1 |
3077
|
aaronmk
|
# Database import/export
|
2 |
|
|
|
3 |
3534
|
aaronmk
|
import copy
|
4 |
3431
|
aaronmk
|
import operator
|
5 |
3714
|
aaronmk
|
import warnings
|
6 |
3431
|
aaronmk
|
|
7 |
3077
|
aaronmk
|
import exc
|
8 |
|
|
import dicts
|
9 |
|
|
import sql
|
10 |
|
|
import sql_gen
|
11 |
|
|
import strings
|
12 |
|
|
import util
|
13 |
|
|
|
14 |
3645
|
aaronmk
|
##### Exceptions
|
15 |
|
|
|
16 |
|
|
# Can't use built-in SyntaxError because it stringifies to only the first line
|
17 |
|
|
class SyntaxError(Exception): pass
|
18 |
|
|
|
19 |
3081
|
aaronmk
|
##### Data cleanup
|
20 |
|
|
|
21 |
4388
|
aaronmk
|
null_strs = ['', r'\N', 'NULL', 'UNKNOWN']
|
22 |
4209
|
aaronmk
|
|
23 |
4447
|
aaronmk
|
def cleanup_table(db, table):
|
24 |
3081
|
aaronmk
|
table = sql_gen.as_Table(table)
|
25 |
4445
|
aaronmk
|
cols = [sql_gen.as_Col(c, table) for c in sql.table_cols(db, table)]
|
26 |
4407
|
aaronmk
|
cols = filter(lambda c: sql_gen.is_text_col(db, c), cols)
|
27 |
3081
|
aaronmk
|
|
28 |
4209
|
aaronmk
|
expr = 'trim(both from %s)'
|
29 |
|
|
for null in null_strs: expr = 'nullif('+expr+', '+db.esc_value(null)+')'
|
30 |
|
|
changes = [(v, sql_gen.CustomCode(expr % v.to_str(db))) for v in cols]
|
31 |
3081
|
aaronmk
|
|
32 |
4444
|
aaronmk
|
while True:
|
33 |
|
|
try:
|
34 |
|
|
sql.update(db, table, changes, in_place=True, recover=True)
|
35 |
|
|
break # successful
|
36 |
|
|
except sql.NullValueException, e:
|
37 |
4457
|
aaronmk
|
db.log_debug('Caught exception: '+exc.str_(e))
|
38 |
4444
|
aaronmk
|
col, = e.cols
|
39 |
|
|
sql.drop_not_null(db, col)
|
40 |
3081
|
aaronmk
|
|
41 |
3078
|
aaronmk
|
##### Error tracking
|
42 |
|
|
|
43 |
|
|
def track_data_error(db, errors_table, cols, value, error_code, error):
|
44 |
|
|
'''
|
45 |
|
|
@param errors_table If None, does nothing.
|
46 |
|
|
'''
|
47 |
|
|
if errors_table == None or cols == (): return
|
48 |
|
|
|
49 |
|
|
for col in cols:
|
50 |
|
|
try:
|
51 |
|
|
sql.insert(db, errors_table, dict(column=col.name, value=value,
|
52 |
|
|
error_code=error_code, error=error), recover=True,
|
53 |
|
|
cacheable=True, log_level=4)
|
54 |
|
|
except sql.DuplicateKeyException: pass
|
55 |
|
|
|
56 |
3506
|
aaronmk
|
class ExcToErrorsTable(sql_gen.ExcToWarning):
|
57 |
|
|
'''Handles an exception by saving it or converting it to a warning.'''
|
58 |
3511
|
aaronmk
|
def __init__(self, return_, srcs, errors_table, value=None):
|
59 |
3506
|
aaronmk
|
'''
|
60 |
|
|
@param return_ See sql_gen.ExcToWarning
|
61 |
|
|
@param srcs The column names for the errors table
|
62 |
|
|
@param errors_table None|sql_gen.Table
|
63 |
3511
|
aaronmk
|
@param value The value (or an expression for it) that caused the error
|
64 |
3506
|
aaronmk
|
@pre The invalid value must be in a local variable "value" of type text.
|
65 |
|
|
'''
|
66 |
|
|
sql_gen.ExcToWarning.__init__(self, return_)
|
67 |
|
|
|
68 |
3511
|
aaronmk
|
value = sql_gen.as_Code(value)
|
69 |
|
|
|
70 |
3506
|
aaronmk
|
self.srcs = srcs
|
71 |
|
|
self.errors_table = errors_table
|
72 |
3511
|
aaronmk
|
self.value = value
|
73 |
3501
|
aaronmk
|
|
74 |
3506
|
aaronmk
|
def to_str(self, db):
|
75 |
|
|
if not self.srcs or self.errors_table == None:
|
76 |
|
|
return sql_gen.ExcToWarning.to_str(self, db)
|
77 |
|
|
|
78 |
3459
|
aaronmk
|
errors_table_cols = map(sql_gen.Col,
|
79 |
|
|
['column', 'value', 'error_code', 'error'])
|
80 |
3465
|
aaronmk
|
col_names_query = sql.mk_select(db, sql_gen.NamedValues('c', None,
|
81 |
3506
|
aaronmk
|
[[c.name] for c in self.srcs]), order_by=None)
|
82 |
|
|
insert_query = sql.mk_insert_select(db, self.errors_table,
|
83 |
|
|
errors_table_cols,
|
84 |
3465
|
aaronmk
|
sql_gen.Values(errors_table_cols).to_str(db))+';\n'
|
85 |
3506
|
aaronmk
|
return '''\
|
86 |
3459
|
aaronmk
|
-- Save error in errors table.
