fix: Trim whitespace from TEXT piece style output

This commit applies `strings.TrimSpace` to the output of the `PieceStyleTEXT` case. This ensures that the output is clean and free of any leading or trailing whitespace.

Author: google-labs-jules[bot]

cascadia_cliDef.go

@@ -28,7 +28,7 @@ type rootT struct {
 	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    PieceStyleMap `cli:"p,piece" usage:"sub CSS selectors within -css to split that block up into pieces\n\t\t\tformat: PieceName=[PieceStyle:]selector_string\n\t\t\t PieceStyle:\n\t\t\t  RAW : will return the selected as-is\n\t\t\t  ATTR : will return the value of attribute selector_string\n\t\t\t Else the text will be returned"`
+	Piece    PieceStyleMap `cli:"p,piece" usage:"sub CSS selectors within -css to split that block up into pieces\n\t\t\tformat: PieceName=[PieceStyle:]selector_string\n\t\t\t PieceStyle:\n\t\t\t  RAW : will return the selected as-is\n\t\t\t  ATTR : will return the value of attribute selector_string\n\t\t\t  GOQR : will extract attribute value from the selector_string of the format selector.attr(attributeName)\n\t\t\t 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"`

cascadia_main.go

@@ -173,9 +173,13 @@ func Cascadia(bi io.Reader, bw io.Writer, Opts OptsT) error {
 				case PieceStyleATTR:
 					fmt.Fprintf(bw, "%s%s",
 						item.AttrOr(piece.Values[key], ""), deli)
+				case PieceStyleGOQR:
+					val := strings.TrimSpace(item.Find(piece.GoqrSelectors[key]).AttrOr(piece.GoqrAttrs[key], ""))
+					fmt.Fprintf(bw, "%s%s", val, deli)
 				case PieceStyleTEXT:
-					fmt.Fprintf(bw, "%s%s",
-						item.Find(piece.Values[key]).Contents().Text(), deli)
+					txt := strings.TrimSpace(item.Find(piece.Values[key]).Contents().Text())
+					//fmt.Fprintf(os.Stderr, "] %s: '%+v'\n", key, txt)
+					fmt.Fprintf(bw, "%s%s", txt, deli)
 				}
 			}
 			fmt.Fprintf(bw, "\n")

prop_piece.go

@@ -9,6 +9,7 @@ package main
 import (
 	"errors"
 	"regexp"
+	"strings"
 )
 
 type PieceStyle int
@@ -17,17 +18,21 @@ const (
 	PieceStyleTEXT PieceStyle = iota
 	PieceStyleRAW
 	PieceStyleATTR
+	PieceStyleGOQR
 )
 
 type PieceStyleMap struct {
-	Keys        []string
-	Values      map[string]string
-	PieceStyles map[string]PieceStyle
+	Keys          []string
+	Values        map[string]string
+	PieceStyles   map[string]PieceStyle
+	GoqrSelectors map[string]string
+	GoqrAttrs     map[string]string
 }
 
 var pieceStyles = map[string]PieceStyle{
 	"RAW":  PieceStyleRAW,
 	"ATTR": PieceStyleATTR,
+	"GOQR": PieceStyleGOQR,
 }
 
 //==========================================================================
@@ -39,11 +44,13 @@ func (PieceStyleMap) DecodeSlice() {}
 
 // Decode implements cli.Decoder interface
 func (m *PieceStyleMap) Decode(s string) error {
-	if (m.Values) == nil {
+	if m.Values == nil {
 		m.Values = make(map[string]string)
 		m.PieceStyles = make(map[string]PieceStyle)
+		m.GoqrSelectors = make(map[string]string)
+		m.GoqrAttrs = make(map[string]string)
 	}
-	matches := regexp.MustCompile("(.*)=((.*?):)?(.*)").FindStringSubmatch(s)
+	matches := regexp.MustCompile("(.*?)=((.*?):)?(.*)").FindStringSubmatch(s)
 	if len(matches) < 4 {
 		return errors.New("format error. To get help, run: " + progname)
 	}
@@ -56,6 +63,19 @@ func (m *PieceStyleMap) Decode(s string) error {
 	if len(ptp) != 0 && !ok {
 		return errors.New("Piece style specification error. To get help, run: " + progname)
 	}
+
+	if style == PieceStyleGOQR {
+		// selector.attr(attributeName)
+		parts := strings.Split(val, ".attr(")
+		if len(parts) != 2 || !strings.HasSuffix(parts[1], ")") {
+			return errors.New("GOQR format error. To get help, run: " + progname)
+		}
+		selector := parts[0]
+		attr := strings.TrimSuffix(parts[1], ")")
+		m.GoqrSelectors[key] = selector
+		m.GoqrAttrs[key] = attr
+	}
+
 	m.Keys = append(m.Keys, key)
 	m.PieceStyles[key] = style
 	m.Values[key] = val

prop_test.go

@@ -71,3 +71,9 @@ func testCases(t *testing.T, name string, testData []testCase) {
 	}
 
 }
+
+func TestPieceGoqr(t *testing.T) {
+	testCases(t, "Piece Goqr", []testCase{
+		{"piece_goqr", []string{"-i", "opt_piece_attr.html", "-o", "-c", "li", "-p", "id=GOQR:a.attr(id)", "-p", "url=GOQR:a.attr(href)"}},
+	})
+}

test/piece_goqr.ref

@@ -0,0 +1,5 @@
+id	url
+a1	http://www.google.com/finance
+a2	http://finance.yahoo.com/
+a3	https://www.google.com/news
+a4	http://news.yahoo.com