Metadata reading and writing for TXT/PDF input/output.

src/calibre/ebooks/pdf/input.py

@@ -10,8 +10,7 @@ import os
 from calibre.customize.conversion import InputFormatPlugin
 from calibre.ebooks.pdf.pdftohtml import pdftohtml
 from calibre.ebooks.metadata.opf import OPFCreator
-from calibre.ebooks.metadata import MetaInformation
-#from calibre.ebooks.metadata.meta import metadata_from_formats
+from calibre.customize.builtins import PDFMetadataReader
 
 class PDFInput(InputFormatPlugin):
     
@@ -27,8 +26,7 @@ class PDFInput(InputFormatPlugin):
         with open('index.html', 'wb') as index:
             index.write(html)
             
-        #mi = metadata_from_formats([stream.name])
-        mi = MetaInformation(_('Unknown'), _('Unknown'))
+        mi = PDFMetadataReader(None).get_metadata(stream, 'pdf')
         opf = OPFCreator(os.getcwd(), mi)
         opf.create_manifest([('index.html', None)])
         opf.create_spine(['index.html'])

src/calibre/ebooks/pdf/output.py

@@ -17,7 +17,7 @@ from calibre.customize.conversion import OutputFormatPlugin, \
     OptionRecommendation
 from calibre.ebooks.oeb.output import OEBOutput
 from calibre.ptempfile import TemporaryDirectory
-from calibre.ebooks.pdf.writer import PDFWriter
+from calibre.ebooks.pdf.writer import PDFWriter, PDFMetadata
 from calibre.ebooks.pdf.pageoptions import UNITS, unit, PAPER_SIZES, \
     paper_size, ORIENTATIONS, orientation, PageOptions
 
@@ -88,7 +88,7 @@ class PDFOutput(OutputFormatPlugin):
             
             out_stream.seek(0)
             out_stream.truncate()
-            writer.dump(opf, out_stream)
+            writer.dump(opf, out_stream, PDFMetadata(oeb_book.metadata))
             
             if close:
                 out_stream.close()

src/calibre/ebooks/pdf/writer.py

@@ -13,6 +13,7 @@ import os, shutil, sys
 
 from calibre.ptempfile import PersistentTemporaryDirectory
 from calibre.ebooks.pdf.pageoptions import PageOptions
+from calibre.ebooks.metadata import authors_to_string
 from calibre.ebooks.metadata.opf2 import OPF
 
 from PyQt4 import QtCore
@@ -22,6 +23,18 @@ from PyQt4.QtWebKit import QWebView
 
 from pyPdf import PdfFileWriter, PdfFileReader
         
+class PDFMetadata(object):
+    def __init__(self, oeb_metadata=None):
+        self.title = _('Unknown')
+        self.author = _('Unknown')
+        
+        if oeb_metadata != None:
+            if len(oeb_metadata.title) >= 1:
+                self.title = oeb_metadata.title[0].value
+            if len(oeb_metadata.creator) >= 1:
+                self.author = authors_to_string([x.value for x in oeb_metadata.creator])
+
+
 class PDFWriter(QObject):
     def __init__(self, log, popts=PageOptions()):
         if QApplication.instance() is None:
@@ -38,7 +51,8 @@ class PDFWriter(QObject):
         self.tmp_path = PersistentTemporaryDirectory('_any2pdf_parts')
         self.popts = popts
         
-    def dump(self, opfpath, out_stream):
+    def dump(self, opfpath, out_stream, pdf_metadata):
+        self.metadata = pdf_metadata
         self._delete_tmpdir()
         
         opf = OPF(opfpath, os.path.dirname(opfpath))
@@ -88,7 +102,7 @@ class PDFWriter(QObject):
         self.logger.info('Combining individual PDF parts...')
     
         try:
-            outPDF = PdfFileWriter()
+            outPDF = PdfFileWriter(title=self.metadata.title, author=self.metadata.author)
             for item in self.combine_queue:
                 inputPDF = PdfFileReader(file(item, 'rb'))
                 for page in inputPDF.pages:

src/calibre/ebooks/txt/input.py

@@ -10,7 +10,7 @@ import os
 from calibre.customize.conversion import InputFormatPlugin
 from calibre.ebooks.markdown import markdown
 from calibre.ebooks.metadata.opf import OPFCreator
-#from calibre.ebooks.metadata.meta import metadata_from_formats
+from calibre.customize.builtins import TXTMetadataReader
 
 class TXTInput(InputFormatPlugin):
     
@@ -26,12 +26,11 @@ class TXTInput(InputFormatPlugin):
         md = markdown.Markdown(
             extensions=['footnotes', 'tables', 'toc'],
             safe_mode=False,)
-        html = '<html><body>'+md.convert(txt)+'</body></html>'
+        html = '<html><head><title /></head><body>'+md.convert(txt)+'</body></html>'
         with open('index.html', 'wb') as index:
             index.write(html.encode('utf-8'))
             
-        #mi = metadata_from_formats([stream.name])
-        mi = None
+        mi = TXTMetadataReader(None).get_metadata(stream, 'txt')
         opf = OPFCreator(os.getcwd(), mi)
         opf.create_manifest([('index.html', None)])
         opf.create_spine(['index.html'])

src/calibre/ebooks/txt/output.py

@@ -34,7 +34,7 @@ class TXTOutput(OutputFormatPlugin):
     def convert(self, oeb_book, output_path, input_plugin, opts, log):
         metadata = TxtMetadata()
         if opts.prepend_metadata.lower() == 'true':
-            metadata.author = opts.authors if opts.authors else authors_to_string(oeb_book.metadata.authors.value) if oeb_book.metadata.authors != [] else _('Unknown')
+            metadata.author = opts.authors if opts.authors else authors_to_string([x.value for x in oeb_book.metadata.creator]) if oeb_book.metadata.creator != [] else _('Unknown')
             metadata.title = opts.title if opts.title else oeb_book.metadata.title[0].value if oeb_book.metadata.title != [] else _('Unknown')
             
         writer = TxtWriter(TxtNewlines(opts.newline).newline, log)