Files
TerminalBasic/crates/tb-vm/src/project.rs

1412 lines
52 KiB
Rust

//! Gemeinsame Tabellenauflösung getrennter Modulübersetzungen.
use crate::{
bytecode::{CompiledModule, Instr},
codegen,
library::{Library, LibraryModule, ModuleMetadata},
};
use std::collections::{HashMap, HashSet};
use tb_frontend::{
ast::{Module, Stmt, TypeName},
forms::FormCatalog,
hir::{HProcKind, HTy},
source::{locate_diagnostics, SourceUnit},
Diagnostic, SourcePos,
};
use tb_runtime::value::TypeInit;
use tb_ui::frm::FormFile;
fn diagnostic(message: impl Into<String>) -> Diagnostic {
Diagnostic {
file: None,
pos: SourcePos::default(),
message: message.into(),
}
}
/// Prozesslokaler Cache: unverknüpfte Produkte, niemals ein bereits gelinktes Programm.
#[derive(Default)]
pub struct ProjectCompiler {
parsed: Vec<Parsed>,
products: Vec<Product>,
pub stats: CompileStats,
/// Vollständige Imports im Cache halten, damit spätere Debugkommandos sie binden können.
pub debug_symbols: bool,
debug_maps: Vec<DebugMap>,
debug_names: Vec<String>,
}
#[derive(Debug, Default, Clone, Copy)]
pub struct CompileStats {
pub parsed: usize,
pub compiled: usize,
pub reused: usize,
}
struct Parsed {
unit: SourceUnit,
prefix: Vec<tb_frontend::source::SourceFile>,
sources: Vec<tb_frontend::source::SourceFile>,
module: Module,
diagnostics: Vec<Diagnostic>,
names: HashSet<String>,
}
struct Product {
module: Module,
debug_module: Module,
catalog: FormCatalog,
code: CompiledModule,
commons: Vec<tb_frontend::hir::HCommon>,
}
pub fn compile_project(
name: &str,
units: &[SourceUnit],
catalog: &FormCatalog,
forms: &[FormFile],
) -> Result<CompiledModule, Vec<Diagnostic>> {
ProjectCompiler::default().compile(name, units, catalog, forms)
}
impl ProjectCompiler {
/// Bindungen aus den tatsächlich übersetzten Modulen einschließlich ihrer Imports.
pub fn bound_form_references(
&self,
) -> Result<tb_frontend::sema::BoundFormReferences, Vec<Diagnostic>> {
let mut result = tb_frontend::sema::BoundFormReferences::default();
for product in &self.products {
let bound = tb_frontend::sema::bound_form_references(&product.module, &product.catalog);
result.objects.extend(bound.objects);
// Importierte Prototypen im Cache haben keine physische Quellposition.
result
.procedures
.extend(bound.procedures.into_iter().filter(|(pos, _)| pos.line > 0));
result.diagnostics.extend(bound.diagnostics);
}
if result.diagnostics.is_empty() {
Ok(result)
} else {
Err(result.diagnostics)
}
}
pub fn compile(
&mut self,
name: &str,
units: &[SourceUnit],
catalog: &FormCatalog,
forms: &[FormFile],
) -> Result<CompiledModule, Vec<Diagnostic>> {
let library = self.compile_library(units, catalog, forms, &[])?;
self.link_library(name, library, forms.first().map(|f| f.root.name.as_str()))
}
pub fn link_library(
&mut self,
name: &str,
library: Library,
startup: Option<&str>,
) -> Result<CompiledModule, Vec<Diagnostic>> {
self.link_product(name, library, startup, false)
}
fn link_product(
&mut self,
name: &str,
library: Library,
startup: Option<&str>,
allow_open: bool,
) -> Result<CompiledModule, Vec<Diagnostic>> {
library.validate().map_err(|e| vec![diagnostic(e)])?;
let source_count = library
.modules
.iter()
.try_fold(0usize, |sum, m| sum.checked_add(m.code.sources.len()));
if source_count.is_none_or(|n| n > u32::MAX as usize) {
return Err(vec![diagnostic("Zu viele Quelldateien")]);
}
let mut sources = Vec::new();
let metadata: Vec<_> =
library
.modules
.iter()
.enumerate()
.map(|(id, m)| {
let offset = sources.len() as u32;
sources.extend(m.code.sources.iter().map(|s| {
tb_frontend::source::SourceFile {
module: id as u16,
path: s.path.clone(),
}
}));
let mut meta = m.metadata.clone();
meta.pos.source += offset;
for (_, p) in &mut meta.declarations {
p.source += offset;
}
for c in &mut meta.commons {
c.pos.source += offset;
}
meta
})
.collect();
let parts = library.modules.into_iter().map(|m| m.code).collect();
let (mut result, maps) = link(name, parts, &metadata, allow_open).map_err(|e| {
let mut errors = vec![e];
locate_diagnostics(&mut errors, &sources);
errors
})?;
self.debug_maps = maps;
self.debug_names = metadata.iter().map(|m| m.name.clone()).collect();
result.startup_form = startup.and_then(|name| {
FormCatalog {
objects: result.objects.clone(),
}
.find(name)
.map(|(id, _)| id)
});
result
.validate()
.map_err(|e| vec![diagnostic(e.to_string())])?;
Ok(result)
}
