Merge pull request #325 from fschrader1992/master

RDF Converter, XML Reader Updates

Author: mpsonntag

MANIFEST.in

@@ -1,3 +1,4 @@
 include LICENSE
 include README.rst
 include odml/info.json
+include doc/section_subclasses.yaml

odml/doc.py

@@ -141,21 +141,22 @@ def get_terminology_equivalent(self):
             return None
         term = terminology.load(self.repository)
         return term
-def pprint(self, indent=2, max_depth=1, max_length=80, current_depth=0):
-    """
-    Pretty print method to visualize Document-Section trees.
-
-    :param indent: number of leading spaces for every child Section or Property.
-    :param max_depth: maximum number of hierarchical levels printed from the
-                          starting Section.
-    :param max_length: maximum number of characters printed in one line.
-    :param current_depth: number of hierarchical levels printed from the
-                          starting Section.
-    """
-    doc_str = "[{} [{}] {}, sections: {}, repository: {}]".format(self.author, self.version,
-                                                                  self.date, len(self._sections), self.repository)
-    print(doc_str)
 
-    for s in self._sections:
-        s.pprint(current_depth=current_depth+1, max_depth=max_depth,
-                 indent=indent, max_length=max_length)
+    def pprint(self, indent=2, max_depth=1, max_length=80, current_depth=0):
+        """
+        Pretty print method to visualize Document-Section trees.
+
+        :param indent: number of leading spaces for every child Section or Property.
+        :param max_depth: maximum number of hierarchical levels printed from the
+                              starting Section.
+        :param max_length: maximum number of characters printed in one line.
+        :param current_depth: number of hierarchical levels printed from the
+                              starting Section.
+        """
+        doc_str = "[{} [{}] {}, sections: {}, repository: {}]".format(self.author, self.version,
+                                                                      self.date, len(self._sections), self.repository)
+        print(doc_str)
+
+        for s in self._sections:
+            s.pprint(current_depth=current_depth+1, max_depth=max_depth,
+                     indent=indent, max_length=max_length)

odml/tools/rdf_converter.py

@@ -43,9 +43,6 @@ def __init__(self, odml_documents):
 
         self.section_subclasses = {}
 
-        # TODO doc/section_subclasses.yaml has to be exported on install, otherwise
-        # the RDFWriter is broken. Below is a quick and dirty fix to at least
-        # unbreak on install.
         subclass_path = os.path.join(dirname(dirname(dirname(abspath(__file__)))),
                          'doc', 'section_subclasses.yaml')
 

odml/tools/xmlparser.py

@@ -218,12 +218,12 @@ def from_string(self, string):
     def check_mandatory_arguments(self, data, ArgClass, tag_name, node):
         for k, v in ArgClass.arguments:
             if v != 0 and not ArgClass.map(k) in data:
-                self.error("missing element <%s> within <%s> tag" %
+                self.error("missing element <%s> within <%s> tag\n" %
                            (k, tag_name) + repr(data), node)
 
     def is_valid_argument(self, tag_name, ArgClass, parent_node, child=None):
         if tag_name not in ArgClass.arguments_keys:
-            self.error("Invalid element <%s> inside <%s> tag" %
+            self.error("Invalid element <%s> inside <%s> tag\n" %
                        (tag_name, parent_node.tag),
                        parent_node if child is None else child)
 
@@ -247,7 +247,7 @@ def warn(self, msg, elem):
 
     def parse_element(self, node):
         if node.tag not in self.tags:
-            self.error("Invalid element <%s>" % node.tag, node)
+            self.error("Invalid element <%s> " % node.tag, node)
             return None  # won't be able to parse this one
         return getattr(self, "parse_" + node.tag)(node, self.tags[node.tag])
 
@@ -273,7 +273,7 @@ def parse_tag(self, root, fmt, insert_children=True):
                 continue
 
