csvsql.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

import os
import io
import re
import csv
import sys
import codecs
import argparse
import binascii
import collections

ESCAPE_SEQUENCE_RE = re.compile(r'''
    ( \\U........      # 8-digit hex escapes
    | \\u....          # 4-digit hex escapes
    | \\x..            # 2-digit hex escapes
    | \\[0-7]{1,3}     # Octal escapes
    | \\N\{[^}]+\}     # Unicode characters by name
    | \\[\\'"abfnrtv]  # Single-character escapes
    )''', re.UNICODE | re.VERBOSE)

re_float = re.compile(r'^(-?[0-9]*\.?[0-9]+(e[-+]?[0-9]+)?|nan|[+-]?inf)$', re.I)
# re_int = re.compile(r'^(0|-?[1-9][0-9]*)$')
re_int = re.compile(r'^-?[0-9]+$')
re_null = re.compile(r'^(|null|\\N)$', re.I)
re_null2 = re.compile(r'^( |-|\.|none|na|n/a)$', re.I)
re_null3 = re.compile(r'^(|null|\\N| |-|\.|none|na|n/a)$', re.I)
re_nonp = re.compile(r'[\000-\010\013\014\016-\037]')

geom_types = sorted((
'POINT', 'POINT Z', 'POINT M', 'POINT ZM',
'LINESTRING', 'LINESTRING Z', 'LINESTRING M', 'LINESTRING ZM',
'POLYGON', 'POLYGON Z', 'POLYGON M', 'POLYGON ZM',
'MULTIPOINT', 'MULTIPOINT Z', 'MULTIPOINT M', 'MULTIPOINT ZM',
'MULTILINESTRING', 'MULTILINESTRING Z', 'MULTILINESTRING M', 'MULTILINESTRING ZM',
'MULTIPOLYGON', 'MULTIPOLYGON Z', 'MULTIPOLYGON M', 'MULTIPOLYGON ZM',
'GEOMETRYCOLLECTION', 'GEOMETRYCOLLECTION Z', 'GEOMETRYCOLLECTION ZM', 'GEOMETRYCOLLECTION ZM',
'GEOMETRY', 'GEOMETRY Z', 'GEOMETRY M', 'GEOMETRY ZM'), key=len, reverse=1)

# SMALLINT < INTEGER < BIGINT < DECIMAL < DOUBLE < GEOM < TEXT < BLOB
# We don't detect BOOLEAN

TYPES = {v:k for k, v in enumerate((
    'SMALLINT', 'INTEGER', 'BIGINT', 'DECIMAL', 'DOUBLE', 'GEOM', 'TEXT', 'BLOB'))}

BIGINT_MIN = -9223372036854775808
BIGINT_MAX = 9223372036854775807
INT_MIN = -2147483648
INT_MAX = 2147483647
# be conservative
SMALLINT_MIN = -128
SMALLINT_MAX = 127

class CustumDialect(csv.Dialect):
    pass

def decode_escapes(s):
    def decode_match(match):
        return codecs.decode(match.group(0), 'unicode-escape')

    return ESCAPE_SEQUENCE_RE.sub(decode_match, s)

def check_wkt(s):
    gtype = dimension = None
    s = s.upper()
    for cmd in geom_types:
        if s.startswith(cmd):
            gtype_a = cmd.split(' ')
            gtype = gtype_a[0]
            if len(gtype_a) == 1:
                dimension = 2
            elif gtype_a[1] == 'Z':
                dimension = 3
            elif gtype_a[1] == 'ZM':
                dimension = 4
            else:
                dimension = 'XY' + gtype_a[1]
            break
    return gtype, dimension

