Merge pull request #22417 from MathiasVP/js-parse-weird-module-import

JS: Parse alternative import syntax

Author: asgerf

javascript/extractor/src/com/semmle/jcorn/Parser.java

@@ -118,6 +118,9 @@
  * 4.0.3</a>, but does not support plugins, and always tracks full source locations.
  */
 public class Parser {
+  private static final Pattern LEGACY_MODULE_IMPORT_TAIL =
+      Pattern.compile("\\s+[A-Za-z_$][A-Za-z0-9_$]*\\s+from\\s*['\"]");
+
   protected final Options options;
   protected final Set<String> keywords;
   private final Set<String> reservedWords, reservedWordsStrict, reservedWordsStrictBind;
@@ -2718,6 +2721,18 @@ boolean isUsingDecl() {
         && Identifiers.isIdentifierChar(this.input.codePointAt(next), false);
   }
 
+  /**
+   * Checks for the abandoned ES6 draft namespace-import syntax:
+   * {@code module namespace from "module-name";}.
+   */
+  boolean isLegacyModuleImport() {
+    if (this.type != TokenType.name || !this.value.equals("module")) return false;
+
+    Matcher matcher = LEGACY_MODULE_IMPORT_TAIL.matcher(this.input);
+    matcher.region(this.pos, this.input.length());
+    return matcher.lookingAt();
+  }
+
   /**
    * Parse a single statement.
    *
@@ -2780,6 +2795,10 @@ protected Statement parseStatement(boolean declaration, boolean topLevel, Set<St
       return this.parseBlock(false);
     } else if (starttype == TokenType.semi) {
       return this.parseEmptyStatement(startLoc);
+    } else if (topLevel && this.isLegacyModuleImport()) {
+      if (!this.options.allowImportExportEverywhere() && !this.inModule)
+        this.raise(this.start, "Legacy module imports may appear only with 'sourceType: module'");
+      return this.parseLegacyModuleImport(startLoc);
     } else if (starttype == TokenType._export || starttype == TokenType._import) {
       if (!this.options.allowImportExportEverywhere()) {
         if (!topLevel)
@@ -3577,6 +3596,29 @@ protected Statement parseImport(Position startLoc) {
     return parseImportRest(loc);
   }
 
+  /**
+   * Parses {@code module namespace from "module-name";} as the equivalent namespace import
+   * {@code import * as namespace from "module-name";}.
+   */
+  protected ImportDeclaration parseLegacyModuleImport(Position startLoc) {
+    SourceLocation loc = new SourceLocation(startLoc);
+    this.next();
+
+    SourceLocation specifierLoc = new SourceLocation(this.startLoc);
+    Identifier local = this.parseIdent(false);
+    this.checkLVal(local, true, null);
+    this.expectContextual("from");
+    if (this.type != TokenType.string) this.unexpected();
+    Literal source = (Literal) this.parseExprAtom(null);
+    this.semicolon();
+
+    List<ImportSpecifier> specifiers = new ArrayList<ImportSpecifier>();
+    specifiers.add(this.finishNode(new ImportNamespaceSpecifier(specifierLoc, local)));
+    return this.finishNode(
+        new ImportDeclaration(
+            loc, specifiers, source, null, ImportPhaseModifier.NONE));
+  }
+
   protected Expression parseImportOrExportAttributesAndSemicolon() {
     Expression result = null;
     if (!this.eagerlyTrySemicolon()) {

javascript/extractor/tests/es2015/input/import8.js

@@ -0,0 +1 @@
+module Utils from "./utils";