pub fn compile_library(
&mut self,
units: &[SourceUnit],
catalog: &FormCatalog,
forms: &[FormFile],
libraries: &[Library],
) -> Result<Library, Vec<Diagnostic>> {
self.stats = CompileStats::default();
if (units.is_empty() && libraries.is_empty())
|| units.len() + libraries.iter().map(|l| l.modules.len()).sum::<usize>()
> u16::MAX as usize
{
return Err(vec![diagnostic("Projekt ohne Module oder zu viele Module")]);
}
let mut sources = Vec::new();
let mut diagnostics = Vec::new();
let mut names = HashSet::new();
let mut parsed: Vec<_> = units
.iter()
.enumerate()
.map(|(id, unit)| {
let cached = self
.parsed
.get(id)
.filter(|p| p.unit == *unit && p.prefix == sources);
let (module, errors) = if let Some(p) = cached {
sources = p.sources.clone();
(p.module.clone(), p.diagnostics.clone())
} else {
let prefix = sources.clone();
let (module, errors) = unit.parse(id as u16, &mut sources);
let entry = Parsed {
unit: unit.clone(),
prefix,
sources: sources.clone(),
module: module.clone(),
diagnostics: errors.clone(),
names: unit
.segments
.iter()
.flat_map(|s| tb_frontend::lexer::lex(&s.text).tokens)
.filter_map(|t| match t.kind {
tb_frontend::lexer::TokenKind::Ident { name, .. } => Some(name),
_ => None,
})
.collect(),
};
if id < self.parsed.len() {
self.parsed[id] = entry;
} else {
self.parsed.push(entry);
}
self.stats.parsed += 1;
(module, errors)
};
if !names.insert(unit.name.to_uppercase()) {
let mut error =
diagnostic(format!("Duplicate definition: module {}", unit.name));
error.pos = module_pos(&module);
error.file = unit.segments.first().map(|s| s.file.clone());
diagnostics.push(error);
}
diagnostics.extend(errors);
module
})
.collect();
locate_diagnostics(&mut diagnostics, &sources);
if !diagnostics.is_empty() {
return Err(diagnostics);
}
for library in libraries {
library.validate().map_err(|e| vec![diagnostic(e)])?;
for m in &library.modules {
if !names.insert(m.metadata.name.to_uppercase()) {
return Err(vec![diagnostic(format!(
"Duplicate definition: module {}",
m.metadata.name
))]);
}
parsed.push(m.metadata.declaration_view());
}
}
let mut catalog = catalog.clone();
for library in libraries {
for m in &library.modules {
merge_objects(&mut catalog.objects, &m.code.objects).map_err(|e| vec![e])?;
}
}
if catalog.find("SCREEN").is_none() {
catalog.add(
"SCREEN",
tb_frontend::forms::ObjectClass::Screen,
None,
false,
);
}
for module in &parsed {
if module
.body
.iter()
.any(|s| matches!(s, Stmt::MetaForm { .. }))
&& catalog.find(&module.name).is_none()
{
catalog.add(
&module.name,
tb_frontend::forms::ObjectClass::Form,
None,
false,
);
}
}
let mut exports: Vec<_> = parsed
.iter()
.map(|m| tb_frontend::sema::export_declarations(m, &[]))
.collect();
// Konstantenabhängigkeiten können beliebig über Module verteilt sein.
// Jeder erfolgreiche Durchlauf löst mindestens eine weitere Deklaration auf.
let constant_count: usize = exports
.iter()
.flatten()
.filter(|s| matches!(s, Stmt::ConstDecl { .. }))
.count();
for _ in 0..constant_count {
let mut constants: HashMap<_, Vec<_>> = HashMap::new();
for stmt in exports.iter().flatten() {
if let Stmt::ConstDecl { items, .. } = stmt {
constants.entry(&items[0].0).or_default().push(stmt.clone());
}
}
let constants: Vec<_> = constants
.into_values()
.filter(|s| s.len() == 1)
.flatten()
.collect();
let next: Vec<_> = parsed
.iter()
.map(|m| tb_frontend::sema::export_declarations(m, &constants))
.collect();
if next == exports {
break;
}
exports = next;
}
let mut parts = Vec::new();
let mut commons = Vec::new();
let mut products = Vec::new();
for (index, module) in parsed.iter().take(units.len()).enumerate() {
let mut key = module.clone();
import_declarations(&mut key, &parsed, &exports, Some(&self.parsed[index].names));
let mut module = module.clone();
import_declarations(&mut module, &parsed, &exports, None);
if self.debug_symbols {
key = module.clone();
}
if let Some(product) = self
.products
.iter()
.find(|p| p.module == key && p.catalog == catalog)
{
parts.push(product.code.clone());
commons.push(product.commons.clone());
self.stats.reused += 1;
continue;
}
self.stats.compiled += 1;
let (hir, mut errors) = tb_frontend::sema::lower_with_forms(&module, &catalog);
for error in &mut errors {
if error.message != "Subprogram not defined" {
continue;
}
// Sema kennt absichtlich nur eindeutige Imports. Ergänze den Konflikt am tatsächlichen Aufruftoken.
