1 files changed, 94 insertions, 0 deletions
diff --git a/word_definition.go b/word_definition.go
new file mode 100644
index 0000000..86b05bd
--- /dev/null
+++ b/word_definition.go
@@ -0,0 +1,94 @@
+// Copyright 2020, Shulhan <ms@kilabit.info>. All rights reserved.
+// Use of this source code is governed by a BSD-style
+// license that can be found in the LICENSE file.
+
+package kamusku
+
+import (
+	"fmt"
+	"strings"
+
+	"github.com/shuLhan/share/lib/net/html"
+	libstrings "github.com/shuLhan/share/lib/strings"
+)
+
+//
+// WordDefinition contains the meaning of word in dictionary, and optional
+// attribute for word classifications and examples.
+//
+type WordDefinition struct {
+	Value    string   `json:"isi"`
+	Classes  []string `json:"kelas,omitempty"`
+	Examples []string `json:"contoh,omitempty"`
+}
+
+func parseWordDefinition(in string, li *html.Node) (defKata *WordDefinition, err error) {
+	elFont := li.GetFirstChild()
+	if elFont == nil || elFont.Data != tagNameFont {
+		return nil, nil
+	}
+	elItalic := elFont.GetFirstChild()
+	if elItalic == nil || elItalic.Data != tagNameItalic {
+		return nil, nil
+	}
+
+	defKata = &WordDefinition{}
+
+	elSpan := elItalic.GetFirstChild()
+	for elSpan != nil && elSpan.Data == tagNameSpan {
+		kelas := elSpan.GetAttrValue(attrNameTitle)
+		if len(kelas) > 0 {
+			defKata.Classes = append(defKata.Classes, kelas)
+		}
+		elSpan = elSpan.GetNextSibling()
+	}
+
+	el := elFont.GetNextSibling()
+	if el == nil {
+		return defKata, nil
+	}
+
+	defKata.Value = strings.TrimSpace(libstrings.SingleSpace(el.Data))
+
+	if defKata.Value == "→" {
+		defKata.Value = ""
+		el = el.GetNextSibling()
+		if el == nil || el.Data != tagNameAnchor {
+			return nil, nil
+		}
+		el = el.GetFirstChild()
+		return nil, fmt.Errorf(`%q adalah bentuk tidak baku dari %q`,
+			in, el.Data)
+	}
+
+	if defKata.Value[len(defKata.Value)-1] != ':' {
+		return defKata, nil
+	}
+
+	defKata.Value = defKata.Value[:len(defKata.Value)-1]
+
+	// Parse the example of kata in the next sibling.
+	el = el.GetNextSibling()
+	for el != nil {
+		if el.Data != tagNameFont {
+			break
+		}
+
+		elItalic = el.GetFirstChild()
+		if elItalic.Data != tagNameItalic {
+			break
+		}
+
+		elText := elItalic.GetFirstChild()
+		if elText != nil {
+			contoh := strings.TrimSpace(elText.Data)
+			if len(contoh) > 0 && contoh != ";" {
+				defKata.Examples = append(defKata.Examples, elText.Data)
+			}
+		}
+
+		el = el.GetNextSibling()
+	}
+
+	return defKata, nil
+}