Skip to main content

compiler/
symbol.rs

1use dynamic::{ConstIntOp, Dynamic, Type};
2use parser::Stmt;
3use smol_str::SmolStr;
4use std::{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    // 双 IndexMap:按模块名 + 模块内符号名 O(1) 查找;按插入顺序遍历。
110    modules: IndexMap<SmolStr, IndexMap<SmolStr, u32>>,
111    pub roots: Vec<SmolStr>,
112}
113
114impl SymbolTable {
115    pub fn add_to_module(&mut self, module: &str, name: SmolStr, s: Symbol) -> Result<u32> {
116        let full_name: SmolStr = format!("{}::{}", module, name).into();
117        let id = self.symbols.insert_full(full_name, s).0 as u32;
118        let module_symbols = self.modules.get_mut(module).ok_or_else(|| anyhow!("模块 {} 不存在", module))?;
119        module_symbols.insert(name, id);
120        Ok(id)
121    }
122    pub fn get_symbol(&self, idx: u32) -> Result<(&SmolStr, &Symbol)> {
123        self.symbols.get_index(idx as usize).ok_or(anyhow!("未发现符号 {}", idx))
124    }
125
126    pub fn get_symbol_mut(&mut self, idx: u32) -> Option<(&SmolStr, &mut Symbol)> {
127        self.symbols.get_index_mut(idx as usize)
128    }
129
130    pub fn symbol(&self, name: &str) -> Vec<(SmolStr, u32)> {
131        self.modules.get(name).map(|m| m.iter().map(|(name, id)| (name.clone(), *id)).collect()).unwrap_or(Vec::new())
132    }
133
134    pub fn disassemble(&self, name: &str) -> Result<String> {
135        let id = self.get_id(name)?;
136        let (name, s) = self.get_symbol(id)?;
137        if let Symbol::Fn { ty, args, generic_params: _, cap, body, is_pub } = s {
138            if *is_pub { Ok(format!("pub {} {:?} {:?} {:?}\n{}", name, ty, args, cap, body)) } else { Ok(format!("{} {:?} {:?} {:?}\n{}", name, ty, args, cap, body)) }
139        } else {
140            Err(anyhow!("未发现符号 {}", name))
141        }
142    }
143
144    pub fn get_field(&self, ty: &Type, name: &str) -> Result<(usize, Type)> {
145        //原生类型的函数 is_map is_list 或者 sqrt
146        let id = match ty {
147            Type::Any => {
148                if let Ok(id) = self.get_id("Any")
149                    && let Ok((_, Symbol::Struct(any_ty, _))) = self.get_symbol(id)
150                    && let Ok((idx, field_ty)) = any_ty.get_field(name)
151                {
152                    return Ok((idx, field_ty.clone()));
153                }
154                match name {
155                    "is_map" | "is_list" | "is_string" | "is_null" | "contains" | "starts_with" => return Ok((usize::MAX, Type::Bool)),
156                    "len" => return Ok((usize::MAX, Type::I32)),
157                    _ => return Ok((usize::MAX, Type::Any)),
158                }
159            }
160            Type::Struct { params: _, fields: _ } => {
161                return ty.get_field(name).map(|(idx, ty)| (idx, ty.clone()));
162            }
163            Type::Str => {
164                let any_method = match name {
165                    "len" | "contains" | "split" | "starts_with" | "is_string" | "is_null" => format!("Any::{}", name),
166                    _ => return Err(anyhow!("未发现 symbol {:?} {}", ty, name)),
167                };
168                return Ok((usize::MAX, Type::Symbol { id: self.get_id(&any_method)?, params: Vec::new() }));
169            }
170            Type::List(_) | Type::Array(_, _) => {
171                let any_method = match name {
172                    "len" | "push" | "pop" | "get_idx" | "set_idx" | "slice" | "is_list" | "is_null" => format!("Any::{}", name),
173                    _ => return Err(anyhow!("未发现 symbol {:?} {}", ty, name)),
174                };
175                return Ok((usize::MAX, Type::Symbol { id: self.get_id(&any_method)?, params: Vec::new() }));
176            }
177            Type::Symbol { id, params: _ } => *id,
178            Type::Vec(_, _) => self.get_id("Vec")?,
179            Type::Fn { tys: _, ret } => {
180                return self.get_field(ret, name);
181            }
182            _ => {
183                //增加一个外部函数定义
184                if matches!(name, "is_map" | "is_list" | "is_string" | "is_null") {
185                    return Ok((usize::MAX, Type::Symbol { id: self.get_id(&format!("Any::{}", name))?, params: Vec::new() }));
186                }
187                return Err(anyhow!("未发现 symbol {:?} {}", ty, name));
188            }
189        };
190        let (_, s) = self.get_symbol(id)?;
