Project

General

Profile

1 11 aaronmk
# Database access
2
3 1869 aaronmk
import copy
4 2127 aaronmk
import operator
5 11 aaronmk
import re
6 865 aaronmk
import warnings
7 11 aaronmk
8 300 aaronmk
import exc
9 1909 aaronmk
import dicts
10 1893 aaronmk
import iters
11 1960 aaronmk
import lists
12 1889 aaronmk
from Proxy import Proxy
13 1872 aaronmk
import rand
14 2217 aaronmk
import sql_gen
15 862 aaronmk
import strings
16 131 aaronmk
import util
17 11 aaronmk
18 832 aaronmk
##### Exceptions
19
20 2170 aaronmk
def get_cur_query(cur, input_query=None, input_params=None):
21 2168 aaronmk
    raw_query = None
22
    if hasattr(cur, 'query'): raw_query = cur.query
23
    elif hasattr(cur, '_last_executed'): raw_query = cur._last_executed
24 2170 aaronmk
25
    if raw_query != None: return raw_query
26
    else: return repr(input_query)+' % '+repr(input_params)
27 14 aaronmk
28 2170 aaronmk
def _add_cursor_info(e, *args, **kw_args):
29
    '''For params, see get_cur_query()'''
30
    exc.add_msg(e, 'query: '+str(get_cur_query(*args, **kw_args)))
31 135 aaronmk
32 300 aaronmk
class DbException(exc.ExceptionWithCause):
33 14 aaronmk
    def __init__(self, msg, cause=None, cur=None):
34 2145 aaronmk
        exc.ExceptionWithCause.__init__(self, msg, cause, cause_newline=True)
35 14 aaronmk
        if cur != None: _add_cursor_info(self, cur)
36
37 2143 aaronmk
class ExceptionWithName(DbException):
38
    def __init__(self, name, cause=None):
39 2145 aaronmk
        DbException.__init__(self, 'for name: '+str(name), cause)
40 2143 aaronmk
        self.name = name
41 360 aaronmk
42 2240 aaronmk
class ExceptionWithNameValue(DbException):
43
    def __init__(self, name, value, cause=None):
44
        DbException.__init__(self,
45 2247 aaronmk
            'for name: '+str(name)+'; value: '+strings.ustr(value), cause)
46 2240 aaronmk
        self.name = name
47
        self.value = value
48
49 468 aaronmk
class ExceptionWithColumns(DbException):
50
    def __init__(self, cols, cause=None):
51 2145 aaronmk
        DbException.__init__(self, 'for columns: '+(', '.join(cols)), cause)
52 468 aaronmk
        self.cols = cols
53 11 aaronmk
54 2143 aaronmk
class NameException(DbException): pass
55
56 468 aaronmk
class DuplicateKeyException(ExceptionWithColumns): pass
57 13 aaronmk
58 468 aaronmk
class NullValueException(ExceptionWithColumns): pass
59 13 aaronmk
60 2240 aaronmk
class FunctionValueException(ExceptionWithNameValue): pass
61 2239 aaronmk
62 2143 aaronmk
class DuplicateTableException(ExceptionWithName): pass
63
64 2188 aaronmk
class DuplicateFunctionException(ExceptionWithName): pass
65
66 89 aaronmk
class EmptyRowException(DbException): pass
67
68 865 aaronmk
##### Warnings
69
70
class DbWarning(UserWarning): pass
71
72 1930 aaronmk
##### Result retrieval
73
74
def col_names(cur): return (col[0] for col in cur.description)
75
76
def rows(cur): return iter(lambda: cur.fetchone(), None)
77
78
def consume_rows(cur):
79
    '''Used to fetch all rows so result will be cached'''
80
    iters.consume_iter(rows(cur))
81
82
def next_row(cur): return rows(cur).next()
83
84
def row(cur):
85
    row_ = next_row(cur)
86
    consume_rows(cur)
87
    return row_
88
89
def next_value(cur): return next_row(cur)[0]
90
91
def value(cur): return row(cur)[0]
92
93
def values(cur): return iters.func_iter(lambda: next_value(cur))
94
95
def value_or_none(cur):
96
    try: return value(cur)
97
    except StopIteration: return None
98
99 2101 aaronmk
##### Input validation
100
101 2198 aaronmk
def clean_name(name): return re.sub(r'\W', r'', name).lower()
102 2101 aaronmk
103
def check_name(name):
104
    if re.search(r'\W', name) != None: raise NameException('Name "'+name
105
        +'" may contain only alphanumeric characters and _')
106
107
def esc_name_by_module(module, name, ignore_case=False):
108
    if module == 'psycopg2':
109
        if ignore_case:
110
            # Don't enclose in quotes because this disables case-insensitivity
111
            check_name(name)
112
            return name
113
        else: quote = '"'
114
    elif module == 'MySQLdb': quote = '`'
115
    else: raise NotImplementedError("Can't escape name for "+module+' database')
116
    return quote + name.replace(quote, '') + quote
117
118
def esc_name_by_engine(engine, name, **kw_args):
119
    return esc_name_by_module(db_engines[engine][0], name, **kw_args)
120
121
def esc_name(db, name, **kw_args):
122
    return esc_name_by_module(util.root_module(db.db), name, **kw_args)
123
124
def qual_name(db, schema, table):
125
    def esc_name_(name): return esc_name(db, name)
126
    table = esc_name_(table)
127
    if schema != None: return esc_name_(schema)+'.'+table
128
    else: return table
129
130 1869 aaronmk
##### Database connections
131 1849 aaronmk
132 2097 aaronmk
db_config_names = ['engine', 'host', 'user', 'password', 'database', 'schemas']
133 1926 aaronmk
134 1869 aaronmk
db_engines = {
135
    'MySQL': ('MySQLdb', {'password': 'passwd', 'database': 'db'}),
136
    'PostgreSQL': ('psycopg2', {}),
137
}
138
139
DatabaseErrors_set = set([DbException])
140
DatabaseErrors = tuple(DatabaseErrors_set)
141
142
def _add_module(module):
143
    DatabaseErrors_set.add(module.DatabaseError)
144
    global DatabaseErrors
145
    DatabaseErrors = tuple(DatabaseErrors_set)
146
147
def db_config_str(db_config):
148
    return db_config['engine']+' database '+db_config['database']
149
150 1909 aaronmk
def _query_lookup(query, params): return (query, dicts.make_hashable(params))
151 1894 aaronmk
152 1901 aaronmk
log_debug_none = lambda msg: None
153
154 1849 aaronmk
class DbConn:
155 2190 aaronmk
    def __init__(self, db_config, serializable=True, autocommit=False,
156
        caching=True, log_debug=log_debug_none):
157 1869 aaronmk
