Further on to Himcc master

README.md

@@ -50,8 +50,11 @@ Options:
   -t, --text        Text output for none-block selection mode 
   -R, --Raw         Raw text output, no trimming of leading and trailing white space 
   -p, --piece       sub CSS selectors within -css to split that block up into pieces
-			format: PieceName=[RAW:]selector_string
-			RAW: will return the selected as-is; else the text will be returned 
+			format: PieceName=[OutputStyle:]selector_string
+			OutputStyle:
+				RAW : will return the selected as-is
+				attr[xxx] : will return the value of an attribute named xxx
+			else the text will be returned 
   -d, --delimiter   delimiter for pieces csv output [=	]
   -w, --wrap-html   wrap up the output with html tags 
   -y, --style       style component within the wrapped html head 

cascadia_cli.yaml

@@ -39,9 +39,9 @@ Options:
     Usage: Raw text output, no trimming of leading and trailing white space
 
   - Name: Piece
-    Type: MapStringString
+    Type: OutputStyleMap
     Flag: 'p,piece'
-    Usage: 'sub CSS selectors within -css to split that block up into pieces\n\t\t\tformat: PieceName=[RAW:]selector_string\n\t\t\tRAW: will return the selected as-is; else the text will be returned'
+    Usage: 'sub CSS selectors within -css to split that block up into pieces\n\t\t\tformat: PieceName=[OutputStyle:]selector_string\n\t\t\tOutputStyle:\n\t\t\t\tRAW : will return the selected as-is\n\t\t\t\tattr[xxx] : will return the value of an attribute named xxx \n\t\t\telse the text will be returned'
 
   - Name: Deli
     Type: string

cascadia_cliDef.go

@@ -1,7 +1,7 @@
 ////////////////////////////////////////////////////////////////////////////
 // Program: cascadiaC
 // Purpose: cascadia wrapper
-// Authors: Tong Sun (c) 2021, All rights reserved
+// Authors: Tong Sun (c) 2023, All rights reserved
 ////////////////////////////////////////////////////////////////////////////
 
 package main
@@ -23,23 +23,23 @@ import (
 
 type rootT struct {
 	cli.Helper
-	Filei    *clix.Reader    `cli:"*i,in" usage:"The html/xml file to read from (or stdin)"`
-	Fileo    *clix.Writer    `cli:"*o,out" usage:"The output file (or stdout)"`
-	CSS      []string        `cli:"*c,css" usage:"CSS selectors (can provide more if not using --piece)"`
-	TextOut  bool            `cli:"t,text" usage:"Text output for none-block selection mode"`
-	TextRaw  bool            `cli:"R,Raw" usage:"Raw text output, no trimming of leading and trailing white space"`
-	Piece    MapStringString `cli:"p,piece" usage:"sub CSS selectors within -css to split that block up into pieces\n\t\t\tformat: PieceName=[RAW:]selector_string\n\t\t\tRAW: will return the selected as-is; else the text will be returned"`
-	Deli     string          `cli:"d,delimiter" usage:"delimiter for pieces csv output" dft:"\t"`
-	WrapHTML bool            `cli:"w,wrap-html" usage:"wrap up the output with html tags"`
-	Style    string          `cli:"y,style" usage:"style component within the wrapped html head"`
-	Base     string          `cli:"b,base" usage:"base href tag used in the wrapped up html"`
-	Quiet    bool            `cli:"q,quiet" usage:"be quiet"`
+	Filei    *clix.Reader   `cli:"*i,in" usage:"The html/xml file to read from (or stdin)"`
+	Fileo    *clix.Writer   `cli:"*o,out" usage:"The output file (or stdout)"`
+	CSS      []string       `cli:"*c,css" usage:"CSS selectors (can provide more if not using --piece)"`
+	TextOut  bool           `cli:"t,text" usage:"Text output for none-block selection mode"`
+	TextRaw  bool           `cli:"R,Raw" usage:"Raw text output, no trimming of leading and trailing white space"`
+	Piece    OutputStyleMap `cli:"p,piece" usage:"sub CSS selectors within -css to split that block up into pieces\n\t\t\tformat: PieceName=[OutputStyle:]selector_string\n\t\t\tOutputStyle:\n\t\t\t\tRAW : will return the selected as-is\n\t\t\t\tattr[xxx] : will return the value of an attribute named xxx \n\t\t\telse the text will be returned"`
+	Deli     string         `cli:"d,delimiter" usage:"delimiter for pieces csv output" dft:"\t"`
+	WrapHTML bool           `cli:"w,wrap-html" usage:"wrap up the output with html tags"`
+	Style    string         `cli:"y,style" usage:"style component within the wrapped html head"`
+	Base     string         `cli:"b,base" usage:"base href tag used in the wrapped up html"`
+	Quiet    bool           `cli:"q,quiet" usage:"be quiet"`
 }
 
