add new script 'getdefinition' and fix some tests

pyproject.toml

@@ -16,6 +16,7 @@ dependencies = [
 	"flit",
     "lxml >=4.8.0", #pathlike objects
     "requests >=2.6",
+	"pillow", # for exif in record
 ]
 [project.optional-dependencies]
 test = [
@@ -27,6 +28,7 @@ test = [
 mink = 'mpapi:mink'
 getAttachments = 'mpapi:getAttachments'
 getItem = 'mpapi:getItem'
+getDefinition = 'mpapi:getDefinition'
 updateItem = 'mpapi:updateItem'
 validate = 'mpapi:validate'
 

src/mpapi/__init__.py

@@ -13,6 +13,34 @@
 user, pw, baseURL = get_credentials()
 
 
+def getDefinition():
+    parser = argparse.ArgumentParser(description="getItem for MpApi")
+    parser.add_argument(
+        "-t",
+        "--mtype",
+        help="module type (Object, Multimedia etc.); optional",
+        default=None,
+    )
+    parser.add_argument("-v", "--version", help="Display version information")
+    args = parser.parse_args()
+
+    if args.version:
+        print(f"Version: {__version__}")
+        sys.exit(0)
+
+    c = MpApi(baseURL=baseURL, pw=pw, user=user)
+    print(f"Logging in as '{user}'")
+    m = c.getDefinition2(mtype=args.mtype)
+    if args.mtype is None:
+        fn = f"definition.xml"
+    else:
+        fn = f"definition-{args.mtype}.xml"
+    print(f"About to write to {fn}")
+    if Path(fn).exists():
+        print("   overwriting existing file")
+    m.toFile(path=fn)
+
+
 def getItem():
     parser = argparse.ArgumentParser(description="getItem for MpApi")
     parser.add_argument(
@@ -30,6 +58,7 @@ def getItem():
         sys.exit(0)
 
     c = MpApi(baseURL=baseURL, pw=pw, user=user)
+    print(f"Logging in as '{user}'")
     m = c.getItem2(mtype=args.mtype, ID=args.ID)
     if args.upload:
         fn = f"getItem-{args.mtype}{args.ID}u.xml"

src/mpapi/client.py

@@ -260,6 +260,7 @@ def getItem2(self, *, mtype: str, ID: int) -> Module:
         object.
         """
         r = self.getItem(module=mtype, id=ID)
+        # print(r.text)
         return Module(xml=r.text)
 
     def createItem(self, *, module: str, xml: str) -> requests.Response:

src/mpapi/module.py

@@ -81,6 +81,7 @@
 from collections import namedtuple  # experimenting with namedtuples
 from copy import deepcopy  # for lxml
 from lxml import etree  # type: ignore
+from lxml.etree import XMLSyntaxError
 from mpapi.constants import NSMAP, parser
 from mpapi.helper import Helper
 from pathlib import Path
@@ -153,11 +154,7 @@ def __init__(self, *, file: PathX = None, tree: ET = None, xml: str = None) -> N
         if tree is not None:
             self.etree = tree
         elif xml is not None:
-            # fromstring is for xml without encoding declaration
-            if isinstance(xml, bytes):
-                self.etree = etree.fromstring(xml, parser)
-            else:
-                self.etree = etree.fromstring(bytes(xml, "utf-8"), parser)
+            self.etree = self._from_xml(xml)
         elif file is not None:
             self.etree = etree.parse(str(file), parser)
         else:
@@ -168,6 +165,20 @@ def __init__(self, *, file: PathX = None, tree: ET = None, xml: str = None) -> N
             </application>"""
             self.etree = etree.fromstring(xml, parser)
 
+    def _from_xml(self, xml: str):
+        # fromstring is for xml without encoding declaration
+        if isinstance(xml, bytes):
+            try:
+                ET = etree.fromstring(xml, parser)
+            except XMLSyntaxError:
+                raise SyntaxError("Invalid module XML--!")
+        else:
+            try:
+                ET = etree.fromstring(bytes(xml, "utf-8"), parser)
+            except XMLSyntaxError:
+                raise SyntaxError("Invalid module XML!")
+        return ET
+
     def __iter__(self) -> ET:
         """
         iterates through all moduleItems, see the method iter if you want to

src/mpapi/record.py

@@ -50,7 +50,7 @@
 from lxml import etree
 from mpapi.module import Module
 from pathlib import Path
-import pyexiv2
+import PIL.Image
 
 parser = etree.XMLParser(remove_blank_text=True)
 
