psi: resolve memory corruption segfaults and runtime panics (resolves #80, #110, #152, #162) (#174)

Author: wenxuanjun

src/analyzer/lang/utils.v

@@ -62,7 +62,9 @@ pub fn get_zero_value_for(typ types.Type) string {
 }
 
 pub fn is_same_module(context psi.PsiElement, element psi.PsiElement) bool {
-	context_module_fqn := context.containing_file.module_fqn()
-	element_module_fqn := element.containing_file.module_fqn()
+	context_file := context.containing_file() or { return false }
+	element_file := element.containing_file() or { return false }
+	context_module_fqn := context_file.module_fqn()
+	element_module_fqn := element_file.module_fqn()
 	return context_module_fqn == element_module_fqn
 }

src/analyzer/psi/CallExpression.v

@@ -50,7 +50,7 @@ pub fn (c CallExpression) resolve() ?PsiElement {
 pub fn (c CallExpression) parameter_index_on_offset(offset u32) int {
 	argument_list := c.find_child_by_type(.argument_list) or { return -1 }
 	commas := argument_list.children().filter(it.get_text() == ',')
-	count_commas_before := commas.filter(it.node.start_byte() < offset).len
+	count_commas_before := commas.filter(it.node().start_byte() < offset).len
 	return count_commas_before
 }
 

src/analyzer/psi/ConstantDefinition.v

@@ -66,13 +66,14 @@ pub fn (c ConstantDefinition) visibility_modifiers() ?&VisibilityModifiers {
 
 pub fn (c &ConstantDefinition) expression() ?PsiElement {
 	if stub := c.get_stub() {
+		file := c.containing_file() or { return none }
 		// pretty hacky but it works
 		res := parser.parse_code(stub.additional)
 		root := res.tree.root_node()
 		first_child := root.first_child()?
 		next_first_child := first_child.first_child()?
-		file := new_psi_file(c.containing_file.path, res.tree, res.source_text)
-		return create_element(AstNode(next_first_child), file)
+		synthetic_file := new_psi_file(file.path, res.tree, res.source_text)
+		return create_element(AstNode(next_first_child), synthetic_file)
 	}
 	return c.last_child()
 }

src/analyzer/psi/EnumDeclaration.v

@@ -12,7 +12,11 @@ pub fn (e &EnumDeclaration) is_public() bool {
 }
 
 pub fn (e &EnumDeclaration) get_type() types.Type {
-	module_fqn := stubs_index.get_module_qualified_name(e.containing_file.path)
+	module_fqn := if file := e.containing_file() {
+		stubs_index.get_module_qualified_name(file.path)
+	} else {
+		''
+	}
 	return types.new_enum_type(e.name(), module_fqn)
 }
 

src/analyzer/psi/EnumFieldDeclaration.v

@@ -55,7 +55,8 @@ pub fn (f &EnumFieldDeclaration) get_type() types.Type {
 
 pub fn (f &EnumFieldDeclaration) fingerprint() string {
 	owner := f.owner() or { return '' }
-	return '${f.containing_file.path}:${f.node.start_point()}${owner.name()}.${f.name()}'
+	file := f.containing_file() or { return '' }
+	return '${file.path}:${f.node.start_point()}${owner.name()}.${f.name()}'
 }
 
 pub fn (f &EnumFieldDeclaration) value() ?PsiElement {
@@ -64,12 +65,13 @@ pub fn (f &EnumFieldDeclaration) value() ?PsiElement {
 			return none
 		}
 
+		file := f.containing_file() or { return none }
 		res := parser.parse_code(stub.additional)
 		root := res.tree.root_node()
 		first_child := root.first_child()?
 		next_first_child := first_child.first_child()?
-		file := new_psi_file(f.containing_file.path, res.tree, res.source_text)
-		return create_element(next_first_child, file)
+		synthetic_file := new_psi_file(file.path, res.tree, res.source_text)
+		return create_element(next_first_child, synthetic_file)
 	}
 
 	return f.find_child_by_name('value')

src/analyzer/psi/GenericParameters.v

@@ -8,15 +8,8 @@ pub struct GenericParameters {
 
 fn (_ &GenericParameters) stub() {}
 
-pub fn (n &GenericParameters) parameters() []GenericParameter {
-	params := n.find_children_by_type_or_stub(.generic_parameter)
-	mut result := []GenericParameter{cap: params.len}
-	for param in params {
-		if param is GenericParameter {
-			result << param
-		}
-	}
-	return result
+pub fn (n &GenericParameters) parameters() []PsiElement {
+	return n.find_children_by_type_or_stub(.generic_parameter)
 }
 