let file = sources
.get(error.pos.source as usize)
.map(|s| s.path.as_str());
let called = units[index]
.segments
.iter()
.filter(|s| Some(s.file.as_str()) == file)
.find_map(|s| {
let line = error.pos.line.checked_sub(s.first_line)? as usize;
let text = s.text.lines().nth(line)?;
tb_frontend::lexer::lex(text)
.tokens
.into_iter()
.filter(|t| t.pos.column >= error.pos.column)
.find_map(|t| match t.kind {
tb_frontend::lexer::TokenKind::Ident { name, .. } => Some(name),
_ => None,
})
});
if let Some(called) = called {
let origins: Vec<_> = exports
.iter()
.enumerate()
.filter(|(_, declarations)| {
declarations
.iter()
.any(|s| matches!(s,Stmt::Declare{sig,..} if sig.name==called))
})
.map(|(id, _)| parsed[id].name.as_str())
.collect();
if origins.len() > 1 {
error.message =
format!("Ambiguous subprogram: {called} ({})", origins.join(", "));
}
}
}
diagnostics.extend(errors);
if let Some(hir) = hir {
let code = codegen::compile(&hir);
parts.push(code.clone());
commons.push(hir.commons.clone());
products.push(Product {
module: key,
debug_module: module,
catalog: catalog.clone(),
code,
commons: hir.commons,
});
}
}
locate_diagnostics(&mut diagnostics, &sources);
if !diagnostics.is_empty() {
return Err(diagnostics);
}
self.parsed.truncate(units.len());
self.products.retain(|p| {
parsed.iter().any(|m| m.name == p.module.name)
&& !products.iter().any(|n| n.module.name == p.module.name)
});
self.products.extend(products);
let mut library = Library::default();
for (index, mut code) in parts.into_iter().enumerate() {
let mut metadata = ModuleMetadata::from_source(
&parsed[index],
exports[index].clone(),
commons[index].clone(),
);
// Die Quell-IDs jedes Produkts werden lokal: ein TBL bleibt unabhängig vom Verbraucher.
let local: Vec<_> = sources
.iter()
.enumerate()
.filter(|(_, s)| s.module as usize == index)
.collect();
let source_id = |id: u32| {
local
.iter()
.position(|(old, _)| *old == id as usize)
.unwrap_or(0) as u32
};
code.sources = local
.iter()
.map(|(_, s)| tb_frontend::source::SourceFile {
module: 0,
path: s.path.clone(),
})
.collect();
if code.sources.is_empty() {
code.sources.push(tb_frontend::source::SourceFile {
module: 0,
path: code.name.clone(),
});
}
for proc in &mut code.procs {
for instruction in &mut proc.code {
if let Instr::Source(id, _) = instruction {
*id = source_id(*id);
}
}
}
metadata.pos.source = source_id(metadata.pos.source);
for (_, p) in &mut metadata.declarations {
p.source = source_id(p.source);
}
for c in &mut metadata.commons {
c.pos.source = source_id(c.pos.source);
}
for stmt in &mut metadata.exports {
match stmt {
Stmt::Declare { pos, .. }
| Stmt::TypeDecl { pos, .. }
| Stmt::ConstDecl { pos, .. } => pos.source = source_id(pos.source),
_ => {}
}
}
let objects = FormCatalog {
objects: code.objects.clone(),
};
for form in forms
.iter()
.filter(|f| f.root.name.eq_ignore_ascii_case(&code.name))
{
code.form_initial.extend(
form.initial_values(&objects)
.map_err(|e| vec![diagnostic(e.to_string())])?,
);
}
library.modules.push(LibraryModule { metadata, code });
}
for dependency in libraries {
library.modules.extend(dependency.modules.clone());
}
library.validate().map_err(|e| vec![diagnostic(e)])?;
// Auch offene Produkte müssen vorhandene Definitionen und COMMON-Verträge konsistent binden.
self.link_product("LIBRARY", library.clone(), None, true)?;
Ok(library)
}
}
/// Lokale Definition gewinnt; außerhalb ihres Moduls muss ein Name eindeutig sein.
fn type_definition<'a>(name: &str, origin: usize, all: &'a [Module]) -> Option<(usize, &'a Stmt)> {
let mut candidates = all.iter().enumerate().flat_map(|(id, module)| {
module.body.iter().filter_map(move |stmt| {
matches!(stmt, Stmt::TypeDecl { name: n, .. } if n == name).then_some((id, stmt))
})
});
if let Some(local) = candidates.clone().find(|(id, _)| *id == origin) {
return Some(local);
}
let first = candidates.next()?;
candidates.next().is_none().then_some(first)
}
fn same_type(
name: &str,
a: usize,
b: usize,
all: &[Module],
visiting: &mut HashSet<(usize, usize, String)>,
) -> bool {
let (
Some((a, Stmt::TypeDecl { fields: af, .. })),
Some((b, Stmt::TypeDecl { fields: bf, .. })),
) = (type_definition(name, a, all), type_definition(name, b, all))
else {
return false;
};
if a == b {
return true;
}
if !visiting.insert((a, b, name.into())) {
return false;
}
let equal = af.len() == bf.len()
&& af.iter().zip(bf).all(|((an, at), (bn, bt))| {
an == bn
&& match (at, bt) {
(TypeName::Udt(an), TypeName::Udt(bn)) => {
an == bn && same_type(an, a, b, all, visiting)
}
_ => at == bt,
}
});
visiting.remove(&(a, b, name.into()));
equal
}
fn import_type(
name: &str,
origin: usize,
target: usize,
all: &[Module],
imported: &mut Vec<Stmt>,
seen: &mut HashMap<(usize, String), String>,
) -> String {
let Some((origin, Stmt::TypeDecl { fields, pos, .. })) = type_definition(name, origin, all)
else {
return name.into();
};
if origin == target
|| (type_definition(name, target, all).is_some_and(|(id, _)| id == target)
&& same_type(name, origin, target, all, &mut HashSet::new()))
{
return name.into();
}
let key = (origin, name.to_string());
if let Some(alias) = seen.get(&key) {
return alias.clone();
}
// Öffentliche eindeutige Namen bleiben erhalten. Konfliktbehaftete Abhängigkeiten
// bekommen einen internen Modulnamen, damit das lokale Layout unangetastet bleibt.