        self.db_config = db_config
158
        self.serializable = serializable
159 2190 aaronmk
        self.autocommit = autocommit
160
        self.caching = caching
161 1901 aaronmk
        self.log_debug = log_debug
162 2193 aaronmk
        self.debug = log_debug != log_debug_none
163 1869 aaronmk
164
        self.__db = None
165 1889 aaronmk
        self.query_results = {}
166 2139 aaronmk
        self._savepoint = 0
167 1869 aaronmk
168
    def __getattr__(self, name):
169
        if name == '__dict__': raise Exception('getting __dict__')
170
        if name == 'db': return self._db()
171
        else: raise AttributeError()
172
173
    def __getstate__(self):
174
        state = copy.copy(self.__dict__) # shallow copy
175 1915 aaronmk
        state['log_debug'] = None # don't pickle the debug callback
176 1869 aaronmk
        state['_DbConn__db'] = None # don't pickle the connection
177
        return state
178
179 2165 aaronmk
    def connected(self): return self.__db != None
180
181 1869 aaronmk
    def _db(self):
182
        if self.__db == None:
183
            # Process db_config
184
            db_config = self.db_config.copy() # don't modify input!
185 2097 aaronmk
            schemas = db_config.pop('schemas', None)
186 1869 aaronmk
            module_name, mappings = db_engines[db_config.pop('engine')]
187
            module = __import__(module_name)
188
            _add_module(module)
189
            for orig, new in mappings.iteritems():
190
                try: util.rename_key(db_config, orig, new)
191
                except KeyError: pass
192
193
            # Connect
194
            self.__db = module.connect(**db_config)
195
196
            # Configure connection
197 2234 aaronmk
            if self.serializable and not self.autocommit: run_raw_query(self,
198
                'SET TRANSACTION ISOLATION LEVEL SERIALIZABLE')
199 2101 aaronmk
            if schemas != None:
200
                schemas_ = ''.join((esc_name(self, s)+', '
201
                    for s in schemas.split(',')))
202
                run_raw_query(self, "SELECT set_config('search_path', \
203
%s || current_setting('search_path'), false)", [schemas_])
204 1869 aaronmk
205
        return self.__db
206 1889 aaronmk
207 1891 aaronmk
    class DbCursor(Proxy):
208 1927 aaronmk
        def __init__(self, outer):
209 1891 aaronmk
            Proxy.__init__(self, outer.db.cursor())
210 2191 aaronmk
            self.outer = outer
211 1927 aaronmk
            self.query_results = outer.query_results
212 1894 aaronmk
            self.query_lookup = None
213 1891 aaronmk
            self.result = []
214 1889 aaronmk
215 1894 aaronmk
        def execute(self, query, params=None):
216 1930 aaronmk
            self._is_insert = query.upper().find('INSERT') >= 0
217 1894 aaronmk
            self.query_lookup = _query_lookup(query, params)
218 2148 aaronmk
            try:
219 2191 aaronmk
                try:
220
                    return_value = self.inner.execute(query, params)
221
                    self.outer.do_autocommit()
222 2148 aaronmk
                finally: self.query = get_cur_query(self.inner)
223 1904 aaronmk
            except Exception, e:
224 2170 aaronmk
                _add_cursor_info(e, self, query, params)
225 1904 aaronmk
                self.result = e # cache the exception as the result
226
                self._cache_result()
227
                raise
228 1930 aaronmk
            # Fetch all rows so result will be cached
229
            if self.rowcount == 0 and not self._is_insert: consume_rows(self)
230 1894 aaronmk
            return return_value
231
232 1891 aaronmk
        def fetchone(self):
233
            row = self.inner.fetchone()
234 1899 aaronmk
            if row != None: self.result.append(row)
235
            # otherwise, fetched all rows
236 1904 aaronmk
            else: self._cache_result()
237
            return row
238
239
        def _cache_result(self):
240 1906 aaronmk
            # For inserts, only cache exceptions since inserts are not
241
            # idempotent, but an invalid insert will always be invalid
242 1930 aaronmk
            if self.query_results != None and (not self._is_insert
243 1906 aaronmk
                or isinstance(self.result, Exception)):
244
245 1894 aaronmk
                assert self.query_lookup != None
246 1916 aaronmk
                self.query_results[self.query_lookup] = self.CacheCursor(
247
                    util.dict_subset(dicts.AttrsDictView(self),
248
                    ['query', 'result', 'rowcount', 'description']))
249 1906 aaronmk
250 1916 aaronmk
        class CacheCursor:
251
            def __init__(self, cached_result): self.__dict__ = cached_result
252
253 1927 aaronmk
            def execute(self, *args, **kw_args):
254 1916 aaronmk
                if isinstance(self.result, Exception): raise self.result
255
                # otherwise, result is a rows list
256
                self.iter = iter(self.result)
257
258
            def fetchone(self):
259
                try: return self.iter.next()
260
                except StopIteration: return None
261 1891 aaronmk
262 2212 aaronmk
    def esc_value(self, value):
263 2215 aaronmk
        module = util.root_module(self.db)
264
        if module == 'psycopg2': return self.db.cursor().mogrify('%s', [value])
265 2212 aaronmk
        elif module == 'MySQLdb':
266
            import _mysql
267
            return _mysql.escape_string(value)
268
        else: raise NotImplementedError("Can't escape value for "+module
269
            +' database')
270
271 1894 aaronmk
    def run_query(self, query, params=None, cacheable=False):
272 2148 aaronmk
        '''Translates known DB errors to typed exceptions:
273
        See self.DbCursor.execute().'''
