TS: declassify files with unrecognized shebang line

javascript/extractor/src/com/semmle/js/extractor/FileExtractor.java

@@ -5,6 +5,7 @@
 import java.io.FileInputStream;
 import java.io.FileReader;
 import java.io.IOException;
+import java.nio.charset.Charset;
 import java.util.LinkedHashSet;
 import java.util.Set;
 import java.util.regex.Pattern;
@@ -38,6 +39,11 @@ public class FileExtractor {
 	 */
 	public static final Pattern JSON_OBJECT_START = Pattern.compile("^(?s)\\s*\\{\\s*\"([^\"]|\\\\.)*\"\\s*:.*");
 
+	/**
+	 * The charset for decoding UTF-8 strings.
+	 */
+	private static final Charset UTF8_CHARSET = Charset.forName("UTF-8");
+
 	/**
 	 * Information about supported file types.
 	 */
@@ -169,6 +175,11 @@ private boolean hasBadFileHeader(File f, String lcExt, ExtractorConfig config) {
 					if (isXml(bytes, length))
 						return true;
 
+					// Avoid files with an unrecognized shebang header.
+					if (hasUnrecognizedShebang(bytes, length)) {
+						return true;
+					}
+
 					return false;
 				} catch (IOException e) {
 					Exceptions.ignore(e, "Let extractor handle this one.");
@@ -249,6 +260,38 @@ private boolean hasUnprintableUtf8(byte[] bytes, int length) {
 				return false;
 			}
 
+			/**
+			 * Returns true if the byte sequence starts with a shebang line that is not
+			 * recognized as a JavaScript interpreter.
+			 */
+			private boolean hasUnrecognizedShebang(byte[] bytes, int length) {
+				// Shebangs preceded by a BOM aren't recognized in UNIX, but the BOM might only
+				// be present in the source file, to be stripped out in the build process.
+				int startIndex = skipBOM(bytes, length);
+				if (startIndex + 2 >= length) return false;
+				if (bytes[startIndex] != '#' || bytes[startIndex + 1] != '!') {
+					return false;
+				}
+				int endOfLine = -1;
+				for (int i = startIndex; i < length; ++i) {
+					if (bytes[i] == '\r' || bytes[i] == '\n') {
+						endOfLine = i;
+						break;
+					}
+				}
+				if (endOfLine == -1) {
+					// The shebang is either very long or there are no other lines in the file.
+					// Treat this as unrecognized.
+					return true;
+				}
+				// Extract the shebang text
+				int startOfText = startIndex + "#!".length();
+				int lengthOfText = endOfLine - startOfText;
+				String text = new String(bytes, startOfText, lengthOfText, UTF8_CHARSET);
+				// Check if the shebang is a recognized JavaScript intepreter.
+				return !NODE_INVOCATION.matcher(text).find();
+			}
+
 			@Override
 			public IExtractor mkExtractor(ExtractorConfig config, ExtractorState state) {
 				return new TypeScriptExtractor(config, state.getTypeScriptParser());

javascript/extractor/tests/shebang/input/not-typescript.ts

@@ -0,0 +1,5 @@
+#!/usr/bin/env perl
+
+use strict;
+
+exit 0;