let alias = if type_definition(name, target, all).is_some_and(|(id, _)| id == origin) {
name.to_string()
} else {
format!("{}!{name}", all[origin].name)
};
seen.insert(key, alias.clone());
let fields = fields
.iter()
.map(|(field, ty)| {
let ty = match ty {
TypeName::Udt(name) => {
TypeName::Udt(import_type(name, origin, target, all, imported, seen))
}
_ => ty.clone(),
};
(field.clone(), ty)
})
.collect();
imported.push(Stmt::TypeDecl {
name: alias.clone(),
fields,
pos: *pos,
});
alias
}
fn import_declarations(
module: &mut Module,
all: &[Module],
exports: &[Vec<Stmt>],
signature: Option<&HashSet<String>>,
) {
let target = all.iter().position(|m| m.name == module.name).unwrap();
let mut procedures: HashSet<_> = module.procs.iter().map(|p| p.sig.name.clone()).collect();
let mut local_constants = HashSet::new();
let mut local_types = HashSet::new();
for stmt in &module.body {
match stmt {
Stmt::Declare { sig, .. } => {
procedures.insert(sig.name.clone());
}
Stmt::ConstDecl { items, .. } => {
local_constants.extend(items.iter().map(|i| i.0.clone()))
}
Stmt::TypeDecl { name, .. } => {
local_types.insert(name.clone());
}
_ => {}
}
}
let mut foreign: HashMap<_, Vec<_>> = HashMap::new();
for (origin, declarations) in exports.iter().enumerate().filter(|(id, _)| *id != target) {
for stmt in declarations {
let key = match stmt {
Stmt::Declare { sig, .. } => (0, sig.name.clone()),
Stmt::ConstDecl { items, .. } => (1, items[0].0.clone()),
Stmt::TypeDecl { name, .. } => (2, name.clone()),
_ => continue,
};
foreign.entry(key).or_default().push((origin, stmt));
}
}
let mut names: Vec<_> = foreign.keys().cloned().collect();
names.sort();
let mut imported = Vec::new();
let mut seen_types = HashMap::new();
for key in names {
let candidates = &foreign[&key];
if candidates.len() != 1 {
continue;
}
let (origin, stmt) = candidates[0];
match stmt {
Stmt::Declare { sig, pos } if !procedures.contains(&sig.name) => {
let mut sig = sig.clone();
for param in &mut sig.params {
if let Some(TypeName::Udt(name)) = &param.as_type {
param.as_type = Some(TypeName::Udt(import_type(
name,
origin,
target,
all,
&mut imported,
&mut seen_types,
)));
}
}
imported.push(Stmt::Declare { sig, pos: *pos });
}
Stmt::ConstDecl { items, .. } if !local_constants.contains(&items[0].0) => {
imported.push(stmt.clone())
}
Stmt::TypeDecl { name, .. } if !local_types.contains(name) => {
import_type(name, origin, target, all, &mut imported, &mut seen_types);
}
_ => {}
}
}
if let Some(names) = signature {
// Nicht verwendete importierte Konstanten erzeugen weder Slots noch Tabellen.
// TYPEs und Prozeduren dagegen gehören zu den unverketteten Tabellen.
imported
.retain(|s| !matches!(s, Stmt::ConstDecl {items,..} if !names.contains(&items[0].0)));
// Importverträge enthalten Namen/Typen/Werte, keine verschobenen Rumpfzeilen
// ihres Ursprungs. Eigene Quellorte bleiben Teil des Modulschlüssels.
for stmt in &mut imported {
match stmt {
Stmt::Declare { pos, .. } | Stmt::TypeDecl { pos, .. } => {
*pos = SourcePos::default()
}
Stmt::ConstDecl { items, pos } => {
*pos = SourcePos::default();
for (_, _, expr) in items {
match expr {
tb_frontend::ast::Expr::DoubleLit(_, p)
| tb_frontend::ast::Expr::StrLit(_, p) => *p = SourcePos::default(),
_ => {}
}
}
}
_ => {}
}
}
}
imported.append(&mut module.body);
let mut types = Vec::new();
imported.retain(|stmt| {
if matches!(stmt, Stmt::TypeDecl { .. }) {
types.push(stmt.clone());
false
} else {
true
}
});
let mut known = HashSet::new();
let mut ordered = Vec::new();
while !types.is_empty() {
let next = types.iter().position(|stmt| match stmt {
Stmt::TypeDecl { fields, .. } => fields.iter().all(|(_, ty)| match ty {
TypeName::Udt(name) => known.contains(name),
_ => true,
}),
_ => unreachable!(),
});
let Some(index) = next else { break };
let stmt = types.remove(index);
if let Stmt::TypeDecl { name, .. } = &stmt {
known.insert(name.clone());
}
ordered.push(stmt);
}
ordered.extend(types); // Sema diagnostiziert fehlende oder zyklische Typen.
