Read relationship types

3 files changed, 96 insertions, 27 deletions

diff --git a/mammoth/docx/relationships_xml.py b/mammoth/docx/relationships_xml.py
index 8bf3a89..c30c428 100644
--- a/mammoth/docx/relationships_xml.py
+++ b/mammoth/docx/relationships_xml.py
@@ -3,21 +3,33 @@ import collections
 
 class Relationships(object):
     def __init__(self, relationships):
-        self._relationships = relationships
+        self._targets_by_id = dict(
+            (relationship.relationship_id, relationship.target)
+            for relationship in relationships
+        )
+        self._targets_by_type = collections.defaultdict(list)
+        for relationship in relationships:
+            self._targets_by_type[relationship.type].append(relationship.target)
     
     def find_target_by_relationship_id(self, key):
-        return self._relationships[key].target
+        return self._targets_by_id[key]
+    
+    def find_targets_by_type(self, relationship_type):
+        return self._targets_by_type[relationship_type]
 
 
-Relationship = collections.namedtuple("Relationship", ["target"])
+Relationship = collections.namedtuple("Relationship", ["relationship_id", "target", "type"])
 
 
 def read_relationships_xml_element(element):
     children = element.find_children("relationships:Relationship")
-    return Relationships(dict(map(_read_relationship, children)))
+    return Relationships(list(map(_read_relationship, children)))
 
 
 def _read_relationship(element):
-    relationship_id = element.attributes["Id"]
-    relationship = Relationship(target=element.attributes["Target"])
-    return relationship_id, relationship
+    relationship = Relationship(
+        relationship_id=element.attributes["Id"],
+        target=element.attributes["Target"],
+        type=element.attributes["Type"],
+    )
+    return relationship
diff --git a/tests/docx/body_xml_tests.py b/tests/docx/body_xml_tests.py
index 067037e..537f123 100644
--- a/tests/docx/body_xml_tests.py
+++ b/tests/docx/body_xml_tests.py
@@ -738,9 +738,9 @@ def _assert_children_are_converted_normally(tag_name):
 class HyperlinkTests(object):
     @istest
     def hyperlink_is_read_as_external_hyperlink_if_it_has_a_relationship_id(self):
-        relationships = Relationships({
-            "r42": Relationship(target="http://example.com")
-        })
+        relationships = Relationships([
+            _hyperlink_relationship("r42", "http://example.com"),
+        ])
         run_element = xml_element("w:r")
         element = xml_element("w:hyperlink", {"r:id": "r42"}, [run_element])
         assert_equal(
@@ -750,9 +750,9 @@ class HyperlinkTests(object):
         
     @istest
     def hyperlink_is_read_as_external_hyperlink_if_it_has_a_relationship_id_and_an_anchor(self):
-        relationships = Relationships({
-            "r42": Relationship(target="http://example.com/")
-        })
+        relationships = Relationships([
+            _hyperlink_relationship("r42", "http://example.com/"),
+        ])
         run_element = xml_element("w:r")
         element = xml_element("w:hyperlink", {"r:id": "r42", "w:anchor": "fragment"}, [run_element])
         assert_equal(
@@ -762,9 +762,9 @@ class HyperlinkTests(object):
         
     @istest
     def existing_fragment_is_replaced_when_anchor_is_set_on_external_link(self):
-        relationships = Relationships({
-            "r42": Relationship(target="http://example.com/#previous")
-        })
+        relationships = Relationships([
+            _hyperlink_relationship("r42", "http://example.com/#previous"),
+        ])
         run_element = xml_element("w:r")
         element = xml_element("w:hyperlink", {"r:id": "r42", "w:anchor": "fragment"}, [run_element])
         assert_equal(
@@ -870,9 +870,9 @@ class ImageTests(object):
     IMAGE_RELATIONSHIP_ID = "rId5"
     
     def _read_embedded_image(self, element):
-        relationships = Relationships({
-            self.IMAGE_RELATIONSHIP_ID: Relationship(target="media/hat.png")
-        })
+        relationships = Relationships([
+            _image_relationship(self.IMAGE_RELATIONSHIP_ID, "media/hat.png"),
+        ])
         
         mocks = funk.Mocks()
         docx_file = mocks.mock()
@@ -1002,9 +1002,9 @@ class ImageTests(object):
             description="It's a hat",
         )
         
