Merge branch 'read_term_eof_changes'

src/machine/machine_indices.rs

@@ -227,7 +227,32 @@ impl CodeIndex {
     }
 }
 
-pub(crate) type HeapVarDict = IndexMap<VarPtr, HeapCellValue, FxBuildHasher>;
+#[derive(Debug, Clone, PartialEq, Eq, Hash)]
+pub enum VarKey {
+    AnonVar(usize),
+    VarPtr(VarPtr),
+}
+
+impl VarKey {
+    #[inline]
+    pub(crate) fn to_string(&self) -> String {
+        match self {
+            VarKey::AnonVar(h) => format!("_{}", h),
+            VarKey::VarPtr(var) => var.borrow().to_string(),
+        }
+    }
+
+    #[inline(always)]
+    pub(crate) fn is_anon(&self) -> bool {
+        if let VarKey::AnonVar(_) = self {
+            true
+        } else {
+            false
+        }
+    }
+}
+
+pub(crate) type HeapVarDict = IndexMap<VarKey, HeapCellValue, FxBuildHasher>;
 
 pub(crate) type GlobalVarDir = IndexMap<Atom, (Ball, Option<HeapCellValue>), FxBuildHasher>;
 

src/machine/machine_state.rs

@@ -486,13 +486,13 @@ impl MachineState {
     pub fn read_term_body(&mut self, mut term_write_result: TermWriteResult) -> CallResult {
         fn push_var_eq_functors<'a>(
             heap: &mut Heap,
-            iter: impl Iterator<Item = (&'a VarPtr, &'a HeapCellValue)>,
+            iter: impl Iterator<Item = (&'a VarKey, &'a HeapCellValue)>,
             atom_tbl: &mut AtomTable,
         ) -> Vec<HeapCellValue> {
             let mut list_of_var_eqs = vec![];
 
             for (var, binding) in iter {
-                let var_atom = atom_tbl.build_with(&var.borrow().to_string());
+                let var_atom = atom_tbl.build_with(&var.to_string());
                 let h = heap.len();
 
                 heap.push(atom_as_cell!(atom!("="), 2));
@@ -542,7 +542,11 @@ impl MachineState {
 
         let singleton_var_list = push_var_eq_functors(
             &mut self.heap,
-            term_write_result.var_dict.iter().filter(|(_, binding)| {
+            term_write_result.var_dict.iter().filter(|(var_name, binding)| {
+                if var_name.is_anon() {
+                    return false;
+                }
+
                 if let Some(r) = binding.as_var() {
                     *singleton_var_set.get(&r).unwrap_or(&false)
                 } else {
@@ -565,7 +569,7 @@ impl MachineState {
 
         let list_of_var_eqs = push_var_eq_functors(
             &mut self.heap,
-            var_list.iter().map(|(var_name, var,_)| (var_name,var)),
+            var_list.iter().filter_map(|(var_name, var,_)| if var_name.is_anon() { None } else { Some((var_name,var)) }),
             &mut self.atom_tbl,
         );
 
@@ -616,7 +620,7 @@ impl MachineState {
         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(
             stream,
             StreamType::Text,
@@ -637,8 +641,12 @@ impl MachineState {
             match self.read(stream, &indices.op_dir) {
                 Ok(term_write_result) => return self.read_term_body(term_write_result),
                 Err(err) => {
-                    match err {
+                    match &err {
                         CompilationError::ParserError(e) if e.is_unexpected_eof() => {
+                            if stream.at_end_of_stream() {
+                                unify!(self, self.registers[2], atom_as_cell!(atom!("end_of_file")));
+                                return Ok(());
+                            } else if stream.past_end_of_stream() {
                                 self.eof_action(
                                     self.registers[2],
                                     stream,
@@ -654,6 +662,7 @@ impl MachineState {
 
                                 return Ok(());
                             }
+                        }
                         _ => {}
                     }
 

src/machine/mock_wam.rs

@@ -71,7 +71,12 @@ impl MockWAM {
         printer.var_names = term_write_result
             .var_dict
             .into_iter()
-            .map(|(var, cell)| (cell, var))
+            .map(|(var, cell)| {
+                match var {
+                    VarKey::VarPtr(var) => (cell, var.clone()),
+                    VarKey::AnonVar(_) => (cell, VarPtr::from(var.to_string()))
+                }
+            })
             .collect();
 