@@ -159,11 +159,12 @@ def set_dateexif(self, *, path: str) -> None:  # alternatively pathlib object
         p = Path(path)
         dateDT = False
         try:
-            img = pyexiv2.Image(path)
+            img = PIL.Image.open(path)
         except:
             print("WARNING: Can't access exif info")
             # mtime = p.stat().st_mtime
         else:
+            exif = img._getexif()
             try:
                 date_str = exif["Exif.Image.DateTime"]  # 2011:04:12 16:29:52
             except:

test/mk_test_data.py

@@ -0,0 +1,76 @@
+#
+# Little script that d/l the test data
+#
+
+from pathlib import Path
+from mpapi.constants import get_credentials
+from mpapi.module import Module
+from mpapi.search import Search
+from mpapi.client import MpApi
+
+user, pw, baseURL = get_credentials()
+
+
+def get_m39(c) -> None:
+    """
+    What we need is the join file from mink
+    mink -j m39
+    jobs.dsl
+    m39:
+        getPack exhibit 20222 m39
+    m39-join-exhibit20222.xml
+    """
+    fn = Path("sdata/m39-join-exhibit20222.xml")
+    if not fn.exists():
+        raise TypeError("File missing")
+
+
+def get_item(c, mtype: str, ID: int) -> None:
+    # if mtype not in allowed_types:
+    #    raise SyntaxError("Unknown type")
+    ID = int(ID)
+
+    if mtype == "Exhibition":
+        mtype2 = "exhibit"
+    else:
+        mtype2 = mtype
+
+    fn = Path(f"sdata/{mtype2}{ID}.xml")
+    if not fn.exists():
+        m = c.getItem2(mtype=mtype, ID=ID)
+        print(f"Writing to {fn}")
+        m.toFile(path=fn)
+
+
+def filter_search(c):
+    fn = "sdata/filter-search.xml"
+    if not Path(fn).exists():
+        print("filter search with limit 100...")
+        q = Search(module="Object", limit=100)
+        q.AND()
+        q.addCriterion(
+            operator="equalsField",
+            field="ObjPublicationGrp.TypeVoc",
+            value="2600647",  # Daten freigegeben für SMB-digital
+        )
+        q.addCriterion(
+            operator="equalsField",
+            field="ObjPublicationGrp.PublicationVoc",
+            value="1810139",  # Ja
+        )
+        q.addField(field="__id")
+        q.validate(mode="search")  # raises on error
+        m = c.search2(query=q)
+        print(f"Writing {fn}")
+        m.toFile(path=fn)
+
+
+if __name__ == "__main__":
+    c = MpApi(baseURL=baseURL, pw=pw, user=user)
+    print(f"Logging in as '{user}'")
+    if not Path("sdata"):  # assuming we execute in MpApi/test
+        Path.mkdir()
+    get_item(c, "Exhibition", 20222)
+    get_m39(c)
+    get_item(c, "Object", 739673)
+    filter_search(c)

test/test_module.py

@@ -11,7 +11,7 @@
 
 def test_constructors_only():
     # four different constructors, sort of
-    m: MpApi.Module = Module(file="sdata/exhibit20222.xml")
+    m: MpApi.Module = Module(file="sdata/m39-join-exhibit20222.xml")
     assert m  # len(m)>0
     xml = """
     <application xmlns="http://www.zetcom.com/ria/ws/module">
@@ -40,27 +40,35 @@ def test_nsmap():
 
 
 def test_output():
-    m = Module(file="sdata/exhibit20222.xml")
+    m = Module(file="sdata/m39-join-exhibit20222.xml")
     m.toFile(path="sdata/output.xml")  # not sure how to test
     xml = m.toString()
     assert xml  # not sure how to test
 
 
 def test_inspection():
-    m = Module(file="sdata/exhibit20222.xml")
+    # requires scaffold data
+    m = Module(file="sdata/m39-join-exhibit20222.xml")
     with pytest.raises(TypeError) as exc_info:
-        m.totalSize(module="Object")  # none
-    assert m.totalSize(module="Multimedia") == 642
+        m.totalSize(module="Objectssss")  # none
+    assert m.totalSize(module="Exhibition") == 1
     desc = m.describe()
+    # print (desc)
     assert isinstance(desc, dict)
-    assert desc == {"Multimedia": 642}
+    assert desc == {
+        "Exhibition": 1,
+        "Multimedia": 677,
+        "Object": 194,
+        "Person": 58,
+        "Registrar": 194,
+    }
     # print(desc)
     m.dropUUID()
     assert m.validate()
 
 
 def test_iter():
-    m = Module(file="sdata/exhibit20222.xml")
+    m = Module(file="sdata/m39-join-exhibit20222.xml")
     for itemN in m:
         pass  # not sure how to test
 
