openlp/openlp/plugins/songs/lib/importers/chordpro.py

215 lines
11 KiB
Python

# -*- coding: utf-8 -*-
##########################################################################
# OpenLP - Open Source Lyrics Projection #
# ---------------------------------------------------------------------- #
# Copyright (c) 2008-2022 OpenLP Developers #
# ---------------------------------------------------------------------- #
# This program is free software: you can redistribute it and/or modify #
# it under the terms of the GNU General Public License as published by #
# the Free Software Foundation, either version 3 of the License, or #
# (at your option) any later version. #
# #
# This program is distributed in the hope that it will be useful, #
# but WITHOUT ANY WARRANTY; without even the implied warranty of #
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the #
# GNU General Public License for more details. #
# #
# You should have received a copy of the GNU General Public License #
# along with this program. If not, see <https://www.gnu.org/licenses/>. #
##########################################################################
"""
The :mod:`chordpro` module provides the functionality for importing
ChordPro files into the current database.
"""
import logging
import re
from openlp.core.common.i18n import translate
from openlp.plugins.songs.lib.importers.songimport import SongImport
from openlp.plugins.songs.lib.db import AuthorType
log = logging.getLogger(__name__)
class ChordProImport(SongImport):
"""
The :class:`ChordProImport` class provides OpenLP with the ability to import ChordPro files.
This importer is based on the information available on these webpages:
- https://www.chordpro.org
- http://webchord.sourceforge.net/tech.html
- http://www.vromans.org/johan/projects/Chordii/chordpro/
- http://www.tenbyten.com/software/songsgen/help/HtmlHelp/files_reference.htm
- http://linkesoft.com/songbook/chordproformat.html
"""
def do_import(self):
self.import_wizard.progress_bar.setMaximum(len(self.import_source))
for file_path in self.import_source:
if self.stop_import_flag:
return
with file_path.open('rt', encoding='utf8') as song_file:
self.do_import_file(song_file)
def do_import_file(self, song_file):
"""
Imports the songs in the given file
:param song_file: The file object to be imported from.
"""
self.set_defaults()
# Loop over the lines of the file
try:
file_content = song_file.read()
except UnicodeDecodeError:
self.log_error(song_file.name, translate('SongsPlugin.CCLIFileImport',
'The file contains unreadable characters.'))
return
current_verse = ''
current_verse_type = 'v'
skip_block = False
for line in file_content.splitlines():
line = line.rstrip()
# Detect tags
if line.startswith('{'):
tag_name, tag_value = self.parse_tag(line)
# Detect which tag
if tag_name in ['title', 't']:
self.title = tag_value
elif tag_name in ['subtitle', 'su', 'st']:
self.alternate_title = tag_value
elif tag_name == 'composer':
self.parse_author(tag_value, AuthorType.Music)
elif tag_name in ['lyricist', 'artist', 'author']: # author is not an official directive
self.parse_author(tag_value, AuthorType.Words)
elif tag_name == 'meta':
meta_tag_name, meta_tag_value = tag_value.split(' ', 1)
# Skip, if no value
if not meta_tag_value:
continue
# The meta-tag can contain anything. We check for the ones above and a few more
if meta_tag_name in ['title', 't']:
self.title = meta_tag_value
elif meta_tag_name in ['subtitle', 'su', 'st']:
self.alternate_title = meta_tag_value
elif meta_tag_name == 'composer':
self.parse_author(meta_tag_value, AuthorType.Music)
elif meta_tag_name in ['lyricist', 'artist', 'author']:
self.parse_author(meta_tag_value, AuthorType.Words)
elif meta_tag_name in ['topic', 'topics']:
for topic in meta_tag_value.split(','):
self.topics.append(topic.strip())
elif 'ccli' in meta_tag_name:
