Bible fixes:

- Red letter text on CrossWalk import.
 - Removed text on web download progress, moved to an "indeterminable" progress style.
 - Some unicode optimisations.
This commit is contained in:
Raoul Snyman 2010-03-27 21:59:35 +02:00
parent 08ee7be551
commit d2670b1cb0
5 changed files with 46 additions and 10 deletions

View File

@ -29,6 +29,8 @@ import logging
import urllib2
from datetime import datetime
from PyQt4 import QtCore
import openlp
log = logging.getLogger(__name__)
@ -123,16 +125,25 @@ def check_latest_version(config, current_version):
log.exception(u'Reason for failure: %s', e.reason)
return version_string
def string_to_unicode(string):
"""
Converts a QString to a Python unicode object.
"""
if isinstance(string, QtCore.QString):
string = unicode(string.toUtf8(), u'utf8')
return string
def variant_to_unicode(variant):
"""
Converts a QVariant to a unicode string.
Converts a QVariant to a Python unicode object.
``variant``
The QVariant instance to convert to unicode.
"""
string = variant.toString()
if isinstance(variant, QtCore.QVariant):
string = variant.toString()
if not isinstance(string, unicode):
string = unicode(string, u'utf8')
string = string_to_unicode(string)
return string
from registry import Registry

View File

@ -32,7 +32,7 @@ from PyQt4 import QtCore, QtGui
from bibleimportwizard import Ui_BibleImportWizard
from openlp.core.lib import Receiver
from openlp.core.utils import AppLocation, variant_to_unicode
from openlp.core.utils import AppLocation, variant_to_unicode, string_to_unicode
from openlp.plugins.bibles.lib.manager import BibleFormat
log = logging.getLogger(__name__)
@ -425,3 +425,4 @@ class ImportWizardForm(QtGui.QWizard, Ui_BibleImportWizard):
self.finishButton.setVisible(True)
self.cancelButton.setVisible(False)
Receiver.send_message(u'process_events')

View File

@ -33,7 +33,7 @@ only_verses = re.compile(r'([\w .]+)[ ]+([0-9]+)[ ]*[:|v|V][ ]*([0-9]+)'
r'(?:[ ]*-[ ]*([0-9]+|end))?(?:[ ]*,[ ]*([0-9]+)(?:[ ]*-[ ]*([0-9]+|end))?)?',
re.UNICODE)
chapter_range = re.compile(r'([\w .]+)[ ]+([0-9]+)[ ]*[:|v|V][ ]*'
r'([0-9]+)[ ]*-[ ]*([0-9]+)[ ]*[:|v|V][ ]*([0-9]+)',
r'([0-9]+|end)[ ]*-[ ]*([0-9]+)[ ]*[:|v|V][ ]*([0-9]+|end)',
re.UNICODE)
log = logging.getLogger(__name__)