ordered.extend(imported);
module.body = ordered;
}
fn module_pos(module: &Module) -> SourcePos {
module
.body
.iter()
.map(tb_frontend::sema::stmt_pos)
.find(|p| p.line > 0)
.or_else(|| module.procs.first().map(|p| p.pos))
.unwrap_or_default()
}
fn remap_type(ty: &mut TypeInit, ids: &[u16]) {
if let TypeInit::Udt(id) = ty {
*id = ids[*id as usize];
}
}
fn remap_signature(ty: &mut HTy, ids: &[u16]) {
if let HTy::Udt(id) = ty {
*id = ids[*id as usize];
}
}
fn merge_objects(
target: &mut Vec<tb_frontend::forms::FormObject>,
objects: &[tb_frontend::forms::FormObject],
) -> Result<Vec<u16>, Diagnostic> {
let mut ids = Vec::new();
for object in objects {
let mut object = object.clone();
object.parent = object.parent.map(|id| ids[id as usize]);
let id = if let Some(id) = target
.iter()
.position(|o| o.name == object.name && o.parent == object.parent)
{
if target[id] != object {
return Err(diagnostic(format!(
"Incompatible Forms object: {}",
object.name
)));
}
id
} else {
target.push(object);
target.len() - 1
};
ids.push(u16::try_from(id).map_err(|_| diagnostic("Zu viele Forms-Objekte"))?);
}
Ok(ids)
}
fn link(
name: &str,
mut parts: Vec<CompiledModule>,
ast: &[ModuleMetadata],
allow_open: bool,
) -> Result<(CompiledModule, Vec<DebugMap>), Diagnostic> {
let at = |pos, message| Diagnostic {
file: None,
pos,
message,
};
let mut result = codegen::compile(&tb_frontend::analyze_source(name, "").hir.unwrap());
result.modules = parts
.iter()
.map(|p| (p.name.clone(), p.option_base))
.collect();
result.objects.clear();
result.sources.clear();
result.strings.clear();
result.procs.clear();
result.option_base = parts[0].option_base;
let mut proc_maps = Vec::new();
let mut count = 1usize;
let mut definitions: HashMap<String, Vec<u16>> = HashMap::new();
for (part, module) in parts.iter().zip(ast) {
let defined: HashSet<_> = module.defined.iter().map(String::as_str).collect();
let mut map = vec![0];
for p in part.procs.iter().skip(1) {
if count >= u16::MAX as usize {
return Err(at(module.proc_pos(&p.name), "Zu viele Prozeduren".into()));
}
let id = u16::try_from(count)
.map_err(|_| at(module.proc_pos(&p.name), "Zu viele Prozeduren".into()))?;
count += 1;
map.push(id);
if defined.contains(p.name.as_str()) || p.kind == HProcKind::DefFn {
definitions.entry(p.name.clone()).or_default().push(id);
}
}
proc_maps.push(map);
}
let mut unused_imports = HashSet::new();
// DECLARE-Platzhalter auf die tatsächliche, eindeutig bestimmte Definition binden.
for (module_id, part) in parts.iter().enumerate() {
let defined: HashSet<_> = ast[module_id].defined.iter().map(String::as_str).collect();
for (id, proc) in part.procs.iter().enumerate().skip(1) {
if !defined.contains(proc.name.as_str()) && proc.kind != HProcKind::DefFn {
if let Some(candidates) = definitions.get(&proc.name) {
if candidates.len() != 1 {
return Err(at(
ast[module_id].proc_pos(&proc.name),
format!(
"Ambiguous subprogram: {} ({})",
proc.name,
ast.iter()
.filter(|m| m.defined.contains(&proc.name))
.map(|m| m.name.as_str())
.collect::<Vec<_>>()
.join(", ")
),
));
}
proc_maps[module_id][id] = candidates[0];
} else if !allow_open && !crate::interp::is_runtime_external(&proc.name) {
let referenced = part
.procs
.iter()
.flat_map(|p| &p.code)
.any(|i| matches!(i,Instr::Call(target,_) if *target as usize==id))
|| part.event_procs.iter().any(|e| e.proc as usize == id);
if referenced {
return Err(at(
ast[module_id].proc_pos(&proc.name),
format!("Subprogram not defined: {} ({})", proc.name, part.name),
));
}
// Ein bloßer Prototyp ohne Relokation ist kein offener Import des Endprodukts.
unused_imports.insert(proc_maps[module_id][id]);
proc_maps[module_id][id] = u16::MAX;
}
}
}
}
let splits: Vec<_> = parts
.iter()
.map(|p| {
p.procs[0]
.code
.iter()
.position(|i| matches!(i, Instr::Stmt(0)))
.map_or(0, |i| i.saturating_sub(1))
})
.collect();
let mut init_starts = Vec::new();
let mut body_starts = Vec::new();
let mut pc = 0;
for split in &splits {
init_starts.push(pc);
pc += split;
}
for (part, split) in parts.iter().zip(&splits) {
body_starts.push(pc);
pc += part.procs[0].code.len() - split - 1;
}
if pc >= u32::MAX as usize {
return Err(diagnostic("Zu viele Hauptprogramminstruktionen"));
}
let mut main = parts[0].procs[0].clone();
