cli: add include & exclude selectors

cli/cmd/cmd_convert.go

@@ -4,9 +4,13 @@ import (
 	"bytes"
 	"fmt"
 
+	"github.com/JohannesKaufmann/dom"
 	"github.com/JohannesKaufmann/html-to-markdown/v2/converter"
 	"github.com/JohannesKaufmann/html-to-markdown/v2/plugin/base"
 	"github.com/JohannesKaufmann/html-to-markdown/v2/plugin/commonmark"
+	"github.com/JohannesKaufmann/html-to-markdown/v2/plugin/strikethrough"
+	"github.com/andybalholm/cascadia"
+	"golang.org/x/net/html"
 )
 
 func overrideValidationError(e *commonmark.ValidateConfigError) error {
@@ -26,8 +30,66 @@ func overrideValidationError(e *commonmark.ValidateConfigError) error {
 	e.KeyWithValue = fmt.Sprintf("--%s=%q", e.Key, e.Value)
 	return e
 }
-func (cli *CLI) convert(input []byte) ([]error, error) {
 
+func (cli *CLI) includeNodesFromDoc(doc *html.Node) (*html.Node, error) {
+	if len(cli.config.includeSelector) == 0 {
+		return doc, nil
+	}
+	nodes := cascadia.QueryAll(doc, cli.config.includeSelector)
+
+	root := &html.Node{}
+	for _, n := range nodes {
+		dom.RemoveNode(n)
+		root.AppendChild(n)
+	}
+
+	return root, nil
+}
+func (cli *CLI) excludeNodesFromDoc(doc *html.Node) error {
+	if len(cli.config.excludeSelector) == 0 {
+		return nil
+	}
+
+	var finder func(node *html.Node)
+	finder = func(node *html.Node) {
+		if cli.config.excludeSelector.Match(node) {
+			dom.RemoveNode(node)
+			return
+		}
+
+		for child := node.FirstChild; child != nil; child = child.NextSibling {
+			// Because we are sometimes removing a node, this causes problems
+			// with the for loop. Using `defer` is a cool trick!
+			// https://gist.github.com/loopthrough/17da0f416054401fec355d338727c46e
+			defer finder(child)
+		}
+	}
+	finder(doc)
+
+	return nil
+}
+func (cli *CLI) parseInputWithSelectors(input []byte) (*html.Node, error) {
+	r := bytes.NewReader(input)
+
+	doc, err := html.Parse(r)
+	if err != nil {
+		return nil, fmt.Errorf("error while parsing html: %w", err)
+	}
+
+	doc, err = cli.includeNodesFromDoc(doc)
+	if err != nil {
+		return nil, err
+	}
+
+	err = cli.excludeNodesFromDoc(doc)
+	if err != nil {
+		return nil, err
+	}
+
+	return doc, nil
+}
+
+func (cli *CLI) convert(input []byte) ([]error, error) {
 	conv := converter.NewConverter(
 		converter.WithPlugins(
 			base.NewBasePlugin(),
@@ -36,9 +98,17 @@ func (cli *CLI) convert(input []byte) ([]error, error) {
 			),
 		),
 	)
+	if cli.config.enablePluginStrikethrough {
+		// TODO: while this works, this does not add the `Name` to the internal list
+		strikethrough.NewStrikethroughPlugin().Init(conv)
+	}
 
-	r := bytes.NewReader(input)
-	markdown, err := conv.ConvertReader(r)
+	doc, err := cli.parseInputWithSelectors(input)
+	if err != nil {
+		return nil, err
+	}
+
+	markdown, err := conv.ConvertNode(doc, converter.WithDomain(cli.config.domain))
 	if err != nil {
 		e, ok := err.(*commonmark.ValidateConfigError)
 		if ok {

cli/cmd/exec.go

@@ -7,6 +7,8 @@ import (
 	"io"
 	"os"
 	"strings"
+
+	"github.com/andybalholm/cascadia"
 )
 
 var (
@@ -22,13 +24,18 @@ type Config struct {
 	// args are the positional (non-flag) command-line arguments.
 	args []string
 
+	// - - - - - General - - - - - //
 	version bool
+	domain  string
 
-	// - - - - //
+	includeSelector cascadia.SelectorGroup
+	excludeSelector cascadia.SelectorGroup
 
+	// - - - - - Options - - - - - //
 	strongDelimiter string
 
-	plugins []string
+	// - - - - - Plugins - - - - - //
+	enablePluginStrikethrough bool
 }
 
 // Release holds the information (from the 3 ldflags) that goreleaser sets.