allow to use md5-cached dirs

.gitignore

@@ -289,3 +289,4 @@ tira-web/tira-web/lib/
 
 # TextMate
 frontend/.editorconfig
+python-client/tests/resources/local_cached_zip/extracted_datasets/

python-client/tests/pd_load_data_test.py

@@ -19,3 +19,37 @@ def test_pd_load_truths_wows_re_ranking(self):
         self.assertEqual(3, len(actual))
         self.assertEqual("hubble telescope achievements", actual.iloc[0]["query"])
         self.assertEqual("1", actual.iloc[0]["qid"])
+
+    def test_pd_load_truths_local_cached_zip(self):
+        tira = Client(tira_cache_dir="tests/resources/local_cached_zip")
+        actual = tira.pd.truths("task-does-not-exist", "dataset-does-not-exist-20241201-training")
+        self.assertEqual(13, len(actual))
+        first_line = actual.iloc[0].to_dict()
+        last_line = actual.iloc[12].to_dict()
+
+        self.assertEqual("5a8865b0-19d7-4b33-bbe3-2f64ad54557f", first_line["id"])
+        self.assertEqual("1051399", first_line["query_id"])
+        self.assertEqual(3376628, first_line["unknown_doc_id"])
+        self.assertEqual(0, first_line["qrel_unknown_doc"])
+
+        self.assertEqual("449f69fa-df0e-4c9e-aead-61983ce9eaa8", last_line["id"])
+        self.assertEqual("833860", last_line["query_id"])
+        self.assertEqual(2830558, last_line["unknown_doc_id"])
+        self.assertEqual(0, last_line["qrel_unknown_doc"])
+
+    def test_pd_load_inputs_local_cached_zip(self):
+        tira = Client(tira_cache_dir="tests/resources/local_cached_zip")
+        actual = tira.pd.inputs("task-does-not-exist", "dataset-does-not-exist-20241201-training")
+        self.assertEqual(13, len(actual))
+        first_line = actual.iloc[0].to_dict()
+        last_line = actual.iloc[12].to_dict()
+
+        self.assertEqual("5a8865b0-19d7-4b33-bbe3-2f64ad54557f", first_line["id"])
+        self.assertTrue("query_id" not in first_line)
+        self.assertTrue("unknown_doc_id" not in first_line)
+        self.assertTrue("qrel_unknown_doc" not in first_line)
+
+        self.assertEqual("449f69fa-df0e-4c9e-aead-61983ce9eaa8", last_line["id"])
+        self.assertTrue("query_id" not in last_line)
+        self.assertTrue("unknown_doc_id" not in last_line)
+        self.assertTrue("qrel_unknown_doc" not in last_line)

python-client/tests/resources/local_cached_zip/.archived/v1/datasets/all

@@ -0,0 +1,29 @@
+[
+{
+    "id": "dataset-does-not-exist-20241201-training",
+    "dataset_id": "dataset-does-not-exist-20241201-training",
+    "default_task": "task-does-not-exist",
+    "default_task_name": "task-does-not-exist",
+    "display_name": "Display Name",
+    "is_confidential": false,
+    "is_deprecated": false,
+    "mirrors": {
+      "truths": {
+        "Zenodo": "URL does not exist"
+      },
+      "inputs": {
+        "Zenodo": "URL does not exist"
+      }
+    },
+    "dataset_extraction": {
+      "truths": {
+        "md5sum": "608b4b658c190d9c3bd840d43653f021",
+        "subdirectory": "pointwise/labels"
+      },
+      "inputs": {
+        "md5sum": "608b4b658c190d9c3bd840d43653f021",
+        "subdirectory": "pointwise/inputs"
+      }
+    }
+  }
+]

python-client/tira/rest_api_client.py

@@ -3,6 +3,8 @@
 import json
 import logging
 import os
+import shutil
+import tempfile
 import time
 import zipfile
 from functools import lru_cache
@@ -533,6 +535,8 @@ def download_dataset(self, task, dataset, truth_dataset=False):
         data_type = "training" if dataset.endswith("-training") else "test"
         suffix = "inputs" if not truth_dataset else "truths"
         url = None
+        expected_md5 = None
+        subdirectory = None
         if (
             not meta_data
             or "mirrors" not in meta_data
@@ -543,6 +547,10 @@ def download_dataset(self, task, dataset, truth_dataset=False):
         else:
             url = list(meta_data["mirrors"][suffix].values())[0]
 
+            if "dataset_extraction" in meta_data and suffix in meta_data["dataset_extraction"]:
+                expected_md5 = meta_data["dataset_extraction"][suffix]["md5sum"]
+                subdirectory = meta_data["dataset_extraction"][suffix]["subdirectory"]
+
         target_dir = f"{self.tira_cache_dir}/extracted_datasets/{task}/{dataset}/"
         suffix = "input-data" if not truth_dataset else "truth-data"
         if os.path.isdir(target_dir + suffix):
@@ -551,9 +559,12 @@ def download_dataset(self, task, dataset, truth_dataset=False):
         if not url:
             url = f'{self.base_url}/data-download/{data_type}/input-{("" if not truth_dataset else "truth")}/{dataset}.zip'
 
-        self.download_and_extract_zip(url, target_dir)
+        if expected_md5 and subdirectory:
+            self.download_and_extract_zip_with_md5(url, target_dir + suffix, expected_md5, subdirectory)
+        else:
+            self.download_and_extract_zip(url, target_dir)
 
-        os.rename(target_dir + f"/{dataset}", target_dir + suffix)
+            os.rename(target_dir + f"/{dataset}", target_dir + suffix)
 
         return target_dir + suffix
 
@@ -626,6 +637,35 @@ def evaluate_run(self, team, dataset, run_id):
 
         return ret
 
+    def download_and_extract_zip_with_md5(self, url, target_dir, expected_md5, subdirectory):
+        if expected_md5 is None or not expected_md5:
+            raise ValueError("foo")
+
+        if not (Path(self.tira_cache_dir) / ".archived" / expected_md5).exists():
+            raise ValueError("foo")
+
+        z = zipfile.ZipFile((Path(self.tira_cache_dir) / ".archived" / expected_md5))
+
+        members_to_extract = []
+        for i in z.namelist():
+            if i and not i.endswith("/") and (not subdirectory or i.startswith(subdirectory)):
+                members_to_extract.append(i)
+
+        if len(members_to_extract) == 0:
+            raise ValueError("I found no files in te zip.")
+
+        with tempfile.TemporaryDirectory() as tmpdirname:
+            for i in members_to_extract:
+                z._extract_member(i, Path(tmpdirname), pwd=None)
+
+            src_dir = Path(tmpdirname)
+            if subdirectory:
+                src_dir = src_dir / subdirectory
+            Path(target_dir).parent.mkdir(exist_ok=True, parents=True)
+            shutil.move(src=src_dir, dst=target_dir)
+
+        return
+
     def download_and_extract_zip(self, url, target_dir, extract=True):
         url = redirects(url=url)["urls"][0]
         if url.split("://")[1].startswith("files.webis.de"):