         Ok(printer.print().result())

src/machine/streams.rs

@@ -884,19 +884,38 @@ impl PartialEq 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 {
     #[inline]
     pub(crate) fn position(&mut self) -> Option<(u64, usize)> {
         // returns lines_read, position.
         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) => {
                 file_stream.position()
             }
-            Stream::NamedTcp(..)
+            Stream::NamedTcp(..) | Stream::NamedTls(..) | Stream::Readline(..) => {
-                | Stream::NamedTls(..)
+                Some(0)
-                | Stream::Readline(..)
+            }
-                | Stream::StaticString(..)
-                | Stream::Byte(..) => Some(0),
             _ => None,
         };
 
@@ -971,7 +990,28 @@ impl Stream {
             return AtEndOfStream::Past;
         }
 
-        if let Stream::InputFile(stream_layout) = self {
+        match self {
+            Stream::Byte(stream_layout) => {
+                let StreamLayout {
+                    past_end_of_stream,
+                    stream,
+                    ..
+                } = &mut **stream_layout;
+
+                let cursor_len = stream.get_ref().0.get_ref().len() as u64;
+                cursor_position(past_end_of_stream, &stream.get_ref().0, cursor_len)
+            }
+            Stream::StaticString(stream_layout) => {
+                let StreamLayout {
+                    past_end_of_stream,
+                    stream,
+                    ..
+                } = &mut **stream_layout;
+
+                let cursor_len = stream.stream.get_ref().len() as u64;
+                cursor_position(past_end_of_stream, &stream.stream, cursor_len)
+            }
+            Stream::InputFile(stream_layout) => {
                 let position = stream_layout.position();
 
                 let StreamLayout {
@@ -983,14 +1023,14 @@ impl Stream {
                 match stream.get_ref().file.metadata() {
                     Ok(metadata) => {
                         if let Some(position) = position {
-                        return match position.cmp(&metadata.len()) {
+                            match position.cmp(&metadata.len()) {
                                 Ordering::Equal => AtEndOfStream::At,
                                 Ordering::Less => AtEndOfStream::Not,
                                 Ordering::Greater => {
                                     *past_end_of_stream = true;
                                     AtEndOfStream::Past
                                 }
-                        };
+                            }
                         } else {
                             *past_end_of_stream = true;
                             AtEndOfStream::Past
@@ -1001,10 +1041,12 @@ impl Stream {
                         AtEndOfStream::Past
                     }
                 }
-        } else {
+            }
+            _ => {
                 AtEndOfStream::Not
             }
         }
+    }
 
     #[inline]
     pub(crate) fn file_name(&self) -> Option<Atom> {
@@ -1306,7 +1348,7 @@ impl MachineState {
         match eof_action {
             EOFAction::Error => {
                 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 => {
                 let end_of_stream = if stream.options().stream_type() == StreamType::Binary {

src/machine/system_calls.rs

@@ -5797,14 +5797,21 @@ impl Machine {
     }
 
     #[inline(always)]
-    pub(crate) fn read_from_chars(&mut self) -> CallResult {
+    fn read_term_and_write_to_heap(
-        if let Some(atom_or_string) = self.machine_st.value_to_str_like(self.machine_st.registers[1]) {
+        &mut self,
-            let chars = CharReader::new(ByteStream::from_string(atom_or_string.to_string()));
+        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(CompilationError::from)
+            .map_err(|err| error_after_read_term(err, 0, &parser))
             .and_then(|term| {
                 write_term_to_heap(
                     &term,
@@ -5813,55 +5820,47 @@ impl Machine {
                 )
             });
 
-            let term_write_result = match term_write_result {
+        match term_write_result {
-                Ok(term_write_result) => 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_from_chars"), 2);
+                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));
+                Err(self.machine_st.error_form(e, stub))
+            }
+        }
     }
-            };
 
+    #[inline(always)]
+    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(term_write_result) = self.read_term_and_write_to_heap(atom_or_string)? {
                 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 {
-            unreachable!()
             }
 
