sitemap

A fast, memory-efficient XML sitemap parser for Go. It tokenizes sitemaps and sitemap indexes incrementally, reusing buffers across entries, and streams results via channels without loading entire documents into memory.

• Automatic Recursion: Handles sitemap indexes by fetching and parsing child sitemaps concurrently.
• Streaming & Low-Allocation: Entries are streamed via channels as they are tokenized, avoiding the need to hold the entire document in memory at once.
• Dual Parsing Modes: Defaults to the reliable standard library encoding/xml, with an opt-in ultra-fast custom lexer for maximum throughput.
• Selective Extraction: Extract only the fields you need (like <lastmod>, <changefreq>, or <priority>). Each field is opt-in, so you only pay for the parsing you actually use.
• Flexible Filtering:
  • Date-based: Drop entries older than a specific cutoff.
  • Regex-based: Whitelist or blacklist specific URIs or URLs.
• Jalali Date Support: Automatically detects and parses Jalali dates in both <lastmod> values and WithModifiedSince cutoffs.
• Customizable HTTP Requests: Fully supports custom HTTP clients, allowing you to easily configure proxies, timeouts, and other transport settings.

The following results are from benchmarking a single sitemap containing 50,000 <loc> entries (the maximum URLs allowed in a sitemap per the sitemaps.org protocol).

(full: <loc> + <lastmod> + <changefreq> + <priority>)

For reference, here are benchmark numbers for other Go sitemap packages:

Run it yourself in /benchmarks, with:

go test -bench . -benchtime=5s -benchmem -count=6

go get github.com/HerneHunter/sitemap

Fetch follows sitemap indexes recursively and streams only the final page URLs:

package main

import (
	"context"
	"fmt"

	"github.com/HerneHunter/sitemap"
)

func main() {
	ctx := context.Background()

	for result := range sitemap.Fetch(ctx, "https://example.com/sitemap_index.xml") {
		fmt.Println(result.Loc)
	}
}

FetchAll does the same for several starting URLs concurrently, merging everything into one channel.

for result := range sitemap.ParseFile(ctx, "sitemap.xml") {
	fmt.Println(result.Loc)
}

// reader could be an os.File, strings.Reader, etc.
for result := range sitemap.Parse(ctx, reader) {
	fmt.Println(result.Loc)
}

All streaming/fetching functions take ...Option. They can be combined freely.

Every option follows the same pattern. pass it in as an argument, wherever it fits in the call:

sitemap.Fetch(ctx, url, sitemap.WithLastMod())

The rest work exactly the same way. Combine as many as you need:

cutoff := time.Now().AddDate(0, -1, 0) // last 30 days

sitemap.Fetch(ctx, url,
	sitemap.WithCustomLexer(),
	sitemap.WithChangeFreq(),
	sitemap.WithPriority(),
	sitemap.WithModifiedSince(cutoff),
	sitemap.WithMaxConcurrency(5),
)

WithSitemapFilter and WithSitemapIndexFilter both take a *Filter, which is a bit richer than the other options since it has three independent fields:

// Filter provides rules for filtering sitemap or index URLs.
type Filter struct {
	// Applied to the URI (path + query). If set, the URI must match at least one.
	Whitelist []*regexp.Regexp

	// Applied to the URI (path + query). If set, the URI must not match any.
	Blacklist []*regexp.Regexp

	// Applied to the full loc string, including the domain.
	RawMatch *regexp.Regexp
}

filter := &sitemap.Filter{
	Blacklist: []*regexp.Regexp{
		regexp.MustCompile(`^/admin`),
		regexp.MustCompile(`\?sort=`),
	},
}
sitemap.Fetch(ctx, url, sitemap.WithSitemapFilter(filter))

WithSitemapIndexFilter works the same way, but applies to URLs within a sitemap index. sitemaps filtered out here are never fetched or parsed.

Every function returns/streams ParseResult values. A result carries the parsed fields (like Loc, LastMod, ChangeFreq, Priority) or an Err.

Errors you may see:

• ErrMalformedXML: the document couldn't be tokenized.
• *HTTPError: a fetch returned a non-2xx status code. Exposes URL and StatusCode.

If a <lastmod> value is present but fails to parse, it's not treated as an error: LastMod is simply left as its zero value (time.Time{}), same as when the field is missing entirely. So with WithModifiedSince, URLs with unparsable timestamps are not filtered out. They still appear in the output channel with a zero-value LastMod, leaving it up to your own code to decide whether to keep or drop them.