22 |
import weakref |
import weakref |
23 |
from sqlite import connect |
from sqlite import connect |
24 |
|
|
25 |
|
from base import TitledObject |
26 |
|
|
27 |
import table |
import table |
28 |
|
|
29 |
sql_type_map = { |
sql_type_map = { |
87 |
self.internal_name = internal_name |
self.internal_name = internal_name |
88 |
|
|
89 |
|
|
90 |
class TransientTableBase: |
class TransientTableBase(table.OldTableInterfaceMixin): |
91 |
|
|
92 |
"""Base class for tables in the transient database""" |
"""Base class for tables in the transient database""" |
93 |
|
|
106 |
self.orig_names = [] |
self.orig_names = [] |
107 |
self.internal_to_orig = {} |
self.internal_to_orig = {} |
108 |
self.orig_to_internal = {} |
self.orig_to_internal = {} |
109 |
|
self.column_map = {} |
110 |
|
|
111 |
# Create the column objects and fill various maps and lists |
# Create the column objects and fill various maps and lists |
112 |
for col in self.columns: |
for index in range(len(self.columns)): |
113 |
|
col = self.columns[index] |
114 |
self.name_to_column[col.name] = col |
self.name_to_column[col.name] = col |
115 |
self.orig_names.append(col.name) |
self.orig_names.append(col.name) |
116 |
self.internal_to_orig[col.internal_name] = col.name |
self.internal_to_orig[col.internal_name] = col.name |
117 |
self.orig_to_internal[col.name] = col.internal_name |
self.orig_to_internal[col.name] = col.internal_name |
118 |
|
self.column_map[col.name] = col |
119 |
|
self.column_map[index] = col |
120 |
|
|
121 |
# Build the CREATE TABLE statement and create the table in the |
# Build the CREATE TABLE statement and create the table in the |
122 |
# database |
# database |
123 |
table_types = [] |
table_types = ["id INTEGER PRIMARY KEY"] |
124 |
for col in self.columns: |
for col in self.columns: |
125 |
table_types.append("%s %s" % (col.internal_name, |
table_types.append("%s %s" % (col.internal_name, |
126 |
sql_type_map[col.type])) |
sql_type_map[col.type])) |
144 |
self.db.execute(stmt) |
self.db.execute(stmt) |
145 |
self.indexed_columns[column] = 1 |
self.indexed_columns[column] = 1 |
146 |
|
|
147 |
def field_count(self): |
def NumColumns(self): |
148 |
return len(self.columns) |
return len(self.columns) |
149 |
|
|
150 |
def field_info(self, i): |
def NumRows(self): |
|
col = self.columns[i] |
|
|
return col.type, col.name, 0, 0 |
|
|
|
|
|
def field_info_by_name(self, name): |
|
|
for col in self.columns: |
|
|
if col.name == name: |
|
|
return col.type, col.name, 0, 0 |
|
|
else: |
|
|
return None |
|
|
|
|
|
def record_count(self): |
|
151 |
result = self.db.execute("SELECT count(*) FROM %s;" % self.tablename) |
result = self.db.execute("SELECT count(*) FROM %s;" % self.tablename) |
152 |
return int(result[0]) |
return int(result[0]) |
153 |
|
|
154 |
def read_record(self, index): |
def Columns(self): |
155 |
|
return self.columns |
156 |
|
|
157 |
|
def Column(self, col): |
158 |
|
return self.column_map[col] |
159 |
|
|
160 |
|
def HasColumn(self, col): |
161 |
|
"""Return whether the table has a column with the given name or index |
162 |
|
""" |
163 |
|
return self.column_map.has_key(col) |
164 |
|
|
165 |
|
def ReadRowAsDict(self, index): |
166 |
|
# Implementation Strategy: Executing a completely new select |
167 |
|
# statement every time this method is called is too slow. The |
168 |
|
# most important usage is to read the records more or less |
169 |
|
# sequentially. This happens e.g. when drawing a layer with a |
170 |
|
# classification where the shapes are drawn in order of the |
171 |
|
# shape ids. Another pattern is that the same row is requested |
172 |
|
# several times in a row. This happens in the table view, for |
173 |
|
# instance. |
174 |
|
|
175 |
|
# We can exploit this to make access faster by having one cursor |
176 |
|
# open all the time and keeping the last row read around in case |
177 |
|
# the same row is accessed again the next time and if the row |
178 |
|
# index is larger than the row we have read last we simply fetch |
179 |
|