274 2167 aaronmk
        assert query != None
275
276 2047 aaronmk
        if not self.caching: cacheable = False
277 1903 aaronmk
        used_cache = False
278
        try:
279 1927 aaronmk
            # Get cursor
280
            if cacheable:
281
                query_lookup = _query_lookup(query, params)
282
                try:
283
                    cur = self.query_results[query_lookup]
284
                    used_cache = True
285
                except KeyError: cur = self.DbCursor(self)
286
            else: cur = self.db.cursor()
287
288
            # Run query
289 2148 aaronmk
            cur.execute(query, params)
290 1903 aaronmk
        finally:
291 2193 aaronmk
            if self.debug: # only compute msg if needed
292 1903 aaronmk
                if used_cache: cache_status = 'Cache hit'
293
                elif cacheable: cache_status = 'Cache miss'
294
                else: cache_status = 'Non-cacheable'
295 1927 aaronmk
                self.log_debug(cache_status+': '
296 2170 aaronmk
                    +strings.one_line(str(get_cur_query(cur, query, params))))
297 1903 aaronmk
298
        return cur
299 1914 aaronmk
300
    def is_cached(self, query, params=None):
301
        return _query_lookup(query, params) in self.query_results
302 2139 aaronmk
303
    def with_savepoint(self, func):
304 2171 aaronmk
        savepoint = 'level_'+str(self._savepoint)
305 2139 aaronmk
        self.run_query('SAVEPOINT '+savepoint)
306
        self._savepoint += 1
307
        try:
308
            try: return_val = func()
309
            finally:
310
                self._savepoint -= 1
311
                assert self._savepoint >= 0
312
        except:
313
            self.run_query('ROLLBACK TO SAVEPOINT '+savepoint)
314
            raise
315
        else:
316
            self.run_query('RELEASE SAVEPOINT '+savepoint)
317 2191 aaronmk
            self.do_autocommit()
318 2139 aaronmk
            return return_val
319 2191 aaronmk
320
    def do_autocommit(self):
321
        '''Autocommits if outside savepoint'''
322
        assert self._savepoint >= 0
323
        if self.autocommit and self._savepoint == 0:
324
            self.log_debug('Autocommiting')
325
            self.db.commit()
326 1849 aaronmk
327 1869 aaronmk
connect = DbConn
328
329 832 aaronmk
##### Querying
330
331 1894 aaronmk
def run_raw_query(db, *args, **kw_args):
332 2085 aaronmk
    '''For params, see DbConn.run_query()'''
333 1894 aaronmk
    return db.run_query(*args, **kw_args)
334 11 aaronmk
335 2068 aaronmk
def mogrify(db, query, params):
336
    module = util.root_module(db.db)
337
    if module == 'psycopg2': return db.db.cursor().mogrify(query, params)
338
    else: raise NotImplementedError("Can't mogrify query for "+module+
339
        ' database')
340
341 832 aaronmk
##### Recoverable querying
342 15 aaronmk
343 2139 aaronmk
def with_savepoint(db, func): return db.with_savepoint(func)
344 11 aaronmk
345 1894 aaronmk
def run_query(db, query, params=None, recover=None, cacheable=False):
346 830 aaronmk
    if recover == None: recover = False
347
348 2148 aaronmk
    try:
349
        def run(): return run_raw_query(db, query, params, cacheable)
350
        if recover and not db.is_cached(query, params):
351
            return with_savepoint(db, run)
352
        else: return run() # don't need savepoint if cached
353
    except Exception, e:
354
        if not recover: raise # need savepoint to run index_cols()
355 2247 aaronmk
        msg = exc.str_(e)
356 2240 aaronmk
357 2148 aaronmk
        match = re.search(r'duplicate key value violates unique constraint '
358
            r'"((_?[^\W_]+)_[^"]+)"', msg)
359
        if match:
360
            constraint, table = match.groups()
361
            try: cols = index_cols(db, table, constraint)
362
            except NotImplementedError: raise e
363
            else: raise DuplicateKeyException(cols, e)
364 2240 aaronmk
365 2148 aaronmk
        match = re.search(r'null value in column "(\w+)" violates not-null '
366 2239 aaronmk
            r'constraint', msg)
367 2148 aaronmk
        if match: raise NullValueException([match.group(1)], e)
368 2240 aaronmk
369 2247 aaronmk
        match = re.search(r'invalid input (?:syntax|value)\b.*: "(.+)"\n'
370 2242 aaronmk
            r'(?:(?s).*)\bfunction "(\w+)".*\bat assignment', msg)
371 2240 aaronmk
        if match:
372 2247 aaronmk
            raise FunctionValueException(match.group(2),
373
                strings.to_unicode(match.group(1)), e)
374 2240 aaronmk
375 2148 aaronmk
        match = re.search(r'relation "(\w+)" already exists', msg)
376
        if match: raise DuplicateTableException(match.group(1), e)
377 2240 aaronmk
378 2188 aaronmk
        match = re.search(r'function "(\w+)" already exists', msg)
379
        if match: raise DuplicateFunctionException(match.group(1), e)
380 2240 aaronmk
381 2148 aaronmk
        raise # no specific exception raised
382 830 aaronmk
383 832 aaronmk
##### Basic queries
384
385 2153 aaronmk
def next_version(name):
386
    '''Prepends the version # so it won't be removed if the name is truncated'''
387 2163 aaronmk
    version = 1 # first existing name was version 0
388 2153 aaronmk
    match = re.match(r'^v(\d+)_(.*)$', name)
389
    if match:
390
        version = int(match.group(1))+1
391
        name = match.group(2)
392
    return 'v'+str(version)+'_'+name
393
394 2151 aaronmk
def run_query_into(db, query, params, into_ref=None, *args, **kw_args):
395 2085 aaronmk
    '''Outputs a query to a temp table.