-        relationships = Relationships({
-            "rId5": Relationship(target="media/hat.emf")
-        })
+        relationships = Relationships([
+            _image_relationship("rId5", "media/hat.emf"),
+        ])
         
         docx_file = mocks.mock()
         funk.allows(docx_file).open("word/media/hat.emf").returns(io.BytesIO(self.IMAGE_BYTES))
@@ -1031,9 +1031,9 @@ class ImageTests(object):
             description="It's a hat",
         )
         
-        relationships = Relationships({
-            "rId5": Relationship(target="file:///media/hat.png")
-        })
+        relationships = Relationships([
+            _image_relationship("rId5", "file:///media/hat.png"),
+        ])
         
         files = mocks.mock()
         funk.allows(files).verify("file:///media/hat.png")
@@ -1276,3 +1276,19 @@ def single(values):
         return values[0]
     else:
         raise Exception("Had {0} elements".format(len(values)))
+
+
+def _hyperlink_relationship(relationship_id, target):
+    return Relationship(
+        relationship_id=relationship_id,
+        target=target,
+        type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink",
+    )
+
+
+def _image_relationship(relationship_id, target):
+    return Relationship(
+        relationship_id=relationship_id,
+        target=target,
+        type="http://schemas.openxmlformats.org/officeDocument/2006/relationships/image",
+    )
diff --git a/tests/docx/relationships_xml_tests.py b/tests/docx/relationships_xml_tests.py
index c3d0af0..5e52142 100644
--- a/tests/docx/relationships_xml_tests.py
+++ b/tests/docx/relationships_xml_tests.py
@@ -5,16 +5,57 @@ from mammoth.docx.relationships_xml import read_relationships_xml_element
 
 
 @istest
-def target_is_read_from_relationship_element():
+def relationship_targets_can_be_found_by_id():
     element = xml_element("relationships:Relationships", {}, [
         xml_element("relationships:Relationship", {
             "Id": "rId8",
             "Type": "http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink",
             "Target": "http://example.com",
-        })
+        }),
+        xml_element("relationships:Relationship", {
+            "Id": "rId2",
+            "Type": "http://schemas.openxmlformats.org/officeDocument/2006/relationships/hyperlink",
+            "Target": "http://example.net",
+        }),
     ])
     relationships = read_relationships_xml_element(element)
     assert_equal(
         "http://example.com",
         relationships.find_target_by_relationship_id("rId8"),
     )
+
+
+@istest
+def relationship_targets_can_be_found_by_type():
+    element = xml_element("relationships:Relationships", {}, [
+        xml_element("relationships:Relationship", {
+            "Id": "rId2",
+            "Target": "docProps/core.xml",
+            "Type": "http://schemas.openxmlformats.org/package/2006/relationships/metadata/core-properties",
+        }),
+        xml_element("relationships:Relationship", {
+            "Id": "rId1",
+            "Target": "word/document.xml",
+            "Type": "http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument",
+        }),
+        xml_element("relationships:Relationship", {
+            "Id": "rId3",
+            "Target": "word/document2.xml",
+            "Type": "http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument",
+        }),
+    ])
+    relationships = read_relationships_xml_element(element)
+    assert_equal(
+        ["word/document.xml", "word/document2.xml"],
+        relationships.find_targets_by_type("http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument"),
+    )
+
+
+@istest
+def when_there_are_no_relationships_of_requested_type_then_empty_list_is_returned():
+    element = xml_element("relationships:Relationships", {}, [])
+    relationships = read_relationships_xml_element(element)
+    assert_equal(
+        [],
+        relationships.find_targets_by_type("http://schemas.openxmlformats.org/officeDocument/2006/relationships/officeDocument"),
+    )