Speed up `import pokedex.db` slightly.

[zzz-pokedex.git] / pokedex / db / load.py

diff --git a/pokedex/db/load.py b/pokedex/db/load.py
index 0425dea..6a548b8 100644 (file)
--- a/pokedex/db/load.py
+++ b/pokedex/db/load.py
@@ -1,8 +1,7 @@
 """CSV to database or vice versa."""
 import csv
 """CSV to database or vice versa."""
 import csv

+import fnmatch

 import os.path
 import os.path

-import pkg_resources
-import re

 import sys
 
 from sqlalchemy.orm.attributes import instrumentation_registry
 import sys
 
 from sqlalchemy.orm.attributes import instrumentation_registry


@@ -11,74 +10,92 @@ import sqlalchemy.types
 
 from pokedex.db import metadata
 import pokedex.db.tables as tables
 
 from pokedex.db import metadata
 import pokedex.db.tables as tables

+from pokedex.defaults import get_default_csv_dir

 
 
 
 

-def _wildcard_char_to_regex(char):
-    """Converts a single wildcard character to the regex equivalent."""
-
-    if char == '?':
-        return '.?'
-    elif char == '*':
-        return '.*'
+def _get_table_names(metadata, patterns):
+    """Returns a list of table names from the given metadata.  If `patterns`
+    exists, only tables matching one of the patterns will be returned.
+    """
+    if patterns:
+        table_names = set()
+        for pattern in patterns:
+            if '.' in pattern or '/' in pattern:
+                # If it looks like a filename, pull out just the table name
+                _, filename = os.path.split(pattern)
+                table_name, _ = os.path.splitext(filename)
+                pattern = table_name
+
+            table_names.update(fnmatch.filter(metadata.tables.keys(), pattern))

     else:
     else:

-        return re.escape(char)
-
-def _wildcard_glob_to_regex(glob):
-    """Converts a single wildcard glob to a regex STRING."""
+        table_names = metadata.tables.keys()

 
 

-    # If it looks like a filename, make it not one
-    if '.' in glob or '/' in glob:
-        _, filename = os.path.split(glob)
-        table_name, _ = os.path.splitext(filename)
-        glob = table_name
+    return list(table_names)

 
 

-    return u''.join(map(_wildcard_char_to_regex, glob))
+def _get_verbose_prints(verbose):
+    """If `verbose` is true, returns three functions: one for printing a
+    starting message, one for printing an interim status update, and one for
+    printing a success or failure message when finished.

 
 

-def _wildcards_to_regex(strings):
-    """Converts a list of wildcard globs to a single regex object."""
+    If `verbose` is false, returns no-op functions.
+    """

 
 

-    regex_parts = map(_wildcard_glob_to_regex, strings)
+    if not verbose:
+        # Return dummies
+        def dummy(*args, **kwargs):
+            pass

 
 

-    regex = '^(?:' + '|'.join(regex_parts) + ')$'
+        return dummy, dummy, dummy

 
 

-    return re.compile(regex)
+    ### Okay, verbose == True; print stuff

 
 

+    def print_start(thing):
+        # Truncate to 66 characters, leaving 10 characters for a success
+        # or failure message
+        truncated_thing = thing[0:66]

 
 

-def _get_verbose_prints(verbose):
-    """If `verbose` is true, returns two functions: one for printing a starting
-    message, and the other for printing a success or failure message when
-    finished.
+        # Also, space-pad to keