move gemtext->atom conversion into package atomconv like the other converters

2 files changed, 320 insertions, 0 deletions

diff --git a/gemini/gemtext/atomconv/convert.go b/gemini/gemtext/atomconv/convert.go
new file mode 100644
index 0000000..30228f3
--- /dev/null
+++ b/gemini/gemtext/atomconv/convert.go
@@ -0,0 +1,181 @@
+package atomconv
+
+import (
+	"bytes"
+	"context"
+	"html/template"
+	"io"
+	"mime"
+	"net/url"
+	"regexp"
+	"strconv"
+	"strings"
+	"time"
+
+	"tildegit.org/tjp/sliderule/gemini"
+	"tildegit.org/tjp/sliderule/gemini/gemtext"
+	"tildegit.org/tjp/sliderule/internal/types"
+)
+
+// Convert turns a gemini document to Atom format.
+//
+// It identifies feed fields and entries according to the specification at
+// gemini://gemini.circumlunar.space/docs/companion/subscription.gmi
+func Convert(wr io.Writer, doc gemtext.Document, location *url.URL) error {
+	if location == nil {
+		panic("atomconv.Convert: provided location was nil")
+	}
+
+	if _, err := wr.Write([]byte(`<?xml version="1.0" encoding="utf-8"?>`)); err != nil {
+		return err
+	}
+	if _, err := wr.Write([]byte{'\n'}); err != nil {
+		return err
+	}
+	return atomTmpl.Execute(wr, parseGemSub(doc, location))
+}
+
+// Auto is a middleware which builds atom feeds for any gemtext pages.
+//
+// It looks for requests ending with the '.atom' extension, passes through the request
+// with the extension clipped off, then if the response is in gemtext it converts it to
+// an Atom feed according to the gmisub spec at
+// gemini://gemini.circumlunar.space/docs/companion/subscription.gmi
+var Auto = types.Middleware(func(h types.Handler) types.Handler {
+	return types.HandlerFunc(func(ctx context.Context, request *types.Request) *types.Response {
+		if request.Scheme != "gemini" || !strings.HasSuffix(request.Path, ".atom") {
+			return h.Handle(ctx, request)
+		}
+
+		r := *request
+		u := *request.URL
+		u.Path = u.Path[:len(u.Path)-5]
+		r.URL = &u
+
+		response := h.Handle(ctx, &r)
+		if response.Status != gemini.StatusSuccess {
+			return response
+		}
+
+		mtype, _, err := mime.ParseMediaType(response.Meta.(string))
+		if err != nil || mtype != "text/gemini" {
+			return response
+		}
+
+		defer func() {
+			_ = response.Close()
+		}()
+
+		doc, err := gemtext.Parse(response.Body)
+		if err != nil {
+			return gemini.Failure(err)
+		}
+
+		buf := &bytes.Buffer{}
+		if err := Convert(buf, doc, request.URL); err != nil {
+			return gemini.Failure(err)
+		}
+		return gemini.Success("application/atom+xml; charset=utf-8", buf)
+	})
+})
+
+type gmiSub struct {
+	ID       template.URL
+	Title    string
+	Subtitle string
+	Updated  string
+
+	Entries []gmiSubEntry
+}
+
+type gmiSubEntry struct {
+	ID      template.URL
+	Updated string
+	Title   string
+}
+
+var linkElemRE = regexp.MustCompile(`(\d{4})-([0-1]\d)-([0-3]\d)`)
+
+func parseGemSub(doc gemtext.Document, location *url.URL) *gmiSub {
+	sub := &gmiSub{ID: template.URL(location.String())}
+	updated := time.Time{}
+
+	for i, line := range doc {
+		switch line.Type() {
+		case gemtext.LineTypeHeading1:
+			if sub.Title != "" {
+				continue
+			}
+
+			sub.Title = line.(gemtext.HeadingLine).Body()
+
+			for { // skip any empty lines
+				i += 1
+				if i >= len(doc) || strings.TrimPrefix(doc[i].String(), "\r") != "\n" {
+					break
+				}
+			}
+			if i < len(doc) && doc[i].Type() == gemtext.LineTypeHeading2 {
+				sub.Subtitle = doc[i].(gemtext.HeadingLine).Body()
+			}
+		case gemtext.LineTypeLink:
+			label := line.(gemtext.LinkLine).Label()
+			if len(label) < 10 {
+				continue
+			}
+			match := linkElemRE.FindStringSubmatch(label[:10])
+			if match == nil {
+				continue
+			}
+
+			year, err := strconv.Atoi(match[1])
+			if err != nil {
+				continue
+			}
+			month, err := strconv.Atoi(match[2])
+			if err != nil || month > 12 {
+				continue
+			}
+			day, err := strconv.Atoi(match[3])
+			if err != nil || day > 31 {
+				continue
+			}
+
+			entryUpdated := time.Date(year, time.Month(month), day, 12, 0, 0, 0, time.UTC)
+			entryTitle := strings.TrimLeft(strings.TrimPrefix(strings.TrimLeft(label[10:], " \t"), "-"), " \t")
+
+			sub.Entries = append(sub.Entries, gmiSubEntry{
+				ID:      template.URL(line.(gemtext.LinkLine).URL()),
+				Updated: entryUpdated.Format(time.RFC3339),
+				Title:   entryTitle,
