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 4257 aaronmk
    has_row_num = opts.env_flag('has_row_num', True, env_names)
37 1942 aaronmk
    db_config = opts.get_env_vars(sql.db_config_names, None, env_names)
38 2890 aaronmk
    errors_table_only = opts.env_flag('errors_table_only', False, env_names)
39 3271 aaronmk
    verbosity = util.cast(float, opts.get_env_var('verbosity', 3, env_names))
40 1942 aaronmk
41 4258 aaronmk
    if not ((errors_table_only or input_cmd != []) and table != None
42
        and 'engine' in db_config):
43 2890 aaronmk
        usage_err()
44
45 1942 aaronmk
    # Connect to DB
46 2680 aaronmk
    def log(msg, level=1):
47
        '''Higher level -> more verbose'''
48 3610 aaronmk
        if level <= verbosity:
49
            sys.stderr.write(strings.to_raw_str(msg.rstrip('\n')+'\n'))
50 2680 aaronmk
    db = sql.connect(db_config, log_debug=log)
51 1942 aaronmk
52 3138 aaronmk
    table = sql_gen.Table(table, schema)
53
54 1963 aaronmk
    use_copy_from = [True]
55
56
    # Loads data into the table using the currently-selected approach.
57 2680 aaronmk
    def load():
58 1942 aaronmk
        # Open input stream
59
        proc = subprocess.Popen(input_cmd, stdout=subprocess.PIPE, bufsize=-1)
60
        in_ = proc.stdout
61
62
        # Get format info
63
        info = csvs.stream_info(in_, parse_header=True)
64
        dialect = info.dialect
65 1963 aaronmk
        if csvs.is_tsv(dialect): use_copy_from[0] = False
66 3609 aaronmk
        col_names = map(strings.to_unicode, info.header)
67 2680 aaronmk
        for i, col in enumerate(col_names): # replace empty column names
68
            if col == '': col_names[i] = 'column_'+str(i)
69 1942 aaronmk
70 1963 aaronmk
        # Select schema and escape names
71 3608 aaronmk
        def esc_name(name): return db.esc_name(name)
72 1942 aaronmk
73 4257 aaronmk
        typed_cols = [sql_gen.TypedCol(v, 'text') for v in col_names]
74
        if has_row_num:
75
            typed_cols.insert(0, sql_gen.TypedCol('row_num', 'serial',
76
                nullable=False))
77 2759 aaronmk
78 3149 aaronmk
        log('Creating table')
79 4257 aaronmk
        sql.create_table(db, table, typed_cols, has_pkey=has_row_num,
80
            col_indexes=False)
81 3149 aaronmk
82
        # Remove rows from any failed COPY FROM
83
        sql.truncate(db, table)
84
85 2680 aaronmk
        def load_():
86
            # Create COPY FROM statement
87
            if use_copy_from[0]:
88 3139 aaronmk
                copy_from = ('COPY '+table.to_str(db)+' ('
89 2680 aaronmk
                    +(', '.join(map(esc_name, col_names)))
90 3578 aaronmk
                    +') FROM STDIN DELIMITER '+db.esc_value(dialect.delimiter)
91
                    +' NULL '+db.esc_value(''))
92 2680 aaronmk
                assert not csvs.is_tsv(dialect)
93
                copy_from += ' CSV'
94
                if dialect.quoting != csv.QUOTE_NONE:
95 3578 aaronmk
                    quote_str = db.esc_value(dialect.quotechar)
96
                    copy_from += ' QUOTE '+quote_str
97
                    if dialect.doublequote: copy_from += ' ESCAPE '+quote_str
98 2680 aaronmk
                copy_from += ';\n'
99
100
            # Load the data
101 2892 aaronmk
            line_in = streams.ProgressInputStream(in_, sys.stderr, n=1000)
102 2680 aaronmk
            try:
103
                if use_copy_from[0]:
104
                    log('Using COPY FROM')
105
                    log(copy_from, level=2)
106
                    db.db.cursor().copy_expert(copy_from, line_in)
107
                else:
108
                    log('Using INSERT')
109
                    cols_ct = len(col_names)+1 # +1 for row_num
110
                    for row in csvs.make_reader(line_in, dialect):
111
                        row = map(strings.to_unicode, row)
112
                        row.insert(0, sql.default) # row_num is autogen
113
                        util.list_set_length(row, cols_ct) # truncate extra cols
114 3270 aaronmk
                        sql.insert(db, table, row, cacheable=False, log_level=5)
115 2680 aaronmk
            finally:
116
                line_in.close() # also closes proc.stdout
117
                proc.wait()
118
        sql.with_savepoint(db, load_)
119 1963 aaronmk
120 2680 aaronmk
        log('Cleaning up table')
121 3081 aaronmk
        sql_io.cleanup_table(db, table, col_names)
122 1942 aaronmk
123 3147 aaronmk
        log('Vacuuming and reanalyzing table')
124
        sql.vacuum(db, table)
125 1942 aaronmk
126 3134 aaronmk
    if not errors_table_only:
127 2890 aaronmk
        try: load()
128
        except sql.DatabaseErrors, e:
129
            if use_copy_from[0]: # first try
130
                exc.print_ex(e, plain=True)
131
                use_copy_from[0] = False
132
                load() # try again with different approach
133
            else: raise
134 1942 aaronmk
135
main()