191        if let Symbol::Struct(s, _) = s {
192            return s.get_field(name).and_then(|(idx, ty)| Ok((idx, ty.clone())));
193        };
194        Err(anyhow!("未发现 field {:?} {}", ty, name))
195    }
196
197    pub fn get_type(&self, ty: &Type) -> Result<Type> {
198        match ty {
199            Type::Ident { name, params } => {
200                let params = params.iter().map(|param| self.get_type(param)).collect::<Result<Vec<_>>>()?;
201                if name.as_str() == "Vec" && params.len() == 1 {
202                    return Ok(Type::Vec(Rc::new(params[0].clone()), 0));
203                }
204                if name.as_str() == "List" {
205                    return Ok(if params.is_empty() { Type::list_any() } else { Type::List(Rc::new(params[0].clone())) });
206                }
207                let id = self.get_id(&name)?;
208                if let (_, Symbol::Struct(ty, _)) = self.get_symbol(id)? {
209                    if let Type::Struct { params: generic_params, .. } = ty
210                        && !generic_params.is_empty()
211                        && generic_params.len() == params.len()
212                    {
213                        return self.get_type(&substitute_type(ty, generic_params, &params));
214                    }
215                    return self.get_type(ty);
216                }
217                return Ok(Type::Symbol { id, params });
218            }
219            Type::Symbol { id, params } => {
220                return match self.get_symbol(*id)? {
221                    (_, Symbol::Fn { ty, args: _, generic_params: _, cap: _, body: _, is_pub: _ }) => self.get_type(ty),
222                    (_, Symbol::Native(ty)) => self.get_type(ty),
223                    (_, Symbol::Struct(ty, _)) => {
224                        let params = params.iter().map(|param| self.get_type(param)).collect::<Result<Vec<_>>>()?;
225                        if let Type::Struct { params: generic_params, .. } = ty
226                            && !generic_params.is_empty()
227                            && generic_params.len() == params.len()
228                        {
229                            self.get_type(&substitute_type(ty, generic_params, &params))
230                        } else {
231                            self.get_type(ty)
232                        }
233                    }
234                    (_, s) => {
235                        log::debug!("s-> {:?}", s);
236                        Ok(Type::Symbol { id: *id, params: params.clone() })
237                    }
238                };
239            }
240            Type::Vec(elem, len) => {
241                return Ok(Type::Vec(Rc::new(self.get_type(elem)?), *len));
242            }
243            Type::List(elem) => {
244                return Ok(Type::List(Rc::new(self.get_type(elem)?)));
245            }
246            Type::Array(elem, len) => {
247                return Ok(Type::Array(Rc::new(self.get_type(elem)?), *len));
248            }
249            Type::ArrayParam(elem, len) => {
250                let elem = self.get_type(elem)?;
251                let len = self.get_type(len)?;
252                if let Some(len) = eval_const_int_type(&len) {
253                    let len = u32::try_from(len).map_err(|_| anyhow!("数组长度超出 u32 范围"))?;
254                    return Ok(Type::Array(Rc::new(elem), len));
255                }
256                return Ok(Type::ArrayParam(Rc::new(elem), Rc::new(len)));
257            }
258            Type::ConstBinary { op, left, right } => {
259                let left = self.get_type(left)?;
260                let right = self.get_type(right)?;
261                let ty = Type::ConstBinary { op: *op, left: Rc::new(left), right: Rc::new(right) };
262                return Ok(eval_const_int_type(&ty).map(Type::ConstInt).unwrap_or(ty));
263            }
264            Type::Fn { tys, ret } => {
265                return Ok(Type::Fn { tys: tys.iter().map(|ty| self.get_type(ty)).collect::<Result<Vec<_>>>()?, ret: Rc::new(self.get_type(ret)?) });
266            }
267            Type::Struct { params, fields } => {
268                return Ok(Type::Struct {
269                    params: params.iter().map(|param| self.get_type(param)).collect::<Result<Vec<_>>>()?,