javascript/extractor/tests/es2015/output/trap/import8.js.trap

@@ -0,0 +1,115 @@
+#10000=@"/import8.js;sourcefile"
+files(#10000,"/import8.js")
+#10001=@"/;folder"
+folders(#10001,"/")
+containerparent(#10001,#10000)
+#10002=@"loc,{#10000},0,0,0,0"
+locations_default(#10002,#10000,0,0,0,0)
+hasLocation(#10000,#10002)
+#20000=@"global_scope"
+scopes(#20000,0)
+#20001=@"script;{#10000},1,1"
+#20002=*
+lines(#20002,#20001,"module Utils from ""./utils"";","")
+#20003=@"loc,{#10000},1,1,1,28"
+locations_default(#20003,#10000,1,1,1,28)
+hasLocation(#20002,#20003)
+numlines(#20001,1,1,0)
+#20004=*
+tokeninfo(#20004,6,#20001,0,"module")
+#20005=@"loc,{#10000},1,1,1,6"
+locations_default(#20005,#10000,1,1,1,6)
+hasLocation(#20004,#20005)
+#20006=*
+tokeninfo(#20006,6,#20001,1,"Utils")
+#20007=@"loc,{#10000},1,8,1,12"
+locations_default(#20007,#10000,1,8,1,12)
+hasLocation(#20006,#20007)
+#20008=*
+tokeninfo(#20008,6,#20001,2,"from")
+#20009=@"loc,{#10000},1,14,1,17"
+locations_default(#20009,#10000,1,14,1,17)
+hasLocation(#20008,#20009)
+#20010=*
+tokeninfo(#20010,4,#20001,3,"""./utils""")
+#20011=@"loc,{#10000},1,19,1,27"
+locations_default(#20011,#10000,1,19,1,27)
+hasLocation(#20010,#20011)
+#20012=*
+tokeninfo(#20012,8,#20001,4,";")
+#20013=@"loc,{#10000},1,28,1,28"
+locations_default(#20013,#10000,1,28,1,28)
+hasLocation(#20012,#20013)
+#20014=*
+tokeninfo(#20014,0,#20001,5,"")
+#20015=@"loc,{#10000},1,29,1,28"
+locations_default(#20015,#10000,1,29,1,28)
+hasLocation(#20014,#20015)
+toplevels(#20001,0)
+hasLocation(#20001,#20003)
+#20016=@"var;{Utils};{#20000}"
+variables(#20016,"Utils",#20000)
+#20017=@"local_type_name;{Utils};{#20000}"
+local_type_names(#20017,"Utils",#20000)
+#20018=@"local_namespace_name;{Utils};{#20000}"
+local_namespace_names(#20018,"Utils",#20000)
+variables(#20016,"Utils",#20000)
+local_type_names(#20017,"Utils",#20000)
+local_namespace_names(#20018,"Utils",#20000)
+#20019=@"var;{this};{#20000}"
+variables(#20019,"this",#20000)
+#20020=*
+stmts(#20020,27,#20001,0,"module  ... utils"";")
+hasLocation(#20020,#20003)
+stmt_containers(#20020,#20001)
+#20021=*
+exprs(#20021,4,#20020,-1,"""./utils""")
+hasLocation(#20021,#20011)
+enclosing_stmt(#20021,#20020)
+expr_containers(#20021,#20001)
+literals("./utils","""./utils""",#20021)
+#20022=*
+regexpterm(#20022,1,#20021,0,"./utils")
+#20023=@"loc,{#10000},1,20,1,26"
+locations_default(#20023,#10000,1,20,1,26)
+hasLocation(#20022,#20023)
+#20024=*
+regexpterm(#20024,12,#20022,0,".")
+#20025=@"loc,{#10000},1,20,1,20"
+locations_default(#20025,#10000,1,20,1,20)
+hasLocation(#20024,#20025)
+#20026=*
+regexpterm(#20026,14,#20022,1,"/utils")
+#20027=@"loc,{#10000},1,21,1,26"
+locations_default(#20027,#10000,1,21,1,26)
+hasLocation(#20026,#20027)
+regexp_const_value(#20026,"/utils")
+#20028=*
+exprs(#20028,85,#20020,0,"Utils f ... utils"";")
+#20029=@"loc,{#10000},1,8,1,28"
+locations_default(#20029,#10000,1,8,1,28)
+hasLocation(#20028,#20029)
+enclosing_stmt(#20028,#20020)
+expr_containers(#20028,#20001)
+#20030=*
+exprs(#20030,78,#20028,1,"Utils")
+hasLocation(#20030,#20007)
+enclosing_stmt(#20030,#20020)
+expr_containers(#20030,#20001)
+literals("Utils","Utils",#20030)
+decl(#20030,#20016)
+typedecl(#20030,#20017)
+namespacedecl(#20030,#20018)
+#20031=*
+entry_cfg_node(#20031,#20001)
+#20032=@"loc,{#10000},1,1,1,0"
+locations_default(#20032,#10000,1,1,1,0)
+hasLocation(#20031,#20032)
+#20033=*
+exit_cfg_node(#20033,#20001)
+hasLocation(#20033,#20015)
+successor(#20020,#20033)
+successor(#20028,#20020)
+successor(#20031,#20028)
+numlines(#10000,1,1,0)
+filetype(#10000,"javascript")