[WIP] Add routines for finding AST nodes

Functions from remix-lib:
- sourceLocationFromAstNode
- nodesAtPosition
- getLinebreakPositions
- findNodeAtSourceLocation

Add walkFull that traverses more of the AST nodes. This is non-legacy
only.

.gitignore

@@ -14,3 +14,5 @@ package-lock.json
 TODO
 soljson.js
 lerna-debug.log
+*~
+/tmp

remix-astwalker/src/astWalker.ts

@@ -98,6 +98,52 @@ export class AstWalker extends EventEmitter {
     }
   }
 
+  walkFullInternal(ast: AstNode, callback?: Function | Object) {
+    function isObject(obj: any): boolean {
+      return obj != null && obj.constructor.name === "Object"
+    }
+    function isAstNode(node: Object): boolean {
+      return (
+        isObject(node) &&
+        'id' in node &&
+        'nodeType' in node &&
+        'src' in node
+      );
+    }
+
+    if (isAstNode(ast) && this.manageCallback(ast, callback)) {
+      // console.log(`XXX id ${ast.id}, nodeType: ${ast.nodeType}, src: ${ast.src}`);
+      for (let k of Object.keys(ast)) {
+        const astItem = ast[k];
+        if (Array.isArray(astItem)) {
+          for (let child of astItem) {
+            if (child) {
+              this.walkFullInternal(child, callback);
+            }
+          }
+        } else {
+          this.walkFullInternal(astItem, callback);
+        }
+      }
+    }
+  }
+
+  // Normalizes parameter callback and calls walkFullInternal
+  walkFull(ast: AstNode, callback?: Function | Object) {
+    if (callback) {
+      if (callback instanceof Function) {
+        callback = Object({ "*": callback });
+      }
+      if (!("*" in callback)) {
+        callback["*"] = function() {
+          return true;
+        };
+      }
+    }
+    return this.walkFullInternal(ast, callback);
+  }
+
+
   walkAstList(sourcesList: Node, cb?: Function) {
     if (cb) {
       if (sourcesList.ast) {

remix-astwalker/src/index.ts

@@ -1,2 +1,3 @@
 export * from './types'
 export * from './astWalker'
+export * from './sourceMappings'

remix-astwalker/src/sourceMappings.ts

@@ -0,0 +1,105 @@
+import { AstWalker } from './astWalker';
+import { AstNode, Location } from "./index";
+
+/**
+ * Break out fields of an AST's "src" attribute string (s:l:f)
+ * into its "start", "length", and "file index" components.
+ *
+ * @param {AstNode} astNode  - the object to convert.
+ */
+export function sourceLocationFromAstNode(astNode: AstNode): Location | null {
+  if (astNode.src) {
+    var split = astNode.src.split(':')
+    return <Location>{
+      start: parseInt(split[0], 10),
+      length: parseInt(split[1], 10),
+      file: parseInt(split[2], 10)
+    }
+  }
+  return null;
+}
+
+/**
+ * Routines for retrieving AST object(s) using some criteria, usually
+ * includng "src' information.
+ */
+export class SourceMappings {
+
+  readonly source: string;
+  readonly lineBreaks: Array<number>;
+
+  constructor(source: string) {
+    this.source = source;
+    this.lineBreaks = this.getLinebreakPositions();
+  };
+
+  /**
+   * get a list of nodes that are at the given @arg offset
+   *
+   * @param {String} astNodeType - type of node to return or null
+   * @param {Int} position     - character offset
+   * @return {Object} ast object given by the compiler
+   */
+  nodesAtPosition(astNodeType: string | null, position: number, ast: AstNode): Array<AstNode> {
+    const astWalker = new AstWalker()
+    const callback = {}
+    let found: Array<AstNode> = [];
+
+    /* FIXME: Looking at AST walker code,
+       I don't understand a need to return a boolean. */
+    callback['*'] = function(node: AstNode): boolean {
+      let nodeLocation = sourceLocationFromAstNode(node);
+      if (nodeLocation &&
+        nodeLocation.start <= position &&
+        nodeLocation.start + nodeLocation.length >= position) {
+        if (!astNodeType || astNodeType === node.name) {
+          found.push(node)
+        }
+      }
+      return true;
+    }
+    astWalker.walk(ast, callback);
+    return found;
+  }
+
+  /**
+   * Retrieve line/column position of each source char
+   *
+   * @param {String} source - contract source code
+   * @return {Array} returns an array containing offset of line breaks
+   */
+  getLinebreakPositions(source: string = this.source): Array<number> {
+    let ret: Array<number> = [];
+    for (var pos = source.indexOf('\n'); pos >= 0; pos = source.indexOf('\n', pos + 1)) {
+      ret.push(pos)
+    }
+    return ret;
+  }
+
+  findNodeAtSourceLocation(astNodeType: string, sourceLocation: Location, ast: AstNode | null) {
+    const astWalker = new AstWalker()
+    const callback = {};
+    let found = null;
+    /* FIXME: Looking at AST walker code,
+       I don't understand a need to return a boolean. */
+    callback['*'] = function(node: AstNode) {
+      let nodeLocation = sourceLocationFromAstNode(node);
+      if (nodeLocation &&
+        nodeLocation.start <= sourceLocation.start &&
+        nodeLocation.start + nodeLocation.length >= sourceLocation.start + sourceLocation.length) {
+        if (astNodeType === node.nodeType) {
+          found = node;
+        }
+      }
+      return true;
+    }
+
+    astWalker.walkFull(ast, callback);
+    return found;
+  }
+}
+
+module.exports = {
+  SourceMappings,
+  sourceLocationFromAstNode
+};

