feat(html): parse link rel in html

2 files changed, 185 insertions, 0 deletions

diff --git a/html.go b/html.go
new file mode 100644
index 0000000..8c968f5
--- /dev/null
+++ b/html.go
@@ -0,0 +1,88 @@
+package human
+
+import (
+	"net/url"
+	"strings"
+)
+
+// GetURLFromHTML returns the [url.URL] extracted from the raw HTML.
+// Base is the URL containing the HTML.
+//
+// Returns nil if nothing is found.
+func GetURLFromHTML(b []byte, base *url.URL) (*url.URL, error) {
+	content := string(b)
+	i := strings.Index(content, `<link `)
+	for i >= 0 {
+		if i+6 >= len(content) {
+			return nil, nil
+		}
+		args := parseArgs(string([]rune(content)[i+6:]))
+		if args["rel"] == "human-json" {
+			href, ok := args["href"]
+			if ok {
+				u, err := url.Parse(href)
+				if err != nil {
+					return nil, err
+				}
+				if u.Host != "" {
+					return u, nil
+				}
+				path := u.Path
+				*u = *base
+				if strings.HasPrefix(path, "/") {
+					u.Path = path
+				} else {
+					u = u.JoinPath(path)
+				}
+				return u, nil
+			}
+		}
+		i = strings.IndexAny(content[i:], `<link `)
+	}
+	return nil, nil
+}
+
+func parseArgs(base string) map[string]string {
+	content := []rune(base)
+	i := 0
+	res := map[string]string{}
+
+	var sep bool
+	var key strings.Builder
+	var value strings.Builder
+	var quote bool
+	for i < len(content) && content[i] != '>' && (i+1 == len(content) || string(content[i:i+2]) != "/>") {
+		curr := content[i]
+		if !sep {
+			switch curr {
+			case '=':
+				sep = true
+			case ' ':
+				sep = false
+				if key.Len() > 0 {
+					res[key.String()] = ""
+				}
+				key.Reset()
+			default:
+				key.WriteRune(curr)
+			}
+		} else {
+			if value.Len() == 0 && curr == '"' {
+				quote = true
+			} else if (curr == '"' && quote) || curr == ' ' && !quote {
+				quote = false
+				sep = false
+				res[key.String()] = value.String()
+				key.Reset()
+				value.Reset()
+			} else {
+				value.WriteRune(curr)
+			}
+		}
+		i++
+	}
+	if key.Len() > 0 {
+		res[key.String()] = value.String()
+	}
+	return res
+}
diff --git a/html_test.go b/html_test.go
new file mode 100644
index 0000000..ab2bda7
--- /dev/null
+++ b/html_test.go
@@ -0,0 +1,97 @@
+package human
+
+import (
+	"net/url"
+	"testing"
+)
+
+func TestGetURLFromHTML(t *testing.T) {
+	base, _ := url.Parse(`https://example.org/foo/`)
+	u, err := GetURLFromHTML([]byte(`<link rel=human-json href=/human.json>`), base)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if u == nil {
+		t.Fatal("not found")
+	}
+	if u.Path != "/human.json" {
+		t.Errorf("invalid path: %s", u.Path)
+	}
+	if u.Host != "example.org" {
+		t.Errorf("invalid host: %s", u.Host)
+	}
+
+	u, err = GetURLFromHTML([]byte(`<link rel="human-json" href="/human.json">`), base)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if u == nil {
+		t.Fatal("not found")
+	}
+	if u.Path != "/human.json" {
+		t.Errorf("invalid path: %s", u.Path)
+	}
+	if u.Host != "example.org" {
+		t.Errorf("invalid host: %s", u.Host)
+	}
+
+	u, err = GetURLFromHTML([]byte(`<link rel="human-json" href="human.json">`), base)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if u == nil {
+		t.Fatal("not found")
+	}
+	if u.Path != "/foo/human.json" {
+		t.Errorf("invalid path: %s", u.Path)
+	}
+	if u.Host != "example.org" {
+		t.Errorf("invalid host: %s", u.Host)
+	}
+
+	u, err = GetURLFromHTML([]byte(`<link rel="human-json" href="../human.json">`), base)
+	if err != nil {
+		t.Fatal(err)
+	}
+	if u == nil {
+		t.Fatal("not found")
+	}
+	if u.Path != "/human.json" {
+		t.Errorf("invalid path: %s", u.Path)
+	}
+	if u.Host != "example.org" {
+		t.Errorf("invalid host: %s", u.Host)
+	}
+}
+
+func TestParseArgs(t *testing.T) {
+	args := parseArgs(`key="hello world">`)
+	if args["key"] != "hello world" {
+		t.Errorf("invalid arg: %v", args)
+	}
+
+	args = parseArgs(`key=hello>`)
+	if args["key"] != "hello" {
+		t.Errorf("invalid arg: %v", args)
+	}
+
+	args = parseArgs(`key=hello world>`)
+	if args["key"] != "hello" {
+		t.Errorf("invalid arg: %v", args)
+	}
+
+	args = parseArgs(`key=hello/>`)
+	if args["key"] != "hello" {
+		t.Errorf("invalid arg: %v", args)
+	}
+
+	args = parseArgs(`key=hello`)
+	if args["key"] != "hello" {
+		t.Errorf("invalid arg: %v", args)
+	}
+
+	args = parseArgs(`key=word foo=bar`)
+	if args["key"] != "word" || args["foo"] != "bar" {
+		t.Errorf("invalid args: %v", args)
+	}
+}