Add strip_components to extract/download_and_extract `http_arch…

src/main/java/com/google/devtools/build/lib/bazel/repository/decompressor/CompressedFunction.java

@@ -30,8 +30,9 @@
  * formats while all formats that collect multiple entries inside a single (potentially compressed)
  * archive are archiver formats. This class handles the former, compressor formats.
  *
- * <p>It ignores the {@link DecompressorDescriptor#prefix()} setting because compressed files cannot
- * contain directories.
+ * <p>It ignores the {@link DecompressorDescriptor#prefix()} and {@link
+ * DecompressorDescriptor#stripComponents()} setting because compressed files cannot contain
+ * directories.
  */
 public abstract class CompressedFunction implements Decompressor {
 

src/main/java/com/google/devtools/build/lib/bazel/repository/decompressor/CompressedTarFunction.java

@@ -108,6 +108,11 @@ public Path decompress(DecompressorDescriptor descriptor)
                   "Failed to extract %s, tarred paths cannot be absolute", strippedRelativePath));
         }
 
+        strippedRelativePath = strippedRelativePath.stripComponents(descriptor.stripComponents());
+        if (strippedRelativePath == PathFragment.EMPTY_FRAGMENT) {
+          continue;
+        }
+
         Path filePath = descriptor.destinationPath().getRelative(strippedRelativePath);
         if (!filePath.startsWith(descriptor.destinationPath())) {
           throw new IOException(

src/main/java/com/google/devtools/build/lib/bazel/repository/decompressor/DecompressorDescriptor.java

@@ -33,6 +33,7 @@ public record DecompressorDescriptor(
     Path archivePath,
     Path destinationPath,
     Optional<String> prefix,
+    int stripComponents,
     ImmutableMap<String, String> renameFiles) {
   public DecompressorDescriptor {
     requireNonNull(context, "context");
@@ -45,6 +46,7 @@ public record DecompressorDescriptor(
   public static Builder builder() {
     return new AutoBuilder_DecompressorDescriptor_Builder()
         .setContext("")
+        .setStripComponents(0)
         .setRenameFiles(ImmutableMap.of());
   }
 
@@ -60,8 +62,22 @@ public abstract static class Builder {
 
     public abstract Builder setPrefix(String prefix);
 
+    public abstract Builder setStripComponents(int stripComponents);
+
     public abstract Builder setRenameFiles(Map<String, String> renameFiles);
 
-    public abstract DecompressorDescriptor build();
+    public abstract DecompressorDescriptor autoBuild();
+
+    public DecompressorDescriptor build() {
+      DecompressorDescriptor d = autoBuild();
+      if (d.stripComponents() < 0) {
+        throw new IllegalArgumentException("'strip_components' must be non-negative");
+      }
+      if (d.stripComponents() != 0 && d.prefix().isPresent() && !d.prefix().get().isEmpty()) {
+        throw new IllegalArgumentException(
+            "Only one of 'strip_prefix' or 'strip_components' can be set");
+      }
+      return d;
+    }
   }
 }

src/main/java/com/google/devtools/build/lib/bazel/repository/decompressor/SevenZDecompressor.java

@@ -86,7 +86,12 @@ public Path decompress(DecompressorDescriptor descriptor)
         if (entryPath.skip()) {
           continue;
         }
-        extract7zEntry(sevenZFile, entry, destinationDirectory, entryPath.getPathFragment());
+        PathFragment pathFragment =
+            entryPath.getPathFragment().stripComponents(descriptor.stripComponents());
+        if (pathFragment == PathFragment.EMPTY_FRAGMENT) {
+          continue;
+        }
+        extract7zEntry(sevenZFile, entry, destinationDirectory, pathFragment);
       }
 
       if (prefix.isPresent() && !foundPrefix) {

src/main/java/com/google/devtools/build/lib/bazel/repository/decompressor/ZipDecompressor.java

@@ -95,8 +95,13 @@ public Path decompress(DecompressorDescriptor descriptor)
         if (entryPath.skip()) {
           continue;
         }
+        PathFragment pathFragment =
+            entryPath.getPathFragment().stripComponents(descriptor.stripComponents());
+        if (pathFragment == PathFragment.EMPTY_FRAGMENT) {
+          continue;
+        }
         extractZipEntry(
-            reader, entry, destinationDirectory, entryPath.getPathFragment(), prefix, symlinks);
+            reader, entry, destinationDirectory, pathFragment, prefix, symlinks);
       }
 
       if (prefix.isPresent() && !foundPrefix) {

src/main/java/com/google/devtools/build/lib/bazel/repository/starlark/StarlarkBaseExternalContext.java

@@ -977,7 +977,8 @@ field is for debugging purposes only and should not be relied upon as a stable A
                 be used to strip it from extracted files.
 
                 <p>For compatibility, this parameter may also be used under the deprecated name
-                <code>stripPrefix</code>.
+                <code>stripPrefix</code>. Only one of <code>strip_prefix</code> or
+                <code>strip_components</code> can be used.
                 """),
         @Param(
             name = "allow_fail",
@@ -1043,6 +1044,16 @@ field is for debugging purposes only and should not be relied upon as a stable A
             positional = false,
             named = true,
             defaultValue = "''"),
+        @Param(
+            name = "strip_components",
+            positional = false,
+            named = true,
+            defaultValue = "0",
+            doc =
+"""
+Strip the given number of leading components from file paths on extraction. Only one of
+<code>strip_components</code> or <code>strip_prefix</code> can be used.
+"""),
       })
   public StructImpl downloadAndExtract(
       Object url,
@@ -1057,9 +1068,12 @@ public StructImpl downloadAndExtract(
       String integrity,
       Dict<?, ?> renameFiles, // <String, String> expected
       String oldStripPrefix,
+      StarlarkInt stripComponentsI,
       StarlarkThread thread)
       throws RepositoryFunctionException, InterruptedException, EvalException {
     stripPrefix = renamedStripPrefix("download_and_extract", stripPrefix, oldStripPrefix);
+    int stripComponents = Starlark.toInt(stripComponentsI, "strip_components");
+    validateStripping("download_and_extract", stripPrefix, stripComponents);
     ImmutableMap<URI, Map<String, List<String>>> authHeaders =
         getAuthHeaders(getAuthContents(authUnchecked, "auth"));
 
@@ -1162,6 +1176,7 @@ public StructImpl downloadAndExtract(
               .setArchivePath(downloadedPath)
               .setDestinationPath(outputPath.getPath())
               .setPrefix(stripPrefix)
+              .setStripComponents(stripComponents)
               .setRenameFiles(renameFilesMap)
               .build(),
           // Type does NOT need to be passed here, as the existing code renames the archive path to
@@ -1260,7 +1275,8 @@ private static void deleteTreeWithRetries(Path downloadDirectory)
                 used to strip it from extracted files.
 
                 <p>For compatibility, this parameter may also be used under the deprecated name
-                <code>stripPrefix</code>.
+                <code>stripPrefix</code>. Only one of <code>strip_prefix</code> or
+                <code>strip_components</code> can be set.
                 """),
         @Param(
             name = "rename_files",
@@ -1291,6 +1307,16 @@ private static void deleteTreeWithRetries(Path downloadDirectory)
             positional = false,
             named = true,
             defaultValue = "''"),
+        @Param(
+            name = "strip_components",
+            positional = false,
+            named = true,
+            defaultValue = "0",
+            doc =
+"""
+Strip the given number of leading components from file paths on extraction. Only one of
+<code>strip_components</code> or <code>strip_prefix</code> can be set.
+"""),
         @Param(
             name = "type",
             defaultValue = "''",
@@ -1314,10 +1340,13 @@ public void extract(
       Dict<?, ?> renameFiles, // <String, String> expected
       String watchArchive,
       String oldStripPrefix,
+      StarlarkInt stripComponentsI,
       String type,
       StarlarkThread thread)
       throws RepositoryFunctionException, InterruptedException, EvalException {
     stripPrefix = renamedStripPrefix("extract", stripPrefix, oldStripPrefix);
+    int stripComponents = Starlark.toInt(stripComponentsI, "strip_components");
+    validateStripping("extract", stripPrefix, stripComponents);
     StarlarkPath archivePath = getPath(archive);
 
     if (!archivePath.exists()) {
@@ -1355,6 +1384,7 @@ public void extract(
             .setArchivePath(archivePath.getPath())
             .setDestinationPath(outputPath.getPath())
             .setPrefix(stripPrefix)
+            .setStripComponents(stripComponents)
             .setRenameFiles(renameFilesMap)
             .build(),
         Optional.ofNullable(type).filter(s -> !s.isBlank()));
@@ -1409,6 +1439,21 @@ private static String renamedStripPrefix(String method, String stripPrefix, Stri
         method);
   }
 
+  private static void validateStripping(String method, String stripPrefix, int stripComponents)
+      throws EvalException {
+    if (stripComponents < 0) {
+      throw Starlark.errorf(
+          "%s() has an invalid argument for 'strip_components': %d. Must be non-negative.",
+          method, stripComponents);
+    }
+
+    if (!stripPrefix.isEmpty() && stripComponents > 0) {
+      throw Starlark.errorf(
+          "%s() got multiple strip values. Only one of 'strip_prefix' or 'strip_components' can be set",
+          method);
+    }
+  }
+
   @StarlarkMethod(
       name = "file",
       doc = "Generates a file in the repository directory with the provided content.",

src/main/java/com/google/devtools/build/lib/vfs/PathFragment.java

@@ -668,6 +668,21 @@ private PathFragment subFragmentImpl(int beginIndex, int endIndex) {
     return makePathFragment(normalizedPath.substring(starti, endi), driveStrLength);
   }
 
+  /** Strip <code>numComponents</code> leading components from file names on extraction. */
+  public PathFragment stripComponents(int numComponents) {
+    if (numComponents == 0) {
+      return this;
+    }
+    if (numComponents < 0) {
+      throw new IllegalArgumentException(
+          String.format("Invalid number of components (%d)", numComponents));
+    }
+    if (numComponents >= this.segmentCount()) {
+      return EMPTY_FRAGMENT;
+    }
+    return this.subFragment(numComponents);
+  }
+
   /**
    * Returns an {@link Iterable} that lazily yields the segments of this path.
    *

src/test/java/com/google/devtools/build/lib/bazel/repository/decompressor/CompressedTarFunctionTest.java

@@ -80,6 +80,19 @@ public void testDecompressWithPrefix() throws Exception {
     archiveDescriptor.assertOutputFiles(outputDir, INNER_FOLDER_NAME);
   }
 
+  /**
+   * Test decompressing a tar.gz file with hard link file and symbolic link file inside and
+   * stripping the first component.
+   */
+  @Test
+  public void testDecompressWithStripComponents() throws Exception {
+    DecompressorDescriptor.Builder descriptorBuilder =
+        archiveDescriptor.createDescriptorBuilder().setStripComponents(1);
+    Path outputDir = decompress(descriptorBuilder.build());
+
+    archiveDescriptor.assertOutputFiles(outputDir, INNER_FOLDER_NAME);
+  }
+
   /**
    * Test decompressing a tar.gz file, with some entries being renamed during the extraction
    * process.
@@ -116,6 +129,24 @@ public void testDecompressWithRenamedFilesAndPrefix() throws Exception {
     assertThat(innerDir.getRelative("renamedFile").exists()).isTrue();
   }
 
+  /** Test that entry renaming is applied prior to component stripping. */
+  @Test
+  public void testDecompressWithRenamedFilesAndStripComponents() throws Exception {
+    String innerDirName = ROOT_FOLDER_NAME + "/" + INNER_FOLDER_NAME;
+
+    HashMap<String, String> renameFiles = new HashMap<>();
+    renameFiles.put(innerDirName + "/hardLinkFile", innerDirName + "/renamedFile");
+    DecompressorDescriptor.Builder descriptorBuilder =
+        archiveDescriptor
+            .createDescriptorBuilder()
+            .setStripComponents(1)
+            .setRenameFiles(renameFiles);
+    Path outputDir = decompress(descriptorBuilder.build());
+
+    Path innerDir = outputDir.getRelative(INNER_FOLDER_NAME);
+    assertThat(innerDir.getRelative("renamedFile").exists()).isTrue();
+  }
+
   private Path decompress(DecompressorDescriptor descriptor) throws Exception {
     return new CompressedTarFunction() {
       @Override

src/test/java/com/google/devtools/build/lib/bazel/repository/decompressor/SevenZDecompressorTest.java

@@ -95,6 +95,19 @@ public void testDecompressWithPrefix() throws Exception {
     assertThat(files).contains(REGULAR_FILENAME);
   }
 
+  /** Test decompressing a .7z file and stripping components. */
+  @Test
+  public void testDecompressWithStripComponents() throws Exception {
+    DecompressorDescriptor.Builder descriptorBuilder =
+        archiveDescriptor().createDescriptorBuilder().setStripComponents(1);
+    Path outputDir = decompress(descriptorBuilder.build());
+    Path fileDir = outputDir.getRelative(INNER_FOLDER_NAME);
+
+    ImmutableList<String> files =
+        fileDir.readdir(Symlinks.NOFOLLOW).stream().map(Dirent::getName).collect(toImmutableList());
+    assertThat(files).contains(REGULAR_FILENAME);
+  }
+
   /** Test decompressing a .7z with entries being renamed during the extraction process. */
   @Test
   public void testDecompressWithRenamedFiles() throws Exception {
@@ -136,6 +149,36 @@ public void testDecompressWithRenamedFilesAndPrefix() throws Exception {
     assertThat(fileDir.getRelative("renamedFile").getFileSize()).isNotEqualTo(0);
   }
 
+  /** Test that entry renaming is applied prior to stripping components. */
+  @Test
+  public void testDecompressWithRenamedFilesAndStripComponents() throws Exception {
+    String innerDirName = ROOT_FOLDER_NAME + "/" + INNER_FOLDER_NAME;
+
+    HashMap<String, String> renameFiles = new HashMap<>();
+    renameFiles.put(innerDirName + "/" + REGULAR_FILENAME, innerDirName + "/renamedFile");
+    DecompressorDescriptor.Builder descriptorBuilder =
+        archiveDescriptor()
+            .createDescriptorBuilder()
+            .setStripComponents(1)
+            .setRenameFiles(renameFiles);
+    Path outputDir = decompress(descriptorBuilder.build());
+
+    Path fileDir = outputDir.getRelative(INNER_FOLDER_NAME);
+    ImmutableList<String> files =
+        fileDir.readdir(Symlinks.NOFOLLOW).stream().map(Dirent::getName).collect(toImmutableList());
+    assertThat(files).contains("renamedFile");
+    assertThat(fileDir.getRelative("renamedFile").getFileSize()).isNotEqualTo(0);
+  }
+
+  /** Test decompressing a .7z file where everything is stripped */
+  @Test
+  public void testDecompressStripAllComponents() throws Exception {
+    Path outputDir =
+        decompress(archiveDescriptor().createDescriptorBuilder().setStripComponents(1_000).build());
+
+    assertThat(outputDir.exists()).isFalse();
+  }
+
   private File archiveDir;
   private File extractionDir;
 

src/test/java/com/google/devtools/build/lib/bazel/repository/decompressor/ZipDecompressorTest.java

@@ -80,7 +80,7 @@ public void testDecompressWithoutPrefix() throws Exception {
   }
 
   /**
-   * Test decompressing a tar.gz file with hard link file and symbolic link file inside and
+   * Test decompressing a zip file with hard link file and symbolic link file inside and
    * stripping a prefix
    */
   @Test
@@ -93,6 +93,20 @@ public void testDecompressWithPrefix() throws Exception {
     archiveDescriptor.assertOutputFiles(outputDir, INNER_FOLDER_NAME);
   }
 
+  /**
+   * Test decompressing a zip file with hard link file and symbolic link file inside and
+   * stripping a component
+   */
+  @Test
+  public void testDecompressWithStripComponents() throws Exception {
+    TestArchiveDescriptor archiveDescriptor = new TestArchiveDescriptor(ARCHIVE_NAME, "out", false);
+    DecompressorDescriptor.Builder descriptorBuilder =
+        archiveDescriptor.createDescriptorBuilder().setStripComponents(1);
+    Path outputDir = decompress(descriptorBuilder.build());
+
+    archiveDescriptor.assertOutputFiles(outputDir, INNER_FOLDER_NAME);
+  }
+
   /**
    * Test decompressing a zip file, with some entries being renamed during the extraction process.
    */
@@ -130,6 +144,25 @@ public void testDecompressWithRenamedFilesAndPrefix() throws Exception {
     assertThat(innerDir.getRelative("renamedFile").exists()).isTrue();
   }
 
+  /** Test that entry renaming is applied prior to stripping components. */
+  @Test
+  public void testDecompressWithRenamedFilesAndStripComponents() throws Exception {
+    TestArchiveDescriptor archiveDescriptor = new TestArchiveDescriptor(ARCHIVE_NAME, "out", false);
+    String innerDirName = ROOT_FOLDER_NAME + "/" + INNER_FOLDER_NAME;
+
+    HashMap<String, String> renameFiles = new HashMap<>();
+    renameFiles.put(innerDirName + "/hardLinkFile", innerDirName + "/renamedFile");
+    DecompressorDescriptor.Builder descriptorBuilder =
+        archiveDescriptor
+            .createDescriptorBuilder()
+            .setStripComponents(1)
+            .setRenameFiles(renameFiles);
+    Path outputDir = decompress(descriptorBuilder.build());
+
+    Path innerDir = outputDir.getRelative(INNER_FOLDER_NAME);
+    assertThat(innerDir.getRelative("renamedFile").exists()).isTrue();
+  }
+
   private Path decompress(DecompressorDescriptor descriptor) throws Exception {
     return ZipDecompressor.INSTANCE.decompress(descriptor);
   }