CFG: fix wrong exit with binop and for loop

src/cli/repl/commands/cfg.ts

@@ -1,10 +1,8 @@
 import { ReplCommand } from './main'
 import { SteppingSlicer } from '../../../core'
 import { requestFromInput, RShell } from '../../../r-bridge'
-import {
-	cfgToMermaid, cfgToMermaidUrl
-} from '../../../util/mermaid'
 import { extractCFG } from '../../../util/cfg'
+import { cfgToMermaid, cfgToMermaidUrl } from '../../../util/mermaid'
 
 async function controlflow(shell: RShell, remainingLine: string) {
 	return await new SteppingSlicer({
@@ -15,27 +13,27 @@
 }
 
 export const controlflowCommand: ReplCommand = {
-	description:  'Get mermaid code for the control-flow graph of R code, start with \'file://\' to indicate a file.',
+	description:  'Get mermaid code for the control-flow graph of R code, start with \'file://\' to indicate a file',
 	usageExample: ':controlflow',
 	aliases:      [ 'cfg' ],
 	script:       false,
 	fn:           async(output, shell, remainingLine) => {
 		const result = await controlflow(shell, remainingLine)
 
 		const cfg = extractCFG(result.normalize)
-		output.stdout(cfgToMermaid(cfg))
+		output.stdout(cfgToMermaid(cfg, result.normalize))
 	}
 }
 
 export const controlflowStarCommand: ReplCommand = {
-	description:  'Get a mermaid url of the control-flow graph of R code, start with \'file://\' to indicate a file.',
+	description:  'Get a mermaid url of the control-flow graph of R code, start with \'file://\' to indicate a file',
 	usageExample: ':controlflow',
 	aliases:      [ 'cfg*' ],
 	script:       false,
 	fn:           async(output, shell, remainingLine) => {
 		const result = await controlflow(shell, remainingLine)
 
 		const cfg = extractCFG(result.normalize)
-		output.stdout(cfgToMermaidUrl(cfg))
+		output.stdout(cfgToMermaidUrl(cfg, result.normalize))
 	}
 }

src/cli/repl/commands/dataflow.ts

@@ -4,7 +4,7 @@ import { requestFromInput, RShell } from '../../../r-bridge'
 import {
 	graphToMermaid,
 	graphToMermaidUrl
-} from '../../../util/mermaid'
+} from '../../../util/mermaid/dfg'
 
 async function dataflow(shell: RShell, remainingLine: string) {
 	return await new SteppingSlicer({
@@ -15,7 +15,7 @@ async function dataflow(shell: RShell, remainingLine: string) {
 }
 
 export const dataflowCommand: ReplCommand = {
-	description:  'Get mermaid code for the dataflow graph of R code, start with \'file://\' to indicate a file.',
+	description:  'Get mermaid code for the dataflow graph of R code, start with \'file://\' to indicate a file',
 	usageExample: ':dataflow',
 	aliases:      [ 'd', 'df' ],
 	script:       false,
@@ -27,7 +27,7 @@ export const dataflowCommand: ReplCommand = {
 }
 
 export const dataflowStarCommand: ReplCommand = {
-	description:  'Get a mermaid url of the dataflow graph of R code, start with \'file://\' to indicate a file.',
+	description:  'Get a mermaid url of the dataflow graph of R code, start with \'file://\' to indicate a file',
 	usageExample: ':dataflow*',
 	aliases:      [ 'd*', 'df*' ],
 	script:       false,

src/cli/repl/commands/normalize.ts

@@ -12,7 +12,7 @@ async function normalize(shell: RShell, remainingLine: string) {
 }
 
 export const normalizeCommand: ReplCommand = {
-	description:  'Get mermaid code for the normalized AST of R code, start with \'file://\' to indicate a file.',
+	description:  'Get mermaid code for the normalized AST of R code, start with \'file://\' to indicate a file',
 	usageExample: ':normalize',
 	aliases:      [ 'n' ],
 	script:       false,
@@ -24,7 +24,7 @@ export const normalizeCommand: ReplCommand = {
 }
 
 export const normalizeStarCommand: ReplCommand = {
-	description:  'Get a mermaid url of the normalized AST of R code, start with \'file://\' to indicate a file.',
+	description:  'Get a mermaid url of the normalized AST of R code, start with \'file://\' to indicate a file',
 	usageExample: ':normalize',
 	aliases:      [ 'n*' ],
 	script:       false,

