do a better job handling EOF in read_term (#1887)

This commit is contained in:
Mark
2023-07-16 20:42:40 -06:00
parent a18da368a1
commit 86c90d77dd
4 changed files with 162 additions and 97 deletions

View File

@@ -616,7 +616,7 @@ impl MachineState {
unreachable!("Stream must be a Stream::Readline(_)") unreachable!("Stream must be a Stream::Readline(_)")
} }
pub fn read_term(&mut self, stream: Stream, indices: &mut IndexStore) -> CallResult { pub fn read_term(&mut self, mut stream: Stream, indices: &mut IndexStore) -> CallResult {
self.check_stream_properties( self.check_stream_properties(
stream, stream,
StreamType::Text, StreamType::Text,
@@ -637,22 +637,27 @@ impl MachineState {
match self.read(stream, &indices.op_dir) { match self.read(stream, &indices.op_dir) {
Ok(term_write_result) => return self.read_term_body(term_write_result), Ok(term_write_result) => return self.read_term_body(term_write_result),
Err(err) => { Err(err) => {
match err { match &err {
CompilationError::ParserError(e) if e.is_unexpected_eof() => { CompilationError::ParserError(e) if e.is_unexpected_eof() => {
self.eof_action( if stream.at_end_of_stream() {
self.registers[2], unify!(self, self.registers[2], atom_as_cell!(atom!("end_of_file")));
stream, return Ok(());
atom!("read_term"), } else if stream.past_end_of_stream() {
3, self.eof_action(
)?; self.registers[2],
stream,
atom!("read_term"),
3,
)?;
if stream.options().eof_action() == EOFAction::Reset { if stream.options().eof_action() == EOFAction::Reset {
if self.fail == false { if self.fail == false {
continue; continue;
}
} }
}
return Ok(()); return Ok(());
}
} }
_ => {} _ => {}
} }

View File

@@ -884,19 +884,38 @@ impl PartialEq for Stream {
impl Eq for Stream {} impl Eq for Stream {}
fn cursor_position<T>(past_end_of_stream: &mut bool, cursor: &Cursor<T>, cursor_len: u64) -> AtEndOfStream {
let position = cursor.position();
let at_end_of_stream = match position.cmp(&cursor_len) {
Ordering::Equal => AtEndOfStream::At,
Ordering::Greater => {
*past_end_of_stream = true;
AtEndOfStream::Past
}
Ordering::Less => AtEndOfStream::Not,
};
at_end_of_stream
}
impl Stream { impl Stream {
#[inline] #[inline]
pub(crate) fn position(&mut self) -> Option<(u64, usize)> { pub(crate) fn position(&mut self) -> Option<(u64, usize)> {
// returns lines_read, position. // returns lines_read, position.
let result = match self { let result = match self {
Stream::Byte(byte_stream_layout) => {
Some(byte_stream_layout.stream.get_ref().0.position())
}
Stream::StaticString(string_stream_layout) => {
Some(string_stream_layout.stream.stream.position())
}
Stream::InputFile(file_stream) => { Stream::InputFile(file_stream) => {
file_stream.position() file_stream.position()
} }
Stream::NamedTcp(..) Stream::NamedTcp(..) | Stream::NamedTls(..) | Stream::Readline(..) => {
| Stream::NamedTls(..) Some(0)
| Stream::Readline(..) }
| Stream::StaticString(..)
| Stream::Byte(..) => Some(0),
_ => None, _ => None,
}; };
@@ -971,38 +990,61 @@ impl Stream {
return AtEndOfStream::Past; return AtEndOfStream::Past;
} }
if let Stream::InputFile(stream_layout) = self { match self {
let position = stream_layout.position(); Stream::Byte(stream_layout) => {
let StreamLayout {
past_end_of_stream,
stream,
..
} = &mut **stream_layout;
let StreamLayout { let cursor_len = stream.get_ref().0.get_ref().len() as u64;
past_end_of_stream, cursor_position(past_end_of_stream, &stream.get_ref().0, cursor_len)
stream, }
.. Stream::StaticString(stream_layout) => {
} = &mut **stream_layout; let StreamLayout {
past_end_of_stream,
stream,
..
} = &mut **stream_layout;
match stream.get_ref().file.metadata() { let cursor_len = stream.stream.get_ref().len() as u64;
Ok(metadata) => { cursor_position(past_end_of_stream, &stream.stream, cursor_len)
if let Some(position) = position { }
return match position.cmp(&metadata.len()) { Stream::InputFile(stream_layout) => {
Ordering::Equal => AtEndOfStream::At, let position = stream_layout.position();
Ordering::Less => AtEndOfStream::Not,
Ordering::Greater => { let StreamLayout {
*past_end_of_stream = true; past_end_of_stream,
AtEndOfStream::Past stream,
..
} = &mut **stream_layout;
match stream.get_ref().file.metadata() {
Ok(metadata) => {
if let Some(position) = position {
match position.cmp(&metadata.len()) {
Ordering::Equal => AtEndOfStream::At,
Ordering::Less => AtEndOfStream::Not,
Ordering::Greater => {
*past_end_of_stream = true;
AtEndOfStream::Past
}
} }
}; } else {
} else { *past_end_of_stream = true;
AtEndOfStream::Past
}
}
_ => {
*past_end_of_stream = true; *past_end_of_stream = true;
AtEndOfStream::Past AtEndOfStream::Past
} }
} }
_ => {
*past_end_of_stream = true;
AtEndOfStream::Past
}
} }
} else { _ => {
AtEndOfStream::Not AtEndOfStream::Not
}
} }
} }
@@ -1306,7 +1348,7 @@ impl MachineState {
match eof_action { match eof_action {
EOFAction::Error => { EOFAction::Error => {
stream.set_past_end_of_stream(true); stream.set_past_end_of_stream(true);
return Err(self.open_past_eos_error(stream, caller, arity)); Err(self.open_past_eos_error(stream, caller, arity))
} }
EOFAction::EOFCode => { EOFAction::EOFCode => {
let end_of_stream = if stream.options().stream_type() == StreamType::Binary { let end_of_stream = if stream.options().stream_type() == StreamType::Binary {