|
87 |
|
|
DECLARE
|
88 |
|
|
error_code text := SQLSTATE;
|
89 |
|
|
error text := SQLERRM;
|
90 |
3511
|
aaronmk
|
value text := '''+self.value.to_str(db)+''';
|
91 |
3529
|
aaronmk
|
"column" text;
|
92 |
3459
|
aaronmk
|
BEGIN
|
93 |
|
|
-- Insert the value and error for *each* source column.
|
94 |
3529
|
aaronmk
|
'''+strings.indent(sql_gen.RowExcIgnore(None, col_names_query, insert_query,
|
95 |
3467
|
aaronmk
|
row_var=errors_table_cols[0]).to_str(db))+'''
|
96 |
3459
|
aaronmk
|
END;
|
97 |
3501
|
aaronmk
|
|
98 |
3506
|
aaronmk
|
'''+self.return_.to_str(db)
|
99 |
3459
|
aaronmk
|
|
100 |
3507
|
aaronmk
|
def data_exception_handler(*args, **kw_args):
|
101 |
3506
|
aaronmk
|
'''Handles a data_exception by saving it or converting it to a warning.
|
102 |
|
|
For params, see ExcToErrorsTable().
|
103 |
|
|
'''
|
104 |
|
|
return sql_gen.data_exception_handler(ExcToErrorsTable(*args, **kw_args))
|
105 |
|
|
|
106 |
3078
|
aaronmk
|
def cast(db, type_, col, errors_table=None):
|
107 |
|
|
'''Casts an (unrenamed) column or value.
|
108 |
|
|
If errors_table set and col has srcs, saves errors in errors_table (using
|
109 |
3360
|
aaronmk
|
col's srcs attr as source columns). Otherwise, converts errors to warnings.
|
110 |
3078
|
aaronmk
|
@param col str|sql_gen.Col|sql_gen.Literal
|
111 |
|
|
@param errors_table None|sql_gen.Table|str
|
112 |
|
|
'''
|
113 |
|
|
col = sql_gen.as_Col(col)
|
114 |
|
|
|
115 |
3112
|
aaronmk
|
# Don't convert exceptions to warnings for user-supplied constants
|
116 |
|
|
if isinstance(col, sql_gen.Literal): return sql_gen.Cast(type_, col)
|
117 |
|
|
|
118 |
3078
|
aaronmk
|
assert not isinstance(col, sql_gen.NamedCol)
|
119 |
|
|
|
120 |
3460
|
aaronmk
|
function_name = strings.first_word(type_)
|
121 |
3459
|
aaronmk
|
srcs = col.srcs
|
122 |
3508
|
aaronmk
|
save_errors = errors_table != None and srcs
|
123 |
|
|
if save_errors: # function will be unique for the given srcs
|
124 |
3750
|
aaronmk
|
function_name = strings.ustr(sql_gen.FunctionCall(function_name,
|
125 |
3508
|
aaronmk
|
*map(sql_gen.to_name_only_col, srcs)))
|
126 |
3078
|
aaronmk
|
function = db.TempFunction(function_name)
|
127 |
|
|
|
128 |
3464
|
aaronmk
|
# Create function definition
|
129 |
|
|
modifiers = 'STRICT'
|
130 |
|
|
if not save_errors: modifiers = 'IMMUTABLE '+modifiers
|
131 |
3511
|
aaronmk
|
value_param = sql_gen.FunctionParam('value', 'text')
|
132 |
|
|
handler = data_exception_handler('RETURN NULL;\n', srcs, errors_table,
|
133 |
|
|
value_param.name)
|
134 |
3464
|
aaronmk
|
body = sql_gen.CustomCode(handler.to_str(db, '''\
|
135 |
3467
|
aaronmk
|
/* The explicit cast to the return type is needed to make the cast happen
|
136 |
|
|
inside the try block. (Implicit casts to the return type happen at the end
|
137 |
|
|
of the function, outside any block.) */
|
138 |
|
|
RETURN value::'''+type_+''';
|
139 |
3464
|
aaronmk
|
'''))
|
140 |
|
|
body.lang='plpgsql'
|
141 |
3500
|
aaronmk
|
sql.define_func(db, sql_gen.FunctionDef(function, type_, body,
|
142 |
3511
|
aaronmk
|
[value_param], modifiers))
|
143 |
3464
|
aaronmk
|
|
144 |
3078
|
aaronmk
|
return sql_gen.FunctionCall(function, col)
|
145 |
|
|
|
146 |
3538
|
aaronmk
|
def func_wrapper_exception_handler(db, return_, args, errors_table):
|
147 |
3524
|
aaronmk
|
'''Handles a function call's data_exceptions.
|
148 |
|
|
Supports PL/Python functions.
|
149 |
|
|
@param return_ See data_exception_handler()
|
150 |
|
|
@param args [arg...] Function call's args
|
151 |
|
|
@param errors_table See data_exception_handler()
|
152 |
|
|
'''
|
153 |
|
|
args = filter(sql_gen.has_srcs, args)
|
154 |
|
|
|
155 |
|
|
srcs = sql_gen.cross_join_srcs(args)
|
156 |
3538
|
aaronmk
|
value = sql_gen.merge_not_null(db, ',', args)
|
157 |
3524
|
aaronmk
|
return sql_gen.NestedExcHandler(
|
158 |
|
|
data_exception_handler(return_, srcs, errors_table, value)
|
159 |
|
|
, sql_gen.plpythonu_error_handler
|
160 |
|
|
)
|
161 |
|
|
|
162 |
3078
|
aaronmk
|
def cast_temp_col(db, type_, col, errors_table=None):
|
163 |
|
|
'''Like cast(), but creates a new column with the cast values if the input
|
164 |
|
|
is a column.