src/cli/repl/commands/parse.ts

@@ -120,7 +120,7 @@ function depthListToTextTree(list: Readonly<DepthList>, config: XmlParserConfig,
 
 
 export const parseCommand: ReplCommand = {
-	description:  'Prints ASCII Art of the parsed, unmodified AST, start with \'file://\' to indicate a file.',
+	description:  'Prints ASCII Art of the parsed, unmodified AST, start with \'file://\' to indicate a file',
 	usageExample: ':parse',
 	aliases:      [ 'p' ],
 	script:       false,

src/dataflow/index.ts

@@ -5,10 +5,4 @@ export const dataflowLogger = log.getSubLogger({ name: 'dataflow' })
 export * from './graph'
 export * from './extractor'
 export * from './environments/environment'
-export { diffGraphsToMermaidUrl } from '../util/mermaid'
-export { diffGraphsToMermaid } from '../util/mermaid'
-export { LabeledDiffGraph } from '../util/mermaid'
-export { graphToMermaidUrl } from '../util/mermaid'
-export { mermaidCodeToUrl } from '../util/mermaid'
-export { graphToMermaid } from '../util/mermaid'
-export { formatRange } from '../util/mermaid'
+export * from '../util/mermaid/dfg'

src/util/cfg.ts

@@ -17,8 +17,6 @@
 export interface CfgVertex {
 	id:        NodeId
 	name:      string
-	/** the content may be undefined, if the node is an artificial exit point node that i use to mark the exit point of an if condition, a function, etc. */
-	content:   string | undefined
 	/** in case of a function definition */
 	children?: NodeId[]
 }
@@ -155,13 +153,10 @@
 }
 
 
-function getLexeme(n: RNodeWithParent) {
-	return n.info.fullLexeme ?? n.lexeme ?? '<unknown>'
-}
 
 function cfgLeaf(leaf: RNodeWithParent): ControlFlowInformation {
 	const graph = new ControlFlowGraph()
-	graph.addVertex({ id: leaf.info.id, name: leaf.type, content: getLexeme(leaf) })
+	graph.addVertex({ id: leaf.info.id, name: leaf.type })
 	return { graph, breaks: [], nexts: [], returns: [], exitPoints: [leaf.info.id], entryPoints: [leaf.info.id] }
 }
 
@@ -179,8 +174,8 @@
 
 function cfgIfThenElse(ifNode: RNodeWithParent, condition: ControlFlowInformation, then: ControlFlowInformation, otherwise: ControlFlowInformation | undefined): ControlFlowInformation {
 	const graph = new ControlFlowGraph()
-	graph.addVertex({ id: ifNode.info.id, name: ifNode.type, content: getLexeme(ifNode) })
-	graph.addVertex({ id: ifNode.info.id + '-exit', name: 'if-exit', content: undefined })
+	graph.addVertex({ id: ifNode.info.id, name: ifNode.type })
+	graph.addVertex({ id: ifNode.info.id + '-exit', name: 'if-exit' })
 	graph.merge(condition.graph)
 	graph.merge(then.graph)
 	if(otherwise) {
@@ -220,8 +215,8 @@
 
 function cfgRepeat(repeat: RRepeatLoop<ParentInformation>, body: ControlFlowInformation): ControlFlowInformation {
 	const graph = body.graph
-	graph.addVertex({ id: repeat.info.id, name: repeat.type, content: getLexeme(repeat) })
-	graph.addVertex({ id: repeat.info.id + '-exit', name: 'repeat-exit', content: undefined })
+	graph.addVertex({ id: repeat.info.id, name: repeat.type })
+	graph.addVertex({ id: repeat.info.id + '-exit', name: 'repeat-exit' })
 
 	for(const entryPoint of body.entryPoints) {
 		graph.addEdge(repeat.info.id, entryPoint, { label: 'FD' })
@@ -241,8 +236,8 @@
 
 function cfgWhile(whileLoop: RWhileLoop<ParentInformation>, condition: ControlFlowInformation, body: ControlFlowInformation): ControlFlowInformation {
 	const graph = condition.graph
-	graph.addVertex({ id: whileLoop.info.id, name: whileLoop.type, content: getLexeme(whileLoop) })
-	graph.addVertex({ id: whileLoop.info.id + '-exit', name: 'while-exit', content: undefined })
+	graph.addVertex({ id: whileLoop.info.id, name: whileLoop.type })
+	graph.addVertex({ id: whileLoop.info.id + '-exit', name: 'while-exit' })
 
 	graph.merge(body.graph)
 
@@ -277,8 +272,8 @@
 
 function cfgFor(forLoop: RForLoop<ParentInformation>, variable: ControlFlowInformation, vector: ControlFlowInformation, body: ControlFlowInformation): ControlFlowInformation {
 	const graph = variable.graph
-	graph.addVertex({ id: forLoop.info.id, name: forLoop.type, content: getLexeme(forLoop) })
-	graph.addVertex({ id: forLoop.info.id + '-exit', name: 'for-exit', content: undefined })
+	graph.addVertex({ id: forLoop.info.id, name: forLoop.type })
+	graph.addVertex({ id: forLoop.info.id + '-exit', name: 'for-exit' })
 
 	graph.merge(vector.graph)
 	graph.merge(body.graph)
@@ -299,10 +294,6 @@
 		}
 	}
 
-	for(const entryPoint of body.entryPoints) {
-		graph.addEdge(forLoop.info.id, entryPoint, { label: 'FD' })
-	}
-
 	for(const next of [...body.nexts, ...body.exitPoints]) {
 		graph.addEdge(forLoop.info.id, next, { label: 'FD' })
 	}
@@ -321,9 +312,9 @@
 function cfgFunctionDefinition(fn: RFunctionDefinition<ParentInformation>, params: ControlFlowInformation[], body: ControlFlowInformation): ControlFlowInformation {
 	const graph = new ControlFlowGraph()
 	const children: NodeId[] = [fn.info.id + '-params', fn.info.id + '-exit']
-	graph.addVertex({ id: fn.info.id + '-params', name: 'function-parameters', content: undefined }, false)
-	graph.addVertex({ id: fn.info.id + '-exit', name: 'function-exit', content: undefined }, false)
-	graph.addVertex({ id: fn.info.id, name: fn.type, content: getLexeme(fn), children })
+	graph.addVertex({ id: fn.info.id + '-params', name: 'function-parameters' }, false)
+	graph.addVertex({ id: fn.info.id + '-exit', name: 'function-exit' }, false)
+	graph.addVertex({ id: fn.info.id, name: fn.type, children })
 
 	graph.merge(body.graph, true)
 	children.push(...body.graph.rootVertexIds())
@@ -352,9 +343,10 @@
 
 function cfgBinaryOp(binOp: RNodeWithParent, lhs: ControlFlowInformation, rhs: ControlFlowInformation): ControlFlowInformation {
 	const graph = new ControlFlowGraph().merge(lhs.graph).merge(rhs.graph)
-	const result: ControlFlowInformation = { graph, breaks: [...lhs.breaks, ...rhs.breaks], nexts: [...lhs.nexts, ...rhs.nexts], returns: [...lhs.returns, ...rhs.returns], entryPoints: [binOp.info.id], exitPoints: [...rhs.entryPoints] }
+	const result: ControlFlowInformation = { graph, breaks: [...lhs.breaks, ...rhs.breaks], nexts: [...lhs.nexts, ...rhs.nexts], returns: [...lhs.returns, ...rhs.returns], entryPoints: [binOp.info.id], exitPoints: [binOp.info.id + '-exit'] }
 
-	graph.addVertex({ id: binOp.info.id, name: binOp.type, content: getLexeme(binOp) })
+	graph.addVertex({ id: binOp.info.id, name: binOp.type })
+	graph.addVertex({ id: binOp.info.id + '-exit', name: 'binOp-exit' })
 
 	for(const exitPoint of lhs.exitPoints) {
 		for(const entryPoint of rhs.entryPoints) {
@@ -364,15 +356,18 @@
 	for(const entryPoint of lhs.entryPoints) {
 		graph.addEdge(entryPoint, binOp.info.id, { label: 'FD' })
 	}
+	for(const exitPoint of rhs.exitPoints) {
+		graph.addEdge(binOp.info.id + '-exit', exitPoint, { label: 'FD' })
+	}
 
 	return result
 }
 
 function cfgAccess(access: RAccess<ParentInformation>, name: ControlFlowInformation, accessors: string | (ControlFlowInformation | null)[]): ControlFlowInformation {
 	const result = name
 	const graph = result.graph
-	graph.addVertex({ id: access.info.id, name: access.type, content: getLexeme(access) })
-	graph.addVertex({ id: access.info.id + '-exit', name: 'access-exit', content: undefined })
+	graph.addVertex({ id: access.info.id, name: access.type })
+	graph.addVertex({ id: access.info.id + '-exit', name: 'access-exit' })
 	for(const entry of name.entryPoints) {
 		graph.addEdge(entry, access.info.id, { label: 'FD' })
 	}
@@ -403,7 +398,7 @@
 	const graph = operand.graph
 	const result: ControlFlowInformation = { ...operand, graph, exitPoints: [unary.info.id] }
 