+			})
+
+			if entryUpdated.After(updated) {
+				updated = entryUpdated
+				sub.Updated = updated.Format(time.RFC3339)
+			}
+		}
+	}
+
+	return sub
+}
+
+var atomTmpl = template.Must(template.New("atom").Parse(`
+<feed xmlns="http://www.w3.org/2005/Atom">
+  <id>{{.ID}}</id>
+  <link href="{{.ID}}"/>
+  <title>{{.Title}}</title>
+  {{- if .Subtitle }}
+  <subtitle>{{.Subtitle}}</subtitle>
+  {{- end }}
+  <updated>{{.Updated}}</updated>
+{{- range .Entries }}
+  <entry>
+    <id>{{.ID}}</id>
+    <link rel="alternate" href="{{.ID}}"/>
+    <title>{{.Title}}</title>
+    <updated>{{.Updated}}</updated>
+  </entry>
+{{- end }}
+</feed>
+`[1:]))
diff --git a/gemini/gemtext/atomconv/convert_test.go b/gemini/gemtext/atomconv/convert_test.go
new file mode 100644
index 0000000..8adaa2e
--- /dev/null
+++ b/gemini/gemtext/atomconv/convert_test.go
@@ -0,0 +1,139 @@
+package atomconv
+
+import (
+	"bytes"
+	"context"
+	"fmt"
+	"io"
+	"net/url"
+	"testing"
+
+	"tildegit.org/tjp/sliderule"
+	"tildegit.org/tjp/sliderule/gemini"
+	"tildegit.org/tjp/sliderule/gemini/gemtext"
+	"tildegit.org/tjp/sliderule/internal/types"
+)
+
+func TestConvert(t *testing.T) {
+	tests := []struct {
+		url    string
+		input  string
+		output string
+	}{
+		{
+			url: "gemini://sombodys.site/a/page",
+			input: `
+# This is a gemlog page
+
+
+## with a subtitle after empty lines
+
+=> ./first-post.gmi 2023-08-25 - This is my first post
+`[1:],
+			output: `
+<?xml version="1.0" encoding="utf-8"?>
+<feed xmlns="http://www.w3.org/2005/Atom">
+  <id>gemini://sombodys.site/a/page</id>
+  <link href="gemini://sombodys.site/a/page"/>
+  <title>This is a gemlog page</title>
+  <subtitle>with a subtitle after empty lines</subtitle>
+  <updated>2023-08-25T12:00:00Z</updated>
+  <entry>
+    <id>./first-post.gmi</id>
+    <link rel="alternate" href="./first-post.gmi"/>
+    <title>This is my first post</title>
+    <updated>2023-08-25T12:00:00Z</updated>
+  </entry>
+</feed>
+`[1:],
+		},
+	}
+
+	for _, test := range tests {
+		t.Run(test.url, func(t *testing.T) {
+			doc, err := gemtext.Parse(bytes.NewBufferString(test.input))
+			if err != nil {
+				t.Fatal(err)
+			}
+			loc, err := url.Parse(test.url)
+			if err != nil {
+				t.Fatal(err)
+			}
+			out := &bytes.Buffer{}
+			if err := Convert(out, doc, loc); err != nil {
+				t.Fatal(err)
+			}
+			if out.String() != test.output {
+				t.Fatal("mismatched output")
+			}
+		})
+	}
+}
+
+func TestAuto(t *testing.T) {
+	rout := &sliderule.Router{}
+
+	rout.Route("/foo.gmi", types.HandlerFunc(func(ctx context.Context, request *types.Request) *types.Response {
+		return gemini.Success("text/gemini", bytes.NewBufferString(`
+# This is my gemini page
+
+## a subtitle
+
+=> ./first-post.gmi 2023-05-17 - My first post
+=> ./second-post.gmi 2023-06-02 second-ever post
+`[1:]))
+	}))
+
+	rout.Route("/bar.gmi", types.HandlerFunc(func(ctx context.Context, request *types.Request) *types.Response {
+		return gemini.Success("text/gemini", bytes.NewBufferString(`
+# Another homepage
+
+=> ./first-post.gmi 2023-05-17 - first post
+=> ./second-post.gmi 2023-06-02 second post
+`[1:]))
+	}))
+
+	h := Auto(rout.Handler())
+
+	response := h.Handle(context.Background(), &types.Request{URL: &url.URL{
+		Scheme: "gemini",
+		Host:   "127.0.0.1",
+		Path:   "/foo.gmi.atom",
+	}})
+	if response.Status != gemini.StatusSuccess {
+		t.Fatal("bad response code")
+	}
+
+	result, err := io.ReadAll(response.Body)
+	if err != nil {
+		t.Fatal(err)
+	}
+
+	target := `
+<?xml version="1.0" encoding="utf-8"?>
+<feed xmlns="http://www.w3.org/2005/Atom">
+  <id>gemini://127.0.0.1/foo.gmi.atom</id>
+  <link href="gemini://127.0.0.1/foo.gmi.atom"/>
+  <title>This is my gemini page</title>
+  <subtitle>a subtitle</subtitle>
+  <updated>2023-06-02T12:00:00Z</updated>
+  <entry>
+    <id>./first-post.gmi</id>
+    <link rel="alternate" href="./first-post.gmi"/>
+    <title>My first post</title>
+    <updated>2023-05-17T12:00:00Z</updated>
+  </entry>
+  <entry>
+    <id>./second-post.gmi</id>
+    <link rel="alternate" href="./second-post.gmi"/>
+    <title>second-ever post</title>
+    <updated>2023-06-02T12:00:00Z</updated>
+  </entry>
+</feed>
+`[1:]
+	if string(result) != target {
+		fmt.Println(target)
+		fmt.Println(string(result))
+		t.Fatal("response body")
+	}
+}