def add_word(word):
from text_indexer.orm.base import session
db_word = session.query(Word).filter_by(word=word).first()
if not db_word:
db_word = Word(word)
session.add(db_word)
session.commit()
return db_word
def get_groups(name="", type=""):
from text_indexer.orm.base import session
groups = session.query(Group)
if name:
groups = groups.filter_by(name=name)
if type:
groups = groups.filter_by(type=type)
return groups.all()
def import_file(self, name, writer, performer, path):
"""
@param name: The name of the song
@param writer: The song writer.
@param path: The path of the song.
"""
db_song = session.query(Song).filter_by(name=name).first()
if not db_song:
db_song = Song(name=name, writer=writer, performer=performer)
session.add(db_song)
self.load_file_into_song(path, db_song)
def get_stanza(self, stanza_number):
"""
This method returns a stanza from this song.
@param stanza_number: The number of the stanza to return.
"""
from text_indexer.orm.base import session
from text_indexer.orm.word_position import WordPosition
wps = session.query(WordPosition).filter_by(song_id=self.id, stanza_number=stanza_number).order_by(WordPosition.stanza_line_number,WordPosition.row_word_number).all()
stanza_line_number=1
stanza = ''
for w in wps:
if w.stanza_line_number > stanza_line_number:
stanza+='\n'
stanza_line_number+=1
stanza+=w.word.word + ' '
return stanza
def __init__(self, parent):
from text_indexer.orm.base import session
from text_indexer.orm.word_position import WordPosition
wx.Panel.__init__(self, parent, -1)
self.grid = wx.grid.Grid(self, -1, (100, 50), (1220, 400))
self.wps = session.query(WordPosition).order_by(WordPosition.song_id, WordPosition.line_number,WordPosition.row_word_number).all()
self.grid.CreateGrid(len(self.wps), 8)
self.grid.SetColSize(0, 130)
self.grid.SetColSize(1, 130)
self.grid.SetColSize(2, 130)
self.grid.SetColSize(3, 130)
self.grid.SetColSize(4, 130)
self.grid.SetColSize(5, 130)
self.grid.SetColSize(6, 130)
self.grid.SetColSize(7, 130)
self.grid.SetColLabelValue(0, "Song")
self.grid.SetColLabelValue(1, "Words")
self.grid.SetColLabelValue(2, "Number in song")
self.grid.SetColLabelValue(3, "Paragraph")
self.grid.SetColLabelValue(4, "Number in paragraph")
self.grid.SetColLabelValue(5, "Line number")
self.grid.SetColLabelValue(6, "number in line")
self.grid.SetColLabelValue(7, "word occurrence")
group_text = wx.StaticText(self, -1, "Select Group", (100, 480))
self.groups_list = [g.name for g in Group.get_groups(type='group')]
self.select_group = wx.ComboBox(self, 500, "", (100, 500),
(160, -1), self.groups_list, wx.CB_DROPDOWN)
self.group_button = wx.Button(self, -1, "Show Group Index", (320, 500))
self.Bind(wx.EVT_BUTTON, self.groupChosen, self.group_button)
self.refresh_button = wx.Button(self, -1, "Refresh", (500, 500))
self.Bind(wx.EVT_BUTTON, self.refresh, self.refresh_button)
self.show_words()
def get_songs(name="", writer="", performer="", word=""):
"""
This method returns the songs from the db.
@param name: The name of the song
@param writer: The writer of the song
@param word: A word that is in the song.
