Add Tag Files functions (#65)

* Add addTagFile() * Add removeTagFile and tests * Build pushed feature branches * Update action versions * Fix messages * Cover the new private function shared by addTagFile and removeTagFile
whikloj · Apr 17, 2024 · b831fc9 · b831fc9
1 parent d91cf60
commit b831fc9
Show file tree

Hide file tree

Showing 9 changed files with 533 additions and 37 deletions.
diff --git a/.github/workflows/v5.yml b/.github/workflows/v5.yml
@@ -4,6 +4,9 @@ on:
   pull_request:
     branches:
       - "v5"
+  push:
+    branches:
+      - "feature/v5/*"
 
 jobs:
   build:
@@ -19,7 +22,7 @@ jobs:
     name: PHP ${{ matrix.php-versions }} - OS ${{ matrix.host-os }}
     steps:
       - name: Checkout
-        uses: actions/checkout@v3
+        uses: actions/checkout@v4
 
       - name: Setup PHP
         uses: shivammathur/setup-php@v2
@@ -35,7 +38,7 @@ jobs:
         if: ${{ startsWith( matrix.host-os , 'ubuntu') }}
 
       - name: Cache dependencies (Ubuntu)
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.composercache-ubuntu.outputs.dir }}
           key: ${{ runner.os }}-composer-${{ hashFiles('**/composer.lock') }}
@@ -50,7 +53,7 @@ jobs:
         if: ${{ startsWith( matrix.host-os , 'windows') }}
 
       - name: Cache dependencies (Windows)
-        uses: actions/cache@v3
+        uses: actions/cache@v4
         with:
           path: ${{ steps.composercache-windows.outputs.dir }}
           key: ${{ runner.os }}-composer-${{ hashFiles('**/composer.lock') }}
@@ -68,7 +71,7 @@ jobs:
         run: composer phpunit
 
       - name: Upload coverage to Codecov
-        uses: codecov/codecov-action@v3
+        uses: codecov/codecov-action@v4
         with:
           file: ./clover.xml
           fail_ci_if_error: true

diff --git a/composer.json b/composer.json
@@ -50,7 +50,7 @@
       "./vendor/bin/phpcpd --suffix='.php' src"
     ],
     "phpunit": [
-      "phpdbg -qrr ./vendor/bin/phpunit -d memory_limit=-1 --testsuite BagIt"
+      "phpdbg -qrr ./vendor/bin/phpunit -d memory_limit=-1 --verbose --testsuite BagIt"
     ],
     "test": [
       "@check",

diff --git a/src/Bag.php b/src/Bag.php
@@ -670,9 +670,7 @@ public function removeBagInfoTagValue(string $tag, string $value, bool $case_sen
      */
     public function addBagInfoTag(string $tag, string $value): void
     {
-        if (!$this->isExtended) {
-            throw new BagItException("This bag is not extended, you need '\$bag->setExtended(true);'");
-        }
+        $this->setExtended(true);
         $internal_tag = self::trimLower($tag);
         if (in_array($internal_tag, self::BAG_INFO_GENERATED_ELEMENTS)) {
             throw new BagItException("Field $tag is auto-generated and cannot be manually set.");
@@ -690,11 +688,9 @@ public function addBagInfoTag(string $tag, string $value): void
      */
     public function addBagInfoTags(array $tags): void
     {
-        if (!$this->isExtended) {
-            throw new BagItException("This bag is not extended, you need '\$bag->setExtended(true);'");
-        }
+        $this->setExtended(true);
         $normalized_keys = array_keys($tags);
-        $normalized_keys = array_map('self::trimLower', $normalized_keys);
+        $normalized_keys = array_map(self::class . '::trimLower', $normalized_keys);
         $overlap = array_intersect($normalized_keys, self::BAG_INFO_GENERATED_ELEMENTS);
         if (count($overlap) !== 0) {
             throw new BagItException(
@@ -892,10 +888,10 @@ public function setAlgorithm(string $algorithm): void
      */
     public function setAlgorithms(array $algorithms): void
     {
-        $internal_names = array_map('self::getHashName', $algorithms);
+        $internal_names = array_map(self::class . '::getHashName', $algorithms);
         $valid_algorithms = array_filter($internal_names, [$this, 'hashIsSupported']);
         if (count($valid_algorithms) !== count($algorithms)) {
-            throw new BagItException("One or more of the algorithms provided are supported.");
+            throw new BagItException("One or more of the algorithms provided are NOT supported.");
         }
         $this->setAlgorithmsInternal($valid_algorithms);
     }
@@ -1152,16 +1148,14 @@ public function pathInBagData(string $filepath): bool
      *
      * @param string $path
      *   The file just deleted.
+     * @throws FilesystemException If we can't delete the directory.
+     * @throws BagItException If the directory is outside the data directory.
      */
     public function checkForEmptyDir(string $path): void
     {
         $parentPath = dirname($path);
         if (str_starts_with($this->makeRelative($parentPath), "data/")) {
-            $files = scandir($parentPath);
-            $payload = array_diff($files, [".", ".."]);
-            if (count($payload) == 0) {
-                rmdir($parentPath);
-            }
+            BagUtils::deleteEmptyDirTree($parentPath, $this->getDataDirectory());
         }
     }
 
