From 1f0932ad4047395bd5ae11b8ee350b26367f1eea Mon Sep 17 00:00:00 2001 From: John Schember Date: Sun, 17 May 2009 21:35:59 -0400 Subject: [PATCH] Much better html to pml parser, now accounts for style information and produces output that looks more like the input. --- src/calibre/ebooks/pdb/ereader/writer.py | 13 +- src/calibre/ebooks/pml/output.py | 18 +-- src/calibre/ebooks/pml/pmlconverter.py | 82 ----------- src/calibre/ebooks/pml/pmlml.py | 178 +++++++++++++++++++++++ 4 files changed, 190 insertions(+), 101 deletions(-) create mode 100644 src/calibre/ebooks/pml/pmlml.py diff --git a/src/calibre/ebooks/pdb/ereader/writer.py b/src/calibre/ebooks/pdb/ereader/writer.py index f49aa4e125..c99c75a929 100644 --- a/src/calibre/ebooks/pdb/ereader/writer.py +++ b/src/calibre/ebooks/pdb/ereader/writer.py @@ -16,7 +16,7 @@ from calibre.ebooks.pdb.formatwriter import FormatWriter from calibre.ebooks.oeb.base import OEB_IMAGES from calibre.ebooks.pdb.header import PdbHeaderBuilder from calibre.ebooks.pdb.ereader import image_name -from calibre.ebooks.pml.pmlconverter import html_to_pml +from calibre.ebooks.pml.pmlml import PMLMLizer IDENTITY = 'PNRdPPrs' @@ -31,7 +31,7 @@ class Writer(FormatWriter): self.log = log def write_content(self, oeb_book, out_stream, metadata=None): - text = self._text(oeb_book.spine) + text = self._text(oeb_book) images = self._images(oeb_book.manifest) metadata = [self._metadata(metadata)] @@ -41,16 +41,15 @@ class Writer(FormatWriter): lengths = [len(i) for i in sections] - pdbHeaderBuilder = PdbHeaderBuilder(IDENTITY, '') + pdbHeaderBuilder = PdbHeaderBuilder(IDENTITY, metadata[0].parition()[0]) pdbHeaderBuilder.build_header(lengths, out_stream) for item in sections: out_stream.write(item) - def _text(self, pages): - pml = '' - for page in pages: - pml += html_to_pml(unicode(page)).encode('cp1252') + def _text(self, oeb_book): + pmlmlizer = PMLMLizer(ignore_tables=self.opts.linearize_tables) + pml = unicode(pmlmlizer.extract_content(oeb_book, self.opts)).encode('cp1252') pml_pages = [] for i in range(0, (len(pml) / MAX_RECORD_SIZE) + 1): diff --git a/src/calibre/ebooks/pml/output.py b/src/calibre/ebooks/pml/output.py index c5fbc990af..9d07718654 100644 --- a/src/calibre/ebooks/pml/output.py +++ b/src/calibre/ebooks/pml/output.py @@ -12,7 +12,7 @@ from calibre.customize.conversion import OutputFormatPlugin from calibre.ptempfile import TemporaryDirectory from calibre.utils.zipfile import ZipFile from calibre.ebooks.oeb.base import OEB_IMAGES -from calibre.ebooks.pml.pmlconverter import html_to_pml +from calibre.ebooks.pml.pmlml import PMLMLizer class PMLOutput(OutputFormatPlugin): @@ -22,22 +22,16 @@ class PMLOutput(OutputFormatPlugin): def convert(self, oeb_book, output_path, input_plugin, opts, log): with TemporaryDirectory('_pmlz_output') as tdir: - self.process_spine(oeb_book.spine, tdir) + pmlmlizer = PMLMLizer(ignore_tables=opts.linearize_tables) + content = pmlmlizer.extract_content(oeb_book, opts) + with open(os.path.join(tdir, 'index.pml'), 'wb') as out: + out.write(content.encode('utf-8')) + self.write_images(oeb_book.manifest, tdir) pmlz = ZipFile(output_path, 'w') pmlz.add_dir(tdir) - def process_spine(self, spine, out_dir): - for item in spine: - html = html_to_pml(unicode(item)).encode('utf-8') - - name = os.path.splitext(os.path.basename(item.href))[0] + '.pml' - path = os.path.join(out_dir, name) - - with open(path, 'wb') as out: - out.write(html) - def write_images(self, manifest, out_dir): for item in manifest: if item.media_type in OEB_IMAGES: diff --git a/src/calibre/ebooks/pml/pmlconverter.py b/src/calibre/ebooks/pml/pmlconverter.py index dded21c38c..0cd7da8e72 100644 --- a/src/calibre/ebooks/pml/pmlconverter.py +++ b/src/calibre/ebooks/pml/pmlconverter.py @@ -10,7 +10,6 @@ __docformat__ = 'restructuredtext en' import re -from calibre import entity_to_unicode from calibre.ebooks.pdb.ereader import image_name from calibre.ebooks.htmlsymbols import HTML_SYMBOLS @@ -67,75 +66,6 @@ PML_HTML_RULES = [ (re.compile(r'\\\\'), lambda match: '\\'), ] -HTML_PML_RULES = [ - - (re.compile(r'\\'), lambda match: '\\\\'), - (re.compile('(?<=[^\n])[ ]*'), lambda match: '\n

'), - (re.compile('

(?=^\n|^\r\n)'), lambda match: '\n'), - - - # Clean up HTML - (re.compile('@page.*?}'), lambda match: ''), - (re.compile('.*?', re.DOTALL), lambda match: ''), - (re.compile('.*?', re.DOTALL), lambda match: ''), - - # Reflow paragraphs - (re.compile('(?P.*?)