View File

@ -203,7 +203,9 @@ class BGExtract(BibleCommon):
# Let's get the page, and then open it in BeautifulSoup, so as to
# attempt to make "easy" work of bad HTML.
page = urllib2.urlopen(urlstring)
Receiver.send_message(u'process_events')
soup = BeautifulSoup(page)
Receiver.send_message(u'process_events')
verses = soup.find(u'div', u'result-text-style-normal')
verse_number = 0
verse_list = {0: u''}
@ -211,6 +213,7 @@ class BGExtract(BibleCommon):
# This is a PERFECT example of opening the Cthulu tag!
# O Bible Gateway, why doth ye such horrific HTML produce?
for verse in verses:
Receiver.send_message(u'process_events')
if isinstance(verse, Tag) and verse.name == u'div' and filter(lambda a: a[0] == u'class', verse.attrs)[0][1] == u'footnotes':
break
if isinstance(verse, Tag) and verse.name == u'sup' and filter(lambda a: a[0] == u'class', verse.attrs)[0][1] != u'versenum':
@ -219,6 +222,7 @@ class BGExtract(BibleCommon):
continue
if isinstance(verse, Tag) and (verse.name == u'p' or verse.name == u'font') and verse.contents:
for item in verse.contents:
Receiver.send_message(u'process_events')
if isinstance(item, Tag) and (item.name == u'h4' or item.name == u'h5'):
continue
if isinstance(item, Tag) and item.name == u'sup' and filter(lambda a: a[0] == u'class', item.attrs)[0][1] != u'versenum':
@ -231,6 +235,7 @@ class BGExtract(BibleCommon):
continue
if isinstance(item, Tag) and item.name == u'font':
for subitem in item.contents:
Receiver.send_message(u'process_events')
if isinstance(subitem, Tag) and subitem.name == u'sup' and filter(lambda a: a[0] == u'class', subitem.attrs)[0][1] != u'versenum':
continue
if isinstance(subitem, Tag) and subitem.name == u'p' and not subitem.contents:
@ -289,27 +294,42 @@ class CWExtract(BibleCommon):
(version, urlbookname.lower(), chapter)
log.debug(u'URL: %s', chapter_url)
page = urllib2.urlopen(chapter_url)
Receiver.send_message(u'process_events')
if not page:
return None
soup = BeautifulSoup(page)
Receiver.send_message(u'process_events')
htmlverses = soup.findAll(u'span', u'versetext')
verses = {}
reduce_spaces = re.compile(r'[ ]{2,}')
fix_punctuation = re.compile(r'[ ]+([.,;])')
for verse in htmlverses:
Receiver.send_message(u'process_events')
versenumber = int(verse.contents[0].contents[0])
versetext = u''
for part in verse.contents:
Receiver.send_message(u'process_events')
if isinstance(part, NavigableString):
versetext = versetext + part
elif part and part.attrMap and \
(part.attrMap[u'class'] == u'WordsOfChrist' or \
part.attrMap[u'class'] == u'strongs'):
for subpart in part.contents:
Receiver.send_message(u'process_events')
if isinstance(subpart, NavigableString):
versetext = versetext + subpart
elif subpart and subpart.attrMap and \
subpart.attrMap[u'class'] == u'strongs':
for subsub in subpart.contents:
Receiver.send_message(u'process_events')
if isinstance(subsub, NavigableString):
versetext = versetext + subsub
Receiver.send_message(u'process_events')
# Fix up leading and trailing spaces, multiple spaces, and spaces
# between text and , and .
versetext = versetext.strip(u'\n\r\t ')
versetext = reduce_spaces.sub(u' ', versetext)
versetext = fix_punctuation.sub(r'\1', versetext)
verses[versenumber] = versetext
return SearchResults(bookname, chapter, verses)
@ -410,10 +430,12 @@ class HTTPBible(BibleDB):
## we get a correct book. For example it is possible
## to request ac and get Acts back.
bookname = search_results.get_book()
Receiver.send_message(u'process_events')
# check to see if book/chapter exists
db_book = self.get_book(bookname)
self.create_chapter(db_book.id, search_results.get_chapter(),
search_results.get_verselist())
Receiver.send_message(u'process_events')
Receiver.send_message(u'bible_hideprogress')
Receiver.send_message(u'process_events')
return BibleDB.get_verses(self, reference_list)

View File

@ -266,8 +266,9 @@ class BibleMediaItem(MediaManagerItem):
MediaManagerItem.addListViewToToolBar(self)
# Progress Bar
self.SearchProgress = QtGui.QProgressBar(self)
self.SearchProgress.setFormat('%p%')
self.SearchProgress.setMaximum(3)
self.SearchProgress.setFormat('')
self.SearchProgress.setMinimum(0)
self.SearchProgress.setMaximum(0)
self.SearchProgress.setGeometry(self.ListView.geometry().left(),
self.ListView.geometry().top(), 81, 23)
self.SearchProgress.setVisible(False)
@ -351,9 +352,10 @@ class BibleMediaItem(MediaManagerItem):
def onSearchProgressShow(self):
self.SearchProgress.setVisible(True)
self.SearchProgress.setMinimum(0)
self.SearchProgress.setMaximum(2)
self.SearchProgress.setValue(1)
Receiver.send_message(u'process_events')
#self.SearchProgress.setMinimum(0)
#self.SearchProgress.setMaximum(2)
#self.SearchProgress.setValue(1)
def onSearchProgressHide(self):
self.SearchProgress.setVisible(False)