Project

General

Profile

1 1942 aaronmk
#!/usr/bin/env python
2
# Loads a command's CSV output stream into a PostgreSQL table.
3
# The command may be run more than once.
4
5
import csv
6
import os.path
7 1963 aaronmk
import re
8 1942 aaronmk
import subprocess
9
import sys
10
11
sys.path.append(os.path.dirname(__file__)+"/../lib")
12
13
import csvs
14
import exc
15
import opts
16
import sql
17 3080 aaronmk
import sql_io
18 2680 aaronmk
import sql_gen
19 1942 aaronmk
import streams
20 1963 aaronmk
import strings
21 1965 aaronmk
import util
22 1942 aaronmk
23
def main():
24
    # Usage
25
    env_names = []
26
    def usage_err():
27
        raise SystemExit('Usage: '+opts.env_usage(env_names)+' '+sys.argv[0]
28
            +' input_cmd [args...]')
29
30
    # Parse args
31
    input_cmd = sys.argv[1:]
32
33
    # Get config from env vars
34
    table = opts.get_env_var('table', None, env_names)
35
    schema = opts.get_env_var('schema', 'public', env_names)
36
    db_config = opts.get_env_vars(sql.db_config_names, None, env_names)
37 2890 aaronmk
    errors_table_only = opts.env_flag('errors_table_only', False, env_names)
38 3271 aaronmk
    verbosity = util.cast(float, opts.get_env_var('verbosity', 3, env_names))
39 1942 aaronmk
40 2890 aaronmk
    if not (input_cmd != [] and table != None and 'engine' in db_config):
41
        usage_err()
42
43 1942 aaronmk
    # Connect to DB
44 2680 aaronmk
    def log(msg, level=1):
45
        '''Higher level -> more verbose'''
46
        if level <= verbosity: sys.stderr.write(msg.rstrip('\n')+'\n')
47
    db = sql.connect(db_config, log_debug=log)
48 1942 aaronmk
49 3138 aaronmk
    table = sql_gen.Table(table, schema)
50
51 3134 aaronmk
    log('Creating errors table')
52
    errors_table = sql_io.errors_table(db, table, if_exists=False)
53 3440 aaronmk
    if errors_table_only: sql.drop_table(db, errors_table)
54 3134 aaronmk
    typed_cols = [
55
        sql_gen.TypedCol('column', 'text', nullable=False),
56
        sql_gen.TypedCol('value', 'text'),
57
        sql_gen.TypedCol('error_code', 'character varying(5)',
58
            nullable=False),
59
        sql_gen.TypedCol('error', 'text', nullable=False),
60
        ]
61
    sql.create_table(db, errors_table, typed_cols, has_pkey=False)
62
    index_cols = ['column', 'value', 'error_code', 'error']
63
    sql.add_index(db, index_cols, errors_table, unique=True)
64 2890 aaronmk
65 1963 aaronmk
    use_copy_from = [True]
66
67
    # Loads data into the table using the currently-selected approach.
68 2680 aaronmk
    def load():
69 1942 aaronmk
        # Open input stream
70
        proc = subprocess.Popen(input_cmd, stdout=subprocess.PIPE, bufsize=-1)
71
        in_ = proc.stdout
72
73
        # Get format info
74
        info = csvs.stream_info(in_, parse_header=True)
75
        dialect = info.dialect
76 1963 aaronmk
        if csvs.is_tsv(dialect): use_copy_from[0] = False
77 2680 aaronmk
        col_names = info.header
78
        for i, col in enumerate(col_names): # replace empty column names
79
            if col == '': col_names[i] = 'column_'+str(i)
80 1942 aaronmk
81 1963 aaronmk
        # Select schema and escape names
82 2062 aaronmk
        def esc_name(name): return sql.esc_name(db, name)
83 1942 aaronmk
84 2876 aaronmk
        typed_cols = [sql_gen.TypedCol('row_num', 'serial', nullable=False)]+[
85 2759 aaronmk
            sql_gen.TypedCol(v, 'text') for v in col_names]
86
87 3149 aaronmk
        log('Creating table')
88
        sql.create_table(db, table, typed_cols, col_indexes=False)
89
90
        # Remove rows from any failed COPY FROM
91
        sql.truncate(db, table)
92
93 2680 aaronmk
        def load_():
94
            # Create COPY FROM statement
95
            if use_copy_from[0]:
96 3139 aaronmk
                copy_from = ('COPY '+table.to_str(db)+' ('
97 2680 aaronmk
                    +(', '.join(map(esc_name, col_names)))
98 3578 aaronmk
                    +') FROM STDIN DELIMITER '+db.esc_value(dialect.delimiter)
99
                    +' NULL '+db.esc_value(''))
100 2680 aaronmk
                assert not csvs.is_tsv(dialect)
101
                copy_from += ' CSV'
102
                if dialect.quoting != csv.QUOTE_NONE:
103 3578 aaronmk
                    quote_str = db.esc_value(dialect.quotechar)
104
                    copy_from += ' QUOTE '+quote_str
105
                    if dialect.doublequote: copy_from += ' ESCAPE '+quote_str
106 2680 aaronmk
                copy_from += ';\n'
107
108
            # Load the data
109 2892 aaronmk
            line_in = streams.ProgressInputStream(in_, sys.stderr, n=1000)
110 2680 aaronmk
            try:
111
                if use_copy_from[0]:
112
                    log('Using COPY FROM')
113
                    log(copy_from, level=2)
114
                    db.db.cursor().copy_expert(copy_from, line_in)
115
                else:
116
                    log('Using INSERT')
117
                    cols_ct = len(col_names)+1 # +1 for row_num
118
                    for row in csvs.make_reader(line_in, dialect):
119
                        row = map(strings.to_unicode, row)
120
                        row.insert(0, sql.default) # row_num is autogen
121
                        util.list_set_length(row, cols_ct) # truncate extra cols
122 3270 aaronmk
                        sql.insert(db, table, row, cacheable=False, log_level=5)
123 2680 aaronmk
            finally:
124
                line_in.close() # also closes proc.stdout
125
                proc.wait()
126
        sql.with_savepoint(db, load_)
127 1963 aaronmk
128 2680 aaronmk
        log('Cleaning up table')
129 3081 aaronmk
        sql_io.cleanup_table(db, table, col_names)
130 1942 aaronmk
131 3147 aaronmk
        log('Vacuuming and reanalyzing table')
132
        sql.vacuum(db, table)
133 1942 aaronmk
134 3134 aaronmk
    if not errors_table_only:
135 2890 aaronmk
        try: load()
136
        except sql.DatabaseErrors, e:
137
            if use_copy_from[0]: # first try
138
                exc.print_ex(e, plain=True)
139
                use_copy_from[0] = False
140
                load() # try again with different approach
141
            else: raise
142 1942 aaronmk
143
main()