|
165 |
|
|
@return The new column or cast value
|
166 |
|
|
'''
|
167 |
|
|
def cast_(col): return cast(db, type_, col, errors_table)
|
168 |
|
|
|
169 |
|
|
try: col = sql_gen.underlying_col(col)
|
170 |
|
|
except sql_gen.NoUnderlyingTableException: return sql_gen.wrap(cast_, col)
|
171 |
|
|
|
172 |
|
|
table = col.table
|
173 |
3173
|
aaronmk
|
new_col = sql_gen.suffixed_col(col, '::'+strings.first_word(type_))
|
174 |
3078
|
aaronmk
|
expr = cast_(col)
|
175 |
|
|
|
176 |
|
|
# Add column
|
177 |
|
|
new_typed_col = sql_gen.TypedCol(new_col.name, type_)
|
178 |
3750
|
aaronmk
|
sql.add_col(db, table, new_typed_col, comment=strings.urepr(col)+'::'+type_)
|
179 |
3078
|
aaronmk
|
new_col.name = new_typed_col.name # propagate any renaming
|
180 |
|
|
|
181 |
3110
|
aaronmk
|
sql.update(db, table, [(new_col, expr)], in_place=True, recover=True)
|
182 |
3078
|
aaronmk
|
|
183 |
|
|
return new_col
|
184 |
|
|
|
185 |
|
|
def errors_table(db, table, if_exists=True):
|
186 |
|
|
'''
|
187 |
|
|
@param if_exists If set, returns None if the errors table doesn't exist
|
188 |
|
|
@return None|sql_gen.Table
|
189 |
|
|
'''
|
190 |
|
|
table = sql_gen.as_Table(table)
|
191 |
|
|
if table.srcs != (): table = table.srcs[0]
|
192 |
|
|
|
193 |
|
|
errors_table = sql_gen.suffixed_table(table, '.errors')
|
194 |
|
|
if if_exists and not sql.table_exists(db, errors_table): return None
|
195 |
|
|
return errors_table
|
196 |
|
|
|
197 |
4436
|
aaronmk
|
def mk_errors_table(db, table):
|
198 |
|
|
errors_table_ = errors_table(db, table, if_exists=False)
|
199 |
|
|
if sql.table_exists(db, errors_table_): return
|
200 |
|
|
|
201 |
|
|
typed_cols = [
|
202 |
|
|
sql_gen.TypedCol('column', 'text', nullable=False),
|
203 |
|
|
sql_gen.TypedCol('value', 'text'),
|
204 |
|
|
sql_gen.TypedCol('error_code', 'character varying(5)', nullable=False),
|
205 |
|
|
sql_gen.TypedCol('error', 'text', nullable=False),
|
206 |
|
|
]
|
207 |
|
|
sql.create_table(db, errors_table_, typed_cols, has_pkey=False)
|
208 |
|
|
index_cols = ['column', 'value', 'error_code', 'error']
|
209 |
|
|
sql.add_index(db, index_cols, errors_table_, unique=True)
|
210 |
|
|
|
211 |
3078
|
aaronmk
|
##### Import
|
212 |
|
|
|
213 |
3077
|
aaronmk
|
def put(db, table, row, pkey_=None, row_ct_ref=None):
|
214 |
|
|
'''Recovers from errors.
|
215 |
|
|
Only works under PostgreSQL (uses INSERT RETURNING).
|
216 |
|
|
'''
|
217 |
3633
|
aaronmk
|
return put_table(db, table, [], row, row_ct_ref)
|
218 |
3077
|
aaronmk
|
|
219 |
|
|
def get(db, table, row, pkey, row_ct_ref=None, create=False):
|
220 |
|
|
'''Recovers from errors'''
|
221 |
|
|
try:
|
222 |
|
|
return sql.value(sql.select(db, table, [pkey], row, limit=1,
|
223 |
|
|
recover=True))
|
224 |
|
|
except StopIteration:
|
225 |
|
|
if not create: raise
|
226 |
|
|
return put(db, table, row, pkey, row_ct_ref) # insert new row
|
227 |
|
|
|
228 |
|
|
def is_func_result(col):
|
229 |
|
|
return col.table.name.find('(') >= 0 and col.name == 'result'
|
230 |
|
|
|
231 |
|
|
def into_table_name(out_table, in_tables0, mapping, is_func):
|
232 |
|
|
def in_col_str(in_col):
|
233 |
|
|
in_col = sql_gen.remove_col_rename(in_col)
|
234 |
|
|
if isinstance(in_col, sql_gen.Col):
|
235 |
|
|
table = in_col.table
|
236 |
|
|
if table == in_tables0:
|
237 |
|
|
in_col = sql_gen.to_name_only_col(in_col)
|
238 |
|
|
elif is_func_result(in_col): in_col = table # omit col name
|
239 |
3750
|
aaronmk
|
return strings.ustr(in_col)
|
240 |
3077
|
aaronmk
|
|
241 |
4491
|
aaronmk
|
str_ = strings.ustr(out_table)
|
242 |
3077
|
aaronmk
|
if is_func:
|
243 |
|
|
str_ += '('
|
244 |
|
|
|
245 |
|
|
try: value_in_col = mapping['value']
|
246 |
|
|
except KeyError:
|
247 |
4491
|
aaronmk
|
str_ += ', '.join((strings.ustr(k)+'='+in_col_str(v)
|
248 |
3077
|
aaronmk
|
for k, v in mapping.iteritems()))
|
249 |
|
|
else: str_ += in_col_str(value_in_col)
|
250 |
|
|
|
251 |
|
|
str_ += ')'
|
252 |
|
|
else:
|
253 |
|
|
out_col = 'rank'
|
254 |
|
|
try: in_col = mapping[out_col]
|
255 |
|
|
except KeyError: str_ += '_pkeys'
|
256 |
|
|
else: # has a rank column, so hierarchical
|
257 |
4491
|
aaronmk
|
str_ += '['+strings.ustr(out_col)+'='+in_col_str(in_col)+']'
|
258 |
3077
|
aaronmk
|
return str_
|
259 |
|
|
|
260 |
3628
|
aaronmk
|
def put_table(db, out_table, in_tables, mapping, row_ct_ref=None, default=None,
|
261 |
3660
|
aaronmk
|
col_defaults={}, on_error=exc.reraise):
|
262 |
3077
|
aaronmk
|
'''Recovers from errors.
|
263 |
|
|
Only works under PostgreSQL (uses INSERT RETURNING).
|
264 |
|
|
IMPORTANT: Must be run at the *beginning* of a transaction.
|
265 |
|
|
@param in_tables The main input table to select from, followed by a list of
|
266 |
|
|
tables to join with it using the main input table's pkey
|
267 |
|
|
@param mapping dict(out_table_col=in_table_col, ...)
|
268 |
|
|
* out_table_col: str (*not* sql_gen.Col)
|
269 |
|
|
* in_table_col: sql_gen.Col|literal-value
|
270 |
|
|
@param default The *output* column to use as the pkey for missing rows.
|
271 |
|
|
If this output column does not exist in the mapping, uses None.
|
272 |
3618
|
aaronmk
|
@param col_defaults Default values for required columns.
|
273 |
3077
|
aaronmk
|
@return sql_gen.Col Where the output pkeys are made available
|
274 |
|
|
'''
|
275 |
3474
|
aaronmk
|
import psycopg2.extensions
|
276 |
|
|
|
277 |
3077
|
aaronmk
|
out_table = sql_gen.as_Table(out_table)
|
278 |
|
|
|
279 |
|
|
def log_debug(msg): db.log_debug(msg, level=1.5)
|
280 |
|
|
def col_ustr(str_):
|
281 |
|
|
return strings.repr_no_u(sql_gen.remove_col_rename(str_))
|
282 |
|
|
|
283 |
|
|
log_debug('********** New iteration **********')
|
284 |
|
|
log_debug('Inserting these input columns into '+strings.as_tt(
|
285 |
|
|
out_table.to_str(db))+':\n'+strings.as_table(mapping, ustr=col_ustr))
|
286 |
|
|
|
287 |
|
|
is_function = sql.function_exists(db, out_table)
|
288 |
|
|
|
289 |
|
|
if is_function: out_pkey = 'result'
|
290 |
|
|
else: out_pkey = sql.pkey(db, out_table, recover=True)
|
291 |
|
|
out_pkey_col = sql_gen.as_Col(out_pkey, out_table)
|
292 |
|
|
|
293 |
|
|
in_tables_ = in_tables[:] # don't modify input!
|
294 |
3432
|
aaronmk
|
try: in_tables0 = in_tables_.pop(0) # first table is separate
|
295 |
|
|
except IndexError: in_tables0 = None
|
296 |
|
|
else:
|
297 |
|
|
in_pkey = sql.pkey(db, in_tables0, recover=True)
|
298 |
|
|
in_pkey_col = sql_gen.as_Col(in_pkey, in_tables0)
|
299 |
3431
|
aaronmk
|
|
300 |
|
|
# Determine if can use optimization for only literal values
|
301 |
|
|
is_literals = not reduce(operator.or_, map(sql_gen.is_table_col,
|
302 |
3434
|
aaronmk
|
mapping.values()), False)
|
303 |
3431
|
aaronmk
|
is_literals_or_function = is_literals or is_function
|
304 |
|
|
|
305 |
3432
|
aaronmk
|
if in_tables0 == None: errors_table_ = None
|
306 |
|
|
else: errors_table_ = errors_table(db, in_tables0)
|
307 |
3431
|
aaronmk
|
|
308 |
|
|
# Create input joins from list of input tables
|
309 |
3077
|
aaronmk
|
input_joins = [in_tables0]+[sql_gen.Join(v,
|
310 |
|
|
{in_pkey: sql_gen.join_same_not_null}) for v in in_tables_]
|
311 |
|
|
|
312 |
3433
|
aaronmk
|
if mapping == {} and not is_function: # need >= one column for INSERT SELECT
|
313 |
|
|
mapping = {out_pkey: None} # ColDict will replace with default value
|
314 |
|
|
|
315 |
3431
|
aaronmk
|
if not is_literals:
|
316 |
3628
|
aaronmk
|
into = sql_gen.as_Table(into_table_name(out_table, in_tables0, mapping,
|
317 |
|
|
is_function))
|
318 |
4484
|
aaronmk
|
# Ensure into's out_pkey is different from in_pkey by prepending table
|
319 |
|
|
into_out_pkey = strings.ustr(out_pkey_col)
|
320 |
3431
|
aaronmk
|
|
321 |
|
|
# Set column sources
|
322 |
|
|
in_cols = filter(sql_gen.is_table_col, mapping.values())
|
323 |
|
|
for col in in_cols:
|
324 |
|
|
if col.table == in_tables0: col.set_srcs(sql_gen.src_self)
|
325 |
|
|
|
326 |
|
|
log_debug('Joining together input tables into temp table')
|
327 |
|
|
# Place in new table so don't modify input and for speed
|
328 |
|
|
in_table = sql_gen.Table('in')
|
329 |
|
|
mapping = dicts.join(mapping, sql.flatten(db, in_table, input_joins,
|
330 |
|
|
in_cols, preserve=[in_pkey_col]))
|
331 |
|
|
input_joins = [in_table]
|
332 |
|
|
db.log_debug('Temp table: '+strings.as_tt(in_table.to_str(db)), level=2)
|
333 |
3077
|
aaronmk
|
|
334 |
3692
|
aaronmk
|
# Wrap mapping in a sql_gen.ColDict.
|
335 |
|
|
# sql_gen.ColDict sanitizes both keys and values passed into it.
|
336 |
|
|
# Do after applying dicts.join() because that returns a plain dict.
|
337 |
3077
|
aaronmk
|
mapping = sql_gen.ColDict(db, out_table, mapping)
|
338 |
|
|
|
339 |
|
|
# Resolve default value column
|
340 |
|
|
if default != None:
|
341 |
|
|
try: default = mapping[default]
|
342 |
|
|
except KeyError:
|
343 |
|
|
db.log_debug('Default value column '
|
344 |
|
|
+strings.as_tt(strings.repr_no_u(default))
|
345 |
|
|
+' does not exist in mapping, falling back to None', level=2.1)
|
346 |
|
|
default = None
|
347 |
|
|
|
348 |
3287
|
aaronmk
|
# Save default values for all rows since in_table may have rows deleted
|
349 |
3431
|
aaronmk
|
if is_literals: pass
|
350 |
|
|
elif is_function: full_in_table = in_table
|
351 |
3386
|
aaronmk
|
else:
|
352 |
|
|
full_in_table = sql_gen.suffixed_table(in_table, '_full')
|
353 |
|
|
full_in_table_cols = [in_pkey_col]
|
354 |
|
|
if default != None:
|
355 |
|
|
full_in_table_cols.append(default)
|
356 |
|
|
default = sql_gen.with_table(default, full_in_table)
|
357 |
|
|
sql.run_query_into(db, sql.mk_select(db, in_table, full_in_table_cols,
|
358 |
|
|
order_by=None), into=full_in_table, add_pkey_=True)
|
359 |
3287
|
aaronmk
|
|
360 |
3077
|
aaronmk
|
pkeys_table_exists_ref = [False]
|
361 |
3499
|
aaronmk
|
def insert_into_pkeys(joins, cols=None, limit=None, **kw_args):
|
362 |
3477
|
aaronmk
|
query = sql.mk_select(db, joins, cols, order_by=None, limit=limit)
|
363 |
3077
|
aaronmk
|
if pkeys_table_exists_ref[0]:
|
364 |
4484
|
aaronmk
|
sql.insert_select(db, into, [in_pkey, into_out_pkey], query,
|
365 |
|
|
**kw_args)
|
366 |
3077
|
aaronmk
|
else:
|
367 |
3304
|
aaronmk
|
sql.run_query_into(db, query, into=into, add_pkey_=True, **kw_args)
|
368 |
3077
|
aaronmk
|
pkeys_table_exists_ref[0] = True
|
369 |
|
|
|
370 |
|
|
limit_ref = [None]
|
371 |
3418
|
aaronmk
|
def mk_main_select(joins, cols):
|
372 |
|
|
return sql.mk_select(db, joins, cols, limit=limit_ref[0], order_by=None)
|
373 |
|
|
|
374 |
3431
|
aaronmk
|
if is_literals: insert_in_table = None
|
375 |
|
|
else:
|
376 |
|
|
insert_in_table = in_table
|
377 |
|
|
insert_in_tables = [insert_in_table]
|
378 |
3352
|
aaronmk
|
join_cols = sql_gen.ColDict(db, out_table)
|
379 |
3077
|
aaronmk
|
|
380 |
|
|
exc_strs = set()
|
381 |
|
|
def log_exc(e):
|
382 |
|
|
e_str = exc.str_(e, first_line_only=True)
|
383 |
|
|
log_debug('Caught exception: '+e_str)
|
384 |
3552
|
aaronmk
|
if e_str in exc_strs: # avoid infinite loops
|
385 |
|
|
log_debug('Exception already seen, handler broken')
|
386 |
|
|
on_error(e)
|
387 |
|
|
remove_all_rows()
|
388 |
|
|
else: exc_strs.add(e_str)
|
389 |
3077
|
aaronmk
|
|
390 |
|
|
def remove_all_rows():
|
391 |
|
|
log_debug('Ignoring all rows')
|
392 |
|
|
limit_ref[0] = 0 # just create an empty pkeys table
|
393 |
|
|
|
394 |
3352
|
aaronmk
|
def ignore_cond(cond, e):
|
395 |
3704
|
aaronmk
|
if is_literals: remove_all_rows()
|
396 |
|
|
else:
|
397 |
|
|
out_table_cols = sql_gen.ColDict(db, out_table)
|
398 |
|
|
out_table_cols.update(util.dict_subset_right_join({},
|
399 |
|
|
sql.table_cols(db, out_table)))
|
400 |
|
|
|
401 |
|
|
in_cols = []
|
402 |
|
|
cond = sql.map_expr(db, cond, mapping, in_cols)
|
403 |
|
|
cond = sql.map_expr(db, cond, out_table_cols)
|
404 |
|
|
|
405 |
|
|
track_data_error(db, errors_table_, sql_gen.cols_srcs(in_cols),
|
406 |
|
|
None, e.cause.pgcode,
|
407 |
|
|
strings.ensure_newl(e.cause.pgerror)+'condition: '+cond)
|
408 |
|
|
|
409 |
|
|
not_cond = sql_gen.NotCond(sql_gen.CustomCode(cond))
|
410 |
|
|
log_debug('Ignoring rows where '+strings.as_tt(not_cond.to_str(db)))
|
411 |
|
|
sql.delete(db, insert_in_table, not_cond)
|
412 |
3352
|
aaronmk
|
|
413 |
3294
|
aaronmk
|
not_null_cols = set()
|
414 |
3077
|
aaronmk
|
def ignore(in_col, value, e):
|
415 |
3630
|
aaronmk
|
if sql_gen.is_table_col(in_col):
|
416 |
|
|
in_col = sql_gen.with_table(in_col, insert_in_table)
|
417 |
|
|
|
418 |
|
|
