Fix location_area_encounter_rates
[zzz-pokedex.git] / pokedex / db / load.py
1 """CSV to database or vice versa."""
2 import csv
3 import pkg_resources
4 import sys
5
6 from sqlalchemy.orm.attributes import instrumentation_registry
7 import sqlalchemy.types
8
9 from pokedex.db import metadata
10 import pokedex.db.tables as tables
11
12
13 def _get_verbose_prints(verbose):
14 """If `verbose` is true, returns two functions: one for printing a starting
15 message, and the other for printing a success or failure message when
16 finished.
17
18 If `verbose` is false, returns two no-op functions.
19 """
20
21 if verbose:
22 import sys
23 def print_start(thing):
24 # Truncate to 66 characters, leaving 10 characters for a success
25 # or failure message
26 truncated_thing = thing[0:66]
27
28 # Also, space-pad to keep the cursor in a known column
29 num_spaces = 66 - len(truncated_thing)
30
31 print "%s...%s" % (truncated_thing, ' ' * num_spaces),
32 sys.stdout.flush()
33
34 def print_done(msg='ok'):
35 print msg
36 sys.stdout.flush()
37
38 return print_start, print_done
39
40 # Not verbose; return dummies
41 def dummy(*args, **kwargs):
42 pass
43
44 return dummy, dummy
45
46
47 def load(session, directory=None, drop_tables=False, verbose=False):
48 """Load data from CSV files into the given database session.
49
50 Tables are created automatically.
51
52 `session`
53 SQLAlchemy session to use.
54
55 `directory`
56 Directory the CSV files reside in. Defaults to the `pokedex` data
57 directory.
58
59 `drop_tables`
60 If set to True, existing `pokedex`-related tables will be dropped.
61
62 `verbose`
63 If set to True, status messages will be printed to stdout.
64 """
65
66 # First take care of verbosity
67 print_start, print_done = _get_verbose_prints(verbose)
68
69
70 if not directory:
71 directory = pkg_resources.resource_filename('pokedex', 'data/csv')
72
73 # Drop all tables if requested
74 if drop_tables:
75 print_start('Dropping tables')
76 metadata.drop_all()
77 print_done()
78
79 metadata.create_all()
80 connection = session.connection()
81
82 # Okay, run through the tables and actually load the data now
83 for table_obj in metadata.sorted_tables:
84 table_name = table_obj.name
85 insert_stmt = table_obj.insert()
86
87 print_start(table_name)
88
89 try:
90 csvfile = open("%s/%s.csv" % (directory, table_name), 'rb')
91 except IOError:
92 # File doesn't exist; don't load anything!
93 print_done('missing?')
94 continue
95
96 reader = csv.reader(csvfile, lineterminator='\n')
97 column_names = [unicode(column) for column in reader.next()]
98
99 # Self-referential tables may contain rows with foreign keys of other
100 # rows in the same table that do not yet exist. Pull these out and add
101 # them to the session last
102 # ASSUMPTION: Self-referential tables have a single PK called "id"
103 deferred_rows = [] # ( row referring to id, [foreign ids we need] )
104 seen_ids = {} # primary key we've seen => 1
105
106 # Fetch foreign key columns that point at this table, if any
107 self_ref_columns = []
108 for column in table_obj.c:
109 if any(_.references(table_obj) for _ in column.foreign_keys):
110 self_ref_columns.append(column)
111
112 new_rows = []
113 def insert_and_commit():
114 session.connection().execute(insert_stmt, new_rows)
115 session.commit()
116 new_rows[:] = []
117
118 for csvs in reader:
119 row_data = {}
120
121 for column_name, value in zip(column_names, csvs):
122 column = table_obj.c[column_name]
123 if column.nullable and value == '':
124 # Empty string in a nullable column really means NULL
125 value = None
126 elif isinstance(column.type, sqlalchemy.types.Boolean):
127 # Boolean values are stored as string values 0/1, but both
128 # of those evaluate as true; SQLA wants True/False
129 if value == '0':
130 value = False
131 else:
132 value = True
133 else:
134 # Otherwise, unflatten from bytes
135 value = value.decode('utf-8')
136
137 # nb: Dictionaries flattened with ** have to have string keys
138 row_data[ str(column_name) ] = value
139
140 # May need to stash this row and add it later if it refers to a
141 # later row in this table
142 if self_ref_columns:
143 foreign_ids = [row_data[_.name] for _ in self_ref_columns]
144 foreign_ids = [_ for _ in foreign_ids if _] # remove NULL ids
145
146 if not foreign_ids:
147 # NULL key. Remember this row and add as usual.
148 seen_ids[row_data['id']] = 1
149
150 elif all(_ in seen_ids for _ in foreign_ids):
151 # Non-NULL key we've already seen. Remember it and commit
152 # so we know the old row exists when we add the new one
153 insert_and_commit()
154 seen_ids[row_data['id']] = 1
155
156 else:
157 # Non-NULL future id. Save this and insert it later!
158 deferred_rows.append((row_data, foreign_ids))
159 continue
160
161 # Insert row!
162 new_rows.append(row_data)
163
164 # Remembering some zillion rows in the session consumes a lot of
165 # RAM. Let's not do that. Commit every 1000 rows
166 if len(new_rows) > 1000:
167 insert_and_commit()
168
169 insert_and_commit()
170
171 # Attempt to add any spare rows we've collected
172 for row_data, foreign_ids in deferred_rows:
173 if not all(_ in seen_ids for _ in foreign_ids):
174 # Could happen if row A refers to B which refers to C.
175 # This is ridiculous and doesn't happen in my data so far
176 raise ValueError("Too many levels of self-reference! "
177 "Row was: " + str(row))
178
179 session.connection().execute(
180 insert_stmt.values(**row_data)
181 )
182 seen_ids[row_data['id']] = 1
183 session.commit()
184
185 print_done()
186
187
188
189 def dump(session, directory=None, verbose=False):
190 """Dumps the contents of a database to a set of CSV files. Probably not
191 useful to anyone besides a developer.
192
193 `session`
194 SQLAlchemy session to use.
195
196 `directory`
197 Directory the CSV files should be put in. Defaults to the `pokedex`
198 data directory.
199
200 `verbose`
201 If set to True, status messages will be printed to stdout.
202 """
203
204 # First take care of verbosity
205 print_start, print_done = _get_verbose_prints(verbose)
206
207
208 if not directory:
209 directory = pkg_resources.resource_filename('pokedex', 'data/csv')
210
211 for table_name in sorted(metadata.tables.keys()):
212 print_start(table_name)
213 table = metadata.tables[table_name]
214
215 writer = csv.writer(open("%s/%s.csv" % (directory, table_name), 'wb'),
216 lineterminator='\n')
217 columns = [col.name for col in table.columns]
218 writer.writerow(columns)
219
220 primary_key = table.primary_key
221 for row in session.query(table).order_by(*primary_key).all():
222 csvs = []
223 for col in columns:
224 # Convert Pythony values to something more universal
225 val = getattr(row, col)
226 if val == None:
227 val = ''
228 elif val == True:
229 val = '1'
230 elif val == False:
231 val = '0'
232 else:
233 val = unicode(val).encode('utf-8')
234
235 csvs.append(val)
236
237 writer.writerow(csvs)
238
239 print_done()