Add support for transforming style rules to ebook-convert

src/calibre/ebooks/conversion/cli.py

@@ -98,6 +98,15 @@ def option_recommendation_to_cli_option(add_option, rec):
         attrs.pop('type', '')
     if opt.name == 'read_metadata_from_opf':
         switches.append('--from-opf')
+    if opt.name == 'transform_css_rules':
+        attrs['help'] = _(
+            'Path to a file containing rules to transform the CSS styles'
+            ' in this book. The easiest way to create such a file is to'
+            ' use the wizard for creating rules in the calibre GUI. Access'
+            ' it in the "Look & Feel->Transform styles" section of the conversion'
+            ' dialog. Once you create the rules, you can use the Export button'
+            ' to save them to a file.'
+        )
     if opt.name in DEFAULT_TRUE_OPTIONS and rec.recommended_value is True:
         switches = ['--disable-'+opt.long_switch]
     add_option(Option(*switches, **attrs))
@@ -176,7 +185,7 @@ def add_pipeline_options(parser, plumber):
                       'subset_embedded_fonts', 'embed_all_fonts',
                       'line_height', 'minimum_line_height',
                       'linearize_tables',
-                      'extra_css', 'filter_css', 'expand_css',
+                      'extra_css', 'filter_css', 'transform_css_rules', 'expand_css',
                       'smarten_punctuation', 'unsmarten_punctuation',
                       'margin_top', 'margin_left', 'margin_right',
                       'margin_bottom', 'change_justification',
@@ -349,6 +358,17 @@ def main(args=sys.argv):
             setattr(opts, x, abspath(getattr(opts, x)))
     if opts.search_replace:
         opts.search_replace = read_sr_patterns(opts.search_replace, log)
+    if opts.transform_css_rules:
+        from calibre.ebooks.css_transform_rules import import_rules, validate_rule
+        with open(opts.transform_css_rules, 'rb') as tcr:
+            opts.transform_css_rules = rules = list(import_rules(tcr.read()))
+            for rule in rules:
+                title, msg = validate_rule(rule)
+                if title and msg:
+                    log.error('Failed to parse CSS transform rules')
+                    log.error(title)
+                    log.error(msg)
+                    return 1
 
     recommendations = [(n.dest, getattr(opts, n.dest),
                         OptionRecommendation.HIGH)

src/calibre/ebooks/conversion/plumber.py

@@ -3,7 +3,7 @@ __license__ = 'GPL 3'
 __copyright__ = '2009, Kovid Goyal <kovid@kovidgoyal.net>'
 __docformat__ = 'restructuredtext en'
 
-import os, re, sys, shutil, pprint
+import os, re, sys, shutil, pprint, json
 from functools import partial
 
 from calibre.customize.conversion import OptionRecommendation, DummyReporter
@@ -355,6 +355,12 @@ OptionRecommendation(name='extra_css',
                 'rules.')
         ),
 
+OptionRecommendation(name='transform_css_rules',
+            recommended_value=None, level=OptionRecommendation.LOW,
+            help=_('Rules for transforming the styles in this book. These'
+                   ' rules are applied after all other CSS processing is done.')
+        ),
+
 OptionRecommendation(name='filter_css',
             recommended_value=None, level=OptionRecommendation.LOW,
             help=_('A comma separated list of CSS properties that '
@@ -1153,12 +1159,18 @@ OptionRecommendation(name='search_replace',
         mobi_file_type = getattr(self.opts, 'mobi_file_type', 'old')
         needs_old_markup = (self.output_plugin.file_type == 'lit' or
                     (self.output_plugin.file_type == 'mobi' and mobi_file_type == 'old'))
+        transform_css_rules = ()
+        if self.opts.transform_css_rules:
+            transform_css_rules = self.opts.transform_css_rules
+            if isinstance(transform_css_rules, basestring):
+                transform_css_rules = json.loads(transform_css_rules)
         flattener = CSSFlattener(fbase=fbase, fkey=fkey,
                 lineh=line_height,
                 untable=needs_old_markup,
                 unfloat=needs_old_markup,
                 page_break_on_body=self.output_plugin.file_type in ('mobi',
                     'lit'),
+                transform_css_rules=transform_css_rules,
                 specializer=partial(self.output_plugin.specialize_css_for_output,
                     self.log, self.opts))
         flattener(self.oeb, self.opts)

src/calibre/ebooks/css_transform_rules.py

@@ -5,6 +5,7 @@
 from __future__ import (unicode_literals, division, absolute_import,
                         print_function)
 from functools import partial
+from collections import OrderedDict
 import operator
 
 from cssutils.css import Property, CSSRule
@@ -192,7 +193,42 @@ class Rule(object):
         declaration.changed = oval or changed
         return changed
 
