From d0763e99a75f6c1c01029e57c8835b5566659722 Mon Sep 17 00:00:00 2001 From: Marcell Mars Date: Mon, 4 Apr 2022 13:46:55 +0200 Subject: [PATCH] more parsing of zotero's rdf.. --- cmd/build.go | 96 ++++++++++++++++++++++++++++++++++++++-------------- 1 file changed, 70 insertions(+), 26 deletions(-) diff --git a/cmd/build.go b/cmd/build.go index d6ed048..dc68913 100644 --- a/cmd/build.go +++ b/cmd/build.go @@ -4,7 +4,6 @@ import ( "encoding/xml" "fmt" "log" - "os" "accorder/pkg/calibre" @@ -13,16 +12,23 @@ import ( "github.com/spf13/viper" ) -type FileAttachment struct { - XMLName xml.Name `xml:"resource"` - Path string `xml:"resource,attr"` +// best zotero rdf documentation: +// https://github.com/zotero/translators/blob/master/Zotero%20RDF.js +// + +type ZoteroItem struct { + XMLName xml.Name `xml:"zoteroItem"` + FilePath string `xml:"filePath"` + MimeType string `xml:"mimeType"` + Publisher string `xml:"publisher,omitempty"` + Authors []ZoteroAuthor `xml:"authors"` + Title string `xml:"title"` + Description string `xml:"description,omitempty"` } -type ZoteroAttachment struct { - XMLName xml.Name `xml:"Attachment"` - ReferenceID string `xml:"about,attr"` - FileAttachment FileAttachment - FileType string `xml:"type"` +type ZoteroAuthor struct { + XMLName xml.Name `xml:"authors"` + Author string `xml:"author"` } var buildCmd = &cobra.Command{ @@ -74,32 +80,70 @@ should be enough for the next successful build).`, if err := doc.ReadFromFile(bibtexPath); err != nil { panic(err) } - var zoteroAttachment ZoteroAttachment root := doc.SelectElement("rdf:RDF") - for _, z := range root.FindElements("[name()='link:type']") { - doc := etree.NewDocument() - doc.SetRoot(z.Parent().Copy()) - if z.Text() != "application/pdf" { + for _, attachmentNode := range root.FindElements("[name()='link:type']") { + var zoteroItem ZoteroItem + if attachmentNode.Text() != "application/pdf" { continue } - doc.WriteTo(os.Stdout) - b, err := doc.WriteToBytes() + newDoc := etree.NewDocument() + zoteroUnion := newDoc.CreateElement("zoteroItem") + filePathElement := zoteroUnion.CreateElement("filePath") + filePathQuery := attachmentNode.Parent().FindElement("[name()='rdf:resource']").SelectAttr("rdf:resource").Value + filePathElement.CreateText(filePathQuery) + mimeTypeElement := zoteroUnion.CreateElement("mimeType") + mimeType := attachmentNode.Text() + mimeTypeElement.CreateText(mimeType) + + bibliographyNode := root.FindElement(fmt.Sprintf("[@rdf:resource='%s']", attachmentNode.Parent().SelectAttr("rdf:about").Value)).Parent().Copy() + + publisherQuery := bibliographyNode.FindElement("[name()='foaf:name']") + if publisherQuery != nil { + publisherElement := zoteroUnion.CreateElement("publisher") + publisher := publisherQuery.Text() + publisherElement.CreateText(publisher) + } + + authorsQuery := bibliographyNode.FindElements("[name()='foaf:Person']") + authors := zoteroUnion.CreateElement("authors") + for _, authorNode := range authorsQuery { + var firstName, surName string + author := authors.CreateElement("author") + + firstNameNode := authorNode.FindElement("[name()='foaf:givenName']") + if firstNameNode != nil { + firstName = firstNameNode.Text() + } + surNameNode := authorNode.FindElement("[name()='foaf:surname']") + if surNameNode != nil { + surName = surNameNode.Text() + } + author.CreateText(fmt.Sprintf("%s %s", firstName, surName)) + } + + titleQuery := bibliographyNode.FindElement("[name()='dc:title']") + if titleQuery != nil { + titleNode := zoteroUnion.CreateElement("title") + title := titleQuery.Text() + titleNode.CreateText(title) + } + + descriptionQuery := bibliographyNode.FindElement("[name()='dcterms:abstract']") + if descriptionQuery != nil { + descriptionNode := zoteroUnion.CreateElement("description") + description := descriptionQuery.Text() + descriptionNode.CreateText(description) + } + // newDoc.WriteTo(os.Stdout) + b, err := newDoc.WriteToBytes() if err != nil { log.Fatal(err) } - if err := xml.Unmarshal(b, &zoteroAttachment); err != nil { + if err := xml.Unmarshal(b, &zoteroItem); err != nil { log.Fatalln(err) } - fmt.Printf("\nZOTEROATTACHMENT: %#v\n", zoteroAttachment) + fmt.Printf("\nZoteroItem: %#v\n", zoteroItem) - fmt.Println(z.Text()) - fmt.Println("~ ~ ~ ~") - fmt.Println(doc.WriteTo(os.Stdout)) - elm := root.FindElement(fmt.Sprintf("[@rdf:resource='%s']", z.Parent().SelectAttr("rdf:about").Value)) - doc.SetRoot(elm.Parent().Copy()) - fmt.Println("+++++") - fmt.Println(doc.WriteTo(os.Stdout)) - fmt.Println("~ ~ ~ ~") } calibre.RenderStandaloneApp(calibrePath, librarianName, libraryUUID, librarySecret, jsonPath) },