Allow user to specify charset

README.md

@@ -19,16 +19,6 @@ If you're on OS X, use [Brew](http://brew.sh/) to install (no Go required).
 
     brew install https://raw.githubusercontent.com/EricChiang/pup/master/pup.rb
 
-For linux distrubtions use the following commands to install under your `PATH`
-environment variable. You can set `ARCH` to `linux_386` for 32-bit infrastructures.
-
-    ARCH=linux_amd64
-    cd /tmp
-    wget https://github.com/EricChiang/pup/releases/download/v0.3.5/pup_${ARCH}.zip
-    unzip pup_${ARCH}.zip && rm pup_${ARCH}.zip
-    sudo mv pup /usr/local/bin
-    pup --version
-
 ## Quick start
 
 ```bash
@@ -353,5 +343,6 @@ output of pup into a more consumable format.
 -i --indent        number of spaces to use for indent or character
 -n --number        print number of elements selected
 -l --limit         restrict number of levels printed
+--charset          specify the charset for pup to use
 --version          display version
 ```

parse.go

@@ -6,16 +6,41 @@ import (
 	"os"
 	"strconv"
 	"strings"
+
+	"golang.org/x/net/html"
+	"golang.org/x/net/html/charset"
+	"golang.org/x/text/transform"
 )
 
 var (
 	pupIn            io.ReadCloser = os.Stdin
+	pupCharset       string        = ""
 	pupMaxPrintLevel int           = -1
 	pupPrintColor    bool          = false
 	pupIndentString  string        = " "
 	pupDisplayer     Displayer     = TreeDisplayer{}
 )
 
+// Parse the html while handling the charset
+func ParseHTML(r io.Reader, cs string) (*html.Node, error) {
+	var err error
+	if cs == "" {
+		// attempt to guess the charset of the HTML document
+		r, err = charset.NewReader(r, "")
+		if err != nil {
+			return nil, err
+		}
+	} else {
+		// let the user specify the charset
+		e, name := charset.Lookup(cs)
+		if name == "" {
+			return nil, fmt.Errorf("'%s' is not a valid charset", cs)
+		}
+		r = transform.NewReader(r, e.NewDecoder())
+	}
+	return html.Parse(r)
+}
+
 func PrintHelp(w io.Writer, exitCode int) {
 	helpString := `Usage
     pup [flags] [selectors] [optional display function]
@@ -28,6 +53,7 @@ Flags
     -i --indent        number of spaces to use for indent or character
     -n --number        print number of elements selected
     -l --limit         restrict number of levels printed
+    --charset          specify the charset for pup to use
     --version          display version
 `
 	fmt.Fprintf(w, helpString, VERSION)
@@ -81,6 +107,9 @@ func ProcessFlags(cmds []string) (nonFlagCmds []string, err error) {
 				return []string{}, fmt.Errorf("Argument for '%s' must be numeric", cmd)
 			}
 			i++
+		case "--charset":
+			pupCharset = cmds[i+1]
+			i++
 		case "--version":
 			fmt.Println(VERSION)
 			os.Exit(0)

pup.go

@@ -5,7 +5,6 @@ import (
 	"os"
 
 	"golang.org/x/net/html"
-	"golang.org/x/net/html/charset"
 )
 
 //      _=,_
@@ -17,7 +16,7 @@ import (
 //       |/ \_( # |"
 //      C/ ,--___/
 
-var VERSION string = "0.3.6"
+var VERSION string = "0.3.7"
 
 func main() {
 	// process flags and arguments
@@ -27,19 +26,13 @@ func main() {
 		os.Exit(2)
 	}
 
-	// Determine the charset of the input
-	cr, err := charset.NewReader(pupIn, "")
-	if err != nil {
-		fmt.Fprintf(os.Stderr, err.Error())
-		os.Exit(2)
-	}
-
 	// Parse the input and get the root node
-	root, err := html.Parse(cr)
+	root, err := ParseHTML(pupIn, pupCharset)
 	if err != nil {
-		fmt.Fprintf(os.Stderr, err.Error())
+		fmt.Fprintf(os.Stderr, "%s\n", err.Error())
 		os.Exit(2)
 	}
+	pupIn.Close()
 
 	// Parse the selectors
 	selectorFuncs := []SelectorFunc{}