From 139ee013a7ad8c173e126cc8ebd0531a225b3ac6 Mon Sep 17 00:00:00 2001
From: advplyr <advplyr@protonmail.com>
Date: Wed, 18 May 2022 19:25:18 -0500
Subject: [PATCH] Add:Parsing tags from OPF metadata #602, Update:OPF parser to
 check for prefix on package/metadata/meta objects

---
 server/objects/mediaTypes/Book.js | 14 ++++++---
 server/utils/parseOpfMetadata.js  | 52 ++++++++++++++++++++++---------
 2 files changed, 47 insertions(+), 19 deletions(-)

diff --git a/server/objects/mediaTypes/Book.js b/server/objects/mediaTypes/Book.js
index 864dd15a..5f1f507d 100644
--- a/server/objects/mediaTypes/Book.js
+++ b/server/objects/mediaTypes/Book.js
@@ -225,6 +225,7 @@ class Book {
   // Look for desc.txt, reader.txt, metadata.abs and opf file then update details if found
   async syncMetadataFiles(textMetadataFiles, opfMetadataOverrideDetails) {
     var metadataUpdatePayload = {}
+    var tagsUpdated = false
 
     var descTxt = textMetadataFiles.find(lf => lf.metadata.filename === 'desc.txt')
     if (descTxt) {
@@ -264,8 +265,13 @@ class Book {
         var opfMetadata = await parseOpfMetadataXML(xmlText)
         if (opfMetadata) {
           for (const key in opfMetadata) {
-            // Add genres only if genres are empty
-            if (key === 'genres') {
+
+            if (key === 'tags') { // Add tags only if tags are empty
+              if (opfMetadata.tags.length && (!this.tags.length || opfMetadataOverrideDetails)) {
+                this.tags = opfMetadata.tags
+                tagsUpdated = true
+              }
+            } else if (key === 'genres') { // Add genres only if genres are empty
               if (opfMetadata.genres.length && (!this.metadata.genres.length || opfMetadataOverrideDetails)) {
                 metadataUpdatePayload[key] = opfMetadata.genres
               }
@@ -290,9 +296,9 @@ class Book {
     }
 
     if (Object.keys(metadataUpdatePayload).length) {
-      return this.metadata.update(metadataUpdatePayload)
+      return this.metadata.update(metadataUpdatePayload) || tagsUpdated
     }
-    return false
+    return tagsUpdated
   }
 
   searchQuery(query) {
diff --git a/server/utils/parseOpfMetadata.js b/server/utils/parseOpfMetadata.js
index 8a9bb4f3..95d74775 100644
--- a/server/utils/parseOpfMetadata.js
+++ b/server/utils/parseOpfMetadata.js
@@ -70,14 +70,14 @@ function fetchLanguage(metadata) {
   return fetchTagString(metadata, 'dc:language')
 }
 
-function fetchSeries(metadata) {
-  if (typeof metadata.meta == "undefined") return null
-  return fetchTagString(metadata.meta, "calibre:series")
+function fetchSeries(metadataMeta) {
+  if (!metadataMeta) return null
+  return fetchTagString(metadataMeta, "calibre:series")
 }
 
-function fetchVolumeNumber(metadata) {
-  if (typeof metadata.meta == "undefined") return null
-  return fetchTagString(metadata.meta, "calibre:series_index")
+function fetchVolumeNumber(metadataMeta) {
+  if (!metadataMeta) return null
+  return fetchTagString(metadataMeta, "calibre:series_index")
 }
 
 function fetchNarrators(creators, metadata) {
@@ -91,21 +91,42 @@ function fetchNarrators(creators, metadata) {
   }
 }
 
+function fetchTags(metadata) {
+  if (!metadata['dc:tag'] || !metadata['dc:tag'].length) return []
+  return metadata['dc:tag'].filter(tag => (typeof tag === 'string'))
+}
+
+function stripPrefix(str) {
+  if (!str) return ''
+  return str.split(':').pop()
+}
+
 module.exports.parseOpfMetadataXML = async (xml) => {
   var json = await xmlToJSON(xml)
-  if (!json || !json.package || !json.package.metadata) return null
-  var metadata = json.package.metadata
+
+  if (!json) return null
+
+  // Handle <package ...> or with prefix <ns0:package ...>
+  const packageKey = Object.keys(json).find(key => stripPrefix(key) === 'package')
+  if (!packageKey) return null
+  const prefix = packageKey.split(':').shift()
+  var metadata = prefix ? json[packageKey][`${prefix}:metadata`] || json[packageKey].metadata : json[packageKey].metadata
+  if (!metadata) return null
 
   if (Array.isArray(metadata)) {
     if (!metadata.length) return null
     metadata = metadata[0]
   }
 
-  if (typeof metadata.meta != "undefined") {
-    metadata.meta = {}
-    for (var match of xml.matchAll(/<meta name="(?<name>.+)" content="(?<content>.+)"\/>/g)) {
-      metadata.meta[match.groups['name']] = [match.groups['content']]
-    }
+  const metadataMeta = prefix ? metadata[`${prefix}:meta`] || metadata.meta : metadata.meta
+
+  metadata.meta = {}
+  if (metadataMeta && metadataMeta.length) {
+    metadataMeta.forEach((meta) => {
+      if (meta && meta['$'] && meta['$'].name) {
+        metadata.meta[meta['$'].name] = [meta['$'].content || '']
+      }
+    })
   }
 
   var creators = parseCreators(metadata)
@@ -119,8 +140,9 @@ module.exports.parseOpfMetadataXML = async (xml) => {
     description: fetchDescription(metadata),
     genres: fetchGenres(metadata),
     language: fetchLanguage(metadata),
-    series: fetchSeries(metadata),
-    sequence: fetchVolumeNumber(metadata)
+    series: fetchSeries(metadata.meta),
+    sequence: fetchVolumeNumber(metadata.meta),
+    tags: fetchTags(metadata)
   }
   return data
 }
\ No newline at end of file