All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.google.javascript.jscomp.instrumentation.reporter.ReportDecoder Maven / Gradle / Ivy

Go to download

Closure Compiler is a JavaScript optimizing compiler. It parses your JavaScript, analyzes it, removes dead code and rewrites and minimizes what's left. It also checks syntax, variable references, and types, and warns about common JavaScript pitfalls. It is used in many of Google's JavaScript apps, including Gmail, Google Web Search, Google Maps, and Google Docs.

There is a newer version: v20240317
Show newest version
/*
 * Copyright 2020 The Closure Compiler Authors.
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */

package com.google.javascript.jscomp.instrumentation.reporter;

import static com.google.common.base.Preconditions.checkState;
import static java.nio.charset.StandardCharsets.UTF_8;
import static java.util.Comparator.comparing;

import com.google.common.base.Splitter;
import com.google.common.io.CharStreams;
import com.google.debugging.sourcemap.Base64VLQ;
import com.google.gson.Gson;
import com.google.gson.GsonBuilder;
import com.google.gson.reflect.TypeToken;
import com.google.javascript.jscomp.instrumentation.reporter.proto.FileProfile;
import com.google.javascript.jscomp.instrumentation.reporter.proto.InstrumentationPoint;
import com.google.javascript.jscomp.instrumentation.reporter.proto.InstrumentationPointStats;
import com.google.javascript.jscomp.instrumentation.reporter.proto.InstrumentationPointStats.Presence;
import com.google.javascript.jscomp.instrumentation.reporter.proto.ReportProfile;
import java.io.IOException;
import java.lang.reflect.Type;
import java.nio.file.Files;
import java.nio.file.Paths;
import java.util.HashMap;
import java.util.HashSet;
import java.util.LinkedHashMap;
import java.util.List;
import java.util.Map;
import java.util.regex.Matcher;
import java.util.regex.Pattern;
import java.util.stream.Collectors;
import java.util.stream.Stream;

/**
 * Class that helps to decode reports received from JS binaries that have been instrumented. Report
 * is usually a map of String => Long. Where key is encoded instrumentation point and value is
 * number of times that code was executed. To decode report beside report itself we need mapping
 * file that was produced when JS binary was compiled.
 *
 * 