-	graph.addVertex({ id: unary.info.id, name: unary.type, content: getLexeme(unary) })
+	graph.addVertex({ id: unary.info.id, name: unary.type })
 
 	return result
 }
@@ -462,7 +457,7 @@
 	}
 	for(const [id, aInfo] of aVert) {
 		const bInfo = bVert.get(id)
-		if(bInfo === undefined || aInfo.name !== bInfo.name || aInfo.content !== bInfo.content || equalChildren(aInfo.children, bInfo.children)) {
+		if(bInfo === undefined || aInfo.name !== bInfo.name || equalChildren(aInfo.children, bInfo.children)) {
 			return false
 		}
 	}
@@ -501,7 +496,6 @@
 			.map(([id, v]) => ({
 				id,
 				name:     v.name,
-				content:  v.content,
 				children: v.children
 			})),
 		edges: [...cfg.graph.edges()].flatMap(([fromId, targets]) =>

src/util/mermaid/ast.ts

@@ -0,0 +1,24 @@
+import { RNodeWithParent, RoleInParent, visitAst } from '../../r-bridge'
+import { escapeMarkdown, mermaidCodeToUrl } from './mermaid'
+
+export function normalizedAstToMermaid(ast: RNodeWithParent, prefix = ''): string {
+	let output = prefix + 'flowchart TD\n'
+	visitAst(ast, n => {
+		const name = `${n.type} (${n.info.id})\\n${n.lexeme ?? ' '}`
+		output += `    n${n.info.id}(["${escapeMarkdown(name)}"])\n`
+		if(n.info.parent !== undefined) {
+			const context = n.info
+			const roleSuffix = context.role === RoleInParent.ExpressionListChild || context.role === RoleInParent.FunctionCallArgument || context.role === RoleInParent.FunctionDefinitionParameter ? `-${context.index}` : ''
+			output += `    n${n.info.parent} -->|"${context.role}${roleSuffix}"| n${n.info.id}\n`
+		}
+		return false
+	})
+	return output
+}
+
+/**
+ * Use mermaid to visualize the normalized AST.
+ */
+export function normalizedAstToMermaidUrl(ast: RNodeWithParent, prefix = ''): string {
+	return mermaidCodeToUrl(normalizedAstToMermaid(ast, prefix))
+}

src/util/mermaid/cfg.ts

@@ -0,0 +1,38 @@
+import { NormalizedAst, RNodeWithParent } from '../../r-bridge'
+import { ControlFlowInformation } from '../cfg'
+import { escapeMarkdown, mermaidCodeToUrl } from './mermaid'
+
+function getLexeme(n?: RNodeWithParent) {
+	return n ? n.info.fullLexeme ?? n.lexeme ?? '<unknown>' : ''
+}
+
+
+export function cfgToMermaid(cfg: ControlFlowInformation, normalizedAst: NormalizedAst, prefix = ''): string {
+	let output = prefix + 'flowchart TD\n'
+
+	for(const [id, vertex] of cfg.graph.vertices()) {
+		const normalizedVertex = normalizedAst.idMap.get(id)
+		const content = getLexeme(normalizedVertex)
+		if(content.length > 0) {
+			const name = `"\`${escapeMarkdown(vertex.name)} (${id})\n${escapeMarkdown(JSON.stringify(content))}\`"`
+			output += `    n${id}[${name}]\n`
+		} else {
+			output += `    n${id}(( ))\n`
+		}
+	}
+	for(const [from, targets] of cfg.graph.edges()) {
+		for(const [to, edge] of targets) {
+			const edgeType = edge.label === 'CD' ? '-->' : '-.->'
+			const edgeSuffix = edge.label === 'CD' ? ` (${edge.when})` : ''
+			output += `    n${from} ${edgeType}|"${escapeMarkdown(edge.label)}${edgeSuffix}"| n${to}\n`
+		}
+	}
+	return output
+}
+
+/**
+ * Use mermaid to visualize the normalized AST.
+ */
+export function cfgToMermaidUrl(cfg: ControlFlowInformation, normalizedAst: NormalizedAst, prefix = ''): string {
+	return mermaidCodeToUrl(cfgToMermaid(cfg, normalizedAst, prefix))
+}