', re.DOTALL), lambda match: match.group('text').replace('\r\n', ' ').replace('\n', ' ')), - - # HTML to PML - (re.compile('.+?).*?">(?P.+?)'), lambda match: '\\Sd="%s"%s\\Sd' % (match.group('target'), match.group('text'))), - (re.compile('.+?).*?">(?P.+?)'), lambda match: '\\Fn="%s"%s\\Fn' % (match.group('target'), match.group('text'))), - (re.compile('.+?).*?">'), lambda match: '\\\\Q="%s"' % match.group('target')), - (re.compile('#.+?).*?">(?P)', re.DOTALL), lambda match: '\\q="%s"%s\\q' % (match.group('target'), match.group('text'))), - (re.compile('.+?)".*?>(.*?)*'), lambda match: '\\m="%s"' % image_name(match.group('name')).strip('\x00')), - (re.compile('&(?P#\d+);'), lambda match: entity_to_unicode(match)), - (re.compile('&(?P.+);'), lambda match: entity_to_unicode(match)), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\k%s\\k' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\k%s\\k' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\Sb%s\\Sb' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\Sb%s\\Sb' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\Sp%s\\Sp' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\Sp%s\\Sp' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\B%s\\B' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\B%s\\B' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\B%s\\B' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\B%s\\B' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\l%s\\l' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\l%s\\l' % match.group('text')), - (re.compile('\d+)%".*?>'), lambda match: '\\w="%s%%"' % match.group('val')), - (re.compile('\d+)%*;.*?>(?P.+?)', re.MULTILINE), lambda match: '\\T="%s%%"%s$' % (match.group('val'), match.group('text'))), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\t%s\\t' % match.group('text')), - (re.compile('', re.DOTALL), lambda match: '\\v%s\\v' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\o%s\\o' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\o%s\\o' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\u%s\\u' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\\\i%s\\i' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\\\i%s\\i' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\r%s\\r' % match.group('text')), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\c%s\\c' % match.group('text')), - (re.compile('[0-4]).*?>(?P.+?)', re.DOTALL), lambda match: '\\X%s%s\\X%s' % (int(match.group('val')) + 1, match.group('text'), int(match.group('val')) + 1)), - (re.compile('(?P.+?)', re.DOTALL), lambda match: '\\x%s\\x' % match.group('text')), - (re.compile('
'), lambda match: '\n'), - (re.compile('
'), lambda match: '\n'), - - # Remove remaining HTML tags - (re.compile('<.*?>'), lambda match: ''), - - # Remove redundant page break markers - (re.compile(r'(\\p){2,}'), lambda match: r'\p'), - - # Remove whitespace on empty lines - (re.compile('^[\t\r ]$', re.MULTILINE), lambda match: ''), - # Remove excess whitespace in lines - (re.compile('(?<=.)[ ]{2,}(?=.)'), lambda match: ' '), - - # Remove excess newlines at the beginning and end - (re.compile('^(\r\n){1,}'), lambda match: ''), - (re.compile('^\n{1,}'), lambda match: ''), - (re.compile('(\r\n){3,}$'), lambda match: ''), - (re.compile('\n{3,}$'), lambda match: ''), -] - def pml_to_html(pml): html = pml for rule in PML_HTML_RULES: @@ -151,15 +81,3 @@ def footnote_sidebar_to_html(id, pml): html = '
%s
' % (id, id, pml_to_html(pml)) return html -def html_to_pml(html): - pml = '' - - for dom_tree in BeautifulSoup(html).findAll('body'): - body = unicode(dom_tree.prettify()) - - for rule in HTML_PML_RULES: - body = rule[0].sub(rule[1], body) - - pml += body - - return pml diff --git a/src/calibre/ebooks/pml/pmlml.py b/src/calibre/ebooks/pml/pmlml.py new file mode 100644 index 0000000000..a6febdc53f --- /dev/null +++ b/src/calibre/ebooks/pml/pmlml.py @@ -0,0 +1,178 @@ +# -*- coding: utf-8 -*- + +__license__ = 'GPL 3' +__copyright__ = '2009, John Schember ' +__docformat__ = 'restructuredtext en' + +''' +Transform OEB content into PML markup +''' + +import os, re + +from calibre.ebooks.oeb.base import XHTML, XHTML_NS, barename, namespace +from calibre.ebooks.oeb.stylizer import Stylizer +from calibre.ebooks.pdb.ereader import image_name + +TAG_MAP = { + 'b' : 'B', + 'strong' : 'B', + 'i' : 'I', + 'small' : 'k', + 'sub' : 'Sb', + 'sup' : 'Sp', + 'big' : 'l', + 'del' : 'o', + 'h1' : 'x', + 'h2' : 'x0', + 'h3' : 'x1', + 'h4' : 'x2', + 'h5' : 'x3', + 'h6' : 'x4', + '!