# rows from the cursor until we've reached the requested row. If |
180 |
|
# the requested row index is smaller then we start a new cursor. |
181 |
|
|
182 |
|
# FIXME: So far this scheme seems to work well enough. Obvious |
183 |
|
# improvements would be to start the cursor at exactly the |
184 |
|
# requested row (should be efficient and easy to do now that the |
185 |
|
# id is the primary key) and to perhaps to also start a new |
186 |
|
# cursor if the requested index is much larger than the last row |
187 |
|
# so that we don't read and discard lots of the rows. |
188 |
|
|
189 |
|
# Check whether we have to start a new cursor |
190 |
if self.read_record_cursor is None or index <self.read_record_last_row: |
if self.read_record_cursor is None or index <self.read_record_last_row: |
191 |
stmt = ("SELECT %s FROM %s;" |
stmt = ("SELECT %s FROM %s;" |
192 |
% (", ".join([c.internal_name for c in self.columns]), |
% (", ".join([c.internal_name for c in self.columns]), |
202 |
assert index >= self.read_record_last_row |
assert index >= self.read_record_last_row |
203 |
|
|
204 |
if index == self.read_record_last_row: |
if index == self.read_record_last_row: |
205 |
result = self.read_record_last_result |
result = self.read_record_last_result |
206 |
else: |
else: |
207 |
for i in range(index - self.read_record_last_row): |
for i in range(index - self.read_record_last_row): |
208 |
result = self.read_record_cursor.fetchone() |
result = self.read_record_cursor.fetchone() |
210 |
self.read_record_last_row = index |
self.read_record_last_row = index |
211 |
return dict(zip(self.orig_names, result)) |
return dict(zip(self.orig_names, result)) |
212 |
|
|
213 |
def field_range(self, colname): |
def ReadValue(self, row, col): |
214 |
col = self.name_to_column[colname] |
"""Return the value of the specified row and column |
215 |
|
|
216 |
|
The col parameter may be the index of the column or its name. |
217 |
|
""" |
218 |
|
# Depending on the actual access patterns of the table data, it |
219 |
|
# might be a bit faster in some circumstances to not implement |
220 |
|
# this via ReadRowAsDict, but this simple implementation should |
221 |
|
# be fast enough for most purposes. |
222 |
|
return self.ReadRowAsDict(row)[self.column_map[col].name] |
223 |
|
|
224 |
|
def ValueRange(self, col): |
225 |
|
col = self.column_map[col] |
226 |
iname = col.internal_name |
iname = col.internal_name |
227 |
min, max = self.db.execute("SELECT min(%s), max(%s) FROM %s;" |
min, max = self.db.execute("SELECT min(%s), max(%s) FROM %s;" |
228 |
% (iname, iname, self.tablename)) |
% (iname, iname, self.tablename)) |
229 |
converter = type_converter_map[col.type] |
converter = type_converter_map[col.type] |
230 |
return ((converter(min), None), (converter(max), None)) |
return (converter(min), converter(max)) |
231 |
|
|
232 |
def GetUniqueValues(self, colname): |
def UniqueValues(self, col): |
233 |
iname = self.orig_to_internal[colname] |
iname = self.column_map[col].internal_name |
234 |
cursor = self.db.cursor() |
cursor = self.db.cursor() |
235 |
cursor.execute("SELECT %s FROM %s GROUP BY %s;" |
cursor.execute("SELECT %s FROM %s GROUP BY %s;" |
236 |
% (iname, self.tablename, iname)) |
% (iname, self.tablename, iname)) |
242 |
result.append(row[0]) |
result.append(row[0]) |
243 |
return result |
return result |
244 |
|
|
245 |
|
def SimpleQuery(self, left, comparison, right): |
246 |
|
"""Return the indices of all rows that matching a condition. |
247 |
|
|
248 |
|
Parameters: |
249 |
|
left -- The column object for the left side of the comparison |
250 |
|
|
251 |
class TransientTable(TransientTableBase): |
comparison -- The comparison operator as a string. It must be |
252 |
|
one of '==', '!=', '<', '<=', '>=', '>' |
253 |
|
|
254 |
|
right -- The right hand side of the comparison. It must be |
255 |
|
either a column object or a value, i.e. a string, |
256 |
|
int or float. |
257 |
|
|
258 |
|
The return value is a sorted list of the indices of the rows |
259 |
|
where the condition is true. |
260 |
|
""" |
