compiler/
symbol.rs

1use dynamic::{ConstIntOp, Dynamic, Type};
2use parser::Stmt;
3use smol_str::SmolStr;
4use std::{collections::BTreeMap, rc::Rc, sync::Arc};
5
6use super::Capture;
7
8#[derive(Debug, Clone, Default)]
9pub enum Symbol {
10    #[default]
11    Null,
12    Const {
13        value: Dynamic,
14        ty: Type,
15        is_pub: bool,
16    },
17    Static {
18        value: Option<Dynamic>,
19        ty: Type,
20        is_pub: bool,
21    },
22    Struct(Type, bool),
23    Fn {
24        ty: Type,
25        args: Vec<SmolStr>,
26        generic_params: Vec<Type>,
27        cap: Capture,
28        body: Arc<Stmt>,
29        is_pub: bool,
30    },
31    Native(Type),
32}
33
34impl Symbol {
35    pub fn native(tys: Vec<Type>, ret: Type) -> Self {
36        Self::Native(Type::Fn { tys, ret: Rc::new(ret) })
37    }
38
39    pub fn is_pub(&self) -> bool {
40        match self {
41            Self::Const { value: _, ty: _, is_pub } => *is_pub,
42            Self::Static { value: _, ty: _, is_pub } => *is_pub,
43            Self::Struct(_, is_pub) => *is_pub,
44            Self::Fn { ty: _, args: _, generic_params: _, cap: _, body: _, is_pub } => *is_pub,
45            _ => true,
46        }
47    }
48
49    pub fn is_fn(&self) -> bool {
50        match self {
51            Self::Fn { ty: _, args: _, generic_params: _, cap: _, body: _, is_pub: _ } => true,
52            Self::Native(_) => true,
53            _ => false,
54        }
55    }
56}
57
58use anyhow::{Result, anyhow};
59use indexmap::IndexMap;
60
61pub fn eval_const_int_type(ty: &Type) -> Option<i64> {
62    match ty {
63        Type::ConstInt(value) => Some(*value),
64        Type::ConstBinary { op, left, right } => {
65            let left = eval_const_int_type(left)?;
66            let right = eval_const_int_type(right)?;
67            match op {
68                ConstIntOp::Add => Some(left + right),
69                ConstIntOp::Sub => Some(left - right),
70                ConstIntOp::Mul => Some(left * right),
71                ConstIntOp::Div => (right != 0).then_some(left / right),
72                ConstIntOp::Mod => (right != 0).then_some(left % right),
73            }
74        }
75        _ => None,
76    }
77}
78
79pub fn substitute_type(ty: &Type, params: &[Type], args: &[Type]) -> Type {
80    match ty {
81        Type::Ident { name, params: nested } if nested.is_empty() => {
82            params.iter().position(|param| matches!(param, Type::Ident { name: param_name, params } if params.is_empty() && param_name == name)).map(|idx| args[idx].clone()).unwrap_or_else(|| ty.clone())
83        }
84        Type::Ident { name, params: nested } => Type::Ident { name: name.clone(), params: nested.iter().map(|param| substitute_type(param, params, args)).collect() },
85        Type::Struct { params: struct_params, fields } => Type::Struct {
86            params: struct_params.iter().map(|param| substitute_type(param, params, args)).collect(),
87            fields: fields.iter().map(|(name, field_ty)| (name.clone(), substitute_type(field_ty, params, args))).collect(),
88        },
89        Type::List(elem) => Type::List(Rc::new(substitute_type(elem, params, args))),
90        Type::Vec(elem, len) => Type::Vec(Rc::new(substitute_type(elem, params, args)), *len),
91        Type::Array(elem, len) => Type::Array(Rc::new(substitute_type(elem, params, args)), *len),
92        Type::ArrayParam(elem, len) => Type::ArrayParam(Rc::new(substitute_type(elem, params, args)), Rc::new(substitute_type(len, params, args))),
93        Type::ConstBinary { op, left, right } => {
94            let left = substitute_type(left, params, args);
95            let right = substitute_type(right, params, args);
96            let ty = Type::ConstBinary { op: *op, left: Rc::new(left), right: Rc::new(right) };
97            eval_const_int_type(&ty).map(Type::ConstInt).unwrap_or(ty)
98        }
99        Type::Fn { tys, ret } => Type::Fn { tys: tys.iter().map(|ty| substitute_type(ty, params, args)).collect(), ret: Rc::new(substitute_type(ret, params, args)) },
100        Type::Symbol { id, params: nested } => Type::Symbol { id: *id, params: nested.iter().map(|param| substitute_type(param, params, args)).collect() },
101        Type::Tuple(items) => Type::Tuple(items.iter().map(|item| substitute_type(item, params, args)).collect()),
102        _ => ty.clone(),
103    }
104}
105
106#[derive(Clone, Default)]
107pub struct SymbolTable {
108    pub symbols: IndexMap<SmolStr, Symbol>,
109    modules: BTreeMap<SmolStr, BTreeMap<SmolStr, u32>>,
110    pub roots: Vec<SmolStr>,
111}
112
113impl SymbolTable {
114    pub fn add_to_module(&mut self, module: &str, name: SmolStr, s: Symbol) -> Result<u32> {
115        let full_name: SmolStr = format!("{}::{}", module, name).into();
116        let id = self.symbols.insert_full(full_name, s).0 as u32;
117        let module_symbols = self.modules.get_mut(module).ok_or_else(|| anyhow!("模块 {} 不存在", module))?;
118        module_symbols.insert(name, id);
119        Ok(id)
120    }
121    pub fn get_symbol(&self, idx: u32) -> Result<(&SmolStr, &Symbol)> {
122        self.symbols.get_index(idx as usize).ok_or(anyhow!("未发现符号 {}", idx))
123    }
124
125    pub fn get_symbol_mut(&mut self, idx: u32) -> Option<(&SmolStr, &mut Symbol)> {
126        self.symbols.get_index_mut(idx as usize)
127    }
128
129    pub fn symbol(&self, name: &str) -> Vec<(SmolStr, u32)> {
130        self.modules.get(name).map(|m| m.iter().map(|(name, id)| (name.clone(), *id)).collect()).unwrap_or(Vec::new())
131    }
132
133    pub fn disassemble(&self, name: &str) -> Result<String> {
134        let id = self.get_id(name)?;
135        let (name, s) = self.get_symbol(id)?;
136        if let Symbol::Fn { ty, args, generic_params: _, cap, body, is_pub } = s {
137            if *is_pub { Ok(format!("pub {} {:?} {:?} {:?}\n{}", name, ty, args, cap, body)) } else { Ok(format!("{} {:?} {:?} {:?}\n{}", name, ty, args, cap, body)) }
138        } else {
139            Err(anyhow!("未发现符号 {}", name))
140        }
141    }
142
143    pub fn get_field(&self, ty: &Type, name: &str) -> Result<(usize, Type)> {
144        //原生类型的函数 is_map is_list 或者 sqrt
145        let id = match ty {
146            Type::Any => {
147                if let Ok(id) = self.get_id("Any")
148                    && let Ok((_, Symbol::Struct(any_ty, _))) = self.get_symbol(id)
149                    && let Ok((idx, field_ty)) = any_ty.get_field(name)
150                {
151                    return Ok((idx, field_ty.clone()));
152                }
153                match name {
154                    "is_map" | "is_list" | "is_string" | "is_null" | "contains" | "starts_with" => return Ok((usize::MAX, Type::Bool)),
155                    "len" => return Ok((usize::MAX, Type::I32)),
156                    _ => return Ok((usize::MAX, Type::Any)),
157                }
158            }
159            Type::Struct { params: _, fields: _ } => {
160                return ty.get_field(name).map(|(idx, ty)| (idx, ty.clone()));
161            }
162            Type::Str => {
163                let any_method = match name {
164                    "len" | "contains" | "split" | "starts_with" | "is_string" | "is_null" => format!("Any::{}", name),
165                    _ => return Err(anyhow!("未发现 symbol {:?} {}", ty, name)),
166                };
167                return Ok((usize::MAX, Type::Symbol { id: self.get_id(&any_method)?, params: Vec::new() }));
168            }
169            Type::List(_) | Type::Array(_, _) => {
170                let any_method = match name {
171                    "len" | "push" | "pop" | "get_idx" | "set_idx" | "slice" | "is_list" | "is_null" => format!("Any::{}", name),
172                    _ => return Err(anyhow!("未发现 symbol {:?} {}", ty, name)),
173                };
174                return Ok((usize::MAX, Type::Symbol { id: self.get_id(&any_method)?, params: Vec::new() }));
175            }
176            Type::Symbol { id, params: _ } => *id,
177            Type::Vec(_, _) => self.get_id("Vec")?,
178            Type::Fn { tys: _, ret } => {
179                return self.get_field(ret, name);
180            }
181            _ => {
182                //增加一个外部函数定义
183                if matches!(name, "is_map" | "is_list" | "is_string" | "is_null") {
184                    return Ok((usize::MAX, Type::Symbol { id: self.get_id(&format!("Any::{}", name))?, params: Vec::new() }));
185                }