@@ -92,7 +100,7 @@ def test_from_scratch_interface():
 def test_drops():
     m = Module(file="sdata/exhibit20222.xml")
     uuid = m.xpath("/m:application/m:modules/m:module/m:moduleItem/@uuid")[0]
-    assert uuid == "507201"
+    assert uuid == "20222"
     m.dropUUID()
     try:
         uuid = m.xpath("/m:application/m:modules/m:module/m:moduleItem/@uuid")
@@ -103,10 +111,10 @@ def test_drops():
 
 
 def test_join():
-    m = Module(file="sdata/exhibit20222.xml")
+    m = Module(file="sdata/m39-join-exhibit20222.xml")
     MM_before = m.totalSize(module="Multimedia")
     assert isinstance(MM_before, int)
-    assert len(m) == 642
+    assert len(m) == 1124
 
     # first add
     ET = etree.parse("sdata/exhibit20222.xml")
@@ -116,15 +124,15 @@ def test_join():
     MM_after = m.totalSize(module="Multimedia")
     assert MM_before == MM_after
     # print(len(m))
-    assert len(m) == 642
+    assert len(m) == 1124
 
-    # second add
-    # print("2nd ADD")
-    ET2 = etree.parse("data/739673.xml")
+    # second add: add 2 different documents
+    ET2 = etree.parse("sdata/object739673.xml")
     Obj_before = Module(tree=ET2).actualSize(module="Object")
+    # print(f"{Obj_before=}")
     m.add(doc=ET2)
-    assert m.totalSize(module="Object") == 1
-    assert len(m) == 643
+    assert m.totalSize(module="Object") == 195
+    assert len(m) == 1125
 
     # there was a version where add(doc=ET) deleted most of the document, so now we
     # test that doc remains the "same"
@@ -137,12 +145,12 @@ def test_join():
 def test_length():
     m = Module()
     assert len(m) == 0
-    m = Module(file="sdata/exhibit20222.xml")
-    assert len(m) == 642
+    m = Module(file="sdata/m39-join-exhibit20222.xml")
+    assert len(m) == 1124
 
 
 def test_getItem():
-    m = Module(file="sdata/exhibit20222.xml")
+    m = Module(file="sdata/m39-join-exhibit20222.xml")
     # from collections import namedtuple
     # Item = namedtuple('Item', ['type', 'id'])
     itemN = m[("Multimedia", 507201)]
@@ -151,7 +159,7 @@ def test_getItem():
 
 
 def test_str():
-    m1 = Module(file="sdata/exhibit20222.xml")
+    m1 = Module(file="sdata/m39-join-exhibit20222.xml")
     assert isinstance(str(m1), str)
     # print (m1.__repr__)
 
@@ -160,19 +168,17 @@ def test_add():
     m1 = Module(file="sdata/exhibit20222.xml")
     m2 = Module(file="sdata/exhibit20222.xml")
     m3 = m1 + m2
+    m4 = Module(file="sdata/Object739673.xml")
+    m5 = m3 + m4
 
-    m1Len = len(m1)
     assert isinstance(m3, Module)
-    assert len(m3) == m1Len
+    assert len(m3) == len(m1)
     assert m3.__repr__ != m1.__repr__
-
-    m4 = Module(file="data/739673.xml")
-    m5 = m3 + m4
-    assert len(m5) == 643
+    assert len(m5) == 2
     m1 = m1 + m2  # change in place
-    assert len(m1) == 642
+    assert len(m1) == 1
     m5 = m1 + m2 + m3  # more than 2 additions
-    assert len(m5) == 642
+    assert len(m5) == 1
 
 
 def test_toZip():
@@ -198,7 +204,8 @@ def test_filter():
     # user, pw, baseURL = get_credentials()
     # client = MpApi(user=user, baseURL=baseURL, pw=pw)
     # m = client.search2(query=q)
-    m = Module(file="filter-search.xml")
+
+    m = Module(file="sdata/filter-search.xml")
     new_m = m.filter(
         xpath="/m:application/m:modules/m:module/m:moduleItem[@hasAttachments = 'true']"
     )