             Ok(())
+        } else {
+            unreachable!()
+        }
     }
 
     #[inline(always)]
     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]) {
-            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 op_dir = CompositeOpDir::new(&self.indices.op_dir, None);
-
-            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 {
+                Ok(())
+            }
         } else {
             unreachable!()
         }

src/read.rs

@@ -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 {
     pub(crate) fn read(
         &mut self,
@@ -50,7 +69,7 @@ impl MachineState {
             parser.add_lines_read(prior_num_lines_read);
 
             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)
         };
@@ -240,9 +259,9 @@ impl CharRead for ReadlineStream {
 }
 
 #[inline]
-pub(crate) fn write_term_to_heap(
+pub(crate) fn write_term_to_heap<'a, 'b>(
-    term: &Term,
+    term: &'a Term,
-    heap: &mut Heap,
+    heap: &'b mut Heap,
     atom_tbl: &mut AtomTable,
 ) -> Result<TermWriteResult, CompilationError> {
     let term_writer = TermWriter::new(heap, atom_tbl);
@@ -275,7 +294,7 @@ impl<'a, 'b> TermWriter<'a, 'b> {
     }
 
     #[inline]
-    fn modify_head_of_queue(&mut self, term: &TermRef<'a>, h: usize) {
+    fn modify_head_of_queue(&mut self, term: &TermRef, h: usize) {
         if let Some((arity, site_h)) = self.queue.pop_front() {
             self.heap[site_h] = self.term_as_addr(term, h);
 
@@ -291,7 +310,7 @@ impl<'a, 'b> TermWriter<'a, 'b> {
         self.heap.push(heap_loc_as_cell!(h));
     }
 
-    fn term_as_addr(&mut self, term: &TermRef<'a>, h: usize) -> HeapCellValue {
+    fn term_as_addr(&mut self, term: &TermRef, h: usize) -> HeapCellValue {
         match term {
             &TermRef::Cons(..) => list_loc_as_cell!(h),
             &TermRef::AnonVar(_) | &TermRef::Var(..) => heap_loc_as_cell!(h),
@@ -310,7 +329,7 @@ impl<'a, 'b> TermWriter<'a, 'b> {
         }
     }
 
-    fn write_term_to_heap(mut self, term: &'a Term) -> Result<TermWriteResult, CompilationError> {
+    fn write_term_to_heap(mut self, term: &Term) -> Result<TermWriteResult, CompilationError> {
         let heap_loc = self.heap.len();
 
         for term in breadth_first_iter(term, RootIterationPolicy::Iterated) {
@@ -364,17 +383,19 @@ impl<'a, 'b> TermWriter<'a, 'b> {
                         self.push_stub_addr();
                     }
                 }
-                &TermRef::AnonVar(Level::Root) | &TermRef::Literal(Level::Root, ..) => {
+                &TermRef::AnonVar(Level::Root) | TermRef::Literal(Level::Root, ..) => {
                     let addr = self.term_as_addr(&term, h);
                     self.heap.push(addr);
                 }
                 &TermRef::Var(Level::Root, _, ref var_ptr) => {
                     let addr = self.term_as_addr(&term, h);
-                    self.var_dict.insert(var_ptr.clone(), heap_loc_as_cell!(h));
+                    self.var_dict.insert(VarKey::VarPtr(var_ptr.clone()), addr);
                     self.heap.push(addr);
                 }
                 &TermRef::AnonVar(_) => {
                     if let Some((arity, site_h)) = self.queue.pop_front() {
+                        self.var_dict.insert(VarKey::AnonVar(h), heap_loc_as_cell!(site_h));
+
                         if arity > 1 {
                             self.queue.push_front((arity - 1, site_h + 1));
                         }
@@ -403,10 +424,12 @@ impl<'a, 'b> TermWriter<'a, 'b> {
                 }
                 &TermRef::Var(_, _, ref var) => {
                     if let Some((arity, site_h)) = self.queue.pop_front() {
-                        if let Some(addr) = self.var_dict.get(var).cloned() {
+                        let var_key = VarKey::VarPtr(var.clone());
+
+                        if let Some(addr) = self.var_dict.get(&var_key).cloned() {
                             self.heap[site_h] = addr;
                         } else {
-                            self.var_dict.insert(var.clone(), heap_loc_as_cell!(site_h));
+                            self.var_dict.insert(var_key, heap_loc_as_cell!(site_h));
                         }
 
                         if arity > 1 {