261 |
|
if comparison not in ("==", "!=", "<", "<=", ">=", ">"): |
262 |
|
raise ValueError("Comparison operator %r not allowed" % comparison) |
263 |
|
|
264 |
|
if hasattr(right, "internal_name"): |
265 |
|
right_template = right.internal_name |
266 |
|
params = () |
267 |
|
else: |
268 |
|
right_template = "%s" |
269 |
|
params = (right,) |
270 |
|
|
271 |
|
query = "SELECT id FROM %s WHERE %s %s %s ORDER BY id;" \ |
272 |
|
% (self.tablename, left.internal_name, comparison, |
273 |
|
right_template) |
274 |
|
|
275 |
|
cursor = self.db.cursor() |
276 |
|
cursor.execute(query, params) |
277 |
|
result = [] |
278 |
|
while 1: |
279 |
|
row = cursor.fetchone() |
280 |
|
if row is None: |
281 |
|
break |
282 |
|
result.append(row[0]) |
283 |
|
return result |
284 |
|
|
285 |
|
def Dependencies(self): |
286 |
|
"""Placeholder for a method in a derived class. |
287 |
|
|
288 |
|
Return a sequence with the tables and other data objects that |
289 |
|
self depends on. |
290 |
|
""" |
291 |
|
raise NotImplementedError |
292 |
|
|
293 |
|
|
294 |
|
class TransientTable(TitledObject, TransientTableBase): |
295 |
|
|
296 |
"""A Table in a transient DB that starts as the copy of a Thuban Table.""" |
"""A Table in a transient DB that starts as the copy of a Thuban Table.""" |
297 |
|
|
302 |
interface. |
interface. |
303 |
""" |
""" |
304 |
TransientTableBase.__init__(self, transient_db) |
TransientTableBase.__init__(self, transient_db) |
305 |
|
TitledObject.__init__(self, table.Title()) |
306 |
self.create(table) |
self.create(table) |
307 |
|
|
308 |
def create(self, table): |
def create(self, table): |
309 |
columns = [] |
columns = [] |
310 |
for i in range(table.field_count()): |
for col in table.Columns(): |
311 |
type, name = table.field_info(i)[:2] |
columns.append(ColumnReference(col.name, col.type, |
|
columns.append(ColumnReference(name, type, |
|
312 |
self.db.new_column_name())) |
self.db.new_column_name())) |
313 |
TransientTableBase.create(self, columns) |
TransientTableBase.create(self, columns) |
314 |
|
|
315 |
# copy the input table to the transient db |
# copy the input table to the transient db |
316 |
insert_template = "INSERT INTO %s (%s) VALUES (%s);" \ |
|
317 |
|
# A key to insert to use for the formatting of the insert |
318 |
|
# statement. The key must not be equal to any of the column |
319 |
|
# names so we construct one by building a string of x's that is |
320 |
|
# longer than any of the column names |
321 |
|
id_key = max([len(col.name) for col in self.columns]) * "x" |
322 |
|
|
323 |
|
insert_template = "INSERT INTO %s (id, %s) VALUES (%%(%s)s, %s);" \ |
324 |
% (self.tablename, |
% (self.tablename, |
325 |
", ".join([col.internal_name |
", ".join([col.internal_name |
326 |
for col in self.columns]), |
for col in self.columns]), |
327 |
|
id_key, |
328 |
", ".join(["%%(%s)s" % col.name |
", ".join(["%%(%s)s" % col.name |
329 |
for col in self.columns])) |
for col in self.columns])) |
330 |
cursor = self.db.cursor() |
cursor = self.db.cursor() |
331 |
for i in range(table.record_count()): |
for i in range(table.NumRows()): |
332 |
cursor.execute(insert_template, table.read_record(i)) |
row = table.ReadRowAsDict(i) |
333 |
|
row[id_key] = i |
334 |
|
cursor.execute(insert_template, row) |
335 |
self.db.conn.commit() |
self.db.conn.commit() |
336 |
|
|
337 |
|
|
338 |
|
|
339 |
class TransientJoinedTable(TransientTableBase): |
class TransientJoinedTable(TitledObject, TransientTableBase): |
340 |
|
|
341 |
"""A Table in the transient DB that contains a join of two tables""" |
"""A Table in the transient DB that contains a join of two tables""" |
342 |
|
|
343 |
def __init__(self, transient_db, left_table, left_field, |
def __init__(self, transient_db, left_table, left_field, |
344 |
right_table, right_field = None): |
right_table, right_field = None, outer_join = False): |
345 |
"""Create a new table in the transient DB as a join of two tables. |
"""Create a new table in the transient DB as a join of two tables. |
346 |
|
|
347 |
