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

1 files changed, 181 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:]))