+ACTION_MAP = OrderedDict((
+    ('remove', _('Remove the property')),
+    ('append', _('Add extra properties')),
+    ('change', _('Change the value to')),
+    ('*', _('Multiply the value by')),
+    ('/', _('Divide the value by')),
+    ('+', _('Add to the value')),
+    ('-', _('Subtract from the value')),
+))
+
+MATCH_TYPE_MAP = OrderedDict((
+    ('is', _('is')),
+    ('is_not', _('is not')),
+    ('*', _('is any value')),
+    ('matches', _('matches pattern')),
+    ('not_matches', _('does not match pattern')),
+    ('==', _('is the same length as')),
+    ('!=', _('is not the same length as')),
+    ('<', _('is less than')),
+    ('>', _('is greater than')),
+    ('<=', _('is less than or equal to')),
+    ('>=', _('is greater than or equal to')),
+))
+
+allowed_keys = frozenset('property match_type query action action_data'.split())
+
 def validate_rule(rule):
+    keys = frozenset(rule)
+    extra = keys - allowed_keys
+    if extra:
+        return _('Unknown keys'), _(
+            'The rule has unknown keys: %s') % ', '.join(extra)
+    missing = allowed_keys - keys
+    if missing:
+        return _('Missing keys'), _(
+            'The rule has missing keys: %s') % ', '.join(missing)
     mt = rule['match_type']
     if not rule['property']:
         return _('Property required'), _('You must specify a CSS property to match')
@@ -203,6 +239,9 @@ def validate_rule(rule):
     if not rule['query'] and mt != '*':
         _('Query required'), _(
             'You must specify a value for the CSS property to match')
+    if mt not in MATCH_TYPE_MAP:
+        return _('Unknown match type'), _(
+            'The match type %s is not known') % mt
     if 'matches' in mt:
         try:
             compile_pat(rule['query'])
@@ -218,6 +257,9 @@ def validate_rule(rule):
             return _('Query invalid'), _(
                 '%s is not a valid length or number') % rule['query']
     ac, ad = rule['action'], rule['action_data']
+    if ac not in ACTION_MAP:
+        return _('Unknown action type'), _(
+            'The action type %s is not known') % mt
     if not ad and ac != 'remove':
         msg = _('You must specify a number')
         if ac == 'append':
@@ -258,6 +300,47 @@ def transform_container(container, serialized_rules, names=()):
         transform_style=partial(transform_declaration, rules), names=names
     )
 
+def rule_to_text(rule):
+    def get(prop):
+        return rule.get(prop) or ''
+    text = _(
+        'If the property {property} {match_type} {query}\n{action}').format(
+            property=get('property'), action=ACTION_MAP[rule['action']],
+            match_type=MATCH_TYPE_MAP[rule['match_type']], query=get('query'))
+    if get('action_data'):
+        text += get('action_data')
+    return text
+
+def export_rules(serialized_rules):
+    lines = []
+    for rule in serialized_rules:
+        lines.extend('# ' + l for l in rule_to_text(rule).splitlines())
+        lines.extend('%s: %s' % (k, v.replace('\n', ' ')) for k, v in rule.iteritems() if k in allowed_keys)
+        lines.append('')
+    return '\n'.join(lines).encode('utf-8')
+
+def import_rules(raw_data):
+    pat = regex.compile('\s*(\S+)\s*:\s*(.+)', flags=regex.VERSION1)
+    current_rule = {}
+
+    def sanitize(r):
+        return {k:(r.get(k) or '') for k in allowed_keys}
+
+    for line in raw_data.decode('utf-8').splitlines():
+        if not line.strip():
+            if current_rule:
+                yield sanitize(current_rule)
+            current_rule = {}
+            continue
+        if line.lstrip().startswith('#'):
+            continue
+        m = pat.match(line)
+        if m is not None:
+            k, v = m.group(1).lower(), m.group(2)
+            if k in allowed_keys:
+                current_rule[k] = v
+    if current_rule:
+        yield sanitize(current_rule)
 
 def test():  # {{{
     import unittest
@@ -335,6 +418,10 @@ def test():  # {{{
             prop = 'border-top-width'
             m('border-width: 1', 'border-bottom-width: 1;\nborder-left-width: 1;\nborder-right-width: 1;\nborder-top-width: 3', '*', '3')
 
+        def test_export_import(self):
+            rule = {'property':'a', 'match_type':'*', 'query':'some text', 'action':'remove', 'action_data':'color: red; a: b'}
+            self.ae(rule, next(import_rules(export_rules([rule]))))
+
     tests = unittest.defaultTestLoader.loadTestsFromTestCase(TestTransforms)
     unittest.TextTestRunner(verbosity=4).run(tests)
 

