PDF metadata: Support reading/writing of tags

src/calibre/utils/podofo/__init__.py

@@ -13,7 +13,7 @@ from calibre.ebooks.metadata import MetaInformation, string_to_authors, \
     authors_to_string
 from calibre.utils.ipc.job import ParallelJob
 from calibre.utils.ipc.server import Server
-from calibre.ptempfile import PersistentTemporaryFile
+from calibre.ptempfile import PersistentTemporaryFile, TemporaryFile
 from calibre import prints
 
 podofo, podofo_err = plugins['podofo']
@@ -38,7 +38,7 @@ def get_metadata(stream, cpath=None):
     server.close()
     if job.result is None:
         raise ValueError('Failed to read metadata: ' + job.details)
-    title, authors, creator, ok = job.result
+    title, authors, creator, tags, ok = job.result
     if not ok:
         print 'Failed to extract cover:'
         print job.details
@@ -49,6 +49,8 @@ def get_metadata(stream, cpath=None):
     mi = MetaInformation(title, authors)
     if creator:
         mi.book_producer = creator
+    if tags:
+        mi.tags = tags
     if os.path.exists(pt.name): os.remove(pt.name)
     if ok:
         mi.cover = cpath
@@ -63,9 +65,17 @@ def get_metadata_quick(raw):
     author = p.author
     authors = string_to_authors(author) if author else  [_('Unknown')]
     creator = p.creator
+    try:
+        tags = [x.strip() for x in p.keywords.split(u',')]
+        tags = [x for x in tags if x]
+    except:
+        tags = []
+
     mi = MetaInformation(title, authors)
     if creator:
         mi.book_producer = creator
+    if tags:
+        mi.tags = tags
     return mi
 
 def get_metadata_(path, cpath=None):
@@ -77,6 +87,11 @@ def get_metadata_(path, cpath=None):
     author = p.author
     authors = string_to_authors(author) if author else  [_('Unknown')]
     creator = p.creator
+    try:
+        tags = [x.strip() for x in p.keywords.split(u',')]
+        tags = [x for x in tags if x]
+    except:
+        tags = []
     ok = True
     try:
         if cpath is not None:
@@ -93,7 +108,7 @@ def get_metadata_(path, cpath=None):
         traceback.print_exc()
         ok = False
 
-    return (title, authors, creator, ok)
+    return (title, authors, creator, tags, ok)
 
 def prep(val):
     if not val:
@@ -105,58 +120,66 @@ def prep(val):
 def set_metadata(stream, mi):
     if not podofo:
         raise Unavailable(podofo_err)
-    pt = PersistentTemporaryFile('_podofo.pdf')
-    pt.write(stream.read())
-    pt.close()
-    server = Server(pool_size=1)
-    job = ParallelJob('write_pdf_metadata', 'Write pdf metadata',
-        lambda x,y:x,  args=[pt.name, mi.title, mi.authors, mi.book_producer])
-    server.add_job(job)
-    while not job.is_finished:
-        time.sleep(0.1)
+    with TemporaryFile('_podofo_read.pdf') as inputf, \
+            TemporaryFile('_podofo_write.pdf') as outputf:
+        server = Server(pool_size=1)
+        with open(inputf, 'wb') as f:
+            shutil.copyfileobj(stream, f)
+        job = ParallelJob('write_pdf_metadata', 'Write pdf metadata',
+            lambda x,y:x,  args=[inputf, outputf, mi.title, mi.authors,
+                mi.book_producer, mi.tags])
+        server.add_job(job)
+        while not job.is_finished:
+            time.sleep(0.1)
+            job.update()
+
         job.update()
-
-    job.update()
-    server.close()
-    if job.failed:
-        prints(job.details)
-    elif job.result is not None:
-        stream.seek(0)
-        stream.truncate()
-        stream.write(job.result)
-        stream.flush()
-        stream.seek(0)
-    try:
-        os.remove(pt.name)
-    except:
-        pass
+        server.close()
+        if job.failed:
+            prints(job.details)
+        elif job.result:
+            with open(outputf, 'rb') as f:
+                f.seek(0, 2)
+                if f.tell() > 100:
+                    f.seek(0)
+                    stream.seek(0)
+                    stream.truncate()
+                    shutil.copyfileobj(f, stream)
+                    stream.flush()
+    stream.seek(0)
 
 
-
-def set_metadata_(path, title, authors, bkp):
+def set_metadata_(path, opath, title, authors, bkp, tags):
     p = podofo.PDFDoc()
     p.open(path)
     title = prep(title)
     touched = False
-    if title:
+    if title and title != p.title:
         p.title = title
         touched = True
 
     author = prep(authors_to_string(authors))
-    if author:
+    if author and author != p.author:
         p.author = author
         touched = True
 
     bkp = prep(bkp)
-    if bkp:
+    if bkp and bkp != p.creator:
         p.creator = bkp
         touched = True
 
+    try:
+        tags = prep(u', '.join([x.strip() for x in tags if x.strip()]))
+        if tags != p.keywords:
+            p.keywords = tags
+            touched = True
+    except:
+        pass
+
     if touched:
-        from calibre.ptempfile import TemporaryFile
-        with TemporaryFile('_pdf_set_metadata.pdf') as f:
-            p.save(f)
-            return open(f, 'rb').read()
+        p.save(opath)
+        return True
+    return False
 
 if __name__ == '__main__':
     f = '/tmp/t.pdf'