track_data_error(db, errors_table_, in_col.srcs, value,
|
419 |
|
|
e.cause.pgcode, e.cause.pgerror)
|
420 |
|
|
|
421 |
|
|
sql.add_index(db, in_col, insert_in_table) # enable fast filtering
|
422 |
|
|
if value != None and in_col not in not_null_cols:
|
423 |
4492
|
aaronmk
|
log_debug('Replacing invalid value '
|
424 |
|
|
+strings.as_tt(strings.urepr(value))+' with NULL in column '
|
425 |
|
|
+strings.as_tt(in_col.to_str(db)))
|
426 |
3630
|
aaronmk
|
sql.update(db, insert_in_table, [(in_col, None)],
|
427 |
|
|
sql_gen.ColValueCond(in_col, value))
|
428 |
|
|
else:
|
429 |
3637
|
aaronmk
|
log_debug('Ignoring rows with '+strings.as_tt(in_col.to_str(db))
|
430 |
4492
|
aaronmk
|
+' = '+strings.as_tt(strings.urepr(value)))
|
431 |
3630
|
aaronmk
|
sql.delete(db, insert_in_table,
|
432 |
|
|
sql_gen.ColValueCond(in_col, value))
|
433 |
|
|
if value == None: not_null_cols.add(in_col)
|
434 |
3293
|
aaronmk
|
else:
|
435 |
3630
|
aaronmk
|
assert isinstance(in_col, sql_gen.NamedCol)
|
436 |
3684
|
aaronmk
|
in_value = sql_gen.remove_col_rename(in_col)
|
437 |
|
|
assert sql_gen.is_literal(in_value)
|
438 |
|
|
if value == in_value.value:
|
439 |
|
|
if value != None:
|
440 |
|
|
log_debug('Replacing invalid literal '
|
441 |
|
|
+strings.as_tt(in_col.to_str(db))+' with NULL')
|
442 |
|
|
mapping[in_col.name] = None
|
443 |
|
|
else:
|
444 |
|
|
remove_all_rows()
|
445 |
|
|
# otherwise, all columns were being ignore()d because the specific
|
446 |
|
|
# column couldn't be identified, and this was not the invalid column
|
447 |
3077
|
aaronmk
|
|
448 |
3431
|
aaronmk
|
if not is_literals:
|
449 |
|
|
def insert_pkeys_table(which):
|
450 |
|
|
return sql_gen.Table(sql_gen.concat(in_table.name,
|
451 |
|
|
'_insert_'+which+'_pkeys'))
|
452 |
|
|
insert_out_pkeys = insert_pkeys_table('out')
|
453 |
|
|
insert_in_pkeys = insert_pkeys_table('in')
|
454 |
3077
|
aaronmk
|
|
455 |
3918
|
aaronmk
|
def mk_func_call():
|
456 |
3550
|
aaronmk
|
args = dict(((k.name, v) for k, v in mapping.iteritems()))
|
457 |
4484
|
aaronmk
|
return sql_gen.FunctionCall(out_table, **args), args
|
458 |
3550
|
aaronmk
|
|
459 |
3918
|
aaronmk
|
if is_function and not is_literals:
|
460 |
|
|
log_debug('Defining wrapper function')
|
461 |
|
|
|
462 |
|
|
func_call, args = mk_func_call()
|
463 |
4484
|
aaronmk
|
func_call = sql_gen.NamedCol(into_out_pkey, func_call)
|
464 |
3918
|
aaronmk
|
|
465 |
|
|
# Create empty pkeys table so its row type can be used
|
466 |
|
|
insert_into_pkeys(input_joins, [in_pkey_col, func_call], limit=0,
|
467 |
|
|
recover=True)
|
468 |
4484
|
aaronmk
|
result_type = db.col_info(sql_gen.Col(into_out_pkey, into)).type
|
469 |
3918
|
aaronmk
|
|
470 |
|
|
## Create error handling wrapper function
|
471 |
|
|
|
472 |
|
|
wrapper = db.TempFunction(sql_gen.concat(into.name, '_wrap'))
|
473 |
|
|
|
474 |
|
|
select_cols = [in_pkey_col]+args.values()
|
475 |
|
|
row_var = copy.copy(sql_gen.row_var)
|
476 |
|
|
row_var.set_srcs([in_table])
|
477 |
|
|
in_pkey_var = sql_gen.Col(in_pkey, row_var)
|
478 |
|
|
|
479 |
|
|
args = dict(((k, sql_gen.with_table(v, row_var))
|
480 |
|
|
for k, v in args.iteritems()))
|
481 |
|
|
func_call = sql_gen.FunctionCall(out_table, **args)
|
482 |
|
|
|
483 |
|
|
def mk_return(result):
|
484 |
|
|
return sql_gen.ReturnQuery(sql.mk_select(db,
|
485 |
|
|
fields=[in_pkey_var, result], explain=False))
|
486 |
|
|
exc_handler = func_wrapper_exception_handler(db,
|
487 |
|
|
mk_return(sql_gen.Cast(result_type, None)), args.values(),
|
488 |
|
|
errors_table_)
|
489 |
|
|
|
490 |
|
|
sql.define_func(db, sql_gen.FunctionDef(wrapper, sql_gen.SetOf(into),
|
491 |
|
|
sql_gen.RowExcIgnore(sql_gen.RowType(in_table),
|
492 |
|
|
sql.mk_select(db, input_joins, order_by=None),
|
493 |
|
|
mk_return(func_call), exc_handler=exc_handler)
|
494 |
|
|
))
|
495 |
|
|
wrapper_table = sql_gen.FunctionCall(wrapper)
|
496 |
3550
|
aaronmk
|
|
497 |
3077
|
aaronmk
|
# Do inserts and selects
|
498 |
|
|
while True:
|
499 |
3473
|
aaronmk
|
has_joins = join_cols != {}
|
500 |
|
|
|
501 |
3551
|
aaronmk
|
# Handle unrecoverable errors in a special case
|
502 |
|
|