 var root = &cli.Command{
 	Name: "cascadiaC",
 	Desc: "cascadia wrapper\nVersion " + version + " built on " + date +
-		"\nCopyright (C) 2021, Tong Sun",
+		"\nCopyright (C) 2023, Tong Sun",
 	Text: "Command line interface to go cascadia CSS selectors package" +
 		"\n\nUsage:\n  cascadia -i in -c css -o [Options...]",
 	Argv: func() interface{} { return new(rootT) },
@@ -59,7 +59,7 @@ var root = &cli.Command{
 //  	CSS	[]string
 //  	TextOut	bool
 //  	TextRaw	bool
-//  	Piece	MapStringString
+//  	Piece	OutputStyleMap
 //  	Deli	string
 //  	WrapHTML	bool
 //  	Style	string
@@ -74,7 +74,7 @@ var root = &cli.Command{
 //  var (
 //          progname  = "cascadiaC"
 //          version   = "0.1.0"
-//          date = "2021-11-27"
+//          date = "2023-01-08"
 
 //  	rootArgv *rootT
 //  	// Opts store all the configurable options

cascadia_main.go

@@ -26,22 +26,31 @@ import (
 // Constant and data type/structure definitions
 
 const (
-	IsRaw       = "RAW:"
+	IsRaw       = "RAW"
 	WrapHTMLEnd = `</body>`
 )
 
-type MapStringString struct {
-	Keys   []string
-	Values map[string]string
-	Raw    map[string]bool
+type OutputStyle int
+
+const (
+	OutputStyleRAW OutputStyle = iota
+	OutputStyleATTR
+	OutputStyleTEXT
+)
+
+type OutputStyleMap struct {
+	Keys         []string
+	Values       map[string]string
+	OutputStyles map[string]OutputStyle
+	AttrName     map[string]string
 }
 
 // The OptsT type defines all the configurable options from cli.
 type OptsT struct {
 	CSS      []string
 	TextOut  bool
 	TextRaw  bool
-	Piece    MapStringString
+	Piece    OutputStyleMap
 	Deli     string
 	WrapHTML bool
 	Style    string
@@ -174,10 +183,14 @@ func Cascadia(bi io.Reader, bw io.Writer, Opts OptsT) error {
 			//fmt.Printf("] #%d: %s\n", index, item.Text())
 			for _, key := range piece.Keys {
 				//fmt.Printf("] %s: %s\n", key, piece.Values[key])
-				if piece.Raw[key] {
+				switch piece.OutputStyles[key] {
+				case OutputStyleRAW:
 					html.Render(bw, item.Find(piece.Values[key]).Get(0))
 					fmt.Fprintf(bw, deli)
-				} else {
+				case OutputStyleATTR:
+					fmt.Fprintf(bw, "%s%s",
+						item.Find(piece.Values[key]).AttrOr(piece.AttrName[key], ""), deli)
+				case OutputStyleTEXT:
 					fmt.Fprintf(bw, "%s%s",
 						item.Find(piece.Values[key]).Contents().Text(), deli)
 				}
@@ -196,23 +209,35 @@ func Cascadia(bi io.Reader, bw io.Writer, Opts OptsT) error {
 
 // DecodeSlice implements cli.SliceDecoder
 // NOTE: if SliceDecoder not implemented, the Decode method would be only invoked once
-func (MapStringString) DecodeSlice() {}
+func (OutputStyleMap) DecodeSlice() {}
 
 // Decode implements cli.Decoder interface
-func (m *MapStringString) Decode(s string) error {
+func (m *OutputStyleMap) Decode(s string) error {
 	if (m.Values) == nil {
 		m.Values = make(map[string]string)
-		m.Raw = make(map[string]bool)
+		m.OutputStyles = make(map[string]OutputStyle)
+		m.AttrName = make(map[string]string)
 	}
 	matches := regexp.MustCompile("(.*)=(.*)").FindStringSubmatch(s)
 	if len(matches) < 2 {
 		return errors.New("format error. To get help, run: " + progname)
 	}
 	key := matches[1]
 	val := matches[2]
-	if len(val) >= 4 && val[:4] == IsRaw {
-		m.Raw[key] = true
-		val = val[4:]
+	index := strings.Index(val, ":")
+	if index > 0 {
+		style := val[:index]
+		val = val[index+1:]
+		if style == IsRaw {
+			m.OutputStyles[key] = OutputStyleRAW
+		} else if strings.HasPrefix(style, "attr[") && strings.HasSuffix(style, "]") {
+			m.OutputStyles[key] = OutputStyleATTR
+			m.AttrName[key] = style[5 : len(style)-1]
+		} else {
+			m.OutputStyles[key] = OutputStyleTEXT
+		}
+	} else {
+		m.OutputStyles[key] = OutputStyleTEXT
 	}
 	m.Keys = append(m.Keys, key)
 	m.Values[key] = val

