htmltools

commit 96ec22c55d4ec4d04176ea875dadebc9515a9fb7
parent f7df01c6c7c882635891edbc13585262f92c5717
Author: Lukas Henkel <lh@entf.net>
Date:   Mon, 17 Jun 2019 22:03:10 +0200

Add htmlselect

Diffstat:
M
Makefile
 | 
2
+-
A
htmlselect/htmlselect.1.scd
 | 
21
+++++++++++++++++++++
A
htmlselect/main.go
 | 
33
+++++++++++++++++++++++++++++++++

3 files changed, 55 insertions(+), 1 deletion(-)
diff --git a/Makefile b/Makefile
@@ -1,4 +1,4 @@
-TOOLS = htmlremove htmltotext htmlunwrap
+TOOLS = htmlremove htmltotext htmlunwrap htmlselect
 PREFIX = /usr/local
 MANS = $(shell find . -name '*.scd' | sed s/\.scd//)
 
diff --git a/htmlselect/htmlselect.1.scd b/htmlselect/htmlselect.1.scd
@@ -0,0 +1,21 @@
+HTMLREMOVE(1)
+
+# NAME
+
+htmlselect - print all elements matching a selector from an HTML document
+
+# SYNOPSIS
+
+*htmlselect* SELECTOR [_FILE_]...
+
+# DESCRIPTION
+
+Reads each file in sequence and prints all elements matching SELECTOR to
+standard output. If no FILE is given or FILE is -, read standard input.
+
+If any FILE cannot be processed, a message prefixed with the FILE name will be
+written to standard error.
+
+# AUTHOR
+
+Lukas Henkel <lh@entf.net>
diff --git a/htmlselect/main.go b/htmlselect/main.go
@@ -0,0 +1,33 @@
+package main // import "entf.net/htmltools/htmlselect"
+
+import (
+	"fmt"
+	"os"
+
+	"github.com/andybalholm/cascadia"
+	"golang.org/x/net/html"
+
+	"entf.net/htmltools/shared"
+)
+
+func main() {
+	args := os.Args[1:]
+	if len(args) == 0 {
+		fmt.Println("usage: htmlselect SELECTOR [FILES...]")
+		os.Exit(1)
+	}
+	sel, err := cascadia.Compile(args[0])
+	if err != nil {
+		fmt.Fprintf(os.Stderr, "selector invalid: %v\n", err)
+		os.Exit(1)
+	}
+	shared.Main(args[1:], func(doc *html.Node) {
+		dosel(sel, doc)
+	})
+}
+
+func dosel(sel cascadia.Selector, doc *html.Node) {
+	for _, n := range sel.MatchAll(doc) {
+		html.Render(os.Stdout, n)
+	}
+}