396
    For params, see run_query().
397
    '''
398 2151 aaronmk
    if into_ref == None: return run_query(db, query, params, *args, **kw_args)
399 2085 aaronmk
    else: # place rows in temp table
400 2151 aaronmk
        check_name(into_ref[0])
401 2153 aaronmk
        kw_args['recover'] = True
402
        while True:
403
            try:
404 2194 aaronmk
                create_query = 'CREATE'
405
                if not db.debug: create_query += ' TEMP'
406
                create_query += ' TABLE '+into_ref[0]+' AS '+query
407
408
                return run_query(db, create_query, params, *args, **kw_args)
409 2153 aaronmk
                    # CREATE TABLE AS sets rowcount to # rows in query
410
            except DuplicateTableException, e:
411
                into_ref[0] = next_version(into_ref[0])
412
                # try again with next version of name
413 2085 aaronmk
414 2120 aaronmk
order_by_pkey = object() # tells mk_select() to order by the pkey
415
416 2127 aaronmk
join_using = object() # tells mk_select() to join the column with USING
417
418 2187 aaronmk
filter_out = object() # tells mk_select() to filter out rows that match the join
419 2180 aaronmk
420 2199 aaronmk
distinct_on_all = object() # tells mk_select() to SELECT DISTINCT ON all columns
421
422 2233 aaronmk
def mk_select(db, tables, fields=None, conds=None, distinct_on=[], limit=None,
423 2227 aaronmk
    start=None, order_by=order_by_pkey, table_is_esc=False, default_table=None):
424 1981 aaronmk
    '''
425 2121 aaronmk
    @param tables The single table to select from, or a list of tables to join
426 2187 aaronmk
        together: [table0, (table1, joins), ...]
427
428
        joins has the format: dict(right_col=left_col, ...)
429
        * if left_col is join_using, left_col is set to right_col
430
        * if left_col is filter_out, the tables are LEFT JOINed together and the
431
          query is filtered by `right_col IS NULL` (indicating no match)
432 1981 aaronmk
    @param fields Use None to select all fields in the table
433 2218 aaronmk
    @param conds WHERE conditions: dict(compare_left_side=compare_right_side):
434
        * compare_left_side: Code|str (for col name)
435
        * compare_right_side: ValueCond|literal value
436 2199 aaronmk
    @param distinct_on The columns to SELECT DISTINCT ON, or distinct_on_all to
437
        use all columns
438 1981 aaronmk
    @param table_is_esc Whether the table name has already been escaped
439 2054 aaronmk
    @return tuple(query, params)
440 1981 aaronmk
    '''
441 2060 aaronmk
    def esc_name_(name): return esc_name(db, name)
442 2058 aaronmk
443 2121 aaronmk
    if not lists.is_seq(tables): tables = [tables]
444 2141 aaronmk
    tables = list(tables) # don't modify input! (list() copies input)
445 2121 aaronmk
    table0 = tables.pop(0) # first table is separate
446
447 1135 aaronmk
    if conds == None: conds = {}
448 135 aaronmk
    assert limit == None or type(limit) == int
449 865 aaronmk
    assert start == None or type(start) == int
450 2254 aaronmk
    if order_by is order_by_pkey:
451 2121 aaronmk
        order_by = pkey(db, table0, recover=True, table_is_esc=table_is_esc)
452
    if not table_is_esc: table0 = esc_name_(table0)
453 865 aaronmk
454 2056 aaronmk
    params = []
455
456 2227 aaronmk
    def parse_col(field):
457 2056 aaronmk
        '''Parses fields'''
458 2227 aaronmk
        return sql_gen.value2sql_gen(field, default_table, table_is_esc,
459
            assume_col=True).to_str(db)
460 11 aaronmk
    def cond(entry):
461 2056 aaronmk
        '''Parses conditions'''
462 2217 aaronmk
        left, right = entry
463 2226 aaronmk
        return sql_gen.cond2sql_gen(right).to_str(db, sql_gen.col2sql_gen(left))
464 2056 aaronmk
465 2200 aaronmk
    query = 'SELECT'
466
467
    # DISTINCT ON columns
468 2233 aaronmk
    if distinct_on != []:
469 2200 aaronmk
        query += ' DISTINCT'
470 2254 aaronmk
        if distinct_on is not distinct_on_all:
471 2200 aaronmk
            query += ' ON ('+(', '.join(map(parse_col, distinct_on)))+')'
472
473
    # Columns
474
    query += ' '
475 1135 aaronmk
    if fields == None: query += '*'
476 2056 aaronmk
    else: query += ', '.join(map(parse_col, fields))
477 2200 aaronmk
478
    # Main table
479 2121 aaronmk
    query += ' FROM '+table0
480 865 aaronmk
481 2122 aaronmk
    # Add joins
482 2263 aaronmk
    tables = [sql_gen.join2sql_gen(v, table_is_esc) for v in tables]
483
    left_table = sql_gen.table2sql_gen(table0, table_is_esc)
484
    for join_ in tables:
485
        table = join_.table
486
        joins = join_.mapping
487 2122 aaronmk
488 2238 aaronmk
        left_join = False
489 2187 aaronmk
490 2238 aaronmk
        # Parse special values
491 2255 aaronmk
        joins = joins.copy() # don't modify input!
