From afcf4c3e2e88a285c445600d774d444f606a047f Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Mon, 2 May 2022 16:35:31 +0200
Subject: [PATCH 1/6] is_url: allow empty netloc

---
 rocrate/utils.py   |  2 +-
 test/test_model.py |  8 ++++++++
 test/test_write.py | 18 ++++++++++++++++++
 3 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/rocrate/utils.py b/rocrate/utils.py
index 7768804..ad02894 100644
--- a/rocrate/utils.py
+++ b/rocrate/utils.py
@@ -47,7 +47,7 @@ def as_list(list_or_other):
 
 def is_url(string):
     parts = urlsplit(string)
-    return all((parts.scheme, parts.netloc, parts.path))
+    return all((parts.scheme, parts.path))
 
 
 def iso_now():
diff --git a/test/test_model.py b/test/test_model.py
index 3bdc539..6f2620d 100644
--- a/test/test_model.py
+++ b/test/test_model.py
@@ -151,6 +151,14 @@ def test_contextual_entities_hash(test_data_dir):
     }))
     wf["hasPart"] = [step]
     assert step.id == step_id
+    email = "jscarberry@example.org"
+    email_uri = f"mailto:{email}"
+    contact_point = crate.add(ContextEntity(crate, email_uri, properties={
+        "@type": "ContactPoint",
+        "email": email
+    }))
+    crate.root_dataset["contactPoint"] = contact_point
+    assert contact_point.id == email_uri
 
 
 def test_properties():
diff --git a/test/test_write.py b/test/test_write.py
index 8aaefc6..c0f51c2 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -160,6 +160,24 @@ def test_remote_uri(tmpdir, helpers, fetch_remote, validate_url, to_zip):
             assert "sdDatePublished" in props
 
 
+def test_file_uri(tmpdir):
+    f_name = uuid.uuid4().hex
+    f_path = tmpdir / f_name
+    f_uri = f"file://{f_path}"
+    with open(f_path, "wt") as f:
+        f.write("FOO\n")
+    crate = ROCrate()
+    f_entity = crate.add_file(f_uri, fetch_remote=True)
+    assert f_entity.id == f_name
+
+    out_path = tmpdir / 'ro_crate_out'
+    crate.write(out_path)
+
+    out_crate = ROCrate(out_path)
+    assert out_crate.dereference(f_name) is not None
+    assert (out_path / f_name).is_file()
+
+
 @pytest.mark.slow
 @pytest.mark.parametrize("fetch_remote", [False, True])
 def test_ftp_uri(tmpdir, fetch_remote):

From b02e3f25b34d689e37c7d8a6597026faac4e0ddc Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Mon, 2 May 2022 17:37:43 +0200
Subject: [PATCH 2/6] fix possible unwanted file creation

---
 rocrate/model/file.py | 27 ++++++++++++++-------------
 test/test_write.py    | 25 ++++++++++++++++++++++++-
 2 files changed, 38 insertions(+), 14 deletions(-)

diff --git a/rocrate/model/file.py b/rocrate/model/file.py
index 3a1d30d..263677d 100644
--- a/rocrate/model/file.py
+++ b/rocrate/model/file.py
@@ -45,19 +45,20 @@ def write(self, base_path):
             mode = 'w' + ('b' if isinstance(self.source, BytesIO) else 't')
             with open(out_file_path, mode) as out_file:
                 out_file.write(self.source.getvalue())
-        elif is_url(str(self.source)) and (self.fetch_remote or self.validate_url):
-            with urllib.request.urlopen(self.source) as response:
-                if self.validate_url:
-                    if isinstance(response, HTTPResponse):
-                        self._jsonld.update({
-                            'contentSize': response.getheader('Content-Length'),
-                            'encodingFormat': response.getheader('Content-Type')
-                        })
-                    if not self.fetch_remote:
-                        self._jsonld['sdDatePublished'] = iso_now()
-                if self.fetch_remote:
-                    out_file_path.parent.mkdir(parents=True, exist_ok=True)
-                    urllib.request.urlretrieve(response.url, out_file_path)
+        elif is_url(str(self.source)):
+            if self.fetch_remote or self.validate_url:
+                with urllib.request.urlopen(self.source) as response:
+                    if self.validate_url:
+                        if isinstance(response, HTTPResponse):
+                            self._jsonld.update({
+                                'contentSize': response.getheader('Content-Length'),
+                                'encodingFormat': response.getheader('Content-Type')
+                            })
+                        if not self.fetch_remote:
+                            self._jsonld['sdDatePublished'] = iso_now()
+                    if self.fetch_remote:
+                        out_file_path.parent.mkdir(parents=True, exist_ok=True)
+                        urllib.request.urlretrieve(response.url, out_file_path)
         elif os.path.isfile(self.source):
             out_file_path.parent.mkdir(parents=True, exist_ok=True)
             if not out_file_path.exists() or not out_file_path.samefile(self.source):
diff --git a/test/test_write.py b/test/test_write.py
index c0f51c2..d66275a 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -162,7 +162,7 @@ def test_remote_uri(tmpdir, helpers, fetch_remote, validate_url, to_zip):
 
 def test_file_uri(tmpdir):
     f_name = uuid.uuid4().hex
-    f_path = tmpdir / f_name
+    f_path = (tmpdir / f_name).resolve()
     f_uri = f"file://{f_path}"
     with open(f_path, "wt") as f:
         f.write("FOO\n")
@@ -178,6 +178,29 @@ def test_file_uri(tmpdir):
     assert (out_path / f_name).is_file()
 
