func parseRSS(c appengine.Context, body []byte, charsetReader func(string, io.Reader) (io.Reader, error)) (*Feed, []*Story, error) {
var f Feed
var s []*Story
r := rss.Rss{}
d := xml.NewDecoder(bytes.NewReader(body))
d.CharsetReader = charsetReader
d.DefaultSpace = "DefaultSpace"
if err := d.Decode(&r); err != nil {
return nil, nil, err
}
f.Title = r.Title
if t, err := parseDate(c, &f, r.LastBuildDate, r.PubDate); err == nil {
f.Updated = t
} else {
c.Warningf("no rss feed date: %v", f.Link)
}
f.Link = r.BaseLink()
f.Hub = r.Hub()
for _, i := range r.Items {
st := Story{
Link: i.Link,
Author: i.Author,
}
if i.Content != "" {
st.content = i.Content
} else if i.Description != "" {
st.content = i.Description
}
if i.Title != "" {
st.Title = i.Title
} else if i.Description != "" {
st.Title = i.Description
}
if st.content == st.Title {
st.Title = ""
}
st.Title = textTitle(st.Title)
if i.Guid != nil {
st.Id = i.Guid.Guid
}
if i.Enclosure != nil && strings.HasPrefix(i.Enclosure.Type, "audio/") {
st.MediaContent = i.Enclosure.Url
} else if i.Media != nil && strings.HasPrefix(i.Media.Type, "audio/") {
st.MediaContent = i.Media.URL
}
if t, err := parseDate(c, &f, i.PubDate, i.Date, i.Published); err == nil {
st.Published = t
st.Updated = t
}
s = append(s, &st)
}
return &f, s, nil
}
// makeOpf creates an epub opf for the specified rss feed.
func (e FeedEpub) makeOpf(rssFeed rss.Rss) (epub.Opf, []epub.Chapter, error) {
if len(rssFeed.Items) == 0 {
return epub.Opf{}, nil, errors.New("No feed items to build")
}
ids := []epub.Identifier{
{Value: rssFeed.BaseLink(),
IdentifierType: codelist5.URN}}
creator := epub.Creator{
Name: rssFeed.Items[0].Author,
Role: "aut",
}
metadata := epub.Metadata{
Language: "en", // maybe, lol!
Title: rssFeed.Title,
Creator: creator,
Date: rssFeed.LastBuildDate,
}
var manifestItems []epub.ManifestItem
var chapters []epub.Chapter
// TODO: This is a totally naive build of file names that may be non-unique.
for i, item := range rssFeed.Items {
var manifestItem epub.ManifestItem
var chapter epub.Chapter
// TODO: in a better world this would be generated by sanitized filename
manifestItem.Id = fmt.Sprintf("manifest_id_%d", i)
filename := "chapters/" + item.Title + ".xhtml"
manifestItem.Href = filename
manifestItem.Title = &item.Title
chapter.FileName = filename
// TODO: we should fetch all the assets used in each rss item and put them in the
// epub in an assets directory, then rewrite the paths to each asset. At the
// very least we should rewrite relative urls to actually point to the feed.
chapter.Contents = item.Description
manifestItem.MediaType = xhtmlMediaType
manifestItems = append(manifestItems, manifestItem)
chapters = append(chapters, chapter)
}
manifest := epub.Manifest{
ManifestItems: manifestItems,
}
opfRootFile := epub.OpfRootFile{
FullPath: fmt.Sprintf("OEBPS/%s.opf", rssFeed.Title),
MediaType: "application/oebps-package+xml",
Identifiers: ids,
Metadata: metadata,
Manifest: manifest,
}
opf := epub.Opf{
RootFiles: []epub.OpfRootFile{opfRootFile},
}
return opf, chapters, nil
}