def main():
    parser = argparse.ArgumentParser(description="Generate SQL definition from CSV files. Automagically determines field type.", epilog="--delimiter accepts c-style escaped string, eg. '\\t' for tabs.\nThe generated SQL is supposed to be manually checked first.")
    parser.add_argument("-d", "--delimeter", help="A one-character string used to separate fields, defaults to ','.", default=",", metavar="CHAR")
    parser.add_argument("-e", "--escapechar", help="The escapechar removes any special meaning from the following character.", metavar="CHAR")
    parser.add_argument("-q", "--quotechar", help="A one-character string used to quote fields containing special characters.", default='"', metavar="CHAR")
    parser.add_argument("-s", "--skipinitialspace", help="When used, whitespace immediately following the delimiter is ignored.", action='store_true')
    parser.add_argument("-t", "--strict", help="When used, exit on bad CSV input.", action='store_true')
    parser.add_argument("-D", "--no-doublequote", help="The escapechar is used as a prefix to the quotechar.", action="store_true")
    parser.add_argument("-c", "--encoding", help="CSV encoding to validate.", default='utf-8')
    parser.add_argument("-n", "--tablename", help="Table name to import into. (default: filename)")
    parser.add_argument("-r", "--rowno", help="Add an integer primary key row number column", metavar="COLNAME", nargs="?", const='id')
    parser.add_argument("-i", "--insert", help="Output inserts.", action='store_true')
    parser.add_argument("-I", "--insert-only", help="Output inserts without CREATE TABLE statements.", action='store_true')
    parser.add_argument("--begin", help="Command to start a transaction, defaults to BEGIN", default='BEGIN')
    parser.add_argument("--wkt", help="Identify Geometry column (WKT) for SpatiaLite/PostGIS.", action='store_true')
    parser.add_argument("--spindex", help="Create spatial index on Geometry columns. (SpatiaLite only)", action='store_true')
    parser.add_argument("-H", "--no-header", help="First row is not header.", action="store_true")
    parser.add_argument("file", help="Input file")
    args = parser.parse_args()
    dialect = CustumDialect
    dialect.delimiter = decode_escapes(args.delimeter)
    dialect.doublequote = not args.no_doublequote
    dialect.escapechar = args.escapechar
    dialect.lineterminator = '\r\n'
    dialect.quotechar = args.quotechar
    dialect.quoting = csv.QUOTE_MINIMAL
    dialect.skipinitialspace = args.skipinitialspace
    dialect.strict = args.strict
    with open(args.file, 'rb') as fb:
        f = io.TextIOWrapper(fb, encoding='latin1', newline='')
        reader = csv.reader(f, dialect)
        columns = row = next(reader)
        if args.no_header:
            columns = ('c%d' % i for i in range(len(row)))
            f.seek(0)
        header = collections.OrderedDict.fromkeys(columns)
        ids = collections.OrderedDict((x, set()) for x in header if 'id' in x.lower())
        geom_cols = collections.OrderedDict()
        for rowcount, row in enumerate(reader, 1):
            for key, col in zip(header, row):
                coltype = header[key]
                if re_null.match(col):
                    if key in ids:
                        del ids[key]
                elif (coltype and TYPES[coltype] < TYPES['TEXT']
                      and re_null2.match(col)):
                    if key in ids:
                        del ids[key]
                elif re_int.match(col):
                    if coltype and TYPES[coltype] > TYPES['DECIMAL']:
                        continue
                    intval = int(col)
                    if intval != 0 and col.startswith('0'):
                        dtype = 'TEXT'
                    elif intval > BIGINT_MAX or intval < BIGINT_MIN:
                        dtype = 'DECIMAL'
                    elif intval > INT_MAX or intval < INT_MIN:
                        dtype = 'BIGINT'
                    # be conservative, don't use <=
                    elif SMALLINT_MIN < intval < SMALLINT_MAX:
                        dtype = 'SMALLINT'
                    else:
                        dtype = 'INTEGER'
                    if not coltype or TYPES[coltype] < TYPES[dtype]:
                        header[key] = dtype
                    if key in ids:
                        if col in ids[key]:
                            del ids[key]
                        else:
                            ids[key].add(col)
                elif re_float.match(col):
                    if coltype and TYPES[coltype] > TYPES['DOUBLE']:
                        continue
                    elif key in ids:
                        del ids[key]
                    header[key] = 'DOUBLE'
                elif coltype != 'BLOB':
                    if key in ids:
                        del ids[key]
                    if re_nonp.search(col):
                        header[key] = 'BLOB'
                        continue
                    else:
                        try:
                            col.encode('latin1').decode(args.encoding)
                        except UnicodeDecodeError:
                            header[key] = 'BLOB'
                            continue
                    if coltype == 'TEXT':
                        continue
                    if args.wkt:
                        gtype_dimension = check_wkt(col)
                    else:
                        header[key] = 'TEXT'
                        continue
                    if gtype_dimension[0]:
                        if coltype == 'GEOM':
                            if geom_cols[key] != gtype_dimension:
                                header[key] = 'TEXT'
                                del geom_cols[key]
                        else:
                            header[key] = 'GEOM'
                            geom_cols[key] = gtype_dimension
                    else:
                        header[key] = 'TEXT'
        tablename = args.tablename or os.path.basename(os.path.splitext(fb.name)[0])
        if not tablename.isidentifier():
            tablename = '"%s"' % tablename
        pkey = tuple(ids.keys())[0] if ids else None
        result = []
        result.append('CREATE TABLE %s (' % tablename)
        if args.rowno:
            result.append('%s %s PRIMARY KEY,' % (
                args.rowno if args.rowno.isidentifier() else '"%s"' % args.rowno,
                'BIGINT' if rowcount > INT_MAX else 'INTEGER'
            ))
        for k, v in header.items():
            if v == 'GEOM':
                continue
            elif v is None:
                v = header[k] = 'TEXT'
            result.append('%s %s%s,' % (
                k if k.isidentifier() else '"%s"' % k,
                v if v != 'DOUBLE' else 'DOUBLE PRECISION',
                ' PRIMARY KEY' if k == pkey and not args.rowno else ''
            ))
        result[-1] = result[-1][:-1]
        result.append(');')
        for k, v in geom_cols.items():
            gtype, dimension = v
            result.append(
                "SELECT AddGeometryColumn('%s', '%s', 4326, '%s', %r);" %
                (tablename.strip('"'), k, gtype, dimension)
            )
            if args.spindex:
                result.append("SELECT CreateSpatialIndex('%s', '%s');" %
                              (tablename.strip('"'), k))
        if not args.insert_only:
            print('\n'.join(result))
    if not (args.insert or args.insert_only):
        return
    header_idx = tuple(header.values())
    header_name_idx = {v:k for k,v in enumerate(header.keys())}
    print(args.begin + ';')
    with open(args.file, 'rb') as fb:
        f = io.TextIOWrapper(fb, encoding='latin1', newline='')
        reader = csv.reader(f, dialect)
        if not args.no_header:
            next(reader)
        for rowcount, row in enumerate(reader, 1):
            if args.rowno:
                vals = [str(rowcount)]
            else:
                vals = []
            for k, v in enumerate(row):
                if k >= len(header_idx):
                    continue
                coltype = header_idx[k]
                if coltype == 'GEOM':
                    pass
                elif TYPES[coltype] < TYPES['GEOM']:
                    if re_null3.match(v):
                        vals.append('null')
                    else:
                        vals.append(v)
                elif coltype == 'TEXT':
                    text = v.encode('latin1').decode(args.encoding)
                    vals.append("'%s'" % text.replace("'", "''"))
                    # .replace('\r', "'||char(13)||'").replace('\n', "'||char(10)||'"))
                else:
                    vals.append("X'%s'" % binascii.b2a_hex(v.encode('latin1')).decode('ascii'))
            for k in geom_cols:
                v = row[header_name_idx[k]]
                if re_null3.match(v):
                    vals.append('null')
                else:
                    vals.append("ST_GeomFromText('%s',4326)" % v.replace("'", "''"))
            print('INSERT INTO %s VALUES (%s);' % (
                tablename, ','.join(vals)))
    print('COMMIT;')

if __name__ == '__main__':
    main()