             # We currently do not support XML attributes.
-            self.error("Attribute not supported, ignoring '%s=%s'" % (k, v), root)
+            self.error("Attribute not supported, ignoring '%s=%s' " % (k, v), root)
 
         for node in root:
             node.tag = node.tag.lower()
@@ -299,7 +299,7 @@ def parse_tag(self, root, fmt, insert_children=True):
                     else:
                         arguments[tag] = curr_text
             else:
-                self.error("Invalid element <%s> in odML document section <%s>"
+                self.error("Invalid element <%s> in odML document section <%s> "
                            % (node.tag, root.tag), node)
 
         if sys.version_info > (3,):
@@ -310,7 +310,11 @@ def parse_tag(self, root, fmt, insert_children=True):
         self.check_mandatory_arguments(check_args, fmt, root.tag, root)
 
         # Instantiate the current odML object with the parsed attributes.
-        obj = fmt.create(**arguments)
+        obj = fmt.create()
+        try:
+            obj = fmt.create(**arguments)
+        except Exception as e:
+            self.error(str(e), root)
 
         if insert_children:
             for child in children:

test/resources/ignore_errors.xml

@@ -0,0 +1,53 @@
+<?xml version="1.0" encoding="UTF-8"?>
+<?xml-stylesheet  type="text/xsl" href="odmlTerms.xsl"?>
+<?xml-stylesheet  type="text/xsl" href="odml.xsl"?>
+<odML version="1.1">
+    <author>D. N. Adams</author>
+    <repository>hello</repository>
+    <section>
+        <property>
+            <type>int</type>
+            <value>123</value>
+        </property>
+        <property>
+            <type>int</type>
+            <uncertainty>[zergrve sdic, 1]</uncertainty>
+            <value>["string 1", "string 2"]</value>
+            <reference>The Hitchhiker's guide to the Galaxy (novel)</reference>
+            <id>wrongid</id>
+            <name>NoCrewMembers</name>
+            <definition>Number of crew members</definition>
+        </property>
+        <property>
+            <name>NoBones</name>
+            <type></type>
+            <value>235</value>
+            <id>wrongid4</id>
+            <definition>Number of bones within the crew member.</definition>
+        </property>
+        <type>crew</type>
+        <definition>Information on the crew</definition>
+        <id>0f96a050-2d9b-498f-a532-fbfcc6aae55e</id>
+        <name>TheCrew</name>
+        <section>
+            <property>
+                <type>string</type>
+                <value>[1, 2]</value>
+                <id>wrongid2</id>
+                <name>Nickname</name>
+                <definition>Nickname(s) of the subject</definition>
+            </property>
+            <property>
+                <type>int</type>
+                <value>535</value>
+                <id>wrongid3</id>
+                <definition>Nickname(s) of the subject</definition>
+            </property>
+            <type>ADent</type>
+            <definition>Information on Arthur Dent</definition>
+        </section>
+    </section>
+    <date>ztuz</date>
+    <id>79b613eb-a256-46bf-84f6-207df465b8f7</id>
+    <version>new version</version>
+</odML>

test/test_parser_xml.py

@@ -12,6 +12,7 @@ def setUp(self):
         self.basepath = os.path.join(dir_path, "resources")
 
         self.xml_reader = xmlparser.XMLReader()
+        self.xml_reader_ignore = xmlparser.XMLReader(ignore_errors=True)
 
     def test_invalid_root(self):
         filename = "invalid_root.xml"
@@ -36,3 +37,12 @@ def test_invalid_version(self):
 
         with self.assertRaises(InvalidVersionException):
             _ = self.xml_reader.from_file(os.path.join(self.basepath, filename))
+
+    def test_ignore_errors(self):
+        filename = "ignore_errors.xml"
+
+        with self.assertRaises(ParserException):
+            _ = self.xml_reader.from_file(os.path.join(self.basepath, filename))
+
+        doc = self.xml_reader_ignore.from_file(os.path.join(self.basepath, filename))
+        doc.pprint()