            }
85        }
86    }
87
88    pub fn lexer_matchers(&self) -> &[Matcher] {
89        match &self.lexer_matchers {
90            Some(lexer_matchers) => lexer_matchers,
91            None => panic!("Lexing struct has not been set for dialect {self:?}"),
92        }
93    }
94
95    pub fn insert_lexer_matchers(&mut self, lexer_patch: Vec<Matcher>, before: &str) {
96        let mut buff = Vec::new();
97        let mut found = false;
98
99        if self.lexer_matchers.is_none() {
100            panic!("Lexer struct must be defined before it can be patched!");
101        }
102
103        for elem in self.lexer_matchers.take().unwrap() {
104            if elem.name() == before {
105                found = true;
106                for patch in lexer_patch.clone() {
107                    buff.push(patch);
108                }
109                buff.push(elem);
110            } else {
111                buff.push(elem);
112            }
113        }
114
115        if !found {
116            panic!("Lexer struct insert before '{before}' failed because tag never found.");
117        }
118
119        self.lexer_matchers = Some(buff);
120    }
121
122    pub fn patch_lexer_matchers(&mut self, lexer_patch: Vec<Matcher>) {
123        let mut buff = Vec::with_capacity(self.lexer_matchers.as_ref().map_or(0, Vec::len));
124        if self.lexer_matchers.is_none() {
125            panic!("Lexer struct must be defined before it can be patched!");
126        }
127
128        let patch_dict: AHashMap<&'static str, Matcher> = lexer_patch
129            .into_iter()
130            .map(|elem| (elem.name(), elem))
131            .collect();
132
133        for elem in self.lexer_matchers.take().unwrap() {
134            if let Some(patch) = patch_dict.get(elem.name()) {
135                buff.push(patch.clone());
136            } else {
137                buff.push(elem);
138            }
139        }
140
141        self.lexer_matchers = Some(buff);
142    }
143
144    pub fn set_lexer_matchers(&mut self, lexer_matchers: Vec<Matcher>) {
145        self.lexer_matchers = lexer_matchers.into();
146    }
147
148    pub fn sets(&self, label: &str) -> AHashSet<&'static str> {
149        match label {
150            "bracket_pairs" | "angle_bracket_pairs" => {
151                panic!("Use `bracket_sets` to retrieve {label} set.");
152            }
153            _ => (),
154        }
155
156        self.sets.get(label).cloned().unwrap_or_default()
157    }
158
159    pub fn sets_mut(&mut self, label: &'static str) -> &mut AHashSet<&'static str> {
160        assert!(
161            label != "bracket_pairs" && label != "angle_bracket_pairs",
162            "Use `bracket_sets` to retrieve {label} set."
163        );
164
165        match self.sets.entry(label) {
166            Entry::Occupied(entry) => entry.into_mut(),
167            Entry::Vacant(entry) => entry.insert(<_>::default()),
168        }
169    }
170
171    pub fn update_keywords_set_from_multiline_string(
172        &mut self,
173        set_label: &'static str,
174        values: &'static str,
175    ) {
176        let keywords = values.lines().map(str::trim);
177        self.sets_mut(set_label).extend(keywords);
178    }
179
180    pub fn add_keyword_to_set(&mut self, set_label: &'static str, value: &'static str) {
181        self.sets_mut(set_label).insert(value);
182    }
183
184    pub fn bracket_sets(&self, label: &str) -> AHashSet<BracketPair> {
185        assert!(
186            label == "bracket_pairs" || label == "angle_bracket_pairs",
187            "Invalid bracket set. Consider using another identifier instead."
188        );
189
190        self.bracket_collections
191            .get(label)
192            .cloned()
193            .unwrap_or_default()
194    }
195
196    pub fn bracket_sets_mut(&mut self, label: &'static str) -> &mut AHashSet<BracketPair> {
197        assert!(
198            label == "bracket_pairs" || label == "angle_bracket_pairs",
199            "Invalid bracket set. Consider using another identifier instead."
200        );
201
202        self.bracket_collections.entry(label).or_default()
203    }
204
205    pub fn update_bracket_sets(&mut self, label: &'static str, pairs: Vec<BracketPair>) {
206        let set = self.bracket_sets_mut(label);
207        for pair in pairs {
208            set.insert(pair);
209        }
210    }
211
212    #[track_caller]
213    pub fn r#ref(&self, name: &str) -> Matchable {
214        match self.library.get(name) {
215            Some(DialectElementType::Matchable(matchable)) => matchable.clone(),
216            Some(DialectElementType::SegmentGenerator(_)) => {
217                panic!("Unexpected SegmentGenerator while fetching '{name}'");
218            }
219            None => {
220                panic!("Grammar refers to '{name}' which was not found in the dialect.",);
221            }
222        }
223    }
224
225    pub fn expand(&mut self) {
226        // Temporarily take ownership of 'library' from 'self' to avoid borrow checker
227        // errors during mutation.
228        let mut library = std::mem::take(&mut self.library);
229        for element in library.values_mut() {
230            if let DialectElementType::SegmentGenerator(generator) = element {
231                *element = DialectElementType::Matchable(generator.expand(self));
232            }
233        }
234        self.library = library;
235
236        for keyword_set in [
237            "unreserved_keywords",
238            "reserved_keywords",
239            "future_reserved_keywords",
240        ] {
241            if let Some(keywords) = self.sets.get(keyword_set) {
242                for &kw in keywords {
243                    if !self.library.contains_key(kw) {
244                        let parser = StringParser::new(kw, SyntaxKind::Keyword);
245
246                        self.library.insert(
247                            kw.into(),
248                            DialectElementType::Matchable(parser.to_matchable()),
249                        );
250                    }
251                }
252            }
253        }
254
255        self.lexer = Lexer::new(self.lexer_matchers()).into();
256    }
257
258    pub fn lexer(&self) -> &Lexer {
259        self.lexer.as_ref().unwrap()
260    }
261}
262
263pub type BracketPair = (&'static str, &'static str, &'static str, bool);