Edit book: Check book: Allow automatic fixing of various simple CSS errors

resources/stylelint.js

@@ -10,9 +10,10 @@
 
 window.stylelint_results = [];
 
-window.check_css =  function(src) {
+window.check_css =  function(src, fix) {
     stylelint.lint({
         code: src,
+        fix: fix,
         config: {
             rules: {
                 'annotation-no-unknown': true,

src/calibre/ebooks/oeb/polish/check/css.py

@@ -22,14 +22,17 @@ from calibre.utils.webengine import secure_webengine, setup_profile
 class CSSParseError(BaseError):
     level = ERROR
     is_parsing_error = True
+    FIXABLE_CSS_ERROR = False
 
 
 class CSSError(BaseError):
     level = ERROR
+    FIXABLE_CSS_ERROR = False
 
 
 class CSSWarning(BaseError):
     level = WARN
+    FIXABLE_CSS_ERROR = False
 
 
 def as_int_or_none(x):
@@ -57,10 +60,15 @@ def message_to_error(message, name, line_offset, rule_metadata):
         line += line_offset
     ans = cls(title, name, line, col)
     ans.HELP = message.get('text') or ''
+    if ans.HELP:
+        ans.HELP += '. '
     ans.css_rule_id = rule_id
-    m = rule_metadata.get(rule_id)
+    m = rule_metadata.get(rule_id) or {}
-    if m and 'url' in m:
+    if 'url' in m:
-        ans.HELP += '. ' + _('See <a href="{}">detailed description</a>.').format(m['url'])
+        ans.HELP += _('See <a href="{}">detailed description</a>.').format(m['url']) + ' '
+    if m.get('fixable'):
+        ans.FIXABLE_CSS_ERROR = True
+        ans.HELP += _('This error will be automatically fixed if you click "Try to correct all fixable errors" below.')
     return ans
 
 
@@ -107,7 +115,7 @@ class Worker(QWebEnginePage):
         if new_title == 'ready':
             self.ready = True
             if self.pending is not None:
-                self.check_css(self.pending)
+                self.check_css(*self.pending)
                 self.pending = None
         elif new_title == 'checked':
             self.runJavaScript('window.get_css_results()', QWebEngineScript.ScriptWorldId.ApplicationWorld, self.check_done)
@@ -119,17 +127,17 @@ class Worker(QWebEnginePage):
         except Exception:
             pass
 
-    def check_css(self, src):
+    def check_css(self, src, fix=False):
         self.working = True
         self.runJavaScript(
-            f'window.check_css({json.dumps(src)})', QWebEngineScript.ScriptWorldId.ApplicationWorld)
+            f'window.check_css({json.dumps(src)}, {"true" if fix else "false"})', QWebEngineScript.ScriptWorldId.ApplicationWorld)
 
-    def check_css_when_ready(self, src):
+    def check_css_when_ready(self, src, fix=False):
         if self.ready:
-            self.check_css(src)
+            self.check_css(src, fix)
         else:
             self.working = True
-            self.pending = src
+            self.pending = src, fix
 
     def check_done(self, results):
         self.working = False
@@ -148,7 +156,8 @@ class Pool:
         w.work_done.connect(self.work_done)
         self.workers.append(w)
 
-    def check_css(self, css_sources):
+    def check_css(self, css_sources, fix=False):
+        self.doing_fix = fix
         self.pending = list(enumerate(css_sources))
         self.results = list(repeat(None, len(css_sources)))
         self.working = True
@@ -166,7 +175,7 @@ class Pool:
                 if not w.working:
                     idx, src = self.pending.pop()
                     w.result_idx = idx
-                    w.check_css_when_ready(src)
+                    w.check_css_when_ready(src, self.doing_fix)
                     break
             else:
                 break
@@ -191,14 +200,14 @@ class Pool:
 pool = Pool()
 shutdown = pool.shutdown
 atexit.register(shutdown)
-Job = namedtuple('Job', 'name css line_offset')
+Job = namedtuple('Job', 'name css line_offset fix_data')
 
 
-def create_job(name, css, line_offset=0, is_declaration=False):
+def create_job(name, css, line_offset=0, is_declaration=False, fix_data=None):
     if is_declaration:
         css = 'div{\n' + css + '\n}'
         line_offset -= 1
-    return Job(name, css, line_offset)
+    return Job(name, css, line_offset, fix_data)
 
 
 def check_css(jobs):

src/calibre/ebooks/oeb/polish/check/main.py

@@ -4,21 +4,23 @@
 __license__ = 'GPL v3'
 __copyright__ = '2013, Kovid Goyal <kovid at kovidgoyal.net>'
 