 pub fn (n &GenericParameters) text_presentation() string {
@@ -27,7 +20,9 @@ pub fn (n &GenericParameters) text_presentation() string {
 	mut sb := strings.new_builder(5)
 	sb.write_string('[')
 	for index, parameter in parameters {
-		sb.write_string(parameter.name())
+		if parameter is PsiNamedElement {
+			sb.write_string(parameter.name())
+		}
 		if index < parameters.len - 1 {
 			sb.write_string(', ')
 		}
@@ -44,7 +39,9 @@ pub fn (n &GenericParameters) parameter_names() []string {
 
 	mut result := []string{cap: parameters.len}
 	for parameter in parameters {
-		result << parameter.name()
+		if parameter is PsiNamedElement {
+			result << parameter.name()
+		}
 	}
 	return result
 }

src/analyzer/psi/Identifier.v

@@ -7,3 +7,28 @@ pub struct Identifier {
 pub fn (i Identifier) value() string {
 	return i.get_text()
 }
+
+pub fn (i Identifier) name() string {
+	return i.get_text()
+}
+
+pub fn (i Identifier) qualifier() ?PsiElement {
+	parent := i.parent()?
+	if parent is SelectorExpression {
+		left := parent.left()?
+		if left.is_equal(i) {
+			return none
+		}
+		return left
+	}
+	return none
+}
+
+pub fn (i Identifier) reference() PsiReference {
+	file := i.containing_file()
+	return new_reference(file, i, false)
+}
+
+pub fn (i Identifier) resolve() ?PsiElement {
+	return i.reference().resolve()
+}

src/analyzer/psi/InterfaceDeclaration.v

@@ -20,7 +20,8 @@ pub fn (s &InterfaceDeclaration) is_public() bool {
 }
 
 pub fn (s &InterfaceDeclaration) module_name() string {
-	return stubs_index.get_module_qualified_name(s.containing_file.path)
+	file := s.containing_file() or { return '' }
+	return stubs_index.get_module_qualified_name(file.path)
 }
 
 pub fn (s &InterfaceDeclaration) get_type() types.Type {

src/analyzer/psi/PrinterVisitor.v

@@ -23,7 +23,7 @@ fn (mut r PrinterVisitor) visit_element(element PsiElement) {
 }
 
 fn (mut r PrinterVisitor) visit_element_impl(element PsiElement) bool {
-	r.lines << '  '.repeat(r.indent) + '${element.node.type_name}'
+	r.lines << '  '.repeat(r.indent) + '${element.node().type_name}'
 	r.text_lines << '${element.get_text()}'
 	return true
 }

src/analyzer/psi/PsiElement.v

@@ -7,21 +7,34 @@ pub type ID = int
 pub type AstNode = bindings.Node[bindings.NodeType]
 
