sqlite_encoder.py

#!/usr/bin/env python2
import sys
import sqlite3
import base64
import argparse
from tempfile import mkstemp
from glob import glob
import os
import filecmp
import csv
from tqdm import tqdm
from utils import safe_makedir


def parse_arguments():
    parser = argparse.ArgumentParser()
    parser.add_argument('--input_path', type=str, required=True,
                        help='''Path to input folder containing segmented data SQL files to import''')
    parser.add_argument('--output_path', type=str, required=True,
                        help='''Path to output folder to generate platform specific files into''')
    parser.add_argument('--reference_width', type=float, required=True,
                        help='''Reference width of which segmented data SQL files were generated''')
    parser.add_argument('--recitation_id', type=int, required=True,
                        help='''Recitation ID of the segmented data''')
    parser.add_argument('--update_previous_recitation', action='store_true',
                        help='''If set, will expect a previously generated tsv file in the input_path''')
    return parser.parse_args()


def create_db():
  print("Creating database...")
  _, db_file_name = mkstemp()
  print("Created database at " + db_file_name)
  conn = sqlite3.connect(db_file_name)
  cursor = conn.cursor()
  cursor.execute('''
    CREATE TABLE "glyphs" (
      'id' INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
      'pageId' INTEGER,
      'lineId' INTEGER,
      'suraId' INTEGER,
      'verseId' INTEGER,
      'indexId' INTEGER,
      'left' INTEGER,
      'right' INTEGER,
      'top' INTEGER,
      'bottom' INTEGER);
  ''')
  cursor.execute('''
    CREATE TABLE "encoded_glyphs" (
      'id' INTEGER PRIMARY KEY AUTOINCREMENT NOT NULL,
      'pageId' INTEGER,
      'lineId' INTEGER,
      'suraId' INTEGER,
      'verseId' INTEGER,
      'data' TEXT);
  ''')
  cursor.execute('''
    CREATE TABLE "results" (
      'pageId' INTEGER,
      'suraId' INTEGER,
      'verseId' INTEGER,
      'lineId' INTEGER,
      'data' TEXT);
  ''')
  return conn


def import_data(conn, sql_dir):
  print("Importing SQL files in " + sql_dir + "...")
  cursor = conn.cursor()
  for file_name in tqdm(glob(sql_dir + "/*.sql")):
    with open(file_name, "rb") as file:
      cursor.executescript(file.read())
  conn.commit()


def group_data(conn):
  print("Grouping data...")
  cursor = conn.cursor()
  cursor.execute('''
    INSERT INTO encoded_glyphs (
      pageId,
      lineId,
      suraId,
      verseId,
      data
    )
    SELECT
      pageId,
      min(lineId) lineId,
      suraId,
      verseId,
      '[' || group_concat('"' || left || ',' || top || ',' || right || ',' || bottom || '"', ',') || ']' data
    FROM glyphs
    GROUP BY
      suraid,
      verseid,
      pageid
    ORDER BY
      suraid,
      verseid,
      pageid;
  ''')
  conn.commit()


def encode_data(conn, width, ref_width):
  encoded_data = []
  ratio = width / ref_width
  cursor = conn.cursor()
  for row in cursor.execute('SELECT * FROM encoded_glyphs ORDER BY pageid, suraid, verseid'):
    data = row[5]
    data = data.replace('[', '').replace(']', '')
    rects = data.split('","')
    data = ""
    for rect in rects:
      coords = rect.replace('"', '')
      coords = coords.split(',')
      if len(data) > 0:
        data += ","
      left = int(round(int(coords[0]) * ratio))
      top = int(round(int(coords[1]) * ratio))
      right = int(round(int(coords[2]) * ratio))
      bottom = int(round(int(coords[3]) * ratio))
      data += str(left << 48 | top << 32 | right << 16 | bottom)
    data = base64.b64encode(data)
    encoded_data.append((row[1], row[2], row[3], row[4], data))
  return encoded_data


def prepare_results(conn, previous_tsv=None, page_ids=[]):
  cursor = conn.cursor()
  cursor.execute('DELETE FROM results')