-from polyglot.builtins import iteritems
+from collections import namedtuple
 
 from calibre.ebooks.oeb.base import OEB_DOCS, OEB_STYLES
-from calibre.ebooks.oeb.polish.utils import guess_type
+from calibre.ebooks.oeb.polish.check.base import WARN, run_checkers
-from calibre.ebooks.oeb.polish.cover import is_raster_image
-from calibre.ebooks.oeb.polish.check.base import run_checkers, WARN
-from calibre.ebooks.oeb.polish.check.parsing import (
-    check_filenames, check_xml_parsing, fix_style_tag,
-    check_html_size, check_ids, check_markup, EmptyFile, check_encoding_declarations)
-from calibre.ebooks.oeb.polish.check.images import check_raster_images
-from calibre.ebooks.oeb.polish.check.links import check_links, check_mimetypes, check_link_destinations
 from calibre.ebooks.oeb.polish.check.fonts import check_fonts
+from calibre.ebooks.oeb.polish.check.images import check_raster_images
+from calibre.ebooks.oeb.polish.check.links import (
+    check_link_destinations, check_links, check_mimetypes,
+)
 from calibre.ebooks.oeb.polish.check.opf import check_opf
-from polyglot.builtins import as_unicode
+from calibre.ebooks.oeb.polish.check.parsing import (
-
+    EmptyFile, check_encoding_declarations, check_filenames, check_html_size, check_ids,
+    check_markup, check_xml_parsing, fix_style_tag,
+)
+from calibre.ebooks.oeb.polish.cover import is_raster_image
+from calibre.ebooks.oeb.polish.utils import guess_type
+from polyglot.builtins import as_unicode, iteritems
 
 XML_TYPES = frozenset(map(guess_type, ('a.xml', 'a.svg', 'a.opf', 'a.ncx'))) | {'application/oebps-page-map+xml'}
 
@@ -105,6 +107,48 @@ def run_checks(container):
     return errors
 
 
+CSSFix = namedtuple('CSSFix', 'original_css elem attribute')
+
+
+def fix_css(container):
+    from calibre.ebooks.oeb.polish.check.css import create_job, pool
+    jobs  = []
+
+    for name, mt in iteritems(container.mime_map):
+        if mt in OEB_STYLES:
+            css = container.raw_data(name, decode=True)
+            jobs.append(create_job(name, css, fix_data=CSSFix(css, None, '')))
+        elif mt in OEB_DOCS:
+            root = container.parsed(name)
+            for style in root.xpath('//*[local-name()="style"]'):
+                if style.get('type', 'text/css') == 'text/css' and style.text:
+                    jobs.append(create_job(name, style.text, fix_data=CSSFix(style.text, style, '')))
+            for elem in root.xpath('//*[@style]'):
+                raw = elem.get('style')
+                if raw:
+                    jobs.append(create_job(name, raw, is_declaration=True, fix_data=CSSFix(raw, elem, 'style')))
+    results = pool.check_css([j.css for j in jobs], fix=True)
+    changed = False
+    for job, result in zip(jobs, results):
+        if result['type'] == 'error':
+            continue
+        fx = job.fix_data
+        fixed_css = result['results']['output']
+        if fixed_css == fx.original_css:
+            continue
+        changed = True
+        if fx.elem is None:
+            with container.open(job.name, 'wb') as f:
+                f.write(fixed_css.encode('utf-8'))
+        else:
+            if fx.attribute:
+                fx.elem.set(fx.attribute, ' '.join(fixed_css.splitlines()[1:-1]))
+            else:
+                fx.elem.text = fixed_css
+            container.dirty(job.name)
+    return changed
+
+
 def fix_errors(container, errors):
     # Fix parsing
     changed = False
@@ -121,11 +165,17 @@ def fix_errors(container, errors):
 
         changed = True
 
+    has_fixable_css_errors = False
     for err in errors:
+        if getattr(err, 'FIXABLE_CSS_ERROR', False):
+            has_fixable_css_errors = True
         if err.INDIVIDUAL_FIX:
             if err(container) is not False:
                 # Assume changed unless fixer explicitly says no change (this
                 # is because sometimes I forget to return True, and it is
                 # better to have a false positive than a false negative)
                 changed = True
+    if has_fixable_css_errors:
+        if fix_css(container):
+            changed = True
     return changed