View File

@@ -5796,72 +5796,71 @@ impl Machine {
self.machine_st.read_term(stream, &mut self.indices) self.machine_st.read_term(stream, &mut self.indices)
} }
#[inline(always)]
fn read_term_and_write_to_heap(
&mut self,
atom_or_string: AtomOrString,
) -> Result<Option<TermWriteResult>, MachineStub> {
let string = match atom_or_string {
AtomOrString::Atom(atom) if atom == atom!("[]") => "".to_owned(),
_ => atom_or_string.to_string(),
};
let chars = CharReader::new(ByteStream::from_string(string));
let mut parser = Parser::new(chars, &mut self.machine_st);
let op_dir = CompositeOpDir::new(&self.indices.op_dir, None);
let term_write_result = parser.read_term(&op_dir, Tokens::Default)
.map_err(|err| error_after_read_term(err, 0, &parser))
.and_then(|term| {
write_term_to_heap(
&term,
&mut self.machine_st.heap,
&mut self.machine_st.atom_tbl,
)
});
match term_write_result {
Ok(term_write_result) => Ok(Some(term_write_result)),
Err(CompilationError::ParserError(e)) if e.is_unexpected_eof() => {
let value = self.machine_st.registers[2];
self.machine_st.unify_atom(atom!("end_of_file"), value);
Ok(None)
}
Err(e) => {
let stub = functor_stub(atom!("read_term_from_chars"), 3);
let e = self.machine_st.session_error(SessionError::from(e));
Err(self.machine_st.error_form(e, stub))
}
}
}
#[inline(always)] #[inline(always)]
pub(crate) fn read_from_chars(&mut self) -> CallResult { pub(crate) fn read_from_chars(&mut self) -> CallResult {
if let Some(atom_or_string) = self.machine_st.value_to_str_like(self.machine_st.registers[1]) { if let Some(atom_or_string) = self.machine_st.value_to_str_like(self.machine_st.registers[1]) {
let chars = CharReader::new(ByteStream::from_string(atom_or_string.to_string())); if let Some(term_write_result) = self.read_term_and_write_to_heap(atom_or_string)? {
let mut parser = Parser::new(chars, &mut self.machine_st); let result = heap_loc_as_cell!(term_write_result.heap_loc);
let op_dir = CompositeOpDir::new(&self.indices.op_dir, None); let var = self.deref_register(2).as_var().unwrap();
let term_write_result = parser.read_term(&op_dir, Tokens::Default) self.machine_st.bind(var, result);
.map_err(CompilationError::from) }
.and_then(|term| {
write_term_to_heap(
&term,
&mut self.machine_st.heap,
&mut self.machine_st.atom_tbl,
)
});
let term_write_result = match term_write_result { Ok(())
Ok(term_write_result) => term_write_result,
Err(e) => {
let stub = functor_stub(atom!("read_from_chars"), 2);
let e = self.machine_st.session_error(SessionError::from(e));
return Err(self.machine_st.error_form(e, stub));
}
};
let result = heap_loc_as_cell!(term_write_result.heap_loc);
let var = self.deref_register(2).as_var().unwrap();
self.machine_st.bind(var, result);
} else { } else {
unreachable!() unreachable!()
} }
Ok(())
} }
#[inline(always)] #[inline(always)]
pub(crate) fn read_term_from_chars(&mut self) -> CallResult { pub(crate) fn read_term_from_chars(&mut self) -> CallResult {
if let Some(atom_or_string) = self.machine_st.value_to_str_like(self.machine_st.registers[1]) { if let Some(atom_or_string) = self.machine_st.value_to_str_like(self.machine_st.registers[1]) {
let chars = CharReader::new(ByteStream::from_string(atom_or_string.to_string())); if let Some(term_write_result) = self.read_term_and_write_to_heap(atom_or_string)? {
let mut parser = Parser::new(chars, &mut self.machine_st); self.machine_st.read_term_body(term_write_result)
let op_dir = CompositeOpDir::new(&self.indices.op_dir, None); } else {
Ok(())
let term_write_result = parser.read_term(&op_dir, Tokens::Default) }
.map_err(CompilationError::from)
.and_then(|term| {
write_term_to_heap(
&term,
&mut self.machine_st.heap,
&mut self.machine_st.atom_tbl,
)
});
let term_write_result = match term_write_result {
Ok(term_write_result) => term_write_result,
Err(e) => {
let stub = functor_stub(atom!("read_term_from_chars"), 3);
let e = self.machine_st.session_error(SessionError::from(e));
return Err(self.machine_st.error_form(e, stub));
}
};
self.machine_st.read_term_body(term_write_result)
} else { } else {
unreachable!() unreachable!()
} }

View File

@@ -36,6 +36,25 @@ pub(crate) fn devour_whitespace<'a, R: CharRead>(parser: &mut Parser<'a, R>) ->
} }
} }
pub(crate) fn error_after_read_term<R>(
err: ParserError,
prior_num_lines_read: usize,
parser: &Parser<R>,
) -> CompilationError {
if err.is_unexpected_eof() {
let line_num = parser.lexer.line_num;
let col_num = parser.lexer.col_num;
// rough overlap with errors 8.14.1.3 k) & l) of the ISO standard here
if !(line_num == prior_num_lines_read && col_num == 0) {
return CompilationError::from(ParserError::IncompleteReduction(line_num, col_num));
}
}
CompilationError::from(err)
}
impl MachineState { impl MachineState {
pub(crate) fn read( pub(crate) fn read(
&mut self, &mut self,
@@ -50,7 +69,7 @@ impl MachineState {
parser.add_lines_read(prior_num_lines_read); parser.add_lines_read(prior_num_lines_read);
let term = parser.read_term(&op_dir, Tokens::Default) let term = parser.read_term(&op_dir, Tokens::Default)
.map_err(CompilationError::from)?; .map_err(|err| error_after_read_term(err, prior_num_lines_read, &parser))?; // CompilationError::from
(term, parser.lines_read() - prior_num_lines_read) (term, parser.lines_read() - prior_num_lines_read)
}; };