lab_server/poem_server/import_poems.py

82 lines
2.6 KiB
Python

import os
import json
import re
from tqdm import tqdm
from banjo.runner import setup_django
import requests
from pathlib import Path
from collections import defaultdict
from argparse import ArgumentParser
import pronouncing
import gzip
corpus_url = "http://static.decontextualize.com/gutenberg-poetry-v001.ndjson.gz"
corpus_file = Path("corpus.gz")
corpus_length = 3085117
def download_corpus():
"Downloads a few million lines of poetry"
if not corpus_file.exists():
print("Downloading a file with three million lines of poetry...")
download = requests.get(corpus_url, stream=True)
with corpus_file.open("wb") as f:
for chunk in download.iter_content(chunk_size=1024):
if chunk:
f.write(chunk)
f.flush()
def read_corpus(limit=None):
"Reads one line at a time from the corpus"
for i, line in enumerate(gzip.open(corpus_file)):
if limit and i == limit:
break
yield json.loads(line.strip())
def clean(line):
"Returns lowercase text without any punctuation"
return re.sub('[^a-z ]', '', line.lower()).strip()
def get_last_word(line):
"""Gets the last word from a line.
Strips out punctuation, then splits the line on spaces and returns the final word.
"""
parts = clean(line).split()
if parts:
return parts[-1]
def import_poems(limit=None):
"Imports each line into the app"
database_file = Path("database.sqlite")
if database_file.exists():
database_file.unlink()
setup_django()
from app.models import Line, Poem, Rhyme
print("Importing lines into the app's database...")
line_counts = defaultdict(int)
for row in tqdm(read_corpus(limit=limit), total=limit or corpus_length):
text = row['s']
last_word = get_last_word(text)
if not last_word:
continue
last_word_phones = pronouncing.phones_for_word(last_word)
if last_word_phones:
rhyming_phones = pronouncing.rhyming_part(last_word_phones[0])
rhyme, _ = Rhyme.objects.get_or_create(phones=rhyming_phones)
else:
rhyme=None
poem, _ = Poem.objects.get_or_create(gutenberg_id=row['gid'])
line = Line.objects.create(
text=text,
clean_text=clean(text),
line_number=line_counts[row['gid']],
poem=poem,
rhyme=rhyme,
)
line_counts[row['gid']] += 1
parser = ArgumentParser()
parser.add_argument('-l', '--limit', type=int)
args = parser.parse_args()
download_corpus()
import_poems(args.limit)