javascript/extractor/tests/shebang/input/typescript-with-shebang.ts

@@ -0,0 +1,4 @@
+#!/usr/bin/env node
+interface Foo {
+  x: number;
+}

javascript/extractor/tests/shebang/input/typescript.ts

@@ -0,0 +1,3 @@
+interface Foo {
+  x: number;
+}

javascript/extractor/tests/shebang/options.json

@@ -0,0 +1,3 @@
+{
+  "typescript": true
+}

javascript/extractor/tests/shebang/output/trap/typescript-with-shebang.ts.trap

@@ -0,0 +1,129 @@
+#10000=@"/typescript-with-shebang.ts;sourcefile"
+files(#10000,"/typescript-with-shebang.ts","typescript-with-shebang","ts",0)
+#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"
+toplevels(#20001,0)
+#20002=@"loc,{#10000},1,1,5,0"
+locations_default(#20002,#10000,1,1,5,0)
+hasLocation(#20001,#20002)
+#20003=@"local_type_name;{Foo};{#20000}"
+local_type_names(#20003,"Foo",#20000)
+#20004=*
+stmts(#20004,34,#20001,0,"#!/usr/ ... mber;\n}")
+#20005=@"loc,{#10000},1,1,4,1"
+locations_default(#20005,#10000,1,1,4,1)
+hasLocation(#20004,#20005)
+stmtContainers(#20004,#20001)
+#20006=*
+typeexprs(#20006,1,#20004,0,"Foo")
+#20007=@"loc,{#10000},2,11,2,13"
+locations_default(#20007,#10000,2,11,2,13)
+hasLocation(#20006,#20007)
+enclosingStmt(#20006,#20004)
+exprContainers(#20006,#20001)
+literals("Foo","Foo",#20006)
+typedecl(#20006,#20003)
+#20008=*
+properties(#20008,#20004,2,8,"x: number;")
+#20009=@"loc,{#10000},3,3,3,12"
+locations_default(#20009,#10000,3,3,3,12)
+hasLocation(#20008,#20009)
+#20010=*
+exprs(#20010,0,#20008,0,"x")
+#20011=@"loc,{#10000},3,3,3,3"
+locations_default(#20011,#10000,3,3,3,3)
+hasLocation(#20010,#20011)
+enclosingStmt(#20010,#20004)
+exprContainers(#20010,#20001)
+literals("x","x",#20010)
+isAbstractMember(#20008)
+#20012=*
+typeexprs(#20012,2,#20008,2,"number")
+#20013=@"loc,{#10000},3,6,3,11"
+locations_default(#20013,#10000,3,6,3,11)
+hasLocation(#20012,#20013)
+enclosingStmt(#20012,#20004)
+exprContainers(#20012,#20001)
+literals("number","number",#20012)
+#20014=*
+lines(#20014,#20001,"#!/usr/bin/env node","
+")
+#20015=@"loc,{#10000},1,1,1,19"
+locations_default(#20015,#10000,1,1,1,19)
+hasLocation(#20014,#20015)
+#20016=*
+lines(#20016,#20001,"interface Foo {","
+")
+#20017=@"loc,{#10000},2,1,2,15"
+locations_default(#20017,#10000,2,1,2,15)
+hasLocation(#20016,#20017)
+#20018=*
+lines(#20018,#20001,"  x: number;","
+")
+#20019=@"loc,{#10000},3,1,3,12"
+locations_default(#20019,#10000,3,1,3,12)
+hasLocation(#20018,#20019)
+indentation(#10000,3," ",2)
+#20020=*
+lines(#20020,#20001,"}","
+")
+#20021=@"loc,{#10000},4,1,4,1"
+locations_default(#20021,#10000,4,1,4,1)
+hasLocation(#20020,#20021)
+numlines(#20001,4,3,0)
+#20022=*
+tokeninfo(#20022,7,#20001,0,"interface")
+#20023=@"loc,{#10000},2,1,2,9"
+locations_default(#20023,#10000,2,1,2,9)
+hasLocation(#20022,#20023)
+#20024=*
+tokeninfo(#20024,6,#20001,1,"Foo")
+hasLocation(#20024,#20007)
+#20025=*
+tokeninfo(#20025,8,#20001,2,"{")
+#20026=@"loc,{#10000},2,15,2,15"
+locations_default(#20026,#10000,2,15,2,15)
+hasLocation(#20025,#20026)
+#20027=*
+tokeninfo(#20027,6,#20001,3,"x")
+hasLocation(#20027,#20011)
+#20028=*
+tokeninfo(#20028,8,#20001,4,":")
+#20029=@"loc,{#10000},3,4,3,4"
+locations_default(#20029,#10000,3,4,3,4)
+hasLocation(#20028,#20029)
+#20030=*
+tokeninfo(#20030,7,#20001,5,"number")
+hasLocation(#20030,#20013)
+#20031=*
+tokeninfo(#20031,8,#20001,6,";")
+#20032=@"loc,{#10000},3,12,3,12"
+locations_default(#20032,#10000,3,12,3,12)
+hasLocation(#20031,#20032)
+#20033=*
+tokeninfo(#20033,8,#20001,7,"}")
+hasLocation(#20033,#20021)
+#20034=*
+tokeninfo(#20034,0,#20001,8,"")
+#20035=@"loc,{#10000},5,1,5,0"
+locations_default(#20035,#10000,5,1,5,0)
+hasLocation(#20034,#20035)
+#20036=*
+entry_cfg_node(#20036,#20001)
+#20037=@"loc,{#10000},1,1,1,0"
+locations_default(#20037,#10000,1,1,1,0)
+hasLocation(#20036,#20037)
+#20038=*
+exit_cfg_node(#20038,#20001)
+hasLocation(#20038,#20035)
+successor(#20004,#20038)
+successor(#20036,#20004)
+numlines(#10000,4,3,0)
+filetype(#10000,"typescript")

