better GAV extraction with base path argument

Cornul11 · Mar 1, 2024 · cab741b · cab741b
1 parent 42c5e0c
commit cab741b
Show file tree

Hide file tree

Showing 4 changed files with 27 additions and 19 deletions.
diff --git a/src/main/java/nl/tudelft/cornul11/thesis/corpus/file/JarAndPomInfoExtractor.java b/src/main/java/nl/tudelft/cornul11/thesis/corpus/file/JarAndPomInfoExtractor.java
@@ -5,31 +5,32 @@ public class JarAndPomInfoExtractor {
     String artifactId;
     String version;
 
-    public JarAndPomInfoExtractor(String jarFilePath) {
-        parseJarOrPomFilePath(jarFilePath);
+    public JarAndPomInfoExtractor(String jarFilePath, String basePath) {
+        parseJarOrPomFilePath(jarFilePath, basePath);
     }
 
-    // TODO: this has to be thoroughly tested, maybe also crosscheck with the pom file for certainty
-    private void parseJarOrPomFilePath(String jarFilePath) {
+    private void parseJarOrPomFilePath(String jarFilePath, String basePath) {
+        jarFilePath = jarFilePath.replace("\\", "/");
+        if (basePath != null) {
+            basePath = basePath.replace("\\", "/").endsWith("/") ? basePath : basePath + "/";
+            if (!basePath.isEmpty() && jarFilePath.startsWith(basePath)) {
+                jarFilePath = jarFilePath.substring(basePath.length());
+            }
+        } else {
+            String defaultBase = ".m2/repository/";
+            int index = jarFilePath.indexOf(defaultBase);
+            if (index != -1) {
+                jarFilePath = jarFilePath.substring(index + defaultBase.length());
+            }
+        }
+
         String[] splitPath = jarFilePath.split("/");
 
         String version = splitPath[splitPath.length - 2];
         String artifactID = splitPath[splitPath.length - 3];
 
-        // Find the position of '.m2/repository' by comparing adjacent elements in the 'splitPath' array
-        int repoIndex = -1;
-        for (int i = 0; i < splitPath.length - 1; i++) {
-            if (splitPath[i].equals(".m2") && splitPath[i + 1].equals("repository")) {
-                repoIndex = i;
-                break;
-            }
-        }
-
-        // Start building the 'groupId' from the position just after '.m2/repository'
-        int startIdx = repoIndex + 2;
-
         StringBuilder groupID = new StringBuilder();
-        for (int i = startIdx; i < splitPath.length - 3; i++) {
+        for (int i = 0; i < splitPath.length - 3; i++) {
             groupID.append(splitPath[i]);
             if (i < splitPath.length - 4) {
                 groupID.append('.');

diff --git a/src/main/java/nl/tudelft/cornul11/thesis/corpus/jarfile/FileAnalyzer.java b/src/main/java/nl/tudelft/cornul11/thesis/corpus/jarfile/FileAnalyzer.java
@@ -27,11 +27,13 @@ public class FileAnalyzer {
     private final AtomicInteger processedJars = new AtomicInteger(0);
     private final int totalJars;
     private final long startTime = System.currentTimeMillis();
+    private final String basePath;
 
     public FileAnalyzer(SignatureDAO signatureDao, ConfigurationLoader config) {
         this.config = config;
         this.signatureDao = signatureDao;
         this.totalJars = config.getTotalJars();
+        this.basePath = config.getBasePath();
     }
 
     public void printIgnoredUberJars() {
@@ -57,7 +59,7 @@ public int processJarFile(Path jarFilePath) {
         long jarCrc = jarHandler.getJarCrc();
         long jarCreationDate = jarHandler.getJarCreationDate();
 
-        JarAndPomInfoExtractor jarAndPomInfoExtractor = new JarAndPomInfoExtractor(jarFilePath.toString());
+        JarAndPomInfoExtractor jarAndPomInfoExtractor = new JarAndPomInfoExtractor(jarFilePath.toString(), basePath);
         if (signatures.isEmpty()) { // it's probably an uber-JAR, let's still add it to the db
             insertedUberJars.incrementAndGet();
             return commitLibrary(jarAndPomInfoExtractor, jarHash, jarCrc, jarHandler.isBrokenJar(), jarCreationDate);

diff --git a/src/main/java/nl/tudelft/cornul11/thesis/corpus/util/ConfigurationLoader.java b/src/main/java/nl/tudelft/cornul11/thesis/corpus/util/ConfigurationLoader.java
@@ -1,4 +1,5 @@
 package nl.tudelft.cornul11.thesis.corpus.util;
+
 import nl.tudelft.cornul11.thesis.corpus.database.DatabaseConfig;
 
 import java.io.FileInputStream;
@@ -62,6 +63,10 @@ public int getTotalJars() {
         return config.getProperty("totalJars") == null ? -1 : Integer.parseInt(config.getProperty("totalJars"));
     }
 
+    public String getBasePath() {
+        return config.getProperty("basePath") == null ? null : config.getProperty("basePath");
+    }
+
     public String getMongoDbDatabase() {
         return config.getProperty("mongoDbDatabase");
     }

diff --git a/src/main/java/nl/tudelft/cornul11/thesis/oracle/OracleInformationComparator.java b/src/main/java/nl/tudelft/cornul11/thesis/oracle/OracleInformationComparator.java
@@ -254,7 +254,7 @@ private List<String> inferLibraries(Path jarPath) throws Exception {
     }
 
     private String getGAVFromPath(Path jarPath) {
-        JarAndPomInfoExtractor jarAndPomInfoExtractor = new JarAndPomInfoExtractor(jarPath.toString());
+        JarAndPomInfoExtractor jarAndPomInfoExtractor = new JarAndPomInfoExtractor(jarPath.toString(), null);
         String groupId = jarAndPomInfoExtractor.getGroupId();
         String artifactId = jarAndPomInfoExtractor.getArtifactId();
         String version = jarAndPomInfoExtractor.getVersion();