@@ -1193,10 +1187,84 @@ public function upgrade(): void
         $this->update();
     }
 
+    /**
+     * Add a special tag file to the bag.
+     * @param string $source Full path to the tag file.
+     * @param string $dest Relative path for the destination.
+     *
+     * @throws BagItException Various errors related to the source and destination locations and access.
+     * @throws FilesystemException Issues reading from or writing to the filesystem.
+     */
+    public function addTagFile(string $source, string $dest): void
+    {
+        if (!file_exists($source) || !is_file($source) || !is_readable($source)) {
+            throw new BagItException("$source does not exist, is not a file or is not readable.");
+        }
+        $this->checkTagFileConstraints($dest);
+        $external = $this->makeAbsolute($dest);
+        if (file_exists($external)) {
+            throw new BagItException("Tag file ($dest) already exists in the bag.");
+        }
+        $this->setExtended(true);
+        $parentDirs = dirname($external);
+        if ($parentDirs !== $this->getBagRoot() && !file_exists($parentDirs)) {
+            // Create any missing tag file directories.
+            BagUtils::checkedMkdir($parentDirs, 0777, true);
+        }
+        BagUtils::checkedCopy($source, $external);
+        $this->changed = true;
+    }
+
+    /**
+     * Remove a tag file and any empty directories it leaves behind.
+     * @param string $dest The relative path to the tag file.
+     * @return void
+     * @throws BagItException If the file does not exist, is not inside the bag root or is a reserved file.
+     * @throws FilesystemException If there are issues deleting the file or directories.
+     */
+    public function removeTagFile(string $dest): void
+    {
+        $this->checkTagFileConstraints($dest);
+        $external = $this->makeAbsolute($dest);
+        if (!file_exists($external)) {
+            throw new BagItException("Tag file ($dest) does not exist in the bag.");
+        }
+        BagUtils::checkedUnlink($external);
+        BagUtils::deleteEmptyDirTree(dirname($external), $this->getBagRoot());
+        $this->changed = true;
+    }
+
     /*
      *  XXX: Private functions
      */
 
+    /**
+     * Common checks for interactions with custom tag files.
+     * @param string $tagFilePath The relative path to the tag file.
+     * @return void
+     * @throws BagItException If the tag file is not in the bag root, is in the data directory, or is a reserved file.
+     */
+    private function checkTagFileConstraints(string $tagFilePath): void
+    {
+        $external = $this->makeAbsolute($tagFilePath);
+        $relativePath = $this->makeRelative($external);
+        if ($relativePath === "") {
+            throw new BagItException("Tag files must be inside the bag root.");
+        }
+        if (str_starts_with(strtolower($relativePath), "data/")) {
+            throw new BagItException("Tag files must be in the bag root or a tag file directory, " .
+                "use ->addFile() to add data files.");
+        }
+        if (in_array(strtolower($relativePath), ['bagit.txt', 'bag-info.txt', 'fetch.txt'])) {
+            throw new BagItException("You cannot alter reserved file ($tagFilePath) file with your own tag file.");
+        } elseif (
+            str_starts_with(strtolower($relativePath), 'tagmanifest-') ||
+            str_starts_with(strtolower($relativePath), 'manifest-')
+        ) {
+            throw new BagItException("You cannot alter a manifest or tag manifest file with your own tag file.");
+        }
+    }
+
     /**
      * Load a bag from disk.
      *