remix-astwalker/src/types.ts

@@ -1,3 +1,9 @@
+export interface Location {
+  start: number;
+  length: number;
+  file: number;  // Would it be clearer to call this a file index?
+}
+
 export interface Node {
   ast?: AstNode;
   legacyAST?: AstNodeLegacy;
@@ -6,12 +12,15 @@ export interface Node {
 }
 
 export interface AstNode {
+  /* The following fields are essential, and indicates an that object
+     is an AST node. */
+  id: number;  // This is unique across all nodes in an AST tree
+  nodeType: string;
+  src: string;
+
   absolutePath?: string;
   exportedSymbols?: Object;
-  id: number;
-  nodeType: string;
   nodes?: Array<AstNode>;
-  src: string;
   literals?: Array<string>;
   file?: string;
   scope?: number;
@@ -22,9 +31,9 @@ export interface AstNode {
 
 export interface AstNodeLegacy {
   id: number;
-  name: string;
+  name: string;  // This corresponds to nodeType in current AST
   src: string;
-  children?: Array<AstNodeLegacy>;
+  children?: Array<AstNodeLegacy>;  // This corresponds to nodes in current AST
   attributes?: AstNodeAtt;
 }
 

remix-astwalker/tests/sourceMappings.ts

@@ -0,0 +1,26 @@
+import tape from "tape";
+import { SourceMappings, sourceLocationFromAstNode } from "../src/sourceMappings";
+import node from "./resources/newAST";
+
+tape("SourceMappings", (t: tape.Test) => {
+  const source = node.source;
+  const srcMappings = new SourceMappings(source);
+  t.test("SourceMappings constructor", (st: tape.Test) => {
+    st.plan(2);
+
+    st.equal(source, srcMappings.source);
+    st.deepEqual(srcMappings.lineBreaks,
+      [15, 26, 27, 38, 39, 81, 87, 103, 119, 135, 141, 142, 186, 192, 193, 199]);
+    st.end();
+  });
+  t.test("SourceMappings fns", (st: tape.Test) => {
+    st.plan(2);
+    const ast = node.ast;
+    st.deepEqual(sourceLocationFromAstNode(ast.nodes[0]),
+      { start: 0, length: 31, file: 0 });
+    const loc = { start: 267, length: 20, file: 0 };
+    const rr = srcMappings.findNodeAtSourceLocation('ExpressionStatement', loc, ast);
+    st.ok(rr);
+    st.end();
+  });
+});