[OPT-1018] DrawioToDict improvements - refactor

sl_util/sl_util/xml_to_json.py → sl_util/sl_util/xml_to_dict.py

@@ -43,7 +43,7 @@ def add_children(children, tag_name, separated_attributes):
     return {tag_name: {k: v[0] if len(v) == 1 else v for k, v in dd.items()}}
 
 
-def xml2dict(t, separated_attributes=False):
+def xml2dict(t, separated_attributes=False) -> dict:
     tag_name = get_tag(t)
     d = {tag_name: {} if t.attrib else None}
     children = list(t)
@@ -61,11 +61,11 @@ def xml2dict(t, separated_attributes=False):
     return d
 
 
-class XmlToJson:
+class XmlToDict:
 
     def __init__(self, xml: str):
         self.xml = xml
 
-    def to_json(self):
+    def to_dict(self) -> dict:
         xml_data = ElementTree.XML(self.xml)
         return xml2dict(xml_data)

sl_util/tests/resources/test_resource_paths.py

@@ -0,0 +1,7 @@
+import os
+
+path = os.path.dirname(__file__)
+
+# xml testing files
+random_data_xml = f'{path}/xml/random_data.xml'
+random_data_xml_json = f'{path}/xml/random_data_xml.json'

sl_util/tests/resources/xml/random_data.xml

@@ -0,0 +1,33 @@
+<root>
+    <flower>sunlight</flower>
+    <with waste="hurried">men</with>
+    <reason ship="nuts">-862244492</reason>
+    <exclaimed>
+        <widely>composition</widely>
+        <!--excited-->
+        <regular>spirit</regular>
+        <record>
+            <another experience="twice">
+                <!--sentence rays person-->
+                <teacher food="hay">1184223729.943545</teacher>
+                <count>magic</count>
+                <title>648067907.3277977</title>
+                <strange immediately="bread">change</strange>
+                <!--mainly row-->
+                <instrument>particular</instrument>
+                <finest>-1459099875.7289388</finest>
+            </another>
+            <hay smell="plates">loose</hay>
+            <tobacco>552630041.6824405</tobacco>
+            <peace suppose="motor">-45766333</peace>
+            <!--badly know nobody distance mighty-->
+            <birthday>stretch</birthday>
+            <food native="good">within</food>
+        </record>
+        <cookies tobacco="conversation">-1213510200.834167</cookies>
+        <dinner>-557912696.1980629</dinner>
+        <yard alone="willing">birds</yard>
+    </exclaimed>
+    <identity>459467385.41047287</identity>
+    <clean recognize="dried">1248105488</clean>
+</root>

sl_util/tests/resources/xml/random_data_xml.json

@@ -0,0 +1,62 @@
+{
+  "root": {
+    "clean": {
+      "recognize": "dried",
+      "text": "1248105488"
+    },
+    "exclaimed": {
+      "cookies": {
+        "text": "-1213510200.834167",
+        "tobacco": "conversation"
+      },
+      "dinner": "-557912696.1980629",
+      "record": {
+        "another": {
+          "count": "magic",
+          "experience": "twice",
+          "finest": "-1459099875.7289388",
+          "instrument": "particular",
+          "strange": {
+            "immediately": "bread",
+            "text": "change"
+          },
+          "teacher": {
+            "food": "hay",
+            "text": "1184223729.943545"
+          },
+          "title": "648067907.3277977"
+        },
+        "birthday": "stretch",
+        "food": {
+          "native": "good",
+          "text": "within"
+        },
+        "hay": {
+          "smell": "plates",
+          "text": "loose"
+        },
+        "peace": {
+          "suppose": "motor",
+          "text": "-45766333"
+        },
+        "tobacco": "552630041.6824405"
+      },
+      "regular": "spirit",
+      "widely": "composition",
+      "yard": {
+        "alone": "willing",
+        "text": "birds"
+      }
+    },
+    "flower": "sunlight",
+    "identity": "459467385.41047287",
+    "reason": {
+      "ship": "nuts",
+      "text": "-862244492"
+    },
+    "with": {
+      "text": "men",
+      "waste": "hurried"
+    }
+  }
+}