self.ccli_number = meta_tag_value
elif tag_name in ['comment', 'c', 'comment_italic', 'ci', 'comment_box', 'cb']:
# Detect if the comment is used as a chorus repeat marker
if tag_value.lower().startswith('chorus'):
if current_verse.strip():
# Add collected verse to the lyrics
# Strip out chords if set up to
if not self.settings.value('songs/enable chords') or \
self.settings.value('songs/disable chords import'):
current_verse = re.sub(r'\[.*?\]', '', current_verse)
self.add_verse(current_verse.rstrip(), current_verse_type)
current_verse_type = 'v'
current_verse = ''
self.repeat_verse('c1')
else:
self.add_comment(tag_value)
elif tag_name in ['start_of_chorus', 'soc']:
current_verse_type = 'c'
elif tag_name in ['end_of_chorus', 'eoc']:
# Add collected chorus to the lyrics
# Strip out chords if set up to
if not self.settings.value('songs/enable chords') or \
self.settings.value('songs/disable chords import'):
current_verse = re.sub(r'\[.*?\]', '', current_verse)
self.add_verse(current_verse.rstrip(), current_verse_type)
current_verse_type = 'v'
current_verse = ''
elif tag_name in ['start_of_tab', 'sot']:
if current_verse.strip():
# Add collected verse to the lyrics
# Strip out chords if set up to
if not self.settings.value('songs/enable chords') or \
self.settings.value('songs/disable chords import'):
current_verse = re.sub(r'\[.*?\]', '', current_verse)
self.add_verse(current_verse.rstrip(), current_verse_type)
current_verse_type = 'v'
current_verse = ''
skip_block = True
elif tag_name in ['end_of_tab', 'eot']:
skip_block = False
elif tag_name in ['new_song', 'ns']:
# A new song starts below this tag
if self.verses and self.title:
if current_verse.strip():
# Strip out chords if set up to
if not self.settings.value('songs/enable chords') or \
self.settings.value('songs/disable chords import'):
current_verse = re.sub(r'\[.*?\]', '', current_verse)
self.add_verse(current_verse.rstrip(), current_verse_type)
if not self.finish():
self.log_error(song_file.name)
self.set_defaults()
current_verse_type = 'v'
current_verse = ''
else:
# Unsupported tag
log.debug('unsupported tag: %s' % line)
elif line.startswith('#'):
# Found a comment line, which is ignored...
continue
elif line == "['|]":
# Found a vertical bar
continue
else:
if skip_block:
continue
elif line == '' and current_verse.strip() and current_verse_type != 'c':
# Add collected verse to the lyrics
# Strip out chords if set up to
if not self.settings.value('songs/enable chords') or \
self.settings.value('songs/disable chords import'):
current_verse = re.sub(r'\[.*?\]', '', current_verse)
self.add_verse(current_verse.rstrip(), current_verse_type)
current_verse_type = 'v'
current_verse = ''
else:
if current_verse.strip() == '':
current_verse = line + '\n'
else:
current_verse += line + '\n'
if current_verse.strip():
# Strip out chords if set up to
if not self.settings.value('songs/enable chords') or self.settings.value(
'songs/disable chords import'):
current_verse = re.sub(r'\[.*?\]', '', current_verse)
self.add_verse(current_verse.rstrip(), current_verse_type)
# if no title was in directives, get it from the first line
if not self.title and self.verses:
(verse_def, verse_text, lang) = self.verses[0]
# strip any chords from the title
self.title = re.sub(r'\[.*?\]', '', verse_text.split('\n')[0])
# strip the last char if it a punctuation
self.title = re.sub(r'[^\w\s]$', '', self.title)
if not self.finish():
self.log_error(song_file.name)
def parse_tag(self, line):
"""
:param line: Line with the tag to be parsed
:return: A tuple with tag name and tag value (if any)
"""
# Strip the first '}'
line = line[1:].strip()
colon_idx = line.find(':')
# check if this is a tag without value
if colon_idx < 0:
# strip the final '}' and return the tag name
return line[:-1], None
tag_name = line[:colon_idx]
tag_value = line[colon_idx + 1:-1].strip()
return tag_name, tag_value