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 |
|
|
import streams
|
18 |
1963
|
aaronmk
|
import strings
|
19 |
1965
|
aaronmk
|
import util
|
20 |
1942
|
aaronmk
|
|
21 |
|
|
def main():
|
22 |
|
|
# Usage
|
23 |
|
|
env_names = []
|
24 |
|
|
def usage_err():
|
25 |
|
|
raise SystemExit('Usage: '+opts.env_usage(env_names)+' '+sys.argv[0]
|
26 |
|
|
+' input_cmd [args...]')
|
27 |
|
|
|
28 |
|
|
# Parse args
|
29 |
|
|
input_cmd = sys.argv[1:]
|
30 |
|
|
if input_cmd == []: usage_err()
|
31 |
|
|
|
32 |
|
|
# Get config from env vars
|
33 |
|
|
table = opts.get_env_var('table', None, env_names)
|
34 |
|
|
schema = opts.get_env_var('schema', 'public', env_names)
|
35 |
|
|
db_config = opts.get_env_vars(sql.db_config_names, None, env_names)
|
36 |
|
|
debug = opts.env_flag('debug', False, env_names)
|
37 |
|
|
if not (table != None and 'engine' in db_config): usage_err()
|
38 |
|
|
|
39 |
|
|
# Connect to DB
|
40 |
|
|
db = sql.connect(db_config)
|
41 |
|
|
|
42 |
1963
|
aaronmk
|
use_copy_from = [True]
|
43 |
|
|
|
44 |
|
|
# Loads data into the table using the currently-selected approach.
|
45 |
|
|
def load_():
|
46 |
1942
|
aaronmk
|
# Open input stream
|
47 |
|
|
proc = subprocess.Popen(input_cmd, stdout=subprocess.PIPE, bufsize=-1)
|
48 |
|
|
in_ = proc.stdout
|
49 |
|
|
|
50 |
|
|
# Get format info
|
51 |
|
|
info = csvs.stream_info(in_, parse_header=True)
|
52 |
|
|
dialect = info.dialect
|
53 |
1963
|
aaronmk
|
if csvs.is_tsv(dialect): use_copy_from[0] = False
|
54 |
1965
|
aaronmk
|
cols = info.header
|
55 |
|
|
for i, col in enumerate(cols): # replace empty column names
|
56 |
|
|
if col == '': cols[i] = 'column_'+str(i)
|
57 |
1942
|
aaronmk
|
|
58 |
1963
|
aaronmk
|
# Select schema and escape names
|
59 |
2062
|
aaronmk
|
def esc_name(name): return sql.esc_name(db, name)
|
60 |
1963
|
aaronmk
|
sql.run_query(db, 'SET search_path TO '+esc_name(schema))
|
61 |
|
|
esc_table = esc_name(table)
|
62 |
1965
|
aaronmk
|
esc_cols = map(esc_name, cols)
|
63 |
1942
|
aaronmk
|
|
64 |
|
|
# Create CREATE TABLE statement
|
65 |
|
|
pkey = esc_name(table+'_pkey')
|
66 |
1963
|
aaronmk
|
create_table = 'CREATE TABLE '+esc_table+' (\n'
|
67 |
1965
|
aaronmk
|
create_table += ' row_num serial NOT NULL PRIMARY KEY\n'
|
68 |
|
|
for esc_col in esc_cols: create_table += ' , '+esc_col+' text\n'
|
69 |
1942
|
aaronmk
|
create_table += ');\n'
|
70 |
|
|
if debug: sys.stderr.write(create_table)
|
71 |
|
|
|
72 |
1963
|
aaronmk
|
# Create table
|
73 |
|
|
sql.run_query(db, create_table)
|
74 |
|
|
|
75 |
1942
|
aaronmk
|
# Create COPY FROM statement
|
76 |
1963
|
aaronmk
|
if use_copy_from[0]:
|
77 |
|
|
cur = db.db.cursor()
|
78 |
|
|
copy_from = ('COPY '+esc_table+' ('+(', '.join(esc_cols))
|
79 |
|
|
+') FROM STDIN DELIMITER %(delimiter)s NULL %(null)s')
|
80 |
|
|
assert not csvs.is_tsv(dialect)
|
81 |
1942
|
aaronmk
|
copy_from += ' CSV'
|
82 |
|
|
if dialect.quoting != csv.QUOTE_NONE:
|
83 |
|
|
copy_from += ' QUOTE %(quotechar)s'
|
84 |
|
|
if dialect.doublequote: copy_from += ' ESCAPE %(quotechar)s'
|
85 |
1963
|
aaronmk
|
copy_from += ';\n'
|
86 |
|
|
copy_from = cur.mogrify(copy_from, dict(delimiter=dialect.delimiter,
|
87 |
2116
|
aaronmk
|
null='', quotechar=dialect.quotechar))
|
88 |
1963
|
aaronmk
|
if debug: sys.stderr.write(copy_from)
|
89 |
1942
|
aaronmk
|
|
90 |
1963
|
aaronmk
|
# Load the data
|
91 |
1942
|
aaronmk
|
line_in = streams.ProgressInputStream(in_, sys.stderr,
|
92 |
|
|
'Processed %d row(s)', n=10000)
|
93 |
1963
|
aaronmk
|
try:
|
94 |
|
|
if use_copy_from[0]:
|
95 |
|
|
sys.stderr.write('Using COPY FROM\n')
|
96 |
|
|
db.db.cursor().copy_expert(copy_from, line_in)
|
97 |
|
|
else:
|
98 |
|
|
sys.stderr.write('Using INSERT\n')
|
99 |
1965
|
aaronmk
|
cols_ct = len(cols)+1 # +1 for row_num
|
100 |
1963
|
aaronmk
|
for row in csvs.make_reader(line_in, dialect):
|
101 |
|
|
row = map(strings.to_unicode, row)
|
102 |
|
|
row.insert(0, sql.default) # leave space for autogen row_num
|
103 |
1965
|
aaronmk
|
util.list_set_length(row, cols_ct) # truncate extra cols
|
104 |
1963
|
aaronmk
|
sql.insert(db, esc_table, row, table_is_esc=True)
|
105 |
1942
|
aaronmk
|
finally:
|
106 |
|
|
line_in.close() # also closes proc.stdout
|
107 |
|
|
proc.wait()
|
108 |
2116
|
aaronmk
|
|
109 |
|
|
# Clean up the data
|
110 |
|
|
sys.stderr.write('Cleaning up table\n')
|
111 |
|
|
sql.cleanup_table(db, esc_table, cols, table_is_esc=True)
|
112 |
1963
|
aaronmk
|
load = lambda: sql.with_savepoint(db, load_)
|
113 |
1942
|
aaronmk
|
|
114 |
1963
|
aaronmk
|
try: load()
|
115 |
|
|
except sql.DatabaseErrors, e:
|
116 |
|
|
if use_copy_from[0]: # first try
|
117 |
|
|
exc.print_ex(e, plain=True)
|
118 |
|
|
use_copy_from[0] = False
|
119 |
|
|
load() # try again with different approach
|
120 |
|
|
else: raise e
|
121 |
|
|
db.db.commit()
|
122 |
1942
|
aaronmk
|
|
123 |
|
|
main()
|