Golang: Generate Sitemap

By Xah Lee. Date: . Last updated: .

Here's a script to generate sitemap of a given directory.

// given a dir, generate a sitemap.xml file for all its html files
// version 2018-11-04

package main

import (
	"fmt"
	"log"
	"os"
	"path/filepath"
	"regexp"
	"strings"
)

var dirsToProcess = []string{
	"/Users/xah/web/ergoemacs_org",
	"/Users/xah/web/wordyenglish_com",
	"/Users/xah/web/xaharts_org",
	"/Users/xah/web/xahlee_info",
	"/Users/xah/web/xahlee_org",
	"/Users/xah/web/xahmusic_org",
	"/Users/xah/web/xahsl_org",
}

var destFname = "sitemap.xml"

var dirsToSkip = []string{
	".git",
	"REC-SVG11-20110816",
	"clojure-doc-1.8",
	"css_2.1_spec",
	"css_transitions",
	"javascript_ecma-262_5.1_2011",
	"javascript_ecma-262_6_2015",
	"javascript_es2016",
	"javascript_es6",
	"node_api",
	"ocaml_doc",
}

// fnameRegex. only these are searched
const fnameRegex = `\.html$`

var fnameRegexToSkip = []string{
	`^xx`,
	`^403error.html`,
	`^404error.html`,
}

var dirRegexToSkip = []string{
	`^\.git$`,
	`^xx`,
}

var dirPathToUrl = map[string]string{
	"/Users/xah/web/ergoemacs_org":    "http://ergoemacs.org",
	"/Users/xah/web/wordyenglish_com": "http://wordyenglish.com",
	"/Users/xah/web/xaharts_org":      "http://xaharts.org",
	"/Users/xah/web/xahlee_info":      "http://xahlee.info",
	"/Users/xah/web/xahlee_org":       "http://xahlee.org",
	"/Users/xah/web/xahmusic_org":     "http://xahmusic.org",
	"/Users/xah/web/xahsl_org":        "http://xahsl.org",
}

// getMatched return the pair from mm, whose key is a prefix in ss. If none, panic.
// version 2018-09-02
var getMatched = func(ss string, mm map[string]string) []string {
	var bb = []string{``, ``}
	for k, v := range mm {
		if strings.HasPrefix(ss, k) {
			bb[0] = k
			bb[1] = v
			return bb
		}
	}
	panic("logic error. 83580")
	return nil
}

// equalAny return true if x equals any of y
// version 2018-09-02
func equalAny(x string, y []string) bool {
	for _, v := range y {
		if x == v {
			return true
		}
	}
	return false
}

// matchAny return true if ss is matched by any of regex regexes.
// version 2018-09-01
func matchAny(ss string, regexes []string) bool {
	for _, re := range regexes {
		result, err := regexp.MatchString(re, ss)
		if err != nil {
			panic(err)
		}
		if result {
			return true
		}
	}
	return false
}

// getHeadBytes return the first n bytes in file at path
// version 2018-09-02
func getHeadBytes(path string, n int) []byte {
	file, err := os.Open(path) // For read access.
	if err != nil {
		log.Fatal(err)
	}
	defer file.Close()
	headBytes := make([]byte, n)
	m, err := file.Read(headBytes)
	if err != nil {
		log.Fatal(err)
	}
	return headBytes[:m]
}

func doFile(path string, path2Url []string) (output []byte) {
	var firstLinish = getHeadBytes(path, 200)
	var pmoved, err = regexp.Match("page_moved_64598", firstLinish)
	if err != nil {
		panic(err)
	}
	if !pmoved {
		output = append(output, fmt.Sprintf("<url><loc>%v</loc></url>\n", strings.Replace(path, path2Url[0], path2Url[1], 1))...)

	}
	return output
}

func writeIt(contentX []byte, pathX string) {
	var fileH, err = os.Create(pathX)
	if err != nil {
		panic(err)
	}
	defer fileH.Close()
	var _, errW = fileH.Write(contentX)
	if errW != nil {
		panic(errW)
	}
}

func sitemap(dirX string, path2Url []string) []byte {
	var output = []byte(`<?xml version="1.0" encoding="UTF-8"?>` + "\n")
	output = append(output, (`<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">` + "\n")...)
	var pWalker = func(pathX string, infoX os.FileInfo, errX error) error {
		if errX != nil {
			panic(fmt.Sprintf("error 「%v」 at a path 「%q」\n", errX, pathX))
		}
		if infoX.IsDir() {
			if equalAny(filepath.Base(pathX), dirsToSkip) || matchAny(filepath.Base(pathX), dirRegexToSkip) {
				return filepath.SkipDir
			}
		} else {
			var fname = filepath.Base(pathX)
			var goodExtension, err = regexp.MatchString(fnameRegex, fname)
			if err != nil {
				panic("stupid golang MatchString error")
			}
			if goodExtension && !matchAny(fname, fnameRegexToSkip) {
				output = append(output, doFile(pathX, path2Url)...)
			}
		}
		return nil
	}
	err := filepath.Walk(dirX, pWalker)
	if err != nil {
		fmt.Printf("error walking the path %q: %v\n", dirX, err)
	}
	output = append(output, (`</urlset>` + "\n")...)
	return output
}

func main() {
	var outBytes []byte
	for _, v := range dirsToProcess {
		var path2Url = getMatched(v, dirPathToUrl)
		outBytes = nil
		outBytes = sitemap(v, path2Url)
		var saveToPath = filepath.Join(v, destFname)
		writeIt(outBytes, saveToPath)
		fmt.Printf("file saved to: %v\n", saveToPath)
	}
	fmt.Println("Done")
}

If you have a question, put $5 at patreon and message me.

Golang

  1. Compile, Run
  2. Source Encoding
  3. Package, Import
  4. Comment
  5. Print
  6. String
  7. Rune
  8. String as Chars
  9. Variable
  10. Zero Value
  11. Constant
  12. If Then Else
  13. Switch/Case
  14. Loop
  15. Basic Types
  16. Array
  17. Slice
  18. Map
  19. Struct
  20. Function
  21. regexp
  22. Read File
  23. Write to File
  24. Walk Dir
  25. Check File Exist
  26. System Call
  27. Get Script Path
  28. Pointer
  29. Defer
  30. Random Number

Examples

  1. match any regexp
  2. Validate Links
  3. Generate Sitemap

Reference

  1. Go Spec