492 2238 aaronmk
        for right, left in joins.items():
493 2254 aaronmk
            if left is filter_out:
494 2238 aaronmk
                left_join = True
495
                # Can't use join_using because the columns being joined can have
496
                # different values in a LEFT JOIN.
497
                # Use '~=' because it's OK to pass NULLs through.
498
                joins[right] = sql_gen.CompareCond(sql_gen.col2sql_gen(right,
499 2263 aaronmk
                    left_table), '~=')
500 2238 aaronmk
                conds[(table, right)] = None # filter query by no match
501
502 2122 aaronmk
        def join(entry):
503 2127 aaronmk
            '''Parses non-USING joins'''
504 2124 aaronmk
            right_col, left_col = entry
505 2173 aaronmk
506 2179 aaronmk
            # Parse special values
507 2176 aaronmk
            if left_col == None: left_col = (left_col,)
508
                # for None values, tuple is optional
509 2254 aaronmk
            elif left_col is join_using: left_col = right_col
510 2179 aaronmk
511 2263 aaronmk
            return cond((sql_gen.col2sql_gen(right_col, table),
512
                sql_gen.cond2sql_gen(left_col, left_table, assume_col=True)))
513 2122 aaronmk
514 2187 aaronmk
        # Create join condition and determine join type
515 2254 aaronmk
        if reduce(operator.and_, (v is join_using for v in joins.itervalues())):
516 2179 aaronmk
            # all cols w/ USING, so can use simpler USING syntax
517 2187 aaronmk
            join_cond = 'USING ('+(', '.join(joins.iterkeys()))+')'
518
        else: join_cond = 'ON '+(' AND '.join(map(join, joins.iteritems())))
519 2127 aaronmk
520 2187 aaronmk
        # Create join
521 2238 aaronmk
        if left_join: query += ' LEFT'
522 2263 aaronmk
        query += ' JOIN '+table.to_str(db)+' '+join_cond
523 2187 aaronmk
524 2122 aaronmk
        left_table = table
525
526 865 aaronmk
    missing = True
527 89 aaronmk
    if conds != {}:
528 2122 aaronmk
        query += ' WHERE '+(' AND '.join(map(cond, conds.iteritems())))
529 2056 aaronmk
        params += conds.values()
530 865 aaronmk
        missing = False
531 2227 aaronmk
    if order_by != None:
532
        query += ' ORDER BY '+sql_gen.col2sql_gen(order_by, table0,
533
            table_is_esc).to_str(db)
534 865 aaronmk
    if limit != None: query += ' LIMIT '+str(limit); missing = False
535
    if start != None:
536
        if start != 0: query += ' OFFSET '+str(start)
537
        missing = False
538
    if missing: warnings.warn(DbWarning(
539
        'SELECT statement missing a WHERE, LIMIT, or OFFSET clause: '+query))
540
541 2056 aaronmk
    return (query, params)
542 11 aaronmk
543 2054 aaronmk
def select(db, *args, **kw_args):
544
    '''For params, see mk_select() and run_query()'''
545
    recover = kw_args.pop('recover', None)
546
    cacheable = kw_args.pop('cacheable', True)
547
548
    query, params = mk_select(db, *args, **kw_args)
549
    return run_query(db, query, params, recover, cacheable)
550
551 2066 aaronmk
def mk_insert_select(db, table, cols=None, select_query=None, params=None,
552 2070 aaronmk
    returning=None, embeddable=False, table_is_esc=False):
553 1960 aaronmk
    '''
554
    @param returning str|None An inserted column (such as pkey) to return
555 2070 aaronmk
    @param embeddable Whether the query should be embeddable as a nested SELECT.
556 2073 aaronmk
        Warning: If you set this and cacheable=True when the query is run, the
557
        query will be fully cached, not just if it raises an exception.