sl_util/tests/unit/test_xml_to_dict.py

@@ -0,0 +1,41 @@
+import json
+from unittest import TestCase
+
+import pytest
+
+from sl_util.sl_util.dict_utils import compare_dict
+from sl_util.sl_util.file_utils import get_byte_data
+from sl_util.sl_util.xml_to_dict import XmlToDict
+from sl_util.tests.resources.test_resource_paths import random_data_xml, random_data_xml_json
+
+
+class TestXmlToDict(TestCase):
+
+    def test_to_dict(self):
+        # GIVEN the source MTMT data
+        xml = get_byte_data(random_data_xml).decode()
+        # AND the parser
+        parser = XmlToDict(xml)
+        # AND the expected result
+        expected_data = json.loads(get_byte_data(random_data_xml_json).decode())
+
+        # WHEN we convert to json
+        result_data = parser.to_dict()
+
+        # THEN the result is as expected
+        result, expected = compare_dict(result_data, expected_data)
+        assert result == expected
+
+    def test_to_dict_invalid_data(self):
+        # GIVEN the source MTMT data
+        xml = "INVALID XML CONTENT"
+        # AND the parser
+        parser = XmlToDict(xml)
+
+        # WHEN we convert to json
+        # THEN a ParseError is raised
+        with pytest.raises(Exception) as error:
+            parser.to_dict()
+        # AND the error is as expected
+        assert error.typename == 'ParseError'
+        assert str(error.value.msg) == 'syntax error: line 1, column 0'

slp_drawio/slp_drawio/load/drawio_loader.py

@@ -17,13 +17,18 @@ class DrawioLoader(ProviderLoader):
     Builder for a drawio class from the xml data
     """
 
+    def __init__(self, project_id: str, source):
+        self.project_id = project_id
+        self.source = source
+        self.diagram = None
+
     def load(self):
         try:
-            source = DrawIOToDict(self.source).to_dict()
+            source_dict = DrawIOToDict(self.source).to_dict()
 
-            representation: DiagramRepresentation = DiagramRepresentationLoader(self.project_id, source).load()
-            components: [DiagramComponent] = DiagramComponentLoader(source).load()
-            dataflows: [DiagramDataflow] = DiagramDataflowLoader(source).load()
+            representation: DiagramRepresentation = DiagramRepresentationLoader(self.project_id, source_dict).load()
+            components: [DiagramComponent] = DiagramComponentLoader(source_dict).load()
+            dataflows: [DiagramDataflow] = DiagramDataflowLoader(source_dict).load()
 
             self.diagram: Diagram = Diagram(representation, components, dataflows)
         except Exception as e:
@@ -32,10 +37,5 @@ def load(self):
             message = e.__str__()
             raise LoadingSourceFileError('Source file cannot be loaded', detail, message)
 
-    def __init__(self, project_id: str, source):
-        self.project_id = project_id
-        self.source = source
-        self.diagram = None
-
     def get_diagram(self) -> Diagram:
         return self.diagram

slp_drawio/slp_drawio/load/drawio_to_dict.py

@@ -1,16 +1,41 @@
+import base64
 import logging
+import zlib
 from tempfile import SpooledTemporaryFile
+from urllib.parse import unquote
 
-from slp_drawio.slp_drawio.load.objects.drawio_wrapper import DrawioWrapper
+from defusedxml import ElementTree
+
+from sl_util.sl_util.xml_to_dict import XmlToDict
 
 logger = logging.getLogger(__name__)
 
+DEFAULT_ENCODING = 'utf8'
+DIAGRAM_TAG = 'diagram'
+
 
 class DrawIOToDict:
 
     def __init__(self, source):
         file: SpooledTemporaryFile = source.file
-        self.wrapper = DrawioWrapper(file.read())
+        self.encoding = 'utf8'
+        self.content: str = file.read().decode()
+        self.decode_b64_tag('diagram')
 
     def to_dict(self) -> dict:
-        return self.wrapper.json()
+        return XmlToDict(self.content).to_dict()
+
+    def decode_b64_tag(self, tag):
+        tree = ElementTree.XML(self.content)
+        tree_tag = tree.find(tag)
+        children = list(tree_tag.iter())
+        if len(children) <= 1:
+            logger.debug(f'{tag} tag is encoded')
+            data = base64.b64decode(tree_tag.text, validate=True)
+            xml = zlib.decompress(data, wbits=-15).decode()
+            xml = unquote(xml)
+            diagram_tree = ElementTree.fromstring(xml)
+            tree_tag.append(diagram_tree)
+            tree_tag.text = None
+            self.content = ElementTree.tostring(tree, encoding=self.encoding)
+            logger.debug(f'{tag} tag decoded successfully')