main.code.clear();
main.name = name.into();
let mut initializers = Vec::new();
let mut bodies = Vec::new();
let mut debug_maps = Vec::new();
let mut common: HashMap<_, (u16, tb_frontend::hir::HCommon)> = HashMap::new();
for (module_id, part) in parts.iter_mut().enumerate() {
let objects = merge_objects(&mut result.objects, &part.objects)?;
let source_offset =
u32::try_from(result.sources.len()).map_err(|_| diagnostic("Zu viele Quelldateien"))?;
if result
.sources
.len()
.checked_add(part.sources.len())
.is_none_or(|n| n > u32::MAX as usize)
{
return Err(diagnostic("Zu viele Quelldateien"));
}
result.sources.extend(
part.sources
.iter()
.map(|s| tb_frontend::source::SourceFile {
module: module_id as u16,
path: s.path.clone(),
}),
);
for mut initial in part.form_initial.drain(..) {
initial.object = objects[initial.object as usize];
for value in initial.properties.values_mut() {
if let tb_ui::forms::PropertyValue::Object(Some((id, _))) = value {
*id = objects[*id as usize];
}
}
result.form_initial.push(initial);
}
let mut types = Vec::new();
for udt in &part.udts {
let mut udt = udt.clone();
udt.name = udt.name.rsplit('!').next().unwrap().to_string();
for ty in &mut udt.fields {
remap_type(ty, &types);
}
let id = result
.udts
.iter()
.position(|u| u.name == udt.name && u.fields == udt.fields)
.unwrap_or_else(|| {
result.udts.push(udt);
result.udts.len() - 1
});
if result.udts.len() > u16::MAX as usize {
return Err(at(ast[module_id].pos, "Zu viele TYPEs".into()));
}
types.push(
u16::try_from(id).map_err(|_| at(ast[module_id].pos, "Zu viele TYPEs".into()))?,
);
}
let commons: HashMap<_, _> = ast[module_id].commons.iter().map(|c| (c.slot, c)).collect();
let mut globals = Vec::new();
for (slot, (ty, name)) in part.globals_init.iter().zip(&part.global_names).enumerate() {
let mut ty = ty.clone();
remap_type(&mut ty, &types);
let declaration = commons.get(&(slot as u16));
let key = declaration.map(|c| (c.block.clone(), c.key.clone()));
let mut common_ty = declaration.map(|c| c.ty.clone());
if let Some(ty) = &mut common_ty {
remap_signature(ty, &types);
}
let id = if let Some((id, previous)) = key.as_ref().and_then(|key| common.get_mut(key))
{
let declaration = declaration.unwrap();
let compatible_dims = match (&previous.dims, &declaration.dims) {
(None, None) => true,
(Some(a), Some(b)) if a.is_empty() || b.is_empty() => true,
(Some(a), Some(b)) => {
a.len() == b.len()
&& a.iter().zip(b).all(|((al, ah), (bl, bh))| {
al.zip(*bl).is_none_or(|(a, b)| a == b)
&& ah.zip(*bh).is_none_or(|(a, b)| a == b)
})
}
_ => false,
};
if Some(&previous.ty) != common_ty.as_ref() || !compatible_dims {
return Err(at(
declaration.pos,
format!("COMMON type or bounds mismatch: {name}"),
));
}
if let (Some(previous), Some(current)) = (&mut previous.dims, &declaration.dims) {
if previous.is_empty() {
*previous = current.clone();
} else {
for ((lo, hi), (new_lo, new_hi)) in previous.iter_mut().zip(current) {
*lo = lo.or(*new_lo);
*hi = hi.or(*new_hi);
}
}
}
*id
} else {
let id = u16::try_from(result.globals_init.len())
.map_err(|_| at(ast[module_id].pos, "Zu viele globale Variablen".into()))?;
result.globals_init.push(ty);
result.global_names.push(if ast.len() == 1 {
name.clone()
} else {
format!("{}!{name}", part.name)
});
if let Some(key) = key {
let mut declaration = (*declaration.unwrap()).clone();
declaration.ty = common_ty.unwrap();
common.insert(key, (id, declaration));
}
id
};
globals.push(id);
}
debug_maps.push(DebugMap {
globals: globals.clone(),
types: types.clone(),
procs: proc_maps[module_id].clone(),
});
let string_offset = result.strings.len();
if string_offset + part.strings.len() >= u16::MAX as usize {
return Err(at(ast[module_id].pos, "Zu viele Stringkonstanten".into()));
}
result.strings.append(&mut part.strings);
let data_offset =
u32::try_from(result.data.len()).map_err(|_| diagnostic("Zu viele DATA-Werte"))?;
if result
.data
.len()
.checked_add(part.data.len())
.is_none_or(|n| n > u32::MAX as usize)
{
return Err(diagnostic("Zu viele DATA-Werte"));
}
result.data.append(&mut part.data);
let jump_offset = result.jump_tables.len();
if jump_offset + part.jump_tables.len() > u16::MAX as usize {
return Err(at(ast[module_id].pos, "Zu viele Sprungtabellen".into()));
}
