Add function to cache ipfs files

2 files changed, 62 insertions, 1 deletions

diff --git a/gn3/file_utils.py b/gn3/file_utils.py
index c1c9d53..4dcf296 100644
--- a/gn3/file_utils.py
+++ b/gn3/file_utils.py
@@ -5,12 +5,14 @@ import os
 import random
 import string
 import tarfile
+
 from functools import partial
 from typing import Dict
 from typing import List
-
 from werkzeug.utils import secure_filename
 
+import ipfshttpclient
+
 
 def get_hash_of_files(files: List[str]) -> str:
     """Given a list of valid of FILES, return their hash as a string"""
@@ -75,3 +77,19 @@ contents to TARGET_DIR/<dir-hash>.
     except Exception:
         return {"status": 128, "error": "gzip failed to unpack file"}
     return {"status": 0, "token": token}
+
+
+def cache_ipfs_file(ipfs_file: str,
+                    cache_dir: str,
+                    ipfs_addr: str = "/ip4/127.0.0.1/tcp/5001") -> str:
+    """Check if a file exists in cache; if it doesn't, cache it.  Return the
+    cached file location
+
+    """
+    file_loc = os.path.join(cache_dir, ipfs_file.split("ipfs/")[-1])
+    if not os.path.exists(file_loc):
+        client = ipfshttpclient.connect(ipfs_addr)
+        client.get(ipfs_file,
+                   target=os.path.join(cache_dir,
+                                       ipfs_file.split("ipfs/")[-1]))
+    return file_loc
diff --git a/tests/unit/test_file_utils.py b/tests/unit/test_file_utils.py
index b319693..abe7aa6 100644
--- a/tests/unit/test_file_utils.py
+++ b/tests/unit/test_file_utils.py
@@ -8,6 +8,7 @@ from unittest import mock
 from gn3.file_utils import extract_uploaded_file
 from gn3.file_utils import get_dir_hash
 from gn3.file_utils import jsonfile_to_dict
+from gn3.file_utils import cache_ipfs_file
 
 
 @dataclass
@@ -19,6 +20,7 @@ class MockFile:
 
 class TestFileUtils(unittest.TestCase):
     """Test cases for procedures defined in file_utils.py"""
+
     def test_get_dir_hash(self):
         """Test that a directory is hashed correctly"""
         test_dir = os.path.join(os.path.dirname(__file__), "test_data")
@@ -75,3 +77,44 @@ extracting the file"""
             "status": 128,
             "error": "gzip failed to unpack file"
         })
+
+    def test_cache_ipfs_file_cache_hit(self):
+        """Test that the correct file location is returned if there's a cache hit"""
+        # Create empty file
+        test_dir = "/tmp/QmQPeNsJPyVWPFDVHb77w8G42Fvo15z4bG2X8D2GhfbSXc-test"
+        if not os.path.exists(test_dir):
+            os.mkdir(test_dir)
+        open(f"{test_dir}/genotype.txt", "a").close()
+        file_loc = cache_ipfs_file(
+            ipfs_file=("/ipfs/"
+                       "QmQPeNsJPyVWPFDVHb"
+                       "77w8G42Fvo15z4bG2X8D2GhfbSXc-test/"
+                       "genotype.txt"),
+            cache_dir="/tmp")
+        # Clean up
+        os.remove(f"{test_dir}/genotype.txt")
+        os.rmdir(test_dir)
+        self.assertEqual(file_loc, f"{test_dir}/genotype.txt")
+
+    @mock.patch("gn3.file_utils.ipfshttpclient")
+    def test_cache_ipfs_file_cache_miss(self,
+                                        mock_ipfs):
+        """Test that a file is cached if there's a cache miss"""
+        mock_ipfs_client = mock.MagicMock()
+        mock_ipfs.connect.return_value = mock_ipfs_client
+
+        test_dir = "/tmp/QmQPeNsJPyVWPFDVHb77w8G42Fvo15z4bG2X8D2GhfbSXc-test"
+        self.assertEqual(cache_ipfs_file(
+            ipfs_file=("/ipfs/"
+                       "QmQPeNsJPyVWPFDVHb"
+                       "77w8G42Fvo15z4bG2X8D2GhfbSXc-test/"
+                       "genotype.txt"),
+            cache_dir="/tmp"
+        ), f"{test_dir}/genotype.txt")
+        mock_ipfs_client.get.assert_called_once_with(
+            ("/ipfs/"
+             "QmQPeNsJPyVWPFDVHb"
+             "77w8G42Fvo15z4bG2X8D2GhfbSXc-test/"
+             "genotype.txt"),
+            target=f"{test_dir}/genotype.txt"
+        )