186                return Err(anyhow!("未发现 symbol {:?} {}", ty, name));
187            }
188        };
189        let (_, s) = self.get_symbol(id)?;
190        if let Symbol::Struct(s, _) = s {
191            return s.get_field(name).and_then(|(idx, ty)| Ok((idx, ty.clone())));
192        };
193        Err(anyhow!("未发现 field {:?} {}", ty, name))
194    }
195
196    pub fn get_type(&self, ty: &Type) -> Result<Type> {
197        match ty {
198            Type::Ident { name, params } => {
199                let params = params.iter().map(|param| self.get_type(param)).collect::<Result<Vec<_>>>()?;
200                if name.as_str() == "Vec" && params.len() == 1 {
201                    return Ok(Type::Vec(Rc::new(params[0].clone()), 0));
202                }
203                if name.as_str() == "List" {
204                    return Ok(if params.is_empty() { Type::list_any() } else { Type::List(Rc::new(params[0].clone())) });
205                }
206                let id = self.get_id(&name)?;
207                if let (_, Symbol::Struct(ty, _)) = self.get_symbol(id)? {
208                    if let Type::Struct { params: generic_params, .. } = ty
209                        && !generic_params.is_empty()
210                        && generic_params.len() == params.len()
211                    {
212                        return self.get_type(&substitute_type(ty, generic_params, &params));
213                    }
214                    return self.get_type(ty);
215                }
216                return Ok(Type::Symbol { id, params });
217            }
218            Type::Symbol { id, params } => {
219                return match self.get_symbol(*id)? {
220                    (_, Symbol::Fn { ty, args: _, generic_params: _, cap: _, body: _, is_pub: _ }) => self.get_type(ty),
221                    (_, Symbol::Native(ty)) => self.get_type(ty),
222                    (_, Symbol::Struct(ty, _)) => {
223                        let params = params.iter().map(|param| self.get_type(param)).collect::<Result<Vec<_>>>()?;
224                        if let Type::Struct { params: generic_params, .. } = ty
225                            && !generic_params.is_empty()
226                            && generic_params.len() == params.len()
227                        {
228                            self.get_type(&substitute_type(ty, generic_params, &params))
229                        } else {
230                            self.get_type(ty)
231                        }
232                    }
233                    (_, s) => {
234                        println!("s-> {:?}", s);
235                        Ok(Type::Symbol { id: *id, params: params.clone() })
236                    }
237                };
238            }
239            Type::Vec(elem, len) => {
240                return Ok(Type::Vec(Rc::new(self.get_type(elem)?), *len));
241            }
242            Type::List(elem) => {
243                return Ok(Type::List(Rc::new(self.get_type(elem)?)));
244            }
245            Type::Array(elem, len) => {
246                return Ok(Type::Array(Rc::new(self.get_type(elem)?), *len));
247            }
248            Type::ArrayParam(elem, len) => {
249                let elem = self.get_type(elem)?;
250                let len = self.get_type(len)?;
251                if let Some(len) = eval_const_int_type(&len) {
252                    let len = u32::try_from(len).map_err(|_| anyhow!("数组长度超出 u32 范围"))?;
253                    return Ok(Type::Array(Rc::new(elem), len));
254                }
255                return Ok(Type::ArrayParam(Rc::new(elem), Rc::new(len)));
256            }
257            Type::ConstBinary { op, left, right } => {
258                let left = self.get_type(left)?;
259                let right = self.get_type(right)?;
260                let ty = Type::ConstBinary { op: *op, left: Rc::new(left), right: Rc::new(right) };
261                return Ok(eval_const_int_type(&ty).map(Type::ConstInt).unwrap_or(ty));
262            }
263            Type::Fn { tys, ret } => {
264                return Ok(Type::Fn { tys: tys.iter().map(|ty| self.get_type(ty)).collect::<Result<Vec<_>>>()?, ret: Rc::new(self.get_type(ret)?) });
265            }
266            Type::Struct { params, fields } => {