 pub interface PsiElement {
-	node            AstNode  // base node from Tree Sitter
-	containing_file &PsiFile // file where the element is located
-	stub_id         StubId
-	get_stub() ?&StubBase
-	stub_list() &StubList
+	// node returns the base node from Tree Sitter.
+	// In stub-based elements, this might be a zero-value node.
+	node() AstNode
+	// element_type returns the specific type of the element.
 	element_type() bindings.NodeType
-	node() AstNode                  // return base node from Tree Sitter
-	containing_file() &PsiFile      // return file where the element is located
-	is_equal(other PsiElement) bool // return true if the element is equal to the other element
-	// find_element_at returns the leaf node at the specified position relative to the start of the node.
-	// If the node is not found, none is returned.
-	find_element_at(offset u32) ?PsiElement
-	// find_reference_at returns the reference node at the specified position relative to the start of the node.
-	// If the node is not found, none is returned.
-	find_reference_at(offset u32) ?PsiElement
+	// containing_file returns the file where the element is located.
+	// Returns none if the element is synthetic or the file context is lost.
+	containing_file() ?&PsiFile
+	// is_equal returns true if the element represents the same underlying node/stub as other.
+	is_equal(other PsiElement) bool
+
+	// return the slot id of the stub, or non_stubbed_element if the element is not stub-based
+	stub_id() StubId
+	// return the stub associated with the element, or none if the element is not stub-based
+	get_stub() ?&StubBase
+	// stub_list return the stub list associated with the element, or none if the element is not stub-based
+	stub_list() ?&StubList
+
+	// text_range returns the range of the node in the source file.
+	text_range() TextRange
+	// text_length returns the length of the node's text.
+	text_length() int
+	// get_text returns the text of the node.
+	get_text() string
+	// text_matches returns true if the text of the node matches the specified value.
+	// This method is more efficient than `get_text() == value`.
+	text_matches(value string) bool
+
 	// parent returns the parent node.
 	// If the node is the root, none is returned.
 	parent() ?PsiElement
@@ -35,17 +48,15 @@ pub interface PsiElement {
 	// parent_of_any_type returns the parent node with one of the specified types.
 	// If no such node exists, none is returned.
 	parent_of_any_type(types ...bindings.NodeType) ?PsiElement
-	// inside returns true if the node is inside a node with the specified type.
-	inside(typ bindings.NodeType) bool
-	// is_parent_of returns true if the passed node is a child of the given node.
-	is_parent_of(element PsiElement) bool
-	// sibling_of_type_backward returns the previous node at the same nesting level with the specified type.
-	// If no such node exists, none is returned.
-	sibling_of_type_backward(typ bindings.NodeType) ?PsiElement
 	// parent_of_type_or_self returns the parent node with the specified type, or the
 	// node itself if its type matches the specified one.
 	// If no such node exists, none is returned.
 	parent_of_type_or_self(typ bindings.NodeType) ?PsiElement
+	// is_parent_of returns true if the passed node is a child of the given node.
+	is_parent_of(element PsiElement) bool
+	// inside returns true if the node is inside a node with the specified type.
+	inside(typ bindings.NodeType) bool
+
 	// children returns all child nodes.
 	children() []PsiElement
 	// named_children returns child nodes except unknown nodes.
@@ -62,6 +73,7 @@ pub interface PsiElement {
 	// last_child_or_stub returns the last child node or stub.
 	// If the node has no children or stub, none is returned.
 	last_child_or_stub() ?PsiElement
+
 	// next_sibling returns the next node at the same nesting level.
 	// If the node is the last child node, none is returned.
 	next_sibling() ?PsiElement
@@ -77,11 +89,19 @@ pub interface PsiElement {
 	// prev_sibling_or_stub returns the previous node at the same nesting level or stub.
 	// If the node is the first child node or stub, none is returned.
 	prev_sibling_or_stub() ?PsiElement
+	// sibling_of_type_backward returns the previous node at the same nesting level with the specified type.
+	// If no such node exists, none is returned.
+	sibling_of_type_backward(typ bindings.NodeType) ?PsiElement
+
+	// find_element_at returns the leaf node at the specified position relative to the start of the node.
+	// If the node is not found, none is returned.
+	find_element_at(offset u32) ?PsiElement
+	// find_reference_at returns the reference node at the specified position relative to the start of the node.
+	// If the node is not found, none is returned.
+	find_reference_at(offset u32) ?PsiElement
 	// find_child_by_type returns the first child node with the specified type.
 	// If no such node is found, none is returned.
 	find_child_by_type(typ bindings.NodeType) ?PsiElement
-	// has_child_of_type returns true if the node has a child with the specified type.
-	has_child_of_type(typ bindings.NodeType) bool
 	// find_child_by_type_or_stub returns the first child node with the specified type or stub.
 	// If no such node is found, none is returned.
 	find_child_by_type_or_stub(typ bindings.NodeType) ?PsiElement
@@ -94,18 +114,12 @@ pub interface PsiElement {
 	// find_children_by_type_or_stub returns all child nodes with the specified type or stub.
 	// If no such nodes are found, an empty array is returned.
 	find_children_by_type_or_stub(typ bindings.NodeType) []PsiElement
-	// get_text returns the text of the node.
-	get_text() string
-	// text_matches returns true if the text of the node matches the specified value.
-	// This method is more efficient than `get_text() == value`.
-	text_matches(value string) bool
+	// has_child_of_type returns true if the node has a child with the specified type.
+	has_child_of_type(typ bindings.NodeType) bool
+
 	// accept passes the element to the passed visitor.
 	accept(visitor PsiElementVisitor)
 	// accept_mut passes the element to the passed visitor.
 	// Unlike `accept()`, this method uses a visitor that can mutate its state.
 	accept_mut(mut visitor MutablePsiElementVisitor)
-	// text_range returns the range of the node in the source file.
-	text_range() TextRange
-	// text_length returns the length of the node's text.
-	text_length() int
 }