  if previous_tsv:
    if os.path.isfile(previous_tsv):
      # open file in universal new line mode: U
      with open(previous_tsv, 'rU') as csvfile:
        reader = csv.reader(csvfile, delimiter='\t', quotechar='"')
        # file format: rewayaId	pageId	suraId	verseId	lineId	data
        next(reader) # skip header
        data = [(int(row[1]), int(row[4]), int(row[2]), int(row[3]), row[5]) for row in reader]
        insert_encoded_data(conn, data)
        # delete previous pages matching new pages
        if page_ids:
          cursor.execute('DELETE FROM results WHERE pageId IN (%s)' %
                         (','.join(['?'] * len(page_ids))), page_ids)
    else:
      raise RuntimeError("Could not find previously generated tsv file at %s" % previous_tsv)


def insert_encoded_data(conn, data):
  cursor = conn.cursor()

  cursor.executemany('''
    INSERT INTO results (
      pageId,
      lineId,
      suraId,
      verseId,
      data
    )
    VALUES (?, ?, ?, ?, ?)
  ''', data)
  conn.commit()


def export_data_sql(conn, recitation_id, file_name, page_ids=[]):
  cursor = conn.cursor()
  with open(file_name, 'wb') as file:
    total = 0
    if page_ids:
      file.write('DELETE FROM recitations_data WHERE recitation_id = %s AND page_id IN (%s)\n' %
                 (recitation_id, ', '.join(map(str, page_ids))))
    for row in cursor.execute('SELECT pageId, lineId, suraId, verseId, data FROM results ORDER BY pageId, lineId, suraId, verseId'):
      file.write(
        "INSERT INTO recitations_data (recitation_id, page_id, line_id, sura_id, verse_id, data) VALUES (%s, %s, %s, %s, %s, '%s')\n" \
          % (recitation_id, row[0], row[1], row[2], row[3], row[4]))
      total += 1
    print("Exported %d records to %s" % (total, file_name))


def export_data_tsv(conn, recitation_id, file_name):
  cursor = conn.cursor()
  with open(file_name, 'wb') as file:
    writer = csv.writer(file, delimiter='\t', lineterminator='\r',
                            quotechar='"', quoting=csv.QUOTE_MINIMAL)
    writer.writerow([
      'rewayaId',
      'pageId',
      'suraId',
      'verseId',
      'lineId',
      'data'
    ])
    total = 0
    for row in cursor.execute('SELECT pageId, lineId, suraId, verseId, data FROM results ORDER BY pageId, lineId, suraId, verseId'):
      writer.writerow([
        recitation_id,
        row[0],
        row[2],
        row[3],
        row[1],
        row[4]
      ])
      total += 1
    print("Exported %d records to %s" % (total, file_name))


def get_unique_pages(conn):
  cursor = conn.cursor()
  return [int(row[0]) for row in cursor.execute('SELECT distinct pageId from glyphs ORDER BY pageId')]


def get_tsv_path(base_dir, recitation_id):
  return "%s/RecitationData%d.tsv" % (base_dir, recitation_id)

if __name__ == "__main__":
  args = parse_arguments()

  conn = create_db()

  import_data(conn, args.input_path)
  group_data(conn)
  page_ids = get_unique_pages(conn)
  out_dir = safe_makedir(args.output_path)

  # tsv
  previous_tsv = None
  if args.update_previous_recitation:
    previous_tsv = get_tsv_path(args.input_path, args.recitation_id)
  prepare_results(conn, previous_tsv, page_ids)
  insert_encoded_data(conn, encode_data(conn, 800, args.reference_width))
  tsv_filename = get_tsv_path(out_dir, args.recitation_id)
  export_data_tsv(conn, args.recitation_id, tsv_filename)

  if previous_tsv and filecmp.cmp(previous_tsv, tsv_filename):
    print("No change was detected since the previous recitation, exiting")
    os.remove(tsv_filename)
  else:
    # sql
    if not args.update_previous_recitation:
      page_ids = []
    for width in [320, 480, 800, 1200, 1500]:
      prepare_results(conn, None, page_ids)
      insert_encoded_data(conn, encode_data(conn, width, args.reference_width))
      export_data_sql(conn, args.recitation_id, "%s/%d.sql" %
                      (out_dir, width), page_ids)

  conn.close()