javascript/extractor/tests/shebang/output/trap/typescript.ts.trap

@@ -0,0 +1,123 @@
+#10000=@"/typescript.ts;sourcefile"
+files(#10000,"/typescript.ts","typescript","ts",0)
+#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"
+toplevels(#20001,0)
+#20002=@"loc,{#10000},1,1,4,0"
+locations_default(#20002,#10000,1,1,4,0)
+hasLocation(#20001,#20002)
+#20003=@"local_type_name;{Foo};{#20000}"
+local_type_names(#20003,"Foo",#20000)
+#20004=*
+stmts(#20004,34,#20001,0,"interfa ... mber;\n}")
+#20005=@"loc,{#10000},1,1,3,1"
+locations_default(#20005,#10000,1,1,3,1)
+hasLocation(#20004,#20005)
+stmtContainers(#20004,#20001)
+#20006=*
+typeexprs(#20006,1,#20004,0,"Foo")
+#20007=@"loc,{#10000},1,11,1,13"
+locations_default(#20007,#10000,1,11,1,13)
+hasLocation(#20006,#20007)
+enclosingStmt(#20006,#20004)
+exprContainers(#20006,#20001)
+literals("Foo","Foo",#20006)
+typedecl(#20006,#20003)
+#20008=*
+properties(#20008,#20004,2,8,"x: number;")
+#20009=@"loc,{#10000},2,3,2,12"
+locations_default(#20009,#10000,2,3,2,12)
+hasLocation(#20008,#20009)
+#20010=*
+exprs(#20010,0,#20008,0,"x")
+#20011=@"loc,{#10000},2,3,2,3"
+locations_default(#20011,#10000,2,3,2,3)
+hasLocation(#20010,#20011)
+enclosingStmt(#20010,#20004)
+exprContainers(#20010,#20001)
+literals("x","x",#20010)
+isAbstractMember(#20008)
+#20012=*
+typeexprs(#20012,2,#20008,2,"number")
+#20013=@"loc,{#10000},2,6,2,11"
+locations_default(#20013,#10000,2,6,2,11)
+hasLocation(#20012,#20013)
+enclosingStmt(#20012,#20004)
+exprContainers(#20012,#20001)
+literals("number","number",#20012)
+#20014=*
+lines(#20014,#20001,"interface Foo {","
+")
+#20015=@"loc,{#10000},1,1,1,15"
+locations_default(#20015,#10000,1,1,1,15)
+hasLocation(#20014,#20015)
+#20016=*
+lines(#20016,#20001,"  x: number;","
+")
+#20017=@"loc,{#10000},2,1,2,12"
+locations_default(#20017,#10000,2,1,2,12)
+hasLocation(#20016,#20017)
+indentation(#10000,2," ",2)
+#20018=*
+lines(#20018,#20001,"}","
+")
+#20019=@"loc,{#10000},3,1,3,1"
+locations_default(#20019,#10000,3,1,3,1)
+hasLocation(#20018,#20019)
+numlines(#20001,3,3,0)
+#20020=*
+tokeninfo(#20020,7,#20001,0,"interface")
+#20021=@"loc,{#10000},1,1,1,9"
+locations_default(#20021,#10000,1,1,1,9)
+hasLocation(#20020,#20021)
+#20022=*
+tokeninfo(#20022,6,#20001,1,"Foo")
+hasLocation(#20022,#20007)
+#20023=*
+tokeninfo(#20023,8,#20001,2,"{")
+#20024=@"loc,{#10000},1,15,1,15"
+locations_default(#20024,#10000,1,15,1,15)
+hasLocation(#20023,#20024)
+#20025=*
+tokeninfo(#20025,6,#20001,3,"x")
+hasLocation(#20025,#20011)
+#20026=*
+tokeninfo(#20026,8,#20001,4,":")
+#20027=@"loc,{#10000},2,4,2,4"
+locations_default(#20027,#10000,2,4,2,4)
+hasLocation(#20026,#20027)
+#20028=*
+tokeninfo(#20028,7,#20001,5,"number")
+hasLocation(#20028,#20013)
+#20029=*
+tokeninfo(#20029,8,#20001,6,";")
+#20030=@"loc,{#10000},2,12,2,12"
+locations_default(#20030,#10000,2,12,2,12)
+hasLocation(#20029,#20030)
+#20031=*
+tokeninfo(#20031,8,#20001,7,"}")
+hasLocation(#20031,#20019)
+#20032=*
+tokeninfo(#20032,0,#20001,8,"")
+#20033=@"loc,{#10000},4,1,4,0"
+locations_default(#20033,#10000,4,1,4,0)
+hasLocation(#20032,#20033)
+#20034=*
+entry_cfg_node(#20034,#20001)
+#20035=@"loc,{#10000},1,1,1,0"
+locations_default(#20035,#10000,1,1,1,0)
+hasLocation(#20034,#20035)
+#20036=*
+exit_cfg_node(#20036,#20001)
+hasLocation(#20036,#20033)
+successor(#20004,#20036)
+successor(#20034,#20004)
+numlines(#10000,3,3,0)
+filetype(#10000,"typescript")