let main_pc = |pc: u32| if (pc as usize) < splits[module_id] { init_starts[module_id] + pc as usize } else { body_starts[module_id] + pc as usize - splits[module_id] } as u32;
for (proc_id, proc) in part.procs.iter_mut().enumerate() {
proc.module = module_id as u16;
for ty in &mut proc.locals_init {
remap_type(ty, &types);
}
for param in &mut proc.params {
remap_signature(&mut param.ty, &types);
}
if let Some(ty) = &mut proc.ret_ty {
remap_signature(ty, &types);
}
if ast.len() > 1 {
proc.name = format!("{}!{}", part.name, proc.name);
}
for instruction in &mut proc.code {
use Instr::*;
match instruction {
Source(id, _) => *id += source_offset,
LoadObjectProperty(id, _, _)
| StoreObjectProperty(id, _, _)
| PushObject(id, _)
| ObjectMethod(id, _, _)
| ObjectLoad(id, _, _)
| LoadObjectIndexedProperty(id, _)
| ObjectMethodFn(id, _, _)
| StoreObjectIndexedProperty(id, _) => *id = objects[*id as usize],
PushStr(id)
| Unsupported(id)
| LoadDynamicObjectProperty(id)
| StoreDynamicObjectProperty(id) => *id += string_offset as u16,
Input(_, _, id, _) if *id != u16::MAX => *id += string_offset as u16,
LoadGlobal(id) | StoreGlobal(id) | MakeRefGlobal(id) => {
*id = globals[*id as usize]
}
LoadArr(global, id, _, ty)
| DimArr(global, id, _, ty)
| CommonArr(global, id, _, ty)
| RedimArr(global, id, _, ty) => {
if *global {
*id = globals[*id as usize];
}
remap_type(ty, &types);
}
EraseSlot(true, id) => *id = globals[*id as usize],
GetPut(_, _, 7, id) | PushUdtId(id) => *id = types[*id as usize],
Call(id, _) => *id = proc_maps[module_id][*id as usize],
Restore(id) => *id += data_offset,
OnErrorGoto(pc) => *pc = main_pc(*pc),
Jump(pc)
| JumpIfFalse(pc)
| JumpIfTrue(pc)
| Gosub(pc)
| RetGosubTo(pc)
| OnErrorLocal(pc)
| ResumeLabel(pc)
| TrapDefine(_, pc)
if proc_id == 0 =>
{
*pc = main_pc(*pc)
}
OnJump(id, _) => {
if proc_id == 0 {
for target in &mut part.jump_tables[*id as usize] {
*target = main_pc(*target);
}
}
*id += jump_offset as u16;
}
_ => {}
}
}
}
initializers.extend_from_slice(&part.procs[0].code[..splits[module_id]]);
bodies.extend_from_slice(
&part.procs[0].code[splits[module_id]..part.procs[0].code.len() - 1],
);
result.procs.extend(part.procs.iter().skip(1).cloned());
for mut e in part.event_procs.drain(..) {
e.proc = proc_maps[module_id][e.proc as usize];
e.object = objects[e.object as usize];
result.event_procs.push(e);
}
result.jump_tables.append(&mut part.jump_tables);
}
initializers.extend(bodies);
initializers.push(Instr::End);
main.code = initializers;
result.procs.insert(0, main);
// Vollständige Signaturen zwischen DECLARE-Platzhalter und Ziel vergleichen.
for (module, part) in parts.iter().enumerate() {
for (id, proc) in part.procs.iter().enumerate().skip(1) {
if proc_maps[module][id] == u16::MAX {
continue;
}
let target = &result.procs[proc_maps[module][id] as usize];
if proc.ret_ty != target.ret_ty
|| proc.kind != target.kind
|| proc.params.len() != target.params.len()
|| proc
.params
.iter()
.zip(&target.params)
.any(|(a, b)| a.ty != b.ty || a.array != b.array || a.by_ref != b.by_ref)
{
return Err(at(
ast[module].proc_pos(proc.name.rsplit('!').next().unwrap()),
format!("Parameter type mismatch: {}", proc.name),
));
}
}
}
if !unused_imports.is_empty() {
let mut compact = vec![u16::MAX; result.procs.len()];
let mut next = 0u16;
for (old, id) in compact.iter_mut().enumerate() {
if !unused_imports.contains(&(old as u16)) {
*id = next;
next += 1;
}
}
result.procs = std::mem::take(&mut result.procs)
.into_iter()
.enumerate()
.filter_map(|(id, mut proc)| {
if compact[id] == u16::MAX {
return None;
}
for instruction in &mut proc.code {
if let Instr::Call(id, _) = instruction {
*id = compact[*id as usize];
}
}
Some(proc)
})
.collect();
for e in &mut result.event_procs {
e.proc = compact[e.proc as usize];
}
for map in &mut debug_maps {
for id in &mut map.procs {
if *id != u16::MAX {
*id = compact[*id as usize];
}
}
}
}
Ok((result, debug_maps))
}
// Erst ein ausdrücklich erzeugter DebugCompiler bindet den Quellcompiler ein.
// Die reine Runtime kann dadurch den gleichen VM-Pfad ohne Parser/Codegenerator linken.
type DebugCompileFn = fn(&DebugCompiler, u16, &str, &str, bool) -> Result<DebugCode, String>;
/// Ephemeral symbol/link context, deliberately not part of TBC serialization.