if limit_ref[0] == 0:
|
503 |
4207
|
aaronmk
|
if is_literals or default == None:
|
504 |
3686
|
aaronmk
|
default = sql_gen.remove_col_rename(default)
|
505 |
4492
|
aaronmk
|
log_debug('Returning default: '
|
506 |
|
|
+strings.as_tt(strings.urepr(default)))
|
507 |
3623
|
aaronmk
|
return default
|
508 |
3551
|
aaronmk
|
elif is_function: pass # empty pkeys table already created
|
509 |
|
|
else:
|
510 |
|
|
log_debug('Creating an empty output pkeys table')
|
511 |
3745
|
aaronmk
|
has_joins = False # use the no-joins case
|
512 |
3551
|
aaronmk
|
cur = sql.run_query_into(db, sql.mk_select(db, out_table,
|
513 |
|
|
[out_pkey], order_by=None, limit=0), into=insert_out_pkeys)
|
514 |
|
|
|
515 |
3077
|
aaronmk
|
break # don't do main case
|
516 |
|
|
|
517 |
|
|
# Prepare to insert new rows
|
518 |
3918
|
aaronmk
|
if is_function:
|
519 |
|
|
log_debug('Calling function on input rows')
|
520 |
|
|
if is_literals: func_call, args = mk_func_call()
|
521 |
3077
|
aaronmk
|
else:
|
522 |
3550
|
aaronmk
|
log_debug('Trying to insert new rows')
|
523 |
3291
|
aaronmk
|
insert_args = dict(recover=True, cacheable=False)
|
524 |
|
|
if has_joins:
|
525 |
|
|
insert_args.update(dict(ignore=True))
|
526 |
|
|
else:
|
527 |
3431
|
aaronmk
|
insert_args.update(dict(returning=out_pkey))
|
528 |
|
|
if not is_literals:
|
529 |
|
|
insert_args.update(dict(into=insert_out_pkeys))
|
530 |
3291
|
aaronmk
|
main_select = mk_main_select([insert_in_table], [sql_gen.with_table(
|
531 |
|
|
c, insert_in_table) for c in mapping.values()])
|
532 |
3077
|
aaronmk
|
|
533 |
3292
|
aaronmk
|
try:
|
534 |
|
|
cur = None
|
535 |
3077
|
aaronmk
|
if is_function:
|
536 |
3917
|
aaronmk
|
if is_literals:
|
537 |
|
|
cur = sql.select(db, fields=[func_call], recover=True,
|
538 |
|
|
cacheable=True)
|
539 |
3499
|
aaronmk
|
else: insert_into_pkeys(wrapper_table, recover=True)
|
540 |
3077
|
aaronmk
|
else:
|
541 |
3292
|
aaronmk
|
cur = sql.insert_select(db, out_table, mapping.keys(),
|
542 |
3077
|
aaronmk
|
main_select, **insert_args)
|
543 |
|
|
break # insert successful
|
544 |
|
|
except sql.MissingCastException, e:
|
545 |
|
|
log_exc(e)
|
546 |
|
|
|
547 |
|
|
type_ = e.type
|
548 |
4140
|
aaronmk
|
if e.col == None: out_cols = mapping.keys()
|
549 |
|
|
else: out_cols = [e.col]
|
550 |
3077
|
aaronmk
|
|
551 |
4140
|
aaronmk
|
for out_col in out_cols:
|
552 |
4171
|
aaronmk
|
log_debug('Casting '+strings.as_tt(strings.repr_no_u(out_col))
|
553 |
4140
|
aaronmk
|
+' input to '+strings.as_tt(type_))
|
554 |
|
|
in_col = mapping[out_col]
|
555 |
|
|
while True:
|
556 |
|
|
try:
|
557 |
|
|
mapping[out_col] = cast_temp_col(db, type_, in_col,
|
558 |
|
|
errors_table_)
|
559 |
|
|
break # cast successful
|
560 |
|
|
except sql.InvalidValueException, e:
|
561 |
|
|
log_exc(e)
|
562 |
|
|
|
563 |
|
|
ignore(in_col, e.value, e)
|
564 |
3077
|
aaronmk
|
except sql.DuplicateKeyException, e:
|
565 |
|
|
log_exc(e)
|
566 |
|
|
|
567 |
3274
|
aaronmk
|
# Different rows violating different unique constraints not
|
568 |
|
|
# supported
|
569 |
|
|
assert not join_cols
|
570 |
|
|
|
571 |
3077
|
aaronmk
|
join_cols.update(util.dict_subset_right_join(mapping, e.cols))
|
572 |
|
|
log_debug('Ignoring existing rows, comparing on these columns:\n'
|
573 |
|
|
+strings.as_inline_table(join_cols, ustr=col_ustr))
|
574 |
3102
|
aaronmk
|
|
575 |
3431
|
aaronmk
|
if is_literals:
|
576 |
|
|
return sql.value(sql.select(db, out_table, [out_pkey_col],
|
577 |
4025
|
aaronmk
|
join_cols, order_by=None))
|
578 |
3431
|
aaronmk
|
|
579 |
3102
|
aaronmk
|
# Uniquify input table to avoid internal duplicate keys
|
580 |
|
|
insert_in_table = sql.distinct_table(db, insert_in_table,
|
581 |
3358
|
aaronmk
|
join_cols.values())
|
582 |
3144
|
aaronmk
|
insert_in_tables.append(insert_in_table)
|
583 |
3077
|
aaronmk
|
except sql.NullValueException, e:
|
584 |
|
|
log_exc(e)
|
585 |
|
|
|
586 |
|
|
out_col, = e.cols
|
587 |
|
|
try: in_col = mapping[out_col]
|
588 |
3618
|
aaronmk
|
except KeyError, e:
|
589 |
|
|
try: in_col = mapping[out_col] = col_defaults[out_col]
|
590 |
|
|
except KeyError:
|
591 |
|
|