267                return Ok(Type::Struct {
268                    params: params.iter().map(|param| self.get_type(param)).collect::<Result<Vec<_>>>()?,
269                    fields: fields.iter().map(|(name, ty)| if matches!(ty, Type::Symbol { .. }) { Ok((name.clone(), ty.clone())) } else { self.get_type(ty).map(|ty| (name.clone(), ty)) }).collect::<Result<Vec<_>>>()?,
270                });
271            }
272            _ => {}
273        }
274        Ok(ty.clone())
275    }
276
277    pub fn add_module(&mut self, name: SmolStr) {
278        let len = self.roots.len();
279        if let Some(pos) = self.roots.iter().position(|r| r.as_str() == name.as_str()) {
280            if pos != len - 1 {
281                self.roots.swap(pos, len - 1);
282            }
283        } else {
284            self.roots.push(name.clone());
285        }
286        self.modules.insert(name, BTreeMap::new());
287    }
288
289    pub fn push_module_scope(&mut self, name: SmolStr) {
290        self.roots.push(name);
291    }
292
293    pub fn pop_module_scope(&mut self) {
294        self.roots.pop();
295    }
296
297    pub fn pop_module(&mut self) {
298        //如果不想模块成为全局的 add_module 之后调用 pop_module
299        if let Some(last) = self.roots.pop() {
300            if let Some(names) = self.modules.get(&last).map(|m| {
301                let kvs: Vec<(SmolStr, u32)> = m.iter().map(|kv| (kv.0.clone(), *kv.1)).collect();
302                kvs.iter().filter_map(|kv| if !self.get_symbol(kv.1).map(|s| s.1.is_pub()).unwrap_or(false) { Some(kv.0.clone()) } else { None }).collect::<Vec<_>>()
303            }) {
304                if let Some(m) = self.modules.get_mut(&last) {
305                    for name in names {
306                        m.remove(&name); //删除非 pub 的符号
307                    }
308                }
309            }
310        }
311    }
312
313    pub fn get_id(&self, name: &str) -> Result<u32> {
314        if let Some(idx) = self.symbols.get_index_of(name) {
315            return Ok(idx as u32);
316        }
317        if let Some(id) = self.roots.iter().rev().find_map(|r| self.modules.get(r).and_then(|m| m.get(name))) {
318            return Ok(*id);
319        }
320        for root in self.roots.iter().rev() {
321            if let Some(idx) = self.symbols.get_index_of(format!("{root}::{name}").as_str()) {
322                return Ok(idx as u32);
323            }
324        }
325        if let Some(id) = self.modules.values().find_map(|m| m.get(name)) {
326            return Ok(*id);
327        }
328        if let Some((mod_name, symbol_name)) = name.split_once("::") {
329            if let Some(m) = self.modules.get(mod_name) {
330                return m.get(symbol_name).copied().ok_or(anyhow!("{} 未发现", name));
331            }
332        }
333        self.roots.iter().find_map(|r| self.modules.get(r).and_then(|m| m.get(name))).copied().ok_or(anyhow!("{} 未发现", name))
334    }
335
336    pub fn add(&mut self, name: SmolStr, s: Symbol) -> u32 {
337        let root = self.roots.last().cloned().unwrap();
338        let id = self.symbols.insert_full(format!("{}::{}", root, name).into(), s).0 as u32;
339        self.modules.get_mut(&root).map(|m| m.insert(name, id));
340        id
341    }
342
343    pub fn add_global(&mut self, name: SmolStr, s: Symbol) -> u32 {
344        if let Some(idx) = self.symbols.get_index_of(name.as_str()) {
345            return idx as u32;
346        }
347        if let Some((mod_name, symbol_name)) = name.as_str().split_once("::") {
348            if let Some(m) = self.modules.get_mut(mod_name) {
349                if let Some(&id) = m.get(symbol_name) {
350                    return id;
351                }
352            }
353        }
354        let id = self.symbols.insert_full(name.clone(), s).0 as u32;
355        if let Some((mod_name, symbol_name)) = name.as_str().split_once("::") {
356            if let Some(m) = self.modules.get_mut(mod_name) {
357                m.insert(symbol_name.into(), id);
358            }
359        }
360        id
361    }
362
363    pub fn take(&mut self, id: u32) -> Option<Symbol> {
364        self.symbols.get_index_mut(id as usize).map(|(_, s)| std::mem::take(s))
365    }
366}
compiler/symbol.rs

compiler/
symbol.rs