cascadia_test.go

@@ -11,7 +11,7 @@ func TestSelectors(t *testing.T) {
 		buf := bytes.NewBufferString("")
 		Opts.CSS, Opts.Piece, Opts.Deli,
 			Opts.WrapHTML, Opts.TextOut, Opts.TextRaw, Opts.Quiet =
-			[]string{test.selector}, MapStringString{}, ",",
+			[]string{test.selector}, OutputStyleMap{}, ",",
 			false, false, false, false
 		Cascadia(strings.NewReader(test.HTML), buf, Opts)
 		got := buf.String()
@@ -26,7 +26,7 @@ func TestSelectors(t *testing.T) {
 	}
 }
 
-////////////////////////////////////////////////////////////////////////////
+// //////////////////////////////////////////////////////////////////////////
 // The following is taken from
 // https://github.com/andybalholm/cascadia/blob/c56252c33997d9b9991f4c1e3b5fbc82d6d656b3/selector_test.go
 type selectorTest struct {
@@ -549,3 +549,98 @@ var selectorTests = []selectorTest{
 		},
 	},
 }
+
+type PieceAttrTest struct {
+	HTML, selector string
+	results        []string
+	piece          OutputStyleMap
+}
+
+var PieceAttrTests = []PieceAttrTest{
+	{
+		`<ul>
+			<li><a id="a1" href="http://www.google.com/finance"/>
+			<li><a id="a2" href="http://finance.yahoo.com/"/>
+			<li><a id="a3" href="https://www.google.com/news"></a>
+			<li><a id="a4" href="http://news.yahoo.com"/>
+		</ul>`,
+		`li`,
+		[]string{
+			`id,`,
+			`a1,`,
+			`a2,`,
+			`a3,`,
+			`a4,`,
+		},
+		OutputStyleMap{
+			[]string{"id"},
+			map[string]string{"id": "a"},
+			map[string]OutputStyle{"id": OutputStyleATTR},
+			map[string]string{"id": "id"},
+		},
+	},
+	{
+		`<ul>
+			<li><a id="a1" href="http://www.google.com/finance"/>
+			<li><a id="a2" href="http://finance.yahoo.com/"/>
+			<li><a id="a3" href="https://www.google.com/news"></a>
+			<li><a id="a4" href="http://news.yahoo.com"/>
+		</ul>`,
+		`li`,
+		[]string{
+			`href2,`,
+			`,`,
+			`,`,
+			`,`,
+			`,`,
+		},
+		OutputStyleMap{
+			[]string{"href2"},
+			map[string]string{"href2": "a"},
+			map[string]OutputStyle{"href2": OutputStyleATTR},
+			map[string]string{"href2": "href2"},
+		},
+	},
+	{
+		`<ul>
+			<li><a id="a1" href="http://www.google.com/finance"/>
+			<li><a id="a2" href="http://finance.yahoo.com/"/>
+			<li><a id="a3" href="https://www.google.com/news"></a>
+			<li><a id="a4" href="http://news.yahoo.com"/>
+		</ul>`,
+		`li`,
+		[]string{
+			`href,`,
+			`http://www.google.com/finance,`,
+			`http://finance.yahoo.com/,`,
+			`https://www.google.com/news,`,
+			`http://news.yahoo.com,`,
+		},
+		OutputStyleMap{
+			[]string{"href"},
+			map[string]string{"href": "a"},
+			map[string]OutputStyle{"href": OutputStyleATTR},
+			map[string]string{"href": "href"},
+		},
+	},
+}
+
+func TestPieceAttr(t *testing.T) {
+	for _, test := range PieceAttrTests {
+		buf := bytes.NewBufferString("")
+		Opts.CSS, Opts.Piece, Opts.Deli,
+			Opts.WrapHTML, Opts.TextOut, Opts.TextRaw, Opts.Quiet =
+			[]string{test.selector}, test.piece, ",",
+			false, false, false, false
+		Cascadia(strings.NewReader(test.HTML), buf, Opts)
+		got := buf.String()
+		if len(got) == 0 && len(test.results) == 0 {
+			// correct
+			continue
+		}
+		want := strings.Join(test.results, "\n") + "\n"
+		if got != want {
+			t.Errorf("wanted %s, got %s instead", want, got)
+		}
+	}
+}