558 1960 aaronmk
    @param table_is_esc Whether the table name has already been escaped
559
    '''
560 2063 aaronmk
    if select_query == None: select_query = 'DEFAULT VALUES'
561
    if cols == []: cols = None # no cols (all defaults) = unknown col names
562 1960 aaronmk
    if not table_is_esc: check_name(table)
563 2063 aaronmk
564
    # Build query
565
    query = 'INSERT INTO '+table
566
    if cols != None:
567
        map(check_name, cols)
568
        query += ' ('+', '.join(cols)+')'
569
    query += ' '+select_query
570
571
    if returning != None:
572
        check_name(returning)
573
        query += ' RETURNING '+returning
574
575 2070 aaronmk
    if embeddable:
576
        # Create function
577 2189 aaronmk
        function_name = '_'.join(map(clean_name, ['insert', table] + cols))
578 2070 aaronmk
        return_type = 'SETOF '+table+'.'+returning+'%TYPE'
579 2189 aaronmk
        while True:
580
            try:
581 2194 aaronmk
                function = function_name
582
                if not db.debug: function = 'pg_temp.'+function
583
584 2189 aaronmk
                function_query = '''\
585
CREATE FUNCTION '''+function+'''() RETURNS '''+return_type+'''
586 2070 aaronmk
    LANGUAGE sql
587
    AS $$'''+mogrify(db, query, params)+''';$$;
588
'''
589 2189 aaronmk
                run_query(db, function_query, recover=True, cacheable=True)
590
                break # this version was successful
591
            except DuplicateFunctionException, e:
592
                function_name = next_version(function_name)
593
                # try again with next version of name
594 2070 aaronmk
595
        # Return query that uses function
596 2134 aaronmk
        return mk_select(db, function+'() AS f ('+returning+')', start=0,
597
            order_by=None, table_is_esc=True)# AS clause requires function alias
598 2070 aaronmk
599 2066 aaronmk
    return (query, params)
600
601
def insert_select(db, *args, **kw_args):
602 2085 aaronmk
    '''For params, see mk_insert_select() and run_query_into()
603 2152 aaronmk
    @param into_ref List with name of temp table to place RETURNING values in
604 2072 aaronmk
    '''
605 2151 aaronmk
    into_ref = kw_args.pop('into_ref', None)
606
    if into_ref != None: kw_args['embeddable'] = True
607 2066 aaronmk
    recover = kw_args.pop('recover', None)
608
    cacheable = kw_args.pop('cacheable', True)
609
610
    query, params = mk_insert_select(db, *args, **kw_args)
611 2153 aaronmk
    return run_query_into(db, query, params, into_ref, recover=recover,
612
        cacheable=cacheable)
613 2063 aaronmk
614 2066 aaronmk
default = object() # tells insert() to use the default value for a column
615
616 2063 aaronmk
def insert(db, table, row, *args, **kw_args):
617 2085 aaronmk
    '''For params, see insert_select()'''
618 1960 aaronmk
    if lists.is_seq(row): cols = None
619
    else:
620
        cols = row.keys()
621
        row = row.values()
622
    row = list(row) # ensure that "!= []" works
623
624 1961 aaronmk
    # Check for special values
625
    labels = []
626
    values = []
627
    for value in row:
628 2254 aaronmk
        if value is default: labels.append('DEFAULT')
629 1961 aaronmk
        else:
630
            labels.append('%s')
631
            values.append(value)
632
633
    # Build query
634 2063 aaronmk
    if values != []: query = ' VALUES ('+(', '.join(labels))+')'
635
    else: query = None
636 1554 aaronmk
637 2064 aaronmk
    return insert_select(db, table, cols, query, values, *args, **kw_args)
638 11 aaronmk
639 135 aaronmk
def last_insert_id(db):
640 1849 aaronmk
    module = util.root_module(db.db)
641 135 aaronmk
    if module == 'psycopg2': return value(run_query(db, 'SELECT lastval()'))
642
    elif module == 'MySQLdb': return db.insert_id()
643
    else: return None
644 13 aaronmk
645 1968 aaronmk
def truncate(db, table, schema='public'):
646
    return run_query(db, 'TRUNCATE '+qual_name(db, schema, table)+' CASCADE')
647 832 aaronmk
648
##### Database structure queries
649
650 2084 aaronmk
def pkey(db, table, recover=None, table_is_esc=False):
651 832 aaronmk
    '''Assumed to be first column in table'''
652 2120 aaronmk
    return col_names(select(db, table, limit=0, order_by=None, recover=recover,
653 2084 aaronmk
        table_is_esc=table_is_esc)).next()
654 832 aaronmk
655 853 aaronmk
def index_cols(db, table, index):
656
    '''Can also use this for UNIQUE constraints, because a UNIQUE index is
657
    automatically created. When you don't know whether something is a UNIQUE
658
    constraint or a UNIQUE index, use this function.'''
659
    check_name(table)
660
    check_name(index)
661 1909 aaronmk
    module = util.root_module(db.db)
662
    if module == 'psycopg2':
663
        return list(values(run_query(db, '''\
664 853 aaronmk
SELECT attname
665 866 aaronmk
FROM
666
(
667
        SELECT attnum, attname
668
        FROM pg_index
669
        JOIN pg_class index ON index.oid = indexrelid
670
        JOIN pg_class table_ ON table_.oid = indrelid
671
        JOIN pg_attribute ON attrelid = indrelid AND attnum = ANY (indkey)
672
        WHERE
673
            table_.relname = %(table)s
674
            AND index.relname = %(index)s
675
    UNION
676
        SELECT attnum, attname
677
        FROM
678
        (
679
            SELECT
680
                indrelid
681
                , (regexp_matches(indexprs, E':varattno (\\\\d+)', 'g'))[1]::int
682
                    AS indkey
683
            FROM pg_index
684
            JOIN pg_class index ON index.oid = indexrelid
685
            JOIN pg_class table_ ON table_.oid = indrelid
686
            WHERE
687
                table_.relname = %(table)s
688
                AND index.relname = %(index)s
689
        ) s
690
        JOIN pg_attribute ON attrelid = indrelid AND attnum = indkey
691
) s
692 853 aaronmk
ORDER BY attnum
693
''',
694 1909 aaronmk
            {'table': table, 'index': index}, cacheable=True)))
695
    else: raise NotImplementedError("Can't list index columns for "+module+
696
        ' database')
697 853 aaronmk
698 464 aaronmk
def constraint_cols(db, table, constraint):
699
    check_name(table)
700
    check_name(constraint)
701 1849 aaronmk
    module = util.root_module(db.db)
702 464 aaronmk
    if module == 'psycopg2':
703
        return list(values(run_query(db, '''\
704
SELECT attname
705
FROM pg_constraint
706
JOIN pg_class ON pg_class.oid = conrelid
707
JOIN pg_attribute ON attrelid = conrelid AND attnum = ANY (conkey)
708
WHERE
709
    relname = %(table)s
710
    AND conname = %(constraint)s
711
ORDER BY attnum
712
''',
713
            {'table': table, 'constraint': constraint})))
714
    else: raise NotImplementedError("Can't list constraint columns for "+module+
715
        ' database')
716
717 2096 aaronmk
row_num_col = '_row_num'
718
719 2086 aaronmk
def add_row_num(db, table):
720 2117 aaronmk
    '''Adds a row number column to a table. Its name is in row_num_col. It will
721
    be the primary key.'''