 
+def test_looks_like_file_uri(tmpdir, monkeypatch):
+    f_name = uuid.uuid4().hex
+    f_parent = (tmpdir / "file:")
+    f_parent.mkdir()
+    f_path = f_parent / f_name
+    with open(f_path, "wt") as f:
+        f.write("FOO\n")
+    monkeypatch.chdir(tmpdir)
+    crate = ROCrate()
+    # Missing if interpreted as URI, present if intepreted as path
+    uri = f"file:/{f_name}"
+    entity = crate.add_file(uri, fetch_remote=False)
+    assert entity.id == uri
+
+    out_path = tmpdir / 'ro_crate_out'
+    crate.write(out_path)
+
+    out_crate = ROCrate(out_path)
+    assert out_crate.dereference(uri) is not None
+    assert not (out_path / f_name).is_file()
+    assert not (out_path / uri).is_file()
+
+
 @pytest.mark.slow
 @pytest.mark.parametrize("fetch_remote", [False, True])
 def test_ftp_uri(tmpdir, fetch_remote):

From 17a0737e3f083b8644ea1eed32eca21b520148f7 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Mon, 2 May 2022 17:46:49 +0200
Subject: [PATCH 3/6] check that paths starting with "file:" can be added

---
 test/test_write.py | 11 +++++++++++
 1 file changed, 11 insertions(+)

diff --git a/test/test_write.py b/test/test_write.py
index d66275a..c5e4b46 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -200,6 +200,17 @@ def test_looks_like_file_uri(tmpdir, monkeypatch):
     assert not (out_path / f_name).is_file()
     assert not (out_path / uri).is_file()
 
+    # Check that the file can be added to the crate using its absolute path
+    entity = crate.add_file(f_path.resolve())
+    assert entity.id == f_name
+
+    out_path = tmpdir / 'ro_crate_out_updated'
+    crate.write(out_path)
+
+    out_crate = ROCrate(out_path)
+    assert out_crate.dereference(f_name) is not None
+    assert (out_path / f_name).is_file()
+
 
 @pytest.mark.slow
 @pytest.mark.parametrize("fetch_remote", [False, True])

From 36452934b00bff0df34e2f0cc4e383be00b82411 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Wed, 4 May 2022 16:06:59 +0200
Subject: [PATCH 4/6] windows-specific fixes

---
 rocrate/utils.py   | 2 ++
 test/test_write.py | 4 +++-
 2 files changed, 5 insertions(+), 1 deletion(-)

diff --git a/rocrate/utils.py b/rocrate/utils.py
index ad02894..010b99a 100644
--- a/rocrate/utils.py
+++ b/rocrate/utils.py
@@ -47,6 +47,8 @@ def as_list(list_or_other):
 
 def is_url(string):
     parts = urlsplit(string)
+    if os.name == "nt" and len(parts.scheme) == 1:
+        return False
     return all((parts.scheme, parts.path))
 
 
diff --git a/test/test_write.py b/test/test_write.py
index c5e4b46..d249b8b 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -18,6 +18,7 @@
 
 import io
 import pytest
+import os
 import uuid
 import zipfile
 from itertools import product
@@ -163,7 +164,7 @@ def test_remote_uri(tmpdir, helpers, fetch_remote, validate_url, to_zip):
 def test_file_uri(tmpdir):
     f_name = uuid.uuid4().hex
     f_path = (tmpdir / f_name).resolve()
-    f_uri = f"file://{f_path}"
+    f_uri = f"file://{f_path.as_posix()}"
     with open(f_path, "wt") as f:
         f.write("FOO\n")
     crate = ROCrate()
@@ -178,6 +179,7 @@ def test_file_uri(tmpdir):
     assert (out_path / f_name).is_file()
 
 
+@pytest.mark.skipif(os.name != "posix", reason="':' not allowed in dir name")
 def test_looks_like_file_uri(tmpdir, monkeypatch):
     f_name = uuid.uuid4().hex
     f_parent = (tmpdir / "file:")

From ffec5d6858fd467dedc4bbbe08fc2b1175121649 Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Wed, 4 May 2022 16:33:23 +0200
Subject: [PATCH 5/6] more tweaking for windows

---
 test/test_write.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/test/test_write.py b/test/test_write.py
index d249b8b..4615c6b 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -164,7 +164,7 @@ def test_remote_uri(tmpdir, helpers, fetch_remote, validate_url, to_zip):
 def test_file_uri(tmpdir):
     f_name = uuid.uuid4().hex
     f_path = (tmpdir / f_name).resolve()
-    f_uri = f"file://{f_path.as_posix()}"
+    f_uri = f"file://{f_path.as_posix().split(':', 1)[-1]}"
     with open(f_path, "wt") as f:
         f.write("FOO\n")
     crate = ROCrate()

From e5d10ba940c82fc806ac3691b8ff1af25abf587a Mon Sep 17 00:00:00 2001
From: simleo <simleo@crs4.it>
Date: Wed, 4 May 2022 17:29:07 +0200
Subject: [PATCH 6/6] one more for windows

---
 test/test_write.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/test/test_write.py b/test/test_write.py
index 4615c6b..b806ae5 100644
--- a/test/test_write.py
+++ b/test/test_write.py
@@ -164,7 +164,7 @@ def test_remote_uri(tmpdir, helpers, fetch_remote, validate_url, to_zip):
 def test_file_uri(tmpdir):
     f_name = uuid.uuid4().hex
     f_path = (tmpdir / f_name).resolve()
-    f_uri = f"file://{f_path.as_posix().split(':', 1)[-1]}"
+    f_uri = f"file:///{f_path}"  # extra slash needed on some windows systems
     with open(f_path, "wt") as f:
         f.write("FOO\n")
     crate = ROCrate()