Add HASH tei support as documented

lib/teiparse/__init__.py

@@ -19,6 +19,33 @@ def check_tei_type(type: str, debug: bool):
         print("DEBUG: TEI type not supported: {}".format(type))
     return False
 
+def valid_hash(hashtype: str, teihash: str, debug: bool):
+    """Check if the TEI hash URN is valid.
+
+    Sample syntax:
+    urn:tei:hash:teapot.example.com:sha256:00480065006C006C006F00200077006F0072006C00640021
+
+    the hash value needs to be in hex code
+    """
+    htype = [
+        "sha256",
+        "sha384",
+        "sha512"
+    ]
+    if hashtype not in htype:
+        if debug:
+            print("DEBUG: Invalid hash type: {}\n".format(hashtype))
+        return True
+    try:
+        val = int(teihash, 16)
+    except ValueError:
+        if debug:
+            print("DEBUG: Hash is not a hex value: {}\n".format(teihash))
+        return False
+    if debug:
+        print("DEBUG: Valid TEI hash {}: {}".format(hashtype, teihash))
+    return True
+
 
 def valid_purl(purl: str, debug: bool):
     """Check if the PURL is valid.
@@ -79,12 +106,30 @@ def valid(
         return False
     if not check_tei_type(urn.specific_string.parts[0], debug):
         return False
+    # Check the length of the domain part
+    domlen = len(urn.specific_string.parts[1])
+    if domlen == 0:
+        if debug:
+            print("ERROR: No domain part given.")
+            return False
     if urn.specific_string.parts[0] == "purl":
         if debug:
             print("DEBUG: Checking PURL syntax")
         # Calculate where PURL begins
         start = len("urn:tei:purl:")
-        start += len(urn.specific_string.parts[1]) + 1
+        #remove the domain
+        start += domlen + 1
         if not valid_purl(tei[start:], debug):
             return False
+        return True
+    if urn.specific_string.parts[0] == "hash":
+        # Hash type tei:hash:<domain>:hashtype:<hash>
+        hashtype = urn.specific_string.parts[2]
+        if not valid_hash(
+            hashtype=hashtype,
+            teihash = urn.specific_string.parts[3],
+            debug = debug):
+            return False
+        return True
+
     return True

pytest/test_tei.py

@@ -117,10 +117,7 @@ def test_bad_02(capsys, request):
 
 def test_bad_03(capsys, request):
     """Test bad TEI with bad type"""
-    import os
-    import sys
     import re
-    sys.path.append(os.path.join(os.path.dirname(__file__), '..'))
     from lib.teiparse import valid
 
     tei = "urn:tei:gurka:prod2.example.com:" \
@@ -136,3 +133,62 @@ def test_bad_03(capsys, request):
     assert re.search(
         "DEBUG: TEI type not supported",
         captured.out) is not None
+
+
+def test_bad_03(capsys, request):
+    """Test bad TEI with no domain."""
+    import re
+    from lib.teiparse import valid
+
+    tei = "urn:tei:hash::" \
+        "7bdb4424-612f-11ef-947e-1a52914d44b3"
+
+    tvalid = valid(tei, True)
+    captured = capsys.readouterr()
+    with capsys.disabled():
+        print(
+            "\nDEBUG {}: output: \n{}\n"
+            .format(request.node.name, captured.out))
+    assert tvalid is False
+    assert re.search(
+        "ERROR: No domain part given.",
+        captured.out) is not None
+
+
+def test_hash_00(capsys, request):
+    """Test with invalid hash."""
+    import re
+    from lib.teiparse import valid
+
+    tei = "urn:tei:hash:prod2.example.com:sha256:" \
+        "7bdb4424-612f-11ef-947e-1a52914d44b3"
+
+    tvalid = valid(tei, True)
+    captured = capsys.readouterr()
+    with capsys.disabled():
+        print(
+            "\nDEBUG {}: output: \n{}\n"
+            .format(request.node.name, captured.out))
+    assert tvalid is False
+    assert re.search(
+        "Hash is not a hex value",
+        captured.out) is not None
+
+def test_hash_01(capsys, request):
+    """Test with valid hash."""
+    import re
+    from lib.teiparse import valid
+
+    tei = "urn:tei:hash:prod2.example.com:sha256:" \
+        "726e11c73d62bbc3b7c3f68d25fe98dd8d8de90fb159cfa0a624efed5437435f"
+
+    tvalid = valid(tei, True)
+    captured = capsys.readouterr()
+    with capsys.disabled():
+        print(
+            "\nDEBUG {}: output: \n{}\n"
+            .format(request.node.name, captured.out))
+    assert tvalid is True
+    assert re.search(
+        "DEBUG: Valid TEI hash sha256:",
+        captured.out) is not None