722 2086 aaronmk
    check_name(table)
723 2096 aaronmk
    run_query(db, 'ALTER TABLE '+table+' ADD COLUMN '+row_num_col
724 2117 aaronmk
        +' serial NOT NULL PRIMARY KEY')
725 2086 aaronmk
726 1968 aaronmk
def tables(db, schema='public', table_like='%'):
727 1849 aaronmk
    module = util.root_module(db.db)
728 1968 aaronmk
    params = {'schema': schema, 'table_like': table_like}
729 832 aaronmk
    if module == 'psycopg2':
730 1968 aaronmk
        return values(run_query(db, '''\
731
SELECT tablename
732
FROM pg_tables
733
WHERE
734
    schemaname = %(schema)s
735
    AND tablename LIKE %(table_like)s
736
ORDER BY tablename
737
''',
738
            params, cacheable=True))
739
    elif module == 'MySQLdb':
740
        return values(run_query(db, 'SHOW TABLES LIKE %(table_like)s', params,
741
            cacheable=True))
742 832 aaronmk
    else: raise NotImplementedError("Can't list tables for "+module+' database')
743 830 aaronmk
744 833 aaronmk
##### Database management
745
746 1968 aaronmk
def empty_db(db, schema='public', **kw_args):
747
    '''For kw_args, see tables()'''
748
    for table in tables(db, schema, **kw_args): truncate(db, table, schema)
749 833 aaronmk
750 832 aaronmk
##### Heuristic queries
751
752 2104 aaronmk
def put(db, table, row, pkey_=None, row_ct_ref=None):
753 1554 aaronmk
    '''Recovers from errors.
754 2077 aaronmk
    Only works under PostgreSQL (uses INSERT RETURNING).
755
    '''
756 2104 aaronmk
    if pkey_ == None: pkey_ = pkey(db, table, recover=True)
757
758 471 aaronmk
    try:
759 2149 aaronmk
        cur = insert(db, table, row, pkey_, recover=True)
760 1554 aaronmk
        if row_ct_ref != None and cur.rowcount >= 0:
761
            row_ct_ref[0] += cur.rowcount
762
        return value(cur)
763 471 aaronmk
    except DuplicateKeyException, e:
764 2104 aaronmk
        return value(select(db, table, [pkey_],
765 1069 aaronmk
            util.dict_subset_right_join(row, e.cols), recover=True))
766 471 aaronmk
767 473 aaronmk
def get(db, table, row, pkey, row_ct_ref=None, create=False):
768 830 aaronmk
    '''Recovers from errors'''
769 2209 aaronmk
    try: return value(select(db, table, [pkey], row, limit=1, recover=True))
770 14 aaronmk
    except StopIteration:
771 40 aaronmk
        if not create: raise
772 471 aaronmk
        return put(db, table, row, pkey, row_ct_ref) # insert new row
773 2078 aaronmk
774 2134 aaronmk
def put_table(db, out_table, in_tables, mapping, limit=None, start=0,
775
    row_ct_ref=None, table_is_esc=False):
776 2078 aaronmk
    '''Recovers from errors.
777
    Only works under PostgreSQL (uses INSERT RETURNING).
778 2131 aaronmk
    @param in_tables The main input table to select from, followed by a list of
779
        tables to join with it using the main input table's pkey
780 2133 aaronmk
    @return (table, col) Where the pkeys (from INSERT RETURNING) are made
781 2078 aaronmk
        available
782
    '''
783 2251 aaronmk
    temp_prefix = clean_name(out_table)
784
    pkeys_ref = [temp_prefix+'_pkeys']
785 2131 aaronmk
786 2132 aaronmk
    # Join together input tables
787 2131 aaronmk
    in_tables = in_tables[:] # don't modify input!