Both input tables, left_table and right_table must have a |
Both input tables, left_table and right_table must have a |
348 |
transient_table method that returns a table object for a table |
transient_table method that returns a table object for a table |
349 |
in the trnsient database. The join is performed on the condition |
in the transient database. The join is performed on the condition |
350 |
that the value of the left_field column the the left table is |
that the value of the left_field column the the left table is |
351 |
equal to the value of the right_field in the right_table. |
equal to the value of the right_field in the right_table. |
352 |
|
|
358 |
names of the input tables without having to create prefixes. |
names of the input tables without having to create prefixes. |
359 |
""" |
""" |
360 |
TransientTableBase.__init__(self, transient_db) |
TransientTableBase.__init__(self, transient_db) |
361 |
|
self.dependencies = (left_table, right_table) |
362 |
self.left_table = left_table.transient_table() |
self.left_table = left_table.transient_table() |
363 |
self.left_field = left_field |
self.left_field = left_field |
364 |
self.right_table = right_table.transient_table() |
self.right_table = right_table.transient_table() |
366 |
self.right_field = right_field |
self.right_field = right_field |
367 |
else: |
else: |
368 |
self.right_field = self.left_field |
self.right_field = self.left_field |
369 |
|
self.outer_join = outer_join |
370 |
|
|
371 |
|
title = "Join of %(left)s and %(right)s" \ |
372 |
|
% {"left": self.left_table.Title(), |
373 |
|
"right": self.right_table.Title()} |
374 |
|
TitledObject.__init__(self, title) |
375 |
|
|
376 |
self.create() |
self.create() |
377 |
|
|
378 |
def create(self): |
def create(self): |
386 |
columns = [] |
columns = [] |
387 |
for col in self.left_table.columns + self.right_table.columns: |
for col in self.left_table.columns + self.right_table.columns: |
388 |
if col.name in visited: |
if col.name in visited: |
389 |
|
# We can't allow multiple columns with the same original |
390 |
|
# name, so omit this one. FIXME: There should be a |
391 |
|
# better solution. |
392 |
continue |
continue |
393 |
columns.append(col) |
columns.append(col) |
394 |
TransientTableBase.create(self, columns) |
TransientTableBase.create(self, columns) |
395 |
|
|
396 |
# Copy the joined data to the table. |
# Copy the joined data to the table. |
397 |
internal_names = [col.internal_name for col in self.columns] |
internal_names = [col.internal_name for col in self.columns] |
398 |
stmt = "INSERT INTO %s (%s) SELECT %s FROM %s JOIN %s ON %s = %s;" \ |
if self.outer_join: |
399 |
% (self.tablename, |
join_operator = 'LEFT OUTER JOIN' |
400 |
", ".join(internal_names), |
else: |
401 |
", ".join(internal_names), |
join_operator = 'JOIN' |
402 |
self.left_table.tablename, |
stmt = ("INSERT INTO %s (id, %s) SELECT %s.id, %s FROM %s" |
403 |
self.right_table.tablename, |
" %s %s ON %s = %s;" |
404 |
self.orig_to_internal[self.left_field], |
% (self.tablename, |
405 |
self.orig_to_internal[self.right_field]) |
", ".join(internal_names), |
406 |
|
self.left_table.tablename, |
407 |
|
", ".join(internal_names), |
408 |
|
self.left_table.tablename, |
409 |
|
join_operator, |
410 |
|
self.right_table.tablename, |
411 |
|
self.orig_to_internal[self.left_field], |
412 |
|
self.orig_to_internal[self.right_field])) |
413 |
self.db.execute(stmt) |
self.db.execute(stmt) |
414 |
|
|
415 |
|
def Dependencies(self): |
416 |
|
"""Return a tuple with the two tables the join depends on.""" |
417 |
|
return self.dependencies |
418 |
|
|
419 |
|
|
420 |
class AutoTransientTable: |
class AutoTransientTable(TitledObject, table.OldTableInterfaceMixin): |
421 |
|
|
422 |
"""Table that copies data to a transient table on demand. |
"""Table that copies data to a transient table on demand. |
423 |
|
|
426 |
""" |
""" |
427 |
|
|
428 |
def __init__(self, transient_db, table): |
def __init__(self, transient_db, table): |
429 |
|
TitledObject.__init__(self, table.Title()) |
430 |