#[derive(Clone)]
pub struct DebugCompiler {
compile_fn: DebugCompileFn,
modules: Vec<(u16, Module, FormCatalog, DebugMap)>,
symbols: tb_frontend::sema::DebugSymbols,
slots: Vec<u16>,
error: Option<String>,
}
#[derive(Clone)]
struct DebugMap {
globals: Vec<u16>,
types: Vec<u16>,
procs: Vec<u16>,
}
#[derive(Clone)]
pub struct DebugCode {
pub procedure: crate::bytecode::ProcCode,
pub strings: Vec<std::rc::Rc<str>>,
pub expression: bool,
}
impl ProjectCompiler {
pub fn debug_compiler(&self) -> DebugCompiler {
let modules: Vec<_> = self
.debug_names
.iter()
.zip(&self.debug_maps)
.enumerate()
.filter_map(|(id, (name, map))| {
self.products
.iter()
.find(|p| p.module.name == *name)
.map(|p| {
(
id as u16,
p.debug_module.clone(),
p.catalog.clone(),
map.clone(),
)
})
})
.collect();
let mut symbols = tb_frontend::sema::DebugSymbols::default();
let mut slots = Vec::new();
let mut error = None;
for (_, ast, catalog, map) in &modules {
if let Some(hir) = tb_frontend::sema::lower_with_forms(ast, catalog).0 {
if symbols.udts.len() + hir.udts.len() >= u16::MAX as usize
|| symbols.globals.len() + hir.globals.len() >= u16::MAX as usize
{
error = Some("Debug-Symboltabelle überschreitet die 16-Bit-Slotgrenze".into());
break;
}
let offset = symbols.udts.len() as u16;
for mut udt in hir.udts {
udt.name = format!("<Debug:{}!{}>", ast.name, udt.name);
for (_, ty) in &mut udt.fields {
if let HTy::Udt(id) = ty {
*id += offset;
}
}
symbols.udts.push(udt);
}
for (mut var, slot) in hir.globals.into_iter().zip(&map.globals) {
var.name = format!("{}!{}", ast.name, var.name);
if let HTy::Udt(id) = &mut var.ty {
*id += offset;
}
symbols.globals.push(var);
slots.push(*slot);
}
}
}
DebugCompiler {
compile_fn: DebugCompiler::compile_impl,
modules,
symbols,
slots,
error,
}
}
}
impl DebugCompiler {
pub fn compile(
&self,
module: u16,
procedure: &str,
text: &str,
expression: bool,
) -> Result<DebugCode, String> {
(self.compile_fn)(self, module, procedure, text, expression)
}
fn compile_impl(
&self,
module: u16,
procedure: &str,
text: &str,
expression: bool,
) -> Result<DebugCode, String> {
if let Some(error) = &self.error {
return Err(error.clone());
}
let (_, ast, catalog, map) = self
.modules
.iter()
.find(|(id, _, _, _)| *id == module)
.ok_or("Kein Debug-Quellkontext für dieses Kompilat")?;
let name = if procedure == "<main>" {
&ast.name
} else {
procedure.rsplit('!').next().unwrap_or(procedure)
};
if map.globals.len() + self.symbols.globals.len() >= u16::MAX as usize
|| map.types.len() + self.symbols.udts.len() >= u16::MAX as usize
{
return Err("Debug-Kontext überschreitet die 16-Bit-Slotgrenze".into());
}
let mut symbols = self.symbols.clone();
symbols.original_globals = map.globals.len();
let (mut hir, debug) = tb_frontend::sema::lower_debug(
ast, catalog, name, text, expression, symbols,
)
.map_err(|e| {
e.iter()
.map(ToString::to_string)
.collect::<Vec<_>>()
.join("\n")
})?;
hir.procs.push(debug);
let mut globals = map.globals.clone();
globals.extend(&self.slots);
let mut types = map.types.clone();
for (_, _, _, mapping) in &self.modules {
types.extend(&mapping.types);
}
let mut compiled = codegen::compile(&hir);
let mut proc = compiled.procs.pop().unwrap();
proc.module = module;
for ty in &mut proc.locals_init {
remap_type(ty, &types);
}
if let Some(ty) = &mut proc.ret_ty {
remap_signature(ty, &types);
}
for i in &mut proc.code {
use Instr::*;
match i {
LoadGlobal(id) | StoreGlobal(id) | MakeRefGlobal(id) => {
*id = *globals
.get(*id as usize)
.ok_or("Unbekannter globaler Slot")?
}
LoadArr(global, id, _, ty) => {
if *global {
*id = globals[*id as usize];
}
remap_type(ty, &types);
}
Call(id, _) => {
*id = map.procs[*id as usize];
if *id == u16::MAX {
return Err(
"Subprogram not defined: ungebundene DECLARE-Deklaration".into()
);
}
}
PushUdtId(id) => *id = types[*id as usize],
_ => {}
}
}
Ok(DebugCode {
procedure: proc,
strings: compiled.strings,
expression,
})
}
}
impl DebugCompiler {
pub(crate) fn same_control_context(
&self,
module: u16,
procedure: &str,
from: SourcePos,
to: SourcePos,
) -> bool {
let Some((_, ast, _, _)) = self.modules.iter().find(|(id, _, _, _)| *id == module) else {
return false;
};
let body = if procedure == "<main>" {
&ast.body
} else {
let Some(proc) = ast.procs.iter().find(|p| {
p.sig
.name
.eq_ignore_ascii_case(procedure.rsplit('!').next().unwrap_or(procedure))
}) else {
return false;
};
&proc.body
};
let a = control_path(body, from);
let b = control_path(body, to);
a.is_some() && a == b
}
}
/// Structured arms supplement bytecode loop ranges: ELSE and CASE are distinct
/// even when their forward jump would otherwise look like a top-level GOTO.
fn control_path(body: &[Stmt], target: SourcePos) -> Option<Vec<(SourcePos, usize)>> {
for stmt in body {
let pos = tb_frontend::sema::stmt_pos(stmt);
if pos == target {
return Some(vec![]);
}
let mut children: Vec<(&[Stmt], Option<SourcePos>)> = Vec::new();
match stmt {
Stmt::If {
then_body,
elseifs,
else_body,
..
} => {
children.push((then_body, None));
for (_, body, pos) in elseifs {
children.push((body, Some(*pos)));
}
if let Some(body) = else_body {
children.push((body, None));
}
}
Stmt::Select { arms, .. } => {
for arm in arms {
children.push((&arm.body, Some(arm.pos)));
}
}
Stmt::For { body, end_pos, .. }
| Stmt::DoLoop { body, end_pos, .. }
| Stmt::While { body, end_pos, .. } => children.push((body, Some(*end_pos))),
_ => {}
}
for (index, (body, entry)) in children.into_iter().enumerate() {
if let Some(mut path) = if entry == Some(target) {
Some(vec![])
} else {
control_path(body, target)
} {
path.insert(0, (pos, index));
return Some(path);
}
}
}
None
}