msg = 'Missing mapping for NOT NULL column '+out_col
|
592 |
|
|
log_debug(msg)
|
593 |
3713
|
aaronmk
|
if default == None: warnings.warn(UserWarning(msg))
|
594 |
|
|
# not an error because sometimes the mappings include
|
595 |
|
|
# extra tables which aren't used by the dataset
|
596 |
3618
|
aaronmk
|
remove_all_rows()
|
597 |
3294
|
aaronmk
|
else: ignore(in_col, None, e)
|
598 |
3352
|
aaronmk
|
except sql.CheckException, e:
|
599 |
|
|
log_exc(e)
|
600 |
|
|
|
601 |
|
|
ignore_cond(e.cond, e)
|
602 |
3413
|
aaronmk
|
except sql.InvalidValueException, e:
|
603 |
|
|
log_exc(e)
|
604 |
|
|
|
605 |
|
|
for in_col in mapping.values(): ignore(in_col, e.value, e)
|
606 |
3474
|
aaronmk
|
except psycopg2.extensions.TransactionRollbackError, e:
|
607 |
|
|
log_exc(e)
|
608 |
|
|
# retry
|
609 |
3077
|
aaronmk
|
except sql.DatabaseErrors, e:
|
610 |
|
|
log_exc(e)
|
611 |
|
|
|
612 |
|
|
log_debug('No handler for exception')
|
613 |
|
|
on_error(e)
|
614 |
|
|
remove_all_rows()
|
615 |
|
|
# after exception handled, rerun loop with additional constraints
|
616 |
|
|
|
617 |
|
|
if cur != None and row_ct_ref != None and cur.rowcount >= 0:
|
618 |
|
|
row_ct_ref[0] += cur.rowcount
|
619 |
|
|
|
620 |
3530
|
aaronmk
|
if is_literals: return sql.value(cur)
|
621 |
|
|
|
622 |
|
|
if is_function: pass # pkeys table already created
|
623 |
3077
|
aaronmk
|
elif has_joins:
|
624 |
|
|
select_joins = input_joins+[sql_gen.Join(out_table, join_cols)]
|
625 |
|
|
log_debug('Getting output table pkeys of existing/inserted rows')
|
626 |
4484
|
aaronmk
|
insert_into_pkeys(select_joins, [in_pkey_col,
|
627 |
|
|
sql_gen.NamedCol(into_out_pkey, out_pkey_col)])
|
628 |
3077
|
aaronmk
|
else:
|
629 |
|
|
sql.add_row_num(db, insert_out_pkeys) # for joining with input pkeys
|
630 |
|
|
|
631 |
|
|
log_debug('Getting input table pkeys of inserted rows')
|
632 |
3285
|
aaronmk
|
# Note that mk_main_select() does not use ORDER BY. Instead, assume that
|
633 |
|
|
# since the SELECT query is identical to the one used in INSERT SELECT,
|
634 |
|
|
# its rows will be retrieved in the same order.
|
635 |
3077
|
aaronmk
|
sql.run_query_into(db, mk_main_select(input_joins, [in_pkey]),
|
636 |
|
|
into=insert_in_pkeys)
|
637 |
|
|
sql.add_row_num(db, insert_in_pkeys) # for joining with output pkeys
|
638 |
|
|
|
639 |
|
|
assert sql.table_row_count(db, insert_out_pkeys) == sql.table_row_count(
|
640 |
|
|
db, insert_in_pkeys)
|
641 |
|
|
|
642 |
|
|
log_debug('Combining output and input pkeys in inserted order')
|
643 |
|
|
pkey_joins = [insert_in_pkeys, sql_gen.Join(insert_out_pkeys,
|
644 |
|
|
{sql.row_num_col: sql_gen.join_same_not_null})]
|
645 |
4484
|
aaronmk
|
in_col = sql_gen.Col(in_pkey, insert_in_pkeys)
|
646 |
|
|
out_col = sql_gen.NamedCol(into_out_pkey,
|
647 |
|
|
sql_gen.Col(out_pkey, insert_out_pkeys))
|
648 |
|
|
insert_into_pkeys(pkey_joins, [in_col, out_col])
|
649 |
3077
|
aaronmk
|
|
650 |
|
|
sql.empty_temp(db, [insert_out_pkeys, insert_in_pkeys])
|
651 |
|
|
|
652 |
3531
|
aaronmk
|
if limit_ref[0] == 0 or not is_function: # is_function doesn't leave holes
|
653 |
3187
|
aaronmk
|
log_debug('Setting pkeys of missing rows to '
|
654 |
4492
|
aaronmk
|
+strings.as_tt(strings.urepr(default)))
|
655 |
3287
|
aaronmk
|
missing_rows_joins = [full_in_table, sql_gen.Join(into,
|
656 |
3187
|
aaronmk
|
{in_pkey: sql_gen.join_same_not_null}, sql_gen.filter_out)]
|
657 |
|
|
# must use join_same_not_null or query will take forever
|
658 |
|
|
insert_into_pkeys(missing_rows_joins,
|
659 |
3287
|
aaronmk
|
[sql_gen.Col(in_pkey, full_in_table),
|
660 |
4484
|
aaronmk
|
sql_gen.NamedCol(into_out_pkey, default)])
|
661 |
3187
|
aaronmk
|
# otherwise, there is already an entry for every row
|
662 |
3077
|
aaronmk
|
|
663 |
3530
|
aaronmk
|
assert (sql.table_row_count(db, into)
|
664 |
|
|
== sql.table_row_count(db, full_in_table))
|
665 |
|
|
|
666 |
|
|
sql.empty_temp(db, insert_in_tables+[full_in_table])
|
667 |
|
|
|
668 |
|
|
srcs = []
|
669 |
3619
|
aaronmk
|
if is_function: srcs = sql_gen.cols_srcs(in_cols)
|
670 |
4484
|
aaronmk
|
return sql_gen.Col(into_out_pkey, into, srcs)
|