--' : 'v', +} + +STYLES = [ + ('font-weight', {'bold' : 'B', 'bolder' : 'B'}), + ('font-style', {'italic' : 'I'}), + ('text-decoration', {'underline' : 'u'}), + ('text-align', {'right' : 'r', 'center' : 'c'}), +] + +class PMLMLizer(object): + def __init__(self, ignore_tables=False): + self.ignore_tables = ignore_tables + + def extract_content(self, oeb_book, opts): + oeb_book.logger.info('Converting XHTML to PML markup...') + self.oeb_book = oeb_book + self.opts = opts + return self.pmlmlize_spine() + + def pmlmlize_spine(self): + output = u'' + for item in self.oeb_book.spine: + stylizer = Stylizer(item.data, item.href, self.oeb_book, self.opts.output_profile) + output += self.dump_text(item.data.find(XHTML('body')), stylizer) + output = self.clean_text(output) + + output = re.sub('%s{1,1}' % os.linesep, '%s%s' % (os.linesep, os.linesep), output) + output = re.sub('%s{3,}' % os.linesep, '%s%s' % (os.linesep, os.linesep), output) + output = re.sub('[ ]{2,}', ' ', output) + + return output + + def clean_text(self, text): + return text + + def dump_text(self, elem, stylizer, tag_stack=[]): + if not isinstance(elem.tag, basestring) \ + or namespace(elem.tag) != XHTML_NS: + return u'' + + text = u'' + style = stylizer.style(elem) + + if style['display'] in ('none', 'oeb-page-head', 'oeb-page-foot') \ + or style['visibility'] == 'hidden': + return u'' + + tag = barename(elem.tag) + tag_count = 0 + + # Are we in a paragraph block? + if tag == 'p' or style['display'] in ('block'): + if 'block' not in tag_stack: + tag_count += 1 + tag_stack.append('block') + + # Process tags that need special processing and that do not have inner + # text. Usually these require an argument + if tag == 'img': + text += '\\m="%s"' % image_name(os.path.basename(elem.get('src'))).strip('\x00') + if tag == 'hr': + text += '\\w' + width = elem.get('width') + if width: + text += '="%s%"' % width + else: + text += '="50%"' + + # Process style information that needs holds a single tag + if style['page-break-before'] == 'always': + text += '\\p' + if style['page-break-after'] == 'always': + text += '\\p' + + # Proccess tags that contain text. + if hasattr(elem, 'text') and elem.text != None and elem.text.strip() != '': + pml_tag = TAG_MAP.get(tag, None) + if pml_tag and pml_tag not in tag_stack: + tag_count += 1 + text += '\\%s' % pml_tag + tag_stack.append(pml_tag) + + # Special processing of tags that require an argument. + # Anchors links + if tag == 'a' and 'q' not in tag_stack: + href = elem.get('href') + if href and href.startswith('#'): + tag_count += 1 + text += '\\q="%s"' % href + tag_stack.append('q') + # Anchor ids + id_name = elem.get('id') + if id_name: + text += '\\Q="%s"' % id_name + + # Processes style information + for s in STYLES: + style_tag = s[1].get(style[s[0]], None) + if style_tag and style_tag not in tag_stack: + tag_count += 1 + text += '\\%s' % style_tag + tag_stack.append(style_tag) + # margin + + text += self.elem_text(elem, tag_stack) + + for item in elem: + text += self.dump_text(item, stylizer, tag_stack) + + close_tag_list = [] + for i in range(0, tag_count): + close_tag_list.insert(0, tag_stack.pop()) + text += self.close_tags(close_tag_list) + if tag in ('h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'p', 'div', 'li'): + text += os.linesep + os.linesep + + + if 'block' not in tag_stack: + text += os.linesep + os.linesep + + if hasattr(elem, 'tail') and elem.tail != None and elem.tail.strip() != '': + text += self.elem_tail(elem, tag_stack) + + return text + + def elem_text(self, elem, tag_stack): + return self.block_text(elem.text, 'block' in tag_stack) + + def elem_tail(self, elem, tag_stack): + return self.block_text(elem.tail, 'block' in tag_stack) + + def block_text(self, text, in_block): + if in_block: + text = text.replace('\n\r', ' ') + text = text.replace('\n', ' ') + text = text.replace('\r', ' ') + return text + + def close_tags(self, tags): + text = u'' + for i in range(0, len(tags)): + tag = tags.pop() + if tag != 'block': + text += '\\%s' % tag + return text +