270                    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<_>>>()?,
271                });
272            }
273            _ => {}
274        }
275        Ok(ty.clone())
276    }
277
278    pub fn add_module(&mut self, name: SmolStr) {
279        let len = self.roots.len();
280        if let Some(pos) = self.roots.iter().position(|r| r.as_str() == name.as_str()) {
281            if pos != len - 1 {
282                self.roots.swap(pos, len - 1);
283            }
284        } else {
285            self.roots.push(name.clone());
286        }
287        self.modules.insert(name, IndexMap::new());
288    }
289
290    pub fn push_module_scope(&mut self, name: SmolStr) {
291        self.roots.push(name);
292    }
293
294    pub fn pop_module_scope(&mut self) {
295        self.roots.pop();
296    }
297
298    pub fn pop_module(&mut self) {
299        //如果不想模块成为全局的 add_module 之后调用 pop_module
300        if let Some(last) = self.roots.pop() {
301            if let Some(names) = self.modules.get(&last).map(|m| {
302                let kvs: Vec<(SmolStr, u32)> = m.iter().map(|kv| (kv.0.clone(), *kv.1)).collect();
303                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<_>>()
304            }) {
305                if let Some(m) = self.modules.get_mut(&last) {
306                    for name in names {
307                        m.shift_remove(&name); //删除非 pub 的符号;保持插入顺序
308                    }
309                }
310            }
311        }
312    }
313
314    pub fn get_id(&self, name: &str) -> Result<u32> {
315        // 1) 全名命中 (含 :: 的 id 或 add_global 注册过的全局符号)。
316        if let Some(idx) = self.symbols.get_index_of(name) {
317            return Ok(idx as u32);
318        }
319        // 2) 含 :: 的名字拆分按模块查找,O(1)。
320        if let Some((mod_name, sym_name)) = name.split_once("::") {
321            if let Some(&id) = self.modules.get(mod_name).and_then(|m| m.get(sym_name)) {
322                return Ok(id);
323            }
324            // 即使 modules 里被 pop_module 移除,完整名仍可能在 symbols IndexMap 里。
325            let full = format!("{mod_name}::{sym_name}");
326            if let Some(idx) = self.symbols.get_index_of(full.as_str()) {
327                return Ok(idx as u32);
328            }
329        }
330        // 3) 不含 :: 的短名,按 roots 倒序查找第一个匹配的模块。O(M)。
331        //    pop_module 会清掉非 pub 项,所以 modules 可能没有;再退回按 root 拼全名查 symbols。
332        let short_name = name;
333        for root in self.roots.iter().rev() {
334            if let Some(&id) = self.modules.get(root).and_then(|m| m.get(short_name)) {
335                return Ok(id);
336            }
337            let full = format!("{root}::{short_name}");
338            if let Some(idx) = self.symbols.get_index_of(full.as_str()) {
339                return Ok(idx as u32);
340            }
341        }
342        // 4) 兜底:任意模块下的同名短符号。O(M*K),K 为模块内符号数。
343        for (_, m) in &self.modules {
344            if let Some(&id) = m.get(short_name) {
345                return Ok(id);
346            }
347        }
348        Err(anyhow!("{} 未发现", name))
349    }
350
351    pub fn add(&mut self, name: SmolStr, s: Symbol) -> u32 {
352        let root = self.roots.last().cloned().unwrap();
353        let id = self.symbols.insert_full(format!("{}::{}", root, name).into(), s).0 as u32;
354        self.modules.get_mut(&root).map(|m| m.insert(name, id));
355        id
356    }
357
358    pub fn add_global(&mut self, name: SmolStr, s: Symbol) -> u32 {
359        if let Some(idx) = self.symbols.get_index_of(name.as_str()) {
360            return idx as u32;
361        }
362        if let Some((mod_name, symbol_name)) = name.as_str().split_once("::") {
363            if let Some(m) = self.modules.get_mut(mod_name) {
364                if let Some(&id) = m.get(symbol_name) {
365                    return id;
366                }
367            }
368        }
369        let id = self.symbols.insert_full(name.clone(), s).0 as u32;
370        if let Some((mod_name, symbol_name)) = name.as_str().split_once("::") {
371            if let Some(m) = self.modules.get_mut(mod_name) {
372                m.insert(symbol_name.into(), id);
373            }
374        }
375        id
376    }
377
378    pub fn take(&mut self, id: u32) -> Option<Symbol> {
379        self.symbols.get_index_mut(id as usize).map(|(_, s)| std::mem::take(s))
380    }
381}