788
    in_tables0 = in_tables.pop(0) # first table is separate
789
    in_pkey = pkey(db, in_tables0, recover=True, table_is_esc=table_is_esc)
790 2236 aaronmk
    in_pkey_col = sql_gen.col2sql_gen(in_pkey, in_tables0, table_is_esc)
791 2178 aaronmk
    insert_joins = [in_tables0]+[(t, {in_pkey: join_using}) for t in in_tables]
792 2131 aaronmk
793 2142 aaronmk
    out_pkey = pkey(db, out_table, recover=True, table_is_esc=table_is_esc)
794 2236 aaronmk
    out_pkey_col = sql_gen.col2sql_gen(out_pkey, out_table, table_is_esc)
795 2142 aaronmk
796 2236 aaronmk
    pkeys = [in_pkey, out_pkey]
797
    pkeys_cols = [in_pkey_col, out_pkey_col]
798
799 2201 aaronmk
    pkeys_table_exists_ref = [False]
800
    def run_query_into_pkeys(query, params):
801
        if pkeys_table_exists_ref[0]:
802 2236 aaronmk
            insert_select(db, pkeys_ref[0], pkeys, query, params)
803 2201 aaronmk
        else:
804
            run_query_into(db, query, params, into_ref=pkeys_ref)
805
            pkeys_table_exists_ref[0] = True
806
807 2208 aaronmk
    conds = {}
808 2233 aaronmk
    distinct_on = []
809 2207 aaronmk
    def mk_main_select(cols):
810 2208 aaronmk
        return mk_select(db, insert_joins, cols, conds, distinct_on,
811 2227 aaronmk
            order_by=None, limit=limit, start=start, table_is_esc=table_is_esc,
812
            default_table=in_tables0)
813 2132 aaronmk
814 2245 aaronmk
    def log_ignore(in_col, value):
815
        db.log_debug('Ignoring rows with '+in_col+' = '+value)
816
817 2206 aaronmk
    # Do inserts and selects
818 2257 aaronmk
    join_cols = {}
819
    filter_join_added = False
820 2251 aaronmk
    out_pkeys_ref = [temp_prefix+'_out_pkeys']
821 2252 aaronmk
    in_pkeys_ref = [temp_prefix+'_in_pkeys']
822 2206 aaronmk
    while True:
823
        try:
824 2257 aaronmk
            if join_cols != {}:
825
                select_joins = insert_joins + [(out_table, join_cols)]
826
                db.log_debug('Getting pkeys of already existing rows')
827
                run_query_into_pkeys(*mk_select(db, select_joins, pkeys_cols,
828
                    order_by=None, start=0, table_is_esc=table_is_esc))
829
830
                # Prepare to insert new rows
831
                input_join_cols = join_cols.values()
832
                distinct_on = filter(util.is_str, input_join_cols)
833
                if not filter_join_added:
834
                    insert_joins.append((pkeys_ref[0], {in_pkey: filter_out}))
835
                    filter_join_added = True
836
837 2253 aaronmk
            db.log_debug('Inserting new rows')
838 2206 aaronmk
            cur = insert_select(db, out_table, mapping.keys(),
839 2207 aaronmk
                *mk_main_select(mapping.values()), returning=out_pkey,
840 2206 aaronmk
                into_ref=out_pkeys_ref, recover=True, table_is_esc=table_is_esc)
841
            if row_ct_ref != None and cur.rowcount >= 0:
842
                row_ct_ref[0] += cur.rowcount
843
                add_row_num(db, out_pkeys_ref[0]) # for joining with input pkeys
844
845 2253 aaronmk
            db.log_debug('Getting input pkeys corresponding to rows in insert')
846 2207 aaronmk
            run_query_into(db, *mk_main_select([in_pkey]),
847
                into_ref=in_pkeys_ref)
848 2206 aaronmk
            add_row_num(db, in_pkeys_ref[0]) # for joining with output pkeys
849
850 2253 aaronmk
            db.log_debug('Joining together output and input pkeys')
851 2206 aaronmk
            run_query_into_pkeys(*mk_select(db, [in_pkeys_ref[0],
852 2236 aaronmk
                (out_pkeys_ref[0], {row_num_col: join_using})], pkeys, start=0))
853 2206 aaronmk
854
            break # insert successful
855
        except DuplicateKeyException, e:
856 2258 aaronmk
            old_join_cols = join_cols.copy()
857 2257 aaronmk
            join_cols.update(util.dict_subset_right_join(mapping, e.cols))
858 2245 aaronmk
            db.log_debug('Ignoring existing rows, comparing on '+str(join_cols))
859 2258 aaronmk
            assert join_cols != old_join_cols # avoid infinite loops
860 2206 aaronmk
            # rerun loop with additional constraints
861 2230 aaronmk
        except NullValueException, e:
862
            out_col, = e.cols
863
            try: in_col = mapping[out_col]
864
            except KeyError: # no mapping for missing col, so every row invalid
865 2245 aaronmk
                db.log_debug('Missing mapping for '+out_col)
866 2230 aaronmk
                run_query_into_pkeys(*mk_select(db, insert_joins,
867 2236 aaronmk
                    [in_pkey_col, sql_gen.NamedCode(out_pkey, None)],
868 2230 aaronmk
                    order_by=None, start=0, table_is_esc=table_is_esc))
869
                break
870
            else:
871 2245 aaronmk
                log_ignore(in_col, 'NULL')
872 2230 aaronmk
                conds[in_col] = sql_gen.CompareCond(None, '!=')
873
                # rerun loop with additional constraints
874 2243 aaronmk
        except FunctionValueException, e:
875
            assert e.name == sql_gen.unescape_table(out_table)
876
            out_col = 'value' # assume function param was named "value"
877
            in_col = mapping[out_col]
878 2245 aaronmk
            value = e.value
879
            log_ignore(in_col, value)
880
            conds[in_col] = sql_gen.CompareCond(value, '!=')
881 2243 aaronmk
            # rerun loop with additional constraints
882 2132 aaronmk
883 2154 aaronmk
    return (pkeys_ref[0], out_pkey)
884 2115 aaronmk
885
##### Data cleanup
886
887
def cleanup_table(db, table, cols, table_is_esc=False):
888
    def esc_name_(name): return esc_name(db, name)
889
890
    if not table_is_esc: check_name(table)
891
    cols = map(esc_name_, cols)
892
893
    run_query(db, 'UPDATE '+table+' SET\n'+(',\n'.join(('\n'+col
894
        +' = nullif(nullif(trim(both from '+col+"), %(null0)s), %(null1)s)"
895
            for col in cols))),
896
        dict(null0='', null1=r'\N'))