Expected usage: {@code * Map mapping = ReportDecoder.parseMappingFromFile("mapping.txt"); * Map encodedReport = readFile("report.txt"); * ReportProfile report = ReportDecoder.decodeReport(mapping, encodedReport); * } * */ public final class ReportDecoder { private ReportDecoder() {} public static Map parseMappingFromFile(String mappingFilePath) throws IOException { String fileContent = CharStreams.toString(Files.newBufferedReader(Paths.get(mappingFilePath), UTF_8)); return ReportDecoder.parseMapping(fileContent); } /** * Parses the file found at location mappingFilePath and populates the properties of this class */ public static Map parseMapping(String mappingFileContent) { Map mapping = new LinkedHashMap<>(); List linesOfFile = Splitter.on('\n').omitEmptyStrings().splitToList(mappingFileContent); // Report should contain at least three lines checkState(linesOfFile.size() >= 3, "Malformed report %s", linesOfFile); String fileNamesLine = linesOfFile.get(0).trim(); String functionNamesLine = linesOfFile.get(1).trim(); String typesLine = linesOfFile.get(2).trim(); checkState(fileNamesLine.startsWith("FileNames:")); checkState(functionNamesLine.startsWith("FunctionNames:")); checkState(typesLine.startsWith("Types:")); String fileNamesAsJsonArray = fileNamesLine.substring(fileNamesLine.indexOf(":") + 1); String functionNamesAsJsonArray = functionNamesLine.substring(functionNamesLine.indexOf(":") + 1); String typesAsJsonArray = typesLine.substring(typesLine.indexOf(":") + 1); Type stringListType = new TypeToken>() {}.getType(); Gson gson = new GsonBuilder().disableHtmlEscaping().create(); List localFileNames = gson.fromJson(fileNamesAsJsonArray, stringListType); List functionNames = gson.fromJson(functionNamesAsJsonArray, stringListType); List typesAsStringList = gson.fromJson(typesAsJsonArray, stringListType); List types = typesAsStringList.stream() .map(InstrumentationPoint.Type::valueOf) .collect(Collectors.toList()); for (int i = 3; i < linesOfFile.size(); ++i) { String lineItem = linesOfFile.get(i); String id = lineItem.substring(0, lineItem.indexOf(':')); String encodedDetails = lineItem.substring(lineItem.indexOf(':') + 1); StringCharIterator encodedDetailsAsCharIt = new StringCharIterator(encodedDetails); InstrumentationPoint temp = InstrumentationPoint.newBuilder() .setFileName(localFileNames.get(Base64VLQ.decode(encodedDetailsAsCharIt))) .setFunctionName(functionNames.get(Base64VLQ.decode(encodedDetailsAsCharIt))) .setType(types.get(Base64VLQ.decode(encodedDetailsAsCharIt))) .setLineNumber(Base64VLQ.decode(encodedDetailsAsCharIt)) .setColumnNumber(Base64VLQ.decode(encodedDetailsAsCharIt)) .build(); mapping.putIfAbsent(id, temp); } return mapping; } public static ReportProfile decodeReport( Map mapping, Map frequencies) { // Decode each entry into InstrumentionPoint. Stream instrumentationPoints = mapping.entrySet().stream() .map( (entry) -> InstrumentationPointStats.newBuilder() .setPoint(entry.getValue()) .setTimesExecuted(frequencies.getOrDefault(entry.getKey(), 0L)) .build()); return ReportDecoder.createReportProfile(instrumentationPoints); } /** * This function builds a report for given JS binary analyzing its content. If an instrumentation * point is present in the JS binary - it will be marked as executed once. If the instrumentation * point was removed by the compiler as dead code - it will be marked as executed zero times. This * function doesn't give 100% guarantee for finding unused code but it provides good enough * approximation to be useful. * * @param mapping Parsed instrumentation mapping. Use {@link #parseMapping} function to create it. * @param fileContent Content of a JS binary compiled with production instrumentation enabled. * @param instrumentationArrayName Name of the array that was passed as * --production_instrumentation_array_name flag when compiling JS binary. * @return Report containing all instrumentation points. If a point times_executed is zero - it * means the function or branch was removed as dead code. If it's one - it's present in the JS * binary. */ public static ReportProfile createProfileOfStaticallyUsedCode( Map mapping, String fileContent, String instrumentationArrayName) { Matcher matcher = Pattern.compile(instrumentationArrayName + "\\.push\\(['\"](.*?)['\"]\\)") .matcher(fileContent); final HashSet encodedPoints = new HashSet<>(); while (matcher.find()) { encodedPoints.add(matcher.group(1)); } Stream instrumentationPoints = mapping.entrySet().stream() .map( (entry) -> InstrumentationPointStats.newBuilder() .setPoint(entry.getValue()) .setTimesExecuted(0) .setPointPresence( encodedPoints.contains(entry.getKey()) ? Presence.PRESENT : Presence.STATICALLY_REMOVED) .build()); return ReportDecoder.createReportProfile(instrumentationPoints); } /** * Given list of profiles merges them, summing up execution counts for instrumentation points used * across multiple profiles. Be aware that instrumentation points are compared as equality so * profiles should come from JS binaries compiled with at the same commit, otherwise changes in * source code will cause differences in some instrumentation points. */ public static ReportProfile mergeProfiles(List profiles) { Map frequencies = new HashMap<>(); Map presences = new HashMap<>(); for (ReportProfile profile : profiles) { for (FileProfile fileProfile : profile.getFileProfileList()) { for (InstrumentationPointStats pointStats : fileProfile.getInstrumentationPointsStatsList()) { frequencies.compute( pointStats.getPoint(), (p, existing) -> (existing == null ? 0 : existing) + pointStats.getTimesExecuted()); // Logic of merging presence field. // 1. PRESENT overrides STATICALLY_REMOVED and UNKNOWN. // 2. STATICALLY_REMOVED overrides UNKNOWN. presences.compute( pointStats.getPoint(), (p, existing) -> (pointStats.getPointPresence() == Presence.PRESENT || existing == Presence.PRESENCE_UNKNOWN || existing == null) ? pointStats.getPointPresence() : existing); } } } Stream finalPoints = frequencies.entrySet().stream() .map( (Map.Entry entry) -> InstrumentationPointStats.newBuilder() .setPoint(entry.getKey()) .setTimesExecuted(entry.getValue()) .setPointPresence(presences.get(entry.getKey())) .build()); return ReportDecoder.createReportProfile(finalPoints); } /** * Groups instrumentation points by files and convert them to FileProfile object, one per file and * return as ReportProfile. */ private static ReportProfile createReportProfile( Stream instrumentationPoints) { List fileProfiles = instrumentationPoints .sorted( comparing( (InstrumentationPointStats pointStats) -> pointStats.getPoint().getFileName()) .thenComparingInt( (InstrumentationPointStats pointStats) -> pointStats.getPoint().getLineNumber())) .collect( Collectors.groupingBy( (InstrumentationPointStats pointStats) -> pointStats.getPoint().getFileName())) .entrySet() .stream() .map( (entry) -> FileProfile.newBuilder() .setFileName(entry.getKey()) .addAllInstrumentationPointsStats(entry.getValue()) .build()) .sorted(comparing(FileProfile::getFileName)) .collect(Collectors.toList()); return ReportProfile.newBuilder().addAllFileProfile(fileProfiles).build(); } /** * A implementation of the Base64VLQ CharIterator used for decoding the mappings encoded in the * JSON string. */ private static class StringCharIterator implements Base64VLQ.CharIterator { final String content; final int length; int current = 0; StringCharIterator(String content) { this.content = content; this.length = content.length(); } @Override public char next() { return content.charAt(current++); } @Override public boolean hasNext() { return current < length; } } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy