accorder/cmd/build.go

package cmd

import (
	"accorder/pkg/calibre"
	"encoding/xml"
	"fmt"
	"log"
	"strconv"
	"strings"

	"github.com/araddon/dateparse"
	"github.com/beevik/etree"
	"github.com/spf13/cobra"
	"github.com/spf13/viper"
)

// best zotero rdf documentation:
// https://github.com/zotero/translators/blob/master/Zotero%20RDF.js
//

type ZoteroItem struct {
	XMLName     xml.Name `xml:"zoteroItem"`
	Attachments struct {
		Attachment []struct {
			Path     string `xml:",chardata"`
			MimeType string `xml:"mimeType,attr"`
		} `xml:"attachment"`
	} `xml:"attachments"`
	Date    string `xml:"date"`
	Authors struct {
		Author []string `xml:"author"`
	} `xml:"authors"`
	Title       string `xml:"title"`
	Description string `xml:"description"`
	Publisher   string `xml:"publisher"`
}

var buildCmd = &cobra.Command{
	Use:   "build",
	Short: "Build standalone, portable webapp from Calibre library.",
	Long: `Build searchable, standalone, portable webapp against the local Calibre
library including all the metadata needed. It creates BROWSE_LIBRARY.html in
root directory of the local Calibre library. For search (authors, titles,
tags...) it uses rendered metadata from static/data{1-8}.js files.

Every time the directory path and/or librarian is provided it is saved in
the configuration file for the future use (therefore: 'accorder build SESSION'
should be enough for the next successful build).`,
	Args: cobra.ExactArgs(1),
	PreRun: func(cmd *cobra.Command, args []string) {
		session := args[0]

		for vipFlag, cliFlag := range map[string]string{
			"librarian_name": "librarian",
			"local_upload":   "directory",
			"library_uuid":   "library-uuid",
			"library_secret": "library-secret",
			"server_upload":  "server",
			"bucket_upload":  "bucket",
		} {
			viper.BindPFlag(fmt.Sprintf("%s.%s", session, vipFlag), cmd.Flags().Lookup(cliFlag))
		}
	},
	Run: func(cmd *cobra.Command, args []string) {
		session := args[0]

		MissingRequiredFlags(
			map[string]string{
				"librarian_name": "librarian",
				"local_upload":   "directory",
			},
			session,
			cmd,
		)

		libraryUUID := ViperValue(session, "library_uuid")
		librarySecret := ViperValue(session, "library_secret")
		calibrePath := ViperValue(session, "local_upload")
		librarianName := ViperValue(session, "librarian_name")
		jsonPath := CliFlagValue(cmd, "jsonpath")
		bibtexPath := CliFlagValue(cmd, "import-bibtex")

		doc := etree.NewDocument()
		if err := doc.ReadFromFile(bibtexPath); err != nil {
			panic(err)
		}
		root := doc.SelectElement("rdf:RDF")
		attachmentsIDs := map[string]bool{}
		for _, attachmentNode := range root.FindElements("[name()='link:type']") {
			var bookOpf calibre.BookOpfW
			var zoteroItem ZoteroItem

			attachmentID := attachmentNode.Parent().SelectAttr("rdf:about").Value
			if attachmentNode.Text() == "text/html" || attachmentsIDs[attachmentID] {
				continue
			}

			bibliographyNode := root.FindElement(fmt.Sprintf("[@rdf:resource='%s']", attachmentID)).Parent().Copy()

			newDoc := etree.NewDocument()
			zoteroUnion := newDoc.CreateElement("zoteroItem")

			attachmentsQuery := bibliographyNode.FindElements("[name()='link:link']")
			itemAttachments := zoteroUnion.CreateElement("attachments")
			mimeTypeMap := make(map[string]bool)
			for _, attachment := range attachmentsQuery {
				attachmentID := attachment.SelectAttr("rdf:resource").Value
				zAttachment := root.FindElement(fmt.Sprintf("[@rdf:about='%s']", attachmentID))
				mimeType := zAttachment.FindElement("[name()='link:type']").Text()
				if mimeType != "text/html" {
					if !mimeTypeMap[mimeType] {
						attachmentsIDs[attachmentID] = true
						filePath := zAttachment.FindElement("[name()='rdf:resource']").SelectAttr("rdf:resource").Value
						attachment := itemAttachments.CreateElement("attachment")
						attachment.SetText(filePath)
						attachment.CreateAttr("mimeType", mimeType)
						mimeTypeMap[mimeType] = true
					} else {
						fmt.Println("DUPLICATE mimeType:", mimeType, attachmentID)
					}
				}
			}

			dateQuery := bibliographyNode.FindElement("[name()='dc:date']")
			if dateQuery != nil {
				dateElement := zoteroUnion.CreateElement("date")
				date, err := dateparse.ParseAny(dateQuery.Text())
				if err == nil {
					formattedDate := date.Format("2006-01-02")
					dateElement.CreateText(formattedDate)

					bookOpf.Metadata.Published = formattedDate
				} else {
					newDateQuery := fmt.Sprintf("1 %s", dateQuery.Text())
					newDate, err := dateparse.ParseAny(newDateQuery)
					if err == nil {
						newFormattedDate := newDate.Format("2006-01-02")
						dateElement.CreateText(newFormattedDate)
						bookOpf.Metadata.Published = newFormattedDate
					} else {
						lastChanceDate := dateQuery.Text()[len(dateQuery.Text())-4:]
						year, err := strconv.Atoi(lastChanceDate)
						if err == nil {
							justYear := fmt.Sprintf("%d-01-01", year)
							dateElement.CreateText(justYear)
							bookOpf.Metadata.Published = justYear
						} else {
							fmt.Println("ERROR parsing date...", err)
						}
					}
				}

			}

			publisherQuery := bibliographyNode.FindElement("[name()='foaf:name']")
			if publisherQuery != nil {
				publisherElement := zoteroUnion.CreateElement("publisher")
				publisher := publisherQuery.Text()
				publisherElement.CreateText(publisher)

				bookOpf.Metadata.Publisher = publisher
			}

			authorsQuery := bibliographyNode.FindElements("[name()='foaf:Person']")
			authors := zoteroUnion.CreateElement("authors")
			for _, authorNode := range authorsQuery {
				var firstName, surName string

				firstNameNode := authorNode.FindElement("[name()='foaf:givenName']")
				if firstNameNode != nil {
					firstName = firstNameNode.Text() + " "
				}
				surNameNode := authorNode.FindElement("[name()='foaf:surname']")
				if surNameNode != nil {
					surName = surNameNode.Text() + " "
				}
				fullName := strings.TrimSuffix(fmt.Sprintf("%s%s", firstName, surName), " ")
				authors.CreateElement("author").SetText(fullName)

				bookOpf.Metadata.Creators = append(bookOpf.Metadata.Creators, calibre.Creator{
					Role: "aut",
					Name: fullName,
				})
			}

			titleQuery := bibliographyNode.FindElement("[name()='dc:title']")
			if titleQuery != nil {
				titleNode := zoteroUnion.CreateElement("title")
				title := titleQuery.Text()
				titleNode.CreateText(title)

				bookOpf.Metadata.Title = title
			}

			descriptionQuery := bibliographyNode.FindElement("[name()='dcterms:abstract']")
			if descriptionQuery != nil {
				descriptionNode := zoteroUnion.CreateElement("description")
				description := descriptionQuery.Text()
				descriptionNode.CreateText(description)

				bookOpf.Metadata.Description = description
			}

			b, err := newDoc.WriteToBytes()
			if err != nil {
				log.Fatal(err)
			}

			if err := xml.Unmarshal(b, &zoteroItem); err != nil {
				log.Fatalln(err)
			}

			for _, a := range zoteroItem.Attachments.Attachment {
				fmt.Println(a.Path, a.MimeType)
			}
			zi, _ := xml.MarshalIndent(zoteroItem, " ", " ")
			_ = zi
			// os.Stdout.Write(zi)
			// fmt.Println("\n~+~ ~ ~ ~ ~")
			// fmt.Printf("\nZoteroItem: %#v\n", zoteroItem)

			bookOpf.Version = "2.0"
			bookOpf.Xmlns = "http://www.idpf.org/2007/opf"
			bookOpf.UniqueIdentifier = "uuid_id"
			bookOpf.Metadata.DC = "http://purl.org/dc/elements/1.1/"
			bookOpf.Metadata.OPF = "http://www.idpf.org/2007/opf"

			bookOpf.Metadata.Identifiers = append(bookOpf.Metadata.Identifiers, calibre.Identifier{
				Scheme: "calibre",
				Id:     "calibre_id",
				Value:  "-1",
			})

			bookOpfOutput, err := xml.MarshalIndent(bookOpf, " ", " ")
			if err != nil {
				log.Fatalln(err)
			}
			_ = bookOpfOutput
			// os.Stdout.Write(bookOpfOutput)
			// fmt.Println("\n ~ ~ ~ ~ ~")
		}
		calibre.RenderStandaloneApp(calibrePath, librarianName, libraryUUID, librarySecret, jsonPath)
	},
}

func init() {
	rootCmd.AddCommand(buildCmd)

	buildCmd.Flags().StringP("directory", "d", "", "A local Calibre directory.")
	buildCmd.Flags().StringP("librarian", "l", "", "Librarian's name.")
	buildCmd.Flags().StringP("export-json", "e", "", "Path where to render all metadata into JSON.")
	buildCmd.Flags().StringP("import-bibtex", "i", "", "Import books from BibTex file into Calibre.")

	buildCmd.Flags().StringP("library-uuid", "u", "", "A library's UUID used if part of MotW.")
	buildCmd.Flags().StringP("library-secret", "p", "", "A password used if part of MotW.")
	buildCmd.Flags().StringP("server", "s", "minio.memoryoftheworld.org", "Server.")
	buildCmd.Flags().StringP("bucket", "b", "", "Server.")
	CustomHelpOutput(buildCmd)

	buildCmd.Flags().MarkHidden("library-uuid")
	buildCmd.Flags().MarkHidden("library-secret")
	buildCmd.Flags().MarkHidden("server")
	buildCmd.Flags().MarkHidden("bucket")
}