From 435df84f49737938c1bde44e6ef59c0cf2c9fcad Mon Sep 17 00:00:00 2001
From: hugosousa <hmosousa@gmail.com>
Date: Mon, 22 Aug 2022 17:54:51 +0100
Subject: [PATCH] Added KRAUTS partitions. Closes #5

---
 tests/test_datasets/test_download_and_read.py | 36 +++++++++++++++++++
 tieval/datasets/metadata.py                   | 26 ++++++++++++++
 2 files changed, 62 insertions(+)

diff --git a/tests/test_datasets/test_download_and_read.py b/tests/test_datasets/test_download_and_read.py
index e86a65c..061ffa3 100644
--- a/tests/test_datasets/test_download_and_read.py
+++ b/tests/test_datasets/test_download_and_read.py
@@ -115,6 +115,42 @@ def test_download_and_read_krauts(tmp_path):
                 assert timex.text == doc.text[s: e]
 
 
+def test_download_and_read_krauts_diezeit(tmp_path):
+
+    os.chdir(tmp_path)
+
+    download("krauts_diezeit")
+    data_path = tmp_path / "data/krauts_diezeit"
+    assert data_path.is_dir()
+
+    krauts = read("krauts_diezeit")
+    assert len(krauts.documents) == 50
+
+
+def test_download_and_read_krauts_dolomiten_42(tmp_path):
+
+    os.chdir(tmp_path)
+
+    download("krauts_dolomiten_42")
+    data_path = tmp_path / "data/krauts_dolomiten_42"
+    assert data_path.is_dir()
+
+    krauts = read("krauts_dolomiten_42")
+    assert len(krauts.documents) == 42
+
+
+def test_download_and_read_krauts_dolomiten_100(tmp_path):
+
+    os.chdir(tmp_path)
+
+    download("krauts_dolomiten_100")
+    data_path = tmp_path / "data/krauts_dolomiten_100"
+    assert data_path.is_dir()
+
+    krauts = read("krauts_dolomiten_100")
+    assert len(krauts.documents) == 100
+
+
 def test_download_and_read_matres(tmp_path):
 
     os.chdir(tmp_path)
diff --git a/tieval/datasets/metadata.py b/tieval/datasets/metadata.py
index 726af66..f8faa33 100644
--- a/tieval/datasets/metadata.py
+++ b/tieval/datasets/metadata.py
@@ -94,6 +94,32 @@ def description(self):
         doc_reader=KRAUTSDocumentReader
     ),
 
+    "krauts_diezeit": DatasetMetadata(
+        name="krauts_diezeit",
+        language="german",
+        url="https://drive.inesctec.pt/s/f98ZDGLqEjqCXEp/download",
+        reader=XMLDatasetReader,
+        doc_reader=KRAUTSDocumentReader
+    ),
+
+
+    "krauts_dolomiten_42": DatasetMetadata(
+        name="krauts_dolomiten_42",
+        language="german",
+        url="https://drive.inesctec.pt/s/AEwAod8Wn2L9E9G/download",
+        reader=XMLDatasetReader,
+        doc_reader=KRAUTSDocumentReader
+    ),
+
+
+    "krauts_dolomiten_100": DatasetMetadata(
+        name="krauts_dolomiten_100",
+        language="german",
+        url="https://drive.inesctec.pt/s/x3N7Kf446RFeDse/download",
+        reader=XMLDatasetReader,
+        doc_reader=KRAUTSDocumentReader
+    ),
+
     "matres": DatasetMetadata(
         name="matres",
         language="english",