- [#] update --piece handling; fix #3

Author: suntong

cascadia_cli.yaml

@@ -1,6 +1,7 @@
 # program name, name for the executable
 ProgramName: cascadia
 Authors: Tong Sun
+Since: 2016
 
 PackageName: main
 
@@ -41,7 +42,7 @@ Options:
   - Name: Piece
     Type: PieceStyleMap
     Flag: 'p,piece'
-    Usage: 'sub CSS selectors within -css to split that block up into pieces\n\t\t\t\tformat: PieceName=[PieceStyle:]selector_string\n\t\t\t\t PieceStyle:\n\t\t\t\t  RAW : will return the selected as-is\n\t\t\t\t  attr[xx] : will return the value of xx attribute\n\t\t\t\telse the text will be returned'
+    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'
 
   - Name: Deli
     Type: string

cascadia_cliDef.go

@@ -1,7 +1,7 @@
 ////////////////////////////////////////////////////////////////////////////
 // Program: cascadiaC
 // Purpose: cascadia wrapper
-// Authors: Tong Sun (c) 2023, All rights reserved
+// Authors: Tong Sun (c) 2016-2023, 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    PieceStyleMap `cli:"p,piece" usage:"sub CSS selectors within -css to split that block up into pieces\n\t\t\t\tformat: PieceName=[PieceStyle:]selector_string\n\t\t\t\t PieceStyle:\n\t\t\t\t  RAW : will return the selected as-is\n\t\t\t\t  attr[xx] : will return the value of xx attribute\n\t\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"`
+	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    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"`
+	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) 2023, Tong Sun",
+		"\nCopyright (C) 2016-2023, 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) },
@@ -74,7 +74,7 @@ var root = &cli.Command{
 //  var (
 //          progname  = "cascadiaC"
 //          version   = "0.1.0"
-//          date = "2023-01-08"
+//          date = "2023-06-29"
 
 //  	rootArgv *rootT
 //  	// Opts store all the configurable options

cascadia_main.go

@@ -172,7 +172,7 @@ func Cascadia(bi io.Reader, bw io.Writer, Opts OptsT) error {
 					fmt.Fprintf(bw, deli)
 				case PieceStyleATTR:
 					fmt.Fprintf(bw, "%s%s",
-						item.Find(piece.Values[key]).AttrOr(piece.AttrName[key], ""), deli)
+						item.AttrOr(piece.Values[key], ""), deli)
 				case PieceStyleTEXT:
 					fmt.Fprintf(bw, "%s%s",
 						item.Find(piece.Values[key]).Contents().Text(), deli)

piece_test.go

@@ -7,10 +7,10 @@ import (
 func TestPieceAttr(t *testing.T) {
 	testCases(t, "Piece Attributes", []testCase{
 		{"id",
-			[]string{"-i", "opt_piece_attr.html", "-o", "-c", "li", "-p", "url=attr[id]:a"},
+			[]string{"-i", "opt_piece_attr.html", "-o", "-c", "li > a", "-p", "url=ATTR:id"},
 		},
 		{"href",
-			[]string{"-i", "opt_piece_attr.html", "-o", "-c", "li", "-p", "url=attr[href]:a"},
+			[]string{"-i", "opt_piece_attr.html", "-o", "-c", "li > a", "-p", "url=ATTR:href"},
 		},
 		//{"both_fields",},
 		//{"noexist"},

prop_piece.go

@@ -9,22 +9,25 @@ package main
 import (
 	"errors"
 	"regexp"
-	"strings"
 )
 
 type PieceStyle int
 
 const (
-	PieceStyleRAW PieceStyle = iota
+	PieceStyleTEXT PieceStyle = iota
+	PieceStyleRAW
 	PieceStyleATTR
-	PieceStyleTEXT
 )
 
 type PieceStyleMap struct {
 	Keys        []string
 	Values      map[string]string
 	PieceStyles map[string]PieceStyle
-	AttrName    map[string]string
+}
+
+var pieceStyles = map[string]PieceStyle{
+	"RAW":  PieceStyleRAW,
+	"ATTR": PieceStyleATTR,
 }
 
 //==========================================================================
@@ -39,30 +42,22 @@ func (m *PieceStyleMap) Decode(s string) error {
 	if (m.Values) == nil {
 		m.Values = make(map[string]string)
 		m.PieceStyles = make(map[string]PieceStyle)
-		m.AttrName = make(map[string]string)
 	}
-	matches := regexp.MustCompile("(.*)=(.*)").FindStringSubmatch(s)
-	if len(matches) < 2 {
+	matches := regexp.MustCompile("(.*)=((.*?):)?(.*)").FindStringSubmatch(s)
+	if len(matches) < 3 {
 		return errors.New("format error. To get help, run: " + progname)
 	}
 	key := matches[1]
-	val := matches[2]
-	index := strings.Index(val, ":")
-	if index > 0 {
-		style := val[:index]
-		val = val[index+1:]
-		if style == IsRaw {
-			m.PieceStyles[key] = PieceStyleRAW
-		} else if strings.HasPrefix(style, "attr[") && strings.HasSuffix(style, "]") {
-			m.PieceStyles[key] = PieceStyleATTR
-			m.AttrName[key] = style[5 : len(style)-1]
-		} else {
-			m.PieceStyles[key] = PieceStyleTEXT
-		}
-	} else {
-		m.PieceStyles[key] = PieceStyleTEXT
+	ptp := matches[3] // piece type
+	val := matches[4]
+	style := PieceStyle(0)
+	style, ok := pieceStyles[ptp]
+	//fmt.Println("]", key, ptp, style, ok, val)
+	if len(ptp) != 0 && !ok {
+		return errors.New("Piece style specification error. To get help, run: " + progname)
 	}
 	m.Keys = append(m.Keys, key)
+	m.PieceStyles[key] = style
 	m.Values[key] = val
 	return nil
 }