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

@@ -136,8 +136,13 @@ class EmbedFontsCSSRules(object):
 
 class CSSFlattener(object):
     def __init__(self, fbase=None, fkey=None, lineh=None, unfloat=False,
-                 untable=False, page_break_on_body=False, specializer=None):
+                 untable=False, page_break_on_body=False, specializer=None,
+                 transform_css_rules=()):
         self.fbase = fbase
+        self.transform_css_rules = transform_css_rules
+        if self.transform_css_rules:
+            from calibre.ebooks.css_transform_rules import compile_rules
+            self.transform_css_rules = compile_rules(self.transform_css_rules)
         self.fkey = fkey
         self.lineh = lineh
         self.unfloat = unfloat
@@ -554,8 +559,11 @@ class CSSFlattener(object):
             if item.media_type in OEB_STYLES:
                 manifest.remove(item)
         id, href = manifest.generate('css', 'stylesheet.css')
-        item = manifest.add(id, href, CSS_MIME, data=cssutils.parseString(css,
+        sheet = cssutils.parseString(css, validate=False)
-            validate=False))
+        if self.transform_css_rules:
+            from calibre.ebooks.css_transform_rules import transform_sheet
+            transform_sheet(self.transform_css_rules, sheet)
+        item = manifest.add(id, href, CSS_MIME, data=sheet)
         self.oeb.manifest.main_stylesheet = item
         return href
 
@@ -584,8 +592,11 @@ class CSSFlattener(object):
             href = None
             if css.strip():
                 id_, href = manifest.generate('page_css', 'page_styles.css')
-                manifest.add(id_, href, CSS_MIME, data=cssutils.parseString(css,
+                sheet = cssutils.parseString(css, validate=False)
-                    validate=False))
+                if self.transform_css_rules:
+                    from calibre.ebooks.css_transform_rules import transform_sheet
+                    transform_sheet(self.transform_css_rules, sheet)
+                manifest.add(id_, href, CSS_MIME, data=sheet)
             gc_map[css] = href
 
         ans = {}

src/calibre/gui2/css_transform_rules.py

@@ -4,13 +4,13 @@
 
 from __future__ import (unicode_literals, division, absolute_import,
                         print_function)
-from collections import OrderedDict
 
 from PyQt5.Qt import (
     QWidget, QVBoxLayout, QHBoxLayout, QLabel, QComboBox, QLineEdit, QPushButton, QSize
 )
 
-from calibre.ebooks.css_transform_rules import validate_rule, safe_parser, compile_rules, transform_sheet
+from calibre.ebooks.css_transform_rules import (
+    validate_rule, safe_parser, compile_rules, transform_sheet, ACTION_MAP, MATCH_TYPE_MAP)
 from calibre.gui2 import error_dialog, elided_text
 from calibre.gui2.tag_mapper import (
     RuleEditDialog as RuleEditDialogBase, Rules as RulesBase, RulesDialog as
@@ -20,29 +20,6 @@ from calibre.utils.config import JSONConfig
 
 class RuleEdit(QWidget):  # {{{
 
-    ACTION_MAP = OrderedDict((
-        ('remove', _('Remove the property')),
-        ('append', _('Add extra properties')),
-        ('change', _('Change the value to')),
-        ('*', _('Multiply the value by')),
-        ('/', _('Divide the value by')),
-        ('+', _('Add to the value')),
-        ('-', _('Subtract from the value')),
-    ))
-
-    MATCH_TYPE_MAP = OrderedDict((
-        ('is', _('is')),
-        ('is_not', _('is not')),
-        ('*', _('is any value')),
-        ('matches', _('matches pattern')),
-        ('not_matches', _('does not match pattern')),
-        ('==', _('is the same length as')),
-        ('!=', _('is not the same length as')),
-        ('<', _('is less than')),
-        ('>', _('is greater than')),
-        ('<=', _('is less than or equal to')),
-        ('>=', _('is greater than or equal to')),
-    ))
     MSG = _('Create the rule below, the rule can be used to transform style properties')
 
     def __init__(self, parent=None):
@@ -69,7 +46,7 @@ class RuleEdit(QWidget):  # {{{
                                'For instance use margin-top, not margin.'))
             elif clause == '{match_type}':
                 self.match_type = w = QComboBox(self)
-                for action, text in self.MATCH_TYPE_MAP.iteritems():
+                for action, text in MATCH_TYPE_MAP.iteritems():
                     w.addItem(text, action)
                 w.currentIndexChanged.connect(self.update_state)
             elif clause == '{query}':
@@ -89,7 +66,7 @@ class RuleEdit(QWidget):  # {{{
         for clause in parts:
             if clause == '{action}':
                 self.action = w = QComboBox(self)
-                for action, text in self.ACTION_MAP.iteritems():
+                for action, text in ACTION_MAP.iteritems():
                     w.addItem(text, action)
                 w.currentIndexChanged.connect(self.update_state)
             elif clause == '{action_data}':
@@ -178,8 +155,8 @@ class RuleItem(RuleItemBase):  # {{{
         query = elided_text(rule['query'], font=parent.font(), width=200, pos='right')
         text = _(
             'If the property <i>{property}</i> <b>{match_type}</b> <b>{query}</b><br>{action}').format(
-                property=rule['property'], action=RuleEdit.ACTION_MAP[rule['action']],
+                property=rule['property'], action=ACTION_MAP[rule['action']],
-                match_type=RuleEdit.MATCH_TYPE_MAP[rule['match_type']], query=query)
+                match_type=MATCH_TYPE_MAP[rule['match_type']], query=query)
         if rule['action_data']:
             ad = elided_text(rule['action_data'], font=parent.font(), width=200, pos='right')
             text += ' <code>%s</code>' % ad