"""
from text_indexer.orm.base import session
songs = session.query(Song)
if name:
songs = songs.filter_by(name=name)
if writer:
songs = songs.filter_by(writer=writer)
if performer:
songs = songs.filter_by(performer=performer)
if word:
songs = songs.join(Song.words).filter_by(word=word)
return songs.all()
def phraseChosen(self, evt):
from text_indexer.orm.base import session
text = self.t3.GetValue()
start,end = self.t3.GetSelection()
selected = text[start:end]
words = selected.replace('\n',' ').strip().replace(' ', ' ').split(' ')
matches = set()
number_of_words = len(words)
word = session.query(Word).filter_by(word=words[0])[0]
wps = word.word_positions
for w in wps:
i = 1
wp = w.get_next_word()
while i < number_of_words and wp.word.word == words[i]:
i+=1
wp = wp.get_next_word()
if i == number_of_words:
matches.add(w)
wps = set()
text = ''
songs = [self.lb1.Items[song_selection] for song_selection in self.lb1.Selections]
for song in songs:
added_song_name = False
for wp in matches:
if wp.song.name == song:
if not added_song_name:
text+= song + ':\n\n'
added_song_name = True
if (wp.song.id, wp.stanza_number) not in wps:
wps.add((wp.song.id, wp.stanza_number))
text+= wp.song.get_stanza(wp.stanza_number)
text+= '\n\n\n'
#
# self.t3.SetValue(str(words))
self.t3.SetValue(text)
self.t3.SetScrollPos(1,1)
for m in re.finditer(" " + selected, text):
self.t3.SetStyle(m.start()+1, m.end(), wx.TextAttr("RED", "YELLOW"))
for m in re.finditer("\n" + word.word, text ):
self.t3.SetStyle(m.start()+1, m.end(), wx.TextAttr("RED", "YELLOW"))
def _create_word_in_DB(
self, word, db_song, stanza_number, row_word_number, line_number, stanza_line_number, number_in_song
):
db_word = session.query(Word).filter_by(word=word).first()
if not db_word:
if word in self.words_to_add.keys():
db_word = self.words_to_add[word]
else:
self.words_to_add[word] = Word(word=word)
word_position = WordPosition(
word=db_word,
song=db_song,
row_word_number=row_word_number,
stanza_number=stanza_number,
line_number=line_number,
stanza_line_number=stanza_line_number,
number_in_song=number_in_song,
)
self.word_positions.append(word_position)
def get_text(self):
"""
This method returns the text of the song.
"""
from text_indexer.orm.base import session
from text_indexer.orm.word_position import WordPosition
wps = session.query(WordPosition).filter_by(song_id=self.id).order_by(WordPosition.line_number,WordPosition.row_word_number).all()
song = ''
stanza_number=1
stanza_line_number=1
for w in wps:
if w.stanza_number > stanza_number:
stanza_number+=1
song+='\n\n'
stanza_line_number=1
if w.stanza_line_number > stanza_line_number:
song+='\n'
stanza_line_number+=1
song+=w.word.word + ' '
return song
def onSearch(self, evt):
try:
from text_indexer.orm.base import session
song_name = self.lb1.Items[self.lb1.GetSelection()]
song = Song.get_songs(name=song_name)[0]
query = session.query(WordPosition).filter_by(song_id=song.id)
if self.radio_selected == self.radio1:
query = query.filter_by(number_in_song=self.text1.Value)
wp = query.first()
self.word_text.Value = wp.word.word
elif self.radio_selected == self.radio2:
query = query.filter_by(line_number=self.text2_1.Value, row_word_number=self.text2_2.Value)
wp = query.first()
self.word_text.Value = wp.word.word
elif self.radio_selected == self.radio3:
stanza = song.get_stanza(int(self.text3_1.Value) - 1)
self.word_text.Value = stanza.split(" ")[int(self.text3_2.Value) - 1]
if not self.word_text.Value:
self.word_text.Value = "No word found"
except Exception, e:
self.word_text.Value = "No word found"
def add_expression(expression):
from text_indexer.orm.word import Word
from text_indexer.orm.base import session
db_words = []
for word in expression.split(' '):
db_word = session.query(Word).filter_by(word=word).first()
if not db_word:
db_word = Word(word=word)
session.add(db_word)
db_words.append(db_word)
session.commit()
db_expression = Expression(name=expression)
session.add(db_expression)
session.commit()
number=1
for db_word in db_words:
word_expression = WordExpressionAssocaition(word_id=db_word.id, expression_id=db_expression.id, place=number)
session.add(word_expression)
number+=1
session.commit()
return db_expression
def get_next_word(self):
from text_indexer.orm.base import session
return session.query(WordPosition).filter_by(song_id=self.song_id, number_in_song=self.number_in_song+1).first()
def get_words(word=""):
from text_indexer.orm.base import session
words = session.query(Word).order_by(Word.word)
if word:
words = words.filter_by(word=word)
return words.all()
for word_position in self.word_positions:
session.add(word_position)
session.commit()
def _create_word_in_DB(
self, word, db_song, stanza_number, row_word_number, line_number, stanza_line_number, number_in_song
):
db_word = session.query(Word).filter_by(word=word).first()
if not db_word:
if word in self.words_to_add.keys():
db_word = self.words_to_add[word]
else:
self.words_to_add[word] = Word(word=word)
word_position = WordPosition(
word=db_word,
song=db_song,
row_word_number=row_word_number,
stanza_number=stanza_number,
line_number=line_number,
stanza_line_number=stanza_line_number,
number_in_song=number_in_song,
)
self.word_positions.append(word_position)
if __name__ == "__main__":
# FileImporter().import_file("Call Me Maybe", "Carly Rae Jepsen", r"C:\text\text_indexer\songs\call_me_maybe.txt")
song = session.query(Song).first()
song.get_stanza(2)
def __init__(self, parent):
wx.Panel.__init__(self, parent, -1)
text = "Statistics"
text = wx.StaticText(self, -1, text, (600, 50))
font = wx.Font(24, wx.SWISS, wx.NORMAL, wx.NORMAL, underline=True)
text.SetFont(font)
self.grid = wx.grid.Grid(self, -1, (200, 150), (900, 140))
self.grid.CreateGrid(2, 4)
self.grid.SetColSize(0, 200)
self.grid.SetColSize(1, 200)
self.grid.SetColSize(2, 200)
self.grid.SetColSize(3, 200)
self.grid.SetColLabelValue(0, "Word")
self.grid.SetColLabelValue(1, "Line")
self.grid.SetColLabelValue(2, "Paragraph")
self.grid.SetColLabelValue(3, "Song")
self.grid.SetRowSize(0, 50)
self.grid.SetRowSize(1, 50)
self.grid.SetRowLabelValue(0, "Chars")
self.grid.SetRowLabelValue(1, "Words")
from text_indexer.orm.base import session
number_of_words = 0
number_of_chars = 0
number_of_lines = 0
number_of_paragraphs = 0
line=0
stanza=0
song_id=0
for wp in session.query(WordPosition).order_by(WordPosition.song_id, WordPosition.line_number).all():
number_of_words+=1
number_of_chars+=len(wp.word.word)
if wp.song_id != song_id:
line = wp.line_number
stanza = wp.stanza_number
song_id = wp.song_id
number_of_lines+=1
number_of_paragraphs+=1
elif wp.stanza_number != stanza:
stanza = wp.stanza_number
line= wp.line_number
number_of_lines+=1
number_of_paragraphs+=1
elif wp.line_number != line:
number_of_lines+=1
chars_per_word = float(number_of_chars)/number_of_words
self.grid.SetCellValue(0, 0, str(chars_per_word))
self.grid.SetCellValue(0, 1, str(chars_per_word*number_of_words/number_of_lines))
self.grid.SetCellValue(0, 2, str(chars_per_word*number_of_words/number_of_paragraphs))
self.grid.SetCellValue(0, 3, str(number_of_words * chars_per_word/len(Song.get_songs())))
self.grid.SetCellValue(1, 0, '1')
self.grid.SetCellValue(1, 1, str(float(number_of_words)/number_of_lines))
self.grid.SetCellValue(1, 2, str(float(number_of_words)/number_of_paragraphs))
self.grid.SetCellValue(1, 3, str(float(number_of_words)/len(Song.get_songs())))
def get_expressions(expression=""):
from text_indexer.orm.base import session
expressions = session.query(Expression)
if expression:
expressions = expressions.filter_by(name=expression)
return expressions.all()