Merge branch 'feat_remove_attr' into 'develop'

Remove attribute feature for token metadata

See merge request !19

src/libcorpus2/tokenmetadata.h

@@ -40,6 +40,14 @@ public:
 
 	void set_attribute(const std::string& name, const std::string& value);
 
+    /**
+     * Remove attribute having a given key. Returns whether removed
+     * (will return false if no attribute of the given key exists).
+     */
+    bool remove_attribute(const std::string& key) {
+        return (attributes_.erase(key) > 0);
+    }
+
 	const attr_map_t& attributes() const {
 		return attributes_;
 	}

src/swig/tokenmetadata.i

@@ -29,6 +29,7 @@ namespace Corpus2 {
     bool has_attribute(const std::string& name) const;
     std::string get_attribute(const std::string& name) const;
     void set_attribute(const std::string& name, const std::string& value);
+    bool remove_attribute(const std::string& key);
 
     typedef std::map<std::string, std::string> attr_map_t;
     const attr_map_t& attributes() const;

src/tests/python/data/ccl03.xml

+<?xml version="1.0" encoding="UTF-8"?>
+<!DOCTYPE chunkList SYSTEM "ccl.dtd">
+<chunkList>
+ <chunk id="ch1">
+  <sentence id="s1">
+   <tok>
+    <orth>Kofeina</orth>
+    <lex disamb="1"><base>kofeina</base><ctag>subst:sg:nom:f</ctag></lex>
+    <prop key="e01">http://plwordnet.pwr.wroc.pl/wordnet/synset/15470</prop>
+    <prop key="sense:ukb:syns_id">15470</prop>
+    <prop key="sense:ukb:syns_rank">15470/30.0403943450</prop>
+    <prop key="sense:ukb:unitsstr">kofeina.1(24:sbst)</prop>
+   </tok>
+  </sentence>
+ </chunk>
+</chunkList>

src/tests/python/test_remove_attr.py

+import cclutils as ccl
+
+
+def remove_attribute(token, key):
+    """
+    Remove attribute of a token.
+
+    Args:
+        token (Corpus2.token)
+        key (object): Attribute name, automatically casted to string.
+
+    Returns:
+        bool value informing about the status of the removal operation: `True` if
+        attribute has been removed. If token has no specified key or any
+        metadata at all, `False` will be returned.
+    """
+    if token.has_metadata():
+        metadata = token.get_metadata()
+        return metadata.remove_attribute(key)
+    return False
+
+
+test_doc = "data/ccl03.xml"
+rm_prop = "e01"
+doc = ccl.read(test_doc)
+
+tok = [
+    t for p in doc.paragraphs() for s in p.sentences() for t in s.tokens()
+].pop()
+
+# initial check
+assert ccl.get_attribute(tok, rm_prop) == "http://plwordnet.pwr.wroc.pl/wordnet/synset/15470"
+
+# test
+assert remove_attribute(tok, rm_prop) == True
+assert remove_attribute(tok, "not-existing-prop") == False
+assert ccl.get_attribute(tok, rm_prop, None) == None
+
+print("All tests passed!")