Cutlery/calibre
1
0
Fork 0
mirror of https://github.com/kovidgoyal/calibre.git synced 2025-07-09 03:04:10 -04:00
Code Issues Packages Projects Releases Wiki Activity

Implement matching for HTML transform rules

Browse Source
This commit is contained in:
Kovid Goyal 2021-11-10 15:45:41 +05:30
parent 4f3c5be22d
commit 2a9caad4d3
No known key found for this signature in database
GPG Key ID: 06BC317B515ACE7C
2 changed files with 108 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

110
src/calibre/ebooks/html_transform_rules.py
View File

@ -5,9 +5,9 @@
from functools import partial from functools import partial
from calibre.ebooks.oeb.base import OEB_DOCS, XPath
from calibre.utils.serialize import json_dumps, json_loads from calibre.utils.serialize import json_dumps, json_loads
from calibre.ebooks.oeb.base import XPath from css_selectors.select import Select, get_parsed_selector
from css_selectors.select import get_parsed_selector
def non_empty_validator(label, val): def non_empty_validator(label, val):
@ -132,14 +132,80 @@ def validate_rule(rule):
return None, None return None, None
def compile_rules(serialized_rules): class Action:
raise NotImplementedError('TODO: Implement this')
def __init__(self, serialized_action):
pass
class Rule:
def __init__(self, serialized_rule):
self.sel_type = 'xpath'
mt = serialized_rule['match_type']
q = serialized_rule['query']
if mt == 'xpath':
self.xpath_selector = XPath(q)
self.selector = self.xpath
elif mt in ('is', 'css'):
self.css_selector = q
self.selector = self.css
elif mt == '*':
self.xpath_selector = XPath('//*')
self.selector = self.xpath
elif mt == 'has_class':
self.css_selector = '.' + q
self.selector = self.css
elif mt == 'not_has_class':
self.css_selector = f":not(.{q})"
self.selector = self.css
else:
raise KeyError(f'Unknown match_type: {mt}')
self.actions = tuple(map(Action, serialized_rule['actions']))
def xpath(self, root):
return self.xpath_selector(root)
def css(self, root):
return tuple(Select(root)(self.css_selector))
def __call__(self, root):
changed = False
for tag in self.selector(root):
for action in self.actions:
if action(tag):
changed = True
return changed
def transform_doc(root, rules):
changed = False
for rule in rules:
if rule(root):
changed = True
return changed
def transform_container(container, serialized_rules, names=()): def transform_container(container, serialized_rules, names=()):
rules = compile_rules(serialized_rules) if not names:
rules types = OEB_DOCS
raise NotImplementedError('TODO: Implement this') names = []
for name, mt in container.mime_map.items():
if mt in types:
names.append(name)
doc_changed = False
rules = tuple(Rule(r) for r in serialized_rules)
for name in names:
mt = container.mime_map.get(name)
if mt in OEB_DOCS:
root = container.parsed(name)
if transform_doc(root, rules):
container.dirty(name)
doc_changed = True
return doc_changed
def rule_to_text(rule): def rule_to_text(rule):
@ -172,7 +238,35 @@ def test(return_tests=False): # {{{
ae = unittest.TestCase.assertEqual ae = unittest.TestCase.assertEqual
def test_matching(self): def test_matching(self):
pass from html5_parser import parse
root = parse(namespace_elements=True, html='''
<html id='root'>
<head id='head'></head>
<body id='body'>
<p class="one red" id='p1'>
<p class="two green" id='p2'>
''')
all_ids = root.xpath('//*/@id')
def q(mt, query=''):
r = Rule({'match_type': mt, 'query': query, 'actions':[]})
ans = []
for tag in r.selector(root):
ans.append(tag.get('id'))
return ans
def t(mt, query='', expected=[]):
self.ae(expected, q(mt, query))
t('*', expected=all_ids)
t('is', 'body', ['body'])
t('is', 'p', ['p1', 'p2'])
t('has_class', 'one', ['p1'])
ei = list(all_ids)
ei.remove('p1')
t('not_has_class', 'one', ei)
t('css', '#body > p.red', ['p1'])
t('xpath', '//h:body', ['body'])
def test_validate_rule(self): def test_validate_rule(self):
def av(match_type='*', query='', atype='remove', adata=''): def av(match_type='*', query='', atype='remove', adata=''):

12
src/calibre/gui2/html_transform_rules.py
View File

@ -11,7 +11,7 @@ from qt.core import (
from calibre import prepare_string_for_xml from calibre import prepare_string_for_xml
from calibre.ebooks.html_transform_rules import ( from calibre.ebooks.html_transform_rules import (
ACTION_MAP, MATCH_TYPE_MAP, compile_rules, export_rules, import_rules, ACTION_MAP, MATCH_TYPE_MAP, export_rules, import_rules,
validate_rule validate_rule
) )
from calibre.gui2 import choose_files, choose_save_file, elided_text, error_dialog from calibre.gui2 import choose_files, choose_save_file, elided_text, error_dialog
@ -311,12 +311,12 @@ class Rules(RulesBase): # {{{
class Tester(Dialog): # {{{ class Tester(Dialog): # {{{
DIALOG_TITLE = _('Test style transform rules') DIALOG_TITLE = _('Test HTML transform rules')
PREFS_NAME = 'test-style-transform-rules' PREFS_NAME = 'test-html-transform-rules'
LABEL = _('Enter a CSS stylesheet below to test') LABEL = _('Enter an HTML document below to test')
def __init__(self, rules, parent=None): def __init__(self, rules, parent=None):
self.rules = compile_rules(rules) self.rules = rules
Dialog.__init__(self, self.DIALOG_TITLE, self.PREFS_NAME, parent=parent) Dialog.__init__(self, self.DIALOG_TITLE, self.PREFS_NAME, parent=parent)
def setup_ui(self): def setup_ui(self):
@ -326,7 +326,7 @@ class Tester(Dialog): # {{{
self.la = la = QLabel(self.LABEL) self.la = la = QLabel(self.LABEL)
l.addWidget(la) l.addWidget(la)
self.css = t = TextEdit(self) self.css = t = TextEdit(self)
t.load_text('/* %s */\n' % _('Enter CSS rules below and click the "Test" button'), 'css') t.load_text('<!-- %s -->\n' % _('Enter the HTML below and click the "Test" button'), 'html')
la.setBuddy(t) la.setBuddy(t)
c = t.textCursor() c = t.textCursor()
c.movePosition(QTextCursor.MoveOperation.End) c.movePosition(QTextCursor.MoveOperation.End)