htmltools

commit 8024985b34f00f9f06197e8d6d7a914c8873d6c6
parent 1d97d4af801b78d5bb5445d5e9affd4bed1c82bf
Author: Lukas Henkel <lh@entf.net>
Date:   Tue, 16 Feb 2021 20:10:58 +0100

New tool htmlattr

Diffstat:
M
Makefile
 | 
2
+-
A
htmlattr/htmlattr.1.scd
 | 
31
+++++++++++++++++++++++++++++++
A
htmlattr/main.go
 | 
58
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

3 files changed, 90 insertions(+), 1 deletion(-)
diff --git a/Makefile b/Makefile
@@ -1,4 +1,4 @@
-TOOLS = htmlremove htmltotext htmlunwrap htmlselect htmlindentheadings
+TOOLS = htmlremove htmltotext htmlunwrap htmlselect htmlindentheadings htmlattr
 PREFIX = /usr/local
 MANS = $(shell find . -name '*.scd' | sed s/\.scd//)
 
diff --git a/htmlattr/htmlattr.1.scd b/htmlattr/htmlattr.1.scd
@@ -0,0 +1,31 @@
+HTMLATTR(1)
+
+# NAME
+
+htmlattr - print specific attribute values from top level nodes
+
+# SYNOPSIS
+
+*htmlattr* [-fs FIELD_SEPERATOR] ATTRIBUTES [_FILE_]...
+# OPTIONS
+
+*-fs*
+	character that will be used to both split the ATTRIBUTES argument and
+	seperate the output columns. "," by default.
+
+
+# DESCRIPTION
+
+Reads each file in sequence, iterates over all top level nodes and prints
+requested attributes for each element line by line. If an element does not
+contain any attributes requested, no line is printed. If some but not all
+attributes exist on a node, the respective columns will be empty. htmlattr
+prints the same number of columns for every node. 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/htmlattr/main.go b/htmlattr/main.go
@@ -0,0 +1,58 @@
+package main // import "entf.net/htmltools/htmlattr"
+
+import (
+	"flag"
+	"fmt"
+	"os"
+	"strings"
+
+	"entf.net/htmltools/shared"
+	"golang.org/x/net/html"
+)
+
+func main() {
+	var fs string
+	flag.StringVar(&fs, "fs", ",", "field seperator")
+	flag.Parse()
+	args := flag.Args()
+	if len(args) == 0 {
+		fmt.Println("usage: htmlattr [-fs FIELD_SEPERATOR] ATTRIBUTES [FILES...]")
+		os.Exit(1)
+	}
+	attrs := strings.Split(args[0], fs)
+	for i, attr := range attrs {
+		attrs[i] = strings.ToLower(attr)
+	}
+	shared.Main(args[1:], func(doc *html.Node) {
+		var body *html.Node
+		for n := doc.FirstChild.FirstChild; n != nil; n = n.NextSibling {
+			if strings.ToLower(n.Data) == "body" {
+				body = n
+				break
+			}
+		}
+		if body == nil {
+			fmt.Fprintln(os.Stderr, "document does not contain a body")
+			os.Exit(1)
+		}
+		for n := body.FirstChild; n != nil; n = n.NextSibling {
+			if n.Type != html.ElementNode {
+				continue
+			}
+			list := make([]string, len(attrs))
+			var any bool
+			for i, attrn := range attrs {
+				for _, attr := range n.Attr {
+					if strings.ToLower(attr.Key) == attrn {
+						any = true
+						list[i] = attr.Val
+					}
+				}
+			}
+			line := strings.Join(list, fs)
+			if any {
+				fmt.Println(line)
+			}
+		}
+	})
+}