self.transient_db = transient_db |
self.transient_db = transient_db |
431 |
self.table = table |
self.table = table |
432 |
self.t_table = None |
self.t_table = None |
433 |
|
|
434 |
def record_count(self): |
def Columns(self): |
435 |
"""Return the number of records""" |
return self.table.Columns() |
436 |
return self.table.record_count() |
|
437 |
|
def Column(self, col): |
438 |
def field_count(self): |
return self.table.Column(col) |
|
"""Return the number of fields in a record""" |
|
|
return self.table.field_count() |
|
|
|
|
|
def field_info(self, field): |
|
|
"""Return a tuple (type, name, width, prec) for the field no. field |
|
|
|
|
|
type is the data type of the field, name the name, width the |
|
|
field width in characters and prec the decimal precision. |
|
|
""" |
|
|
info = self.table.field_info(field) |
|
|
if info: |
|
|
info = info[:2] + (0, 0) |
|
|
return info |
|
|
|
|
|
def field_info_by_name(self, fieldName): |
|
|
info = self.table.field_info_by_name(fieldName) |
|
|
if info: |
|
|
info = info[:2] + (0, 0) |
|
|
return info |
|
439 |
|
|
440 |
def read_record(self, record): |
def HasColumn(self, col): |
441 |
|
"""Return whether the table has a column with the given name or index |
442 |
|
""" |
443 |
|
return self.table.HasColumn(col) |
444 |
|
|
445 |
|
def NumRows(self): |
446 |
|
return self.table.NumRows() |
447 |
|
|
448 |
|
def NumColumns(self): |
449 |
|
return self.table.NumColumns() |
450 |
|
|
451 |
|
def ReadRowAsDict(self, record): |
452 |
"""Return the record no. record as a dict mapping field names to values |
"""Return the record no. record as a dict mapping field names to values |
453 |
""" |
""" |
454 |
if self.t_table is not None: |
if self.t_table is not None: |
455 |
return self.t_table.read_record(record) |
return self.t_table.ReadRowAsDict(record) |
456 |
else: |
else: |
457 |
return self.table.read_record(record) |
return self.table.ReadRowAsDict(record) |
458 |
|
|
459 |
def write_record(self, record, values): |
def ReadValue(self, row, col): |
460 |
raise NotImplementedError |
"""Return the value of the specified row and column |
461 |
|
|
462 |
|
The col parameter may be the index of the column or its name. |
463 |
|
""" |
464 |
|
if self.t_table is not None: |
465 |
|
return self.t_table.ReadValue(row, col) |
466 |
|
else: |
467 |
|
return self.table.ReadValue(row, col) |
468 |
|
|
469 |
def copy_to_transient(self): |
def copy_to_transient(self): |
470 |
"""Internal: Create a transient table and copy the data into it""" |
"""Internal: Create a transient table and copy the data into it""" |
478 |
self.copy_to_transient() |
self.copy_to_transient() |
479 |
return self.t_table |
return self.t_table |
480 |
|
|
481 |
def field_range(self, colname): |
def ValueRange(self, col): |
482 |
if self.t_table is None: |
if self.t_table is None: |
483 |
self.copy_to_transient() |
self.copy_to_transient() |
484 |
return self.t_table.field_range(colname) |
return self.t_table.ValueRange(col) |
485 |
|
|
486 |
def GetUniqueValues(self, colname): |
def UniqueValues(self, col): |
487 |
if self.t_table is None: |
if self.t_table is None: |
488 |
self.copy_to_transient() |
self.copy_to_transient() |
489 |
return self.t_table.GetUniqueValues(colname) |
return self.t_table.UniqueValues(col) |
490 |
|
|
491 |
|
def SimpleQuery(self, left, comparison, right): |
492 |
|
if self.t_table is None: |
493 |
|
self.copy_to_transient() |
494 |
|
# Make sure to use the column object of the transient table. The |
495 |
|
# left argument is always a column object so we can just ask the |
496 |
|
# t_table for the right object. |
497 |
|
if hasattr(right, "name"): |
498 |
|
return self.t_table.SimpleQuery(self.t_table.Column(left.name), |
499 |
|
comparison, |
500 |
|
self.t_table.Column(right.name)) |
501 |
|
else: |
502 |
|
return self.t_table.SimpleQuery(self.t_table.Column(left.name), |
503 |
|
comparison, right) |
504 |
|
|
505 |
|
def Dependencies(self): |
506 |
|
"""Return a tuple containing the original table""" |
507 |
|
return (self.table,) |