Don't let lex state merging be fooled by trivial loops

This commit is contained in:
Max Brunsfeld 2019-01-04 11:52:52 -08:00
parent baf7f3603c
commit d0c3e26e84
3 changed files with 21 additions and 16 deletions

View file

@ -168,7 +168,7 @@ impl<'a> LexTableBuilder<'a> {
self.table.states[state_id].advance_actions.push((
CharacterSet::empty().add_char('\0'),
AdvanceAction {
state: next_state_id,
state: Some(next_state_id),
in_main_token: true,
},
));
@ -189,10 +189,15 @@ impl<'a> LexTableBuilder<'a> {
}
}
let (next_state_id, _) = self.add_state(states, eof_valid && is_separator);
let next_state = if next_state_id == state_id {
None
} else {
Some(next_state_id)
};
self.table.states[state_id].advance_actions.push((
characters,
AdvanceAction {
state: next_state_id,
state: next_state,
in_main_token: !is_separator,
},
));
@ -231,10 +236,10 @@ fn minimize_lex_table(table: &mut LexTable, parse_table: &mut ParseTable) {
}
}
for state in table.states.iter_mut() {
for advance_action in state.advance_actions.iter_mut() {
if let Some(new_state_id) = state_replacements.get(&advance_action.1.state) {
advance_action.1.state = *new_state_id;
}
for (_, advance_action) in state.advance_actions.iter_mut() {
advance_action.state = advance_action
.state
.map(|s| state_replacements.get(&s).cloned().unwrap_or(s))
}
}
}
@ -259,8 +264,8 @@ fn minimize_lex_table(table: &mut LexTable, parse_table: &mut ParseTable) {
}
for state in table.states.iter_mut() {
for advance_action in state.advance_actions.iter_mut() {
advance_action.1.state = final_state_replacements[advance_action.1.state];
for (_, advance_action) in state.advance_actions.iter_mut() {
advance_action.state = advance_action.state.map(|s| final_state_replacements[s]);
}
}

View file

@ -342,7 +342,7 @@ impl Generator {
for (i, state) in lex_table.states.into_iter().enumerate() {
add_line!(self, "case {}:", i);
indent!(self);
self.add_lex_state(state);
self.add_lex_state(i, state);
dedent!(self);
}
@ -358,7 +358,7 @@ impl Generator {
add_line!(self, "");
}
fn add_lex_state(&mut self, state: LexState) {
fn add_lex_state(&mut self, index: usize, state: LexState) {
if let Some(accept_action) = state.accept_action {
add_line!(self, "ACCEPT_TOKEN({});", self.symbol_ids[&accept_action]);
}
@ -372,14 +372,14 @@ impl Generator {
if self.add_character_set_condition(&characters, &ruled_out_characters) {
add!(self, ")\n");
indent!(self);
self.add_advance_action(&action);
self.add_advance_action(index, &action);
if let CharacterSet::Include(chars) = characters {
ruled_out_characters.extend(chars.iter().map(|c| *c as u32));
}
dedent!(self);
} else {
self.buffer.truncate(previous_length);
self.add_advance_action(&action);
self.add_advance_action(index, &action);
}
}
@ -491,11 +491,11 @@ impl Generator {
})
}
fn add_advance_action(&mut self, action: &AdvanceAction) {
fn add_advance_action(&mut self, index: usize, action: &AdvanceAction) {
if action.in_main_token {
add_line!(self, "ADVANCE({});", action.state);
add_line!(self, "ADVANCE({});", action.state.unwrap_or(index));
} else {
add_line!(self, "SKIP({});", action.state);
add_line!(self, "SKIP({});", action.state.unwrap_or(index));
}
}

View file

@ -48,7 +48,7 @@ pub(crate) struct ParseTable {
#[derive(Clone, Debug, PartialEq, Eq)]
pub(crate) struct AdvanceAction {
pub state: LexStateId,
pub state: Option<LexStateId>,
pub in_main_token: bool,
}