diff --git a/src/BagUtils.php b/src/BagUtils.php
@@ -5,6 +5,7 @@
 namespace whikloj\BagItTools;
 
 use TypeError;
+use whikloj\BagItTools\Exceptions\BagItException;
 use whikloj\BagItTools\Exceptions\FilesystemException;
 
 /**
@@ -324,6 +325,18 @@ public static function checkedFwrite($fp, string $content): void
         }
     }
 
+    /**
+     * Remove a directory and check if it succeeded.
+     * @param string $path The path to remove.
+     * @throws FilesystemException If the call to rmdir() fails.
+     */
+    public static function checkedRmDir(string $path): void
+    {
+        if (!@rmdir($path)) {
+            throw new FilesystemException("Unable to remove directory $path");
+        }
+    }
+
     /**
      * Decode a file path according to the special rules of the spec.
      *
@@ -410,4 +423,32 @@ public static function standardizePathSeparators(string $path): string
     {
         return str_replace('\\', '/', $path);
     }
+
+    /**
+     * Walk up a path as far as the rootDir and delete empty directories.
+     * @param string $path The path to check.
+     * @param string $rootDir The root to not remove .
+     *
+     * @throws BagItException If the path is not within the bag root.
+     * @throws FilesystemException If we can't remove a directory
+     */
+    public static function deleteEmptyDirTree(string $path, string $rootDir): void
+    {
+        if (rtrim(strtolower($path), '/') === rtrim(strtolower($rootDir), '/')) {
+            return;
+        }
+        if (!str_starts_with($path, $rootDir)) {
+            throw new BagItException("Path is not within the root directory.");
+        }
+        if (file_exists($path) && is_dir($path)) {
+            $parent = dirname($path);
+            $files = array_diff(scandir($path), [".", ".."]);
+            if (count($files) === 0) {
+                self::checkedRmDir($path);
+            }
+            if ($parent !== $rootDir) {
+                self::deleteEmptyDirTree($parent, $rootDir);
+            }
+        }
+    }
 }
diff --git a/tests/BagTest.php b/tests/BagTest.php
@@ -583,7 +583,7 @@ public function testSetAlgorithmsFailure(): void
         $bag = Bag::create($this->tmpdir);
         $this->assertArrayEquals(['sha512'], $bag->getAlgorithms());
         $this->expectException(BagItException::class);
-        $this->expectExceptionMessage("One or more of the algorithms provided are supported.");
+        $this->expectExceptionMessage("One or more of the algorithms provided are NOT supported.");
         $bag->setAlgorithms(['sha1', 'SHA-224', "bad-algorithm"]);
     }
 
@@ -1013,17 +1013,17 @@ public function testFailOnEncodedBagIt(): void
     }
 
     /**
-     * Test that for a non-extended bag, trying to add bag-info tags throws an error.
+     * Test that for a non-extended bag, trying to add bag-info tags no longer throws an error.
      * @group Bag
      * @covers ::addBagInfoTag
      */
     public function testAddBagInfoWhenNotExtended(): void
     {
-        $this->expectException(BagItException::class);
-        $this->expectExceptionMessage("This bag is not extended, you need '\$bag->setExtended(true);'");
-
         $bag = Bag::create($this->tmpdir);
-        $bag->addBagInfoTag("Contact-Name", "Jared Whiklo");
+        $this->assertFalse($bag->isExtended());
+        $bag->addBagInfoTag("Contact-Name", "Bob Smith");
+        $this->assertTrue($bag->isExtended());
+        $this->assertArrayEquals(["Bob Smith"], $bag->getBagInfoByTag("Contact-Name"));
     }
 
     /**