Demonstrable modularization of e-book conversion.

src/calibre/ebooks/lit/reader.py

@@ -802,6 +802,7 @@ class LitFile(object):
 
 
 class LitContainer(object):
+    """Simple Container-interface, read-only accessor for LIT files."""
 
     def __init__(self, filename_or_stream):
         self._litfile = LitFile(filename_or_stream)

src/calibre/ebooks/mobi/mobiml.py

@@ -82,7 +82,15 @@ class MobiMLizer(object):
     def __init__(self, ignore_tables=False):
         self.ignore_tables = ignore_tables
     
-    def transform(self, oeb, context):
+    @classmethod
+    def config(cls, cfg):
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        return cls()
+    
+    def __call__(self, oeb, context):
         oeb.logger.info('Converting XHTML to Mobipocket markup...')
         self.oeb = oeb
         self.profile = profile = context.dest

src/calibre/ebooks/mobi/writer.py

@@ -296,13 +296,43 @@ class Serializer(object):
 class MobiWriter(object):
     COLLAPSE_RE = re.compile(r'[ \t\r\n\v]+')
 
+    DEFAULT_PROFILE = 'CybookG3'
+
+    TRANSFORMS = [HTMLTOCAdder, CaseMangler, CSSFlattener, SVGRasterizer,
+                  ManifestTrimmer, MobiMLizer]
+    
     def __init__(self, compression=None, imagemax=None,
                  prefer_author_sort=False):
         self._compression = compression or UNCOMPRESSED
         self._imagemax = imagemax or OTHER_MAX_IMAGE_SIZE
         self._prefer_author_sort = prefer_author_sort
 
-    def dump(self, oeb, path):
+    @classmethod
+    def config(cls, cfg):
+        """Add any book-writing options to the :class:`Config` object
+        :param:`cfg`.
+        """
+        mobi = cfg.add_group('mobipocket', _('Mobipocket-specific options.'))
+        mobi('compress', ['--compress'], default=False,
+             help=_('Compress file text using PalmDOC compression. '
+                    'Results in smaller files, but takes a long time to run.'))
+        mobi('rescale_images', ['--rescale-images'], default=False, 
+             help=_('Modify images to meet Palm device size limitations.'))
+        mobi('prefer_author_sort', ['--prefer-author-sort'], default=False,
+             help=_('When present, use the author sorting information for '
+                    'generating the Mobipocket author metadata.'))
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        """Generate a Writer instance from command-line options."""
+        compression = PALMDOC if opts.compress else UNCOMPRESSED
+        imagemax = PALM_MAX_IMAGE_SIZE if opts.rescale_images else None
+        prefer_author_sort = opts.prefer_author_sort
+        return cls(compression=compression, imagemax=imagemax,
+                   prefer_author_sort=prefer_author_sort)
+    
+    def __call__(self, oeb, path):
         if hasattr(path, 'write'):
             return self._dump_stream(oeb, path)
         with open(path, 'w+b') as stream:
@@ -533,20 +563,12 @@ def config(defaults=None):
         c = StringConfig(defaults, desc)
         
     mobi = c.add_group('mobipocket', _('Mobipocket-specific options.'))
-    mobi('compress', ['--compress'], default=False,
-         help=_('Compress file text using PalmDOC compression. '
-               'Results in smaller files, but takes a long time to run.'))
-    mobi('rescale_images', ['--rescale-images'], default=False, 
-        help=_('Modify images to meet Palm device size limitations.'))
     mobi('toc_title', ['--toc-title'], default=None, 
          help=_('Title for any generated in-line table of contents.'))
     mobi('ignore_tables', ['--ignore-tables'], default=False,
          help=_('Render HTML tables as blocks of text instead of actual '
                 'tables. This is neccessary if the HTML contains very large '
                 'or complex tables.'))
-    mobi('prefer_author_sort', ['--prefer-author-sort'], default=False,
-         help=_('When present, use the author sorting information for '
-                'generating the Mobipocket author metadata.'))
     profiles = c.add_group('profiles', _('Device renderer profiles. '
         'Affects conversion of font sizes, image rescaling and rasterization '
         'of tables. Valid profiles are: %s.') % ', '.join(_profiles))

src/calibre/ebooks/oeb/base.py

@@ -820,7 +820,9 @@ class Manifest(object):
     def __iter__(self):
         for item in self.items:
             yield item
-    values = __iter__
+
+    def values(self):
+        return list(self.items)
     
     def __contains__(self, item):
         return item in self.items
@@ -1134,7 +1136,7 @@ class TOC(object):
             node.to_opf1(tour)
         return tour
     
-    def to_ncx(self, parent, depth=1):
+    def to_ncx(self, parent):
         for node in self.nodes:
             id = node.id or unicode(uuid.uuid4())
             attrib = {'id': id, 'playOrder': '0'}
@@ -1143,9 +1145,8 @@ class TOC(object):
             point = element(parent, NCX('navPoint'), attrib=attrib)
             label = etree.SubElement(point, NCX('navLabel'))
             element(label, NCX('text')).text = node.title
-            href = node.href if depth > 1 else urldefrag(node.href)[0]
-            element(point, NCX('content'), src=href)
-            node.to_ncx(point, depth+1)
+            element(point, NCX('content'), src=node.href)
+            node.to_ncx(point)
         return parent
 
 

src/calibre/ebooks/oeb/factory.py

@@ -6,20 +6,93 @@ from __future__ import with_statement
 __license__   = 'GPL v3'
 __copyright__ = '2008, Marshall T. Vandegrift <llasram@gmail.com>'
 
-import os
+import sys, os, logging
+from itertools import chain
 from calibre.ebooks.oeb.base import OEBError
 from calibre.ebooks.oeb.reader import OEBReader
+from calibre.ebooks.oeb.writer import OEBWriter
 from calibre.ebooks.lit.reader import LitReader
+from calibre.ebooks.lit.writer import LitWriter
+from calibre.ebooks.mobi.reader import MobiReader
+from calibre.ebooks.mobi.writer import MobiWriter
+from calibre.ebooks.oeb.base import Logger, OEBBook
+from calibre.ebooks.oeb.profile import Context
+from calibre.utils.config import Config
 
 __all__ = ['get_reader']
 
-READER_REGISTRY = {
-    '.opf': OEBReader,
-    '.lit': LitReader,
+REGISTRY = {
+    '.opf': (OEBReader, None),
+    '.lit': (LitReader, LitWriter),
+    '.mobi': (MobiReader, MobiWriter),
     }
 
 def ReaderFactory(path):
-    ext = os.path.splitext(path)[1].lower()
-    if not ext:
+    if os.path.isdir(path):
         return OEBReader
-    return READER_REGISTRY[ext]()
+    ext = os.path.splitext(path)[1].lower()
+    Reader = REGISTRY.get(ext, (None, None))[0]
+    if Reader is None:
+        raise OEBError('Unknown e-book file extension %r' % ext)
+    return Reader
+
+def WriterFactory(path):
+    if os.path.isdir(path):
+        return OEBWriter
+    ext = os.path.splitext(path)[1].lower()
+    if not os.path.exists(path) and not ext:
+        return OEBWriter
+    Writer = REGISTRY.get(ext, (None, None))[1]
+    if Writer is None:
+        raise OEBError('Unknown e-book file extension %r' % ext)
+    return Writer
+
+
+def option_parser(Reader, Writer):
+    cfg = Config('ebook-convert', _('Options to control e-book conversion.'))
+    Reader.config(cfg)
+    for Transform in chain(Reader.TRANSFORMS, Writer.TRANSFORMS):
+        Transform.config(cfg)
+    Writer.config(cfg)
+    parser = cfg.option_parser()
+    parser.add_option('--encoding', default=None,
+        help=_('Character encoding for input. Default is to auto detect.'))
+    parser.add_option('-o', '--output', default=None, 
+        help=_('Output file. Default is derived from input filename.'))
+    parser.add_option('-p', '--pretty-print', action='store_true',
+        default=False, help=_('Produce more human-readable XML output.'))
+    parser.add_option('-v', '--verbose', default=0, action='count',
+        help=_('Useful for debugging.'))
+    return parser
+
+def main(argv=sys.argv):
+    if len(argv) < 3:
+        print _("Usage: ebook-convert INFILE OUTFILE [OPTIONS..]")
+        return 1
+    inpath, outpath = argv[1], argv[2]
+    Reader = ReaderFactory(inpath)
+    Writer = WriterFactory(outpath)
+    parser = option_parser(Reader, Writer)
+    opts, args = parser.parse_args(argv[3:])
+    if len(args) != 0:
+        parser.print_help()
+        return 1
+    logger = Logger(logging.getLogger('ebook-convert'))
+    logger.setup_cli_handler(opts.verbose)
+    encoding = opts.encoding
+    pretty_print = opts.pretty_print
+    oeb = OEBBook(encoding=encoding, pretty_print=pretty_print, logger=logger)
+    context = Context(Reader.DEFAULT_PROFILE, Writer.DEFAULT_PROFILE)
+    reader = Reader.generate(opts)
+    writer = Writer.generate(opts)
+    transforms = []
+    for Transform in chain(Reader.TRANSFORMS, Writer.TRANSFORMS):
+        transforms.append(Transform.generate(opts))
+    reader(oeb, inpath)
+    for transform in transforms:
+        transform(oeb, context)
+    writer(oeb, outpath)
+    return 0
+
+if __name__ == '__main__':
+    sys.exit(main())

src/calibre/ebooks/oeb/reader.py

@@ -31,15 +31,39 @@ from calibre.ptempfile import TemporaryDirectory
 __all__ = ['OEBReader']
 
 class OEBReader(object):
+    """Read an OEBPS 1.x or OPF/OPS 2.0 file collection."""
     
     COVER_SVG_XP    = XPath('h:body//svg:svg[position() = 1]')
     COVER_OBJECT_XP = XPath('h:body//h:object[@data][position() = 1]')
 
     Container = DirContainer
+    """Container type used to access book files.  Override in sub-classes."""
     
     DEFAULT_PROFILE = 'PRS505'
+    """Default renderer profile for content read with this Reader."""
+
+    TRANSFORMS = []
+    """List of transforms to apply to content read with this Reader."""
+
+    def __init__(self):
+        return
+    
+    @classmethod
+    def config(cls, cfg):
+        """Add any book-reading options to the :class:`Config` object
+        :param:`cfg`.
+        """
+        return
+
+    @classmethod
+    def generate(cls, opts):
+        """Generate a Reader instance from command-line options."""
+        return cls()
 
     def __call__(self, oeb, path):
+        """Read the book at :param:`path` into the :class:`OEBBook` object
+        :param:`oeb`.
+        """
         self.oeb = oeb
         self.logger = oeb.logger
         oeb.container = self.Container(path)

src/calibre/ebooks/oeb/transforms/flatcss.py

@@ -94,7 +94,15 @@ class CSSFlattener(object):
         self.unfloat = unfloat
         self.untable = untable
 
-    def transform(self, oeb, context):
+    @classmethod
+    def config(cls, cfg):
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        return cls()
+    
+    def __call__(self, oeb, context):
         oeb.logger.info('Flattening CSS and remapping font sizes...')
         self.oeb = oeb
         self.context = context

src/calibre/ebooks/oeb/transforms/htmltoc.py

@@ -52,7 +52,18 @@ class HTMLTOCAdder(object):
         self.title = title
         self.style = style
     
-    def transform(self, oeb, context):
+    @classmethod
+    def config(cls, cfg):
+        group = cfg.add_group('htmltoc', _('HTML TOC generation options.'))
+        group('toc_title', ['--toc-title'], default=None, 
+              help=_('Title for any generated in-line table of contents.'))
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        return cls(title=opts.toc_title)
+    
+    def __call__(self, oeb, context):
         if 'toc' in oeb.guide:
             return
         oeb.logger.info('Generating in-line TOC...')

src/calibre/ebooks/oeb/transforms/manglecase.py

@@ -29,7 +29,15 @@ CASE_MANGLER_CSS = """
 TEXT_TRANSFORMS = set(['capitalize', 'uppercase', 'lowercase'])
 
 class CaseMangler(object):
-    def transform(self, oeb, context):
+    @classmethod
+    def config(cls, cfg):
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        return cls()
+    
+    def __call__(self, oeb, context):
         oeb.logger.info('Applying case-transforming CSS...')
         self.oeb = oeb
         self.profile = context.source

src/calibre/ebooks/oeb/transforms/rasterize.py

@@ -34,7 +34,15 @@ class SVGRasterizer(object):
         if QApplication.instance() is None:
             QApplication([])
 
-    def transform(self, oeb, context):
+    @classmethod
+    def config(cls, cfg):
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        return cls()
+    
+    def __call__(self, oeb, context):
         oeb.logger.info('Rasterizing SVG images...')
         self.oeb = oeb
         self.profile = context.dest

src/calibre/ebooks/oeb/transforms/trimmanifest.py

@@ -17,7 +17,15 @@ from calibre.ebooks.oeb.base import LINK_SELECTORS, CSSURL_RE
 from calibre.ebooks.oeb.base import urlnormalize
 
 class ManifestTrimmer(object):
-    def transform(self, oeb, context):
+    @classmethod
+    def config(cls, cfg):
+        return cfg
+
+    @classmethod
+    def generate(cls, opts):
+        return cls()
+    
+    def __call__(self, oeb, context):
         oeb.logger.info('Trimming unused files from manifest...')
         used = set()
         hrefs = oeb.manifest.hrefs

src/calibre/ebooks/oeb/writer.py

@@ -9,12 +9,15 @@ __copyright__ = '2008, Marshall T. Vandegrift <llasram@gmail.com>'
 import sys, os, logging
 from calibre.ebooks.oeb.base import OPF_MIME, xml2str
 from calibre.ebooks.oeb.base import Logger, DirContainer, OEBBook
-from calibre.utils.config import Config
 
 __all__ = ['OEBWriter']
 
 class OEBWriter(object):
     DEFAULT_PROFILE = 'PRS505'
+    """Default renderer profile for content written with this Writer."""
+
+    TRANSFORMS = []
+    """List of transforms to apply to content written with this Writer."""
     
     def __init__(self, version='2.0', page_map=False, pretty_print=False):
         self.version = version
@@ -23,6 +26,9 @@ class OEBWriter(object):
 
     @classmethod
     def config(cls, cfg):
+        """Add any book-writing options to the :class:`Config` object
+        :param:`cfg`.
+        """
         oeb = cfg.add_group('oeb', _('OPF/NCX/etc. generation options.'))
         versions = ['1.2', '2.0']
         oeb('opf_version', ['--opf-version'], default='2.0', choices=versions,
@@ -34,6 +40,7 @@ class OEBWriter(object):
 
     @classmethod
     def generate(cls, opts):
+        """Generate a Writer instance from command-line options."""
         version = opts.opf_version
         page_map = opts.adobe_page_map
         pretty_print = opts.pretty_print
@@ -41,6 +48,9 @@ class OEBWriter(object):
                    pretty_print=pretty_print)
     
     def __call__(self, oeb, path):
+        """Read the book in the :class:`OEBBook` object :param:`oeb` to a file
+        at :param:`path`.
+        """
         version = int(self.version[0])
         opfname = None
         if os.path.splitext(path)[1].lower() == '.opf':
@@ -63,48 +73,3 @@ class OEBWriter(object):
                 href = opfname
             output.write(href, xml2str(data, pretty_print=pretty_print))
         return
-
-
-def option_parser():
-    cfg = Config('oeb', _('Options to control OEB conversion.'))
-    OEBWriter.config(cfg)
-    parser = cfg.option_parser()
-    parser.add_option('--encoding', default=None,
-        help=_('Character encoding for files. Default is to auto detect.'))
-    parser.add_option('-o', '--output', default=None, 
-        help=_('Output file. Default is derived from input filename.'))
-    parser.add_option('-p', '--pretty-print', action='store_true',
-        default=False, help=_('Produce more human-readable XML output.'))
-    parser.add_option('-v', '--verbose', default=0, action='count',
-        help=_('Useful for debugging.'))
-    return parser
-    
-def any2oeb(opts, inpath):
-    from calibre.ebooks.oeb.factory import ReaderFactory
-    logger = Logger(logging.getLogger('any2oeb'))
-    logger.setup_cli_handler(opts.verbose)
-    outpath = opts.output
-    if outpath is None:
-        outpath = os.path.basename(inpath)
-        outpath = os.path.splitext(outpath)[0]
-    encoding = opts.encoding
-    pretty_print = opts.pretty_print
-    oeb = OEBBook(encoding=encoding, pretty_print=pretty_print, logger=logger)
-    reader = ReaderFactory(inpath)
-    reader(oeb, inpath)
-    writer = OEBWriter.generate(opts)
-    writer(oeb, outpath)
-    return 0
-
-def main(argv=sys.argv):
-    parser = option_parser()
-    opts, args = parser.parse_args(argv[1:])
-    if len(args) != 1:
-        parser.print_help()
-        return 1
-    inpath = args[0]
-    retval = any2oeb(opts, inpath)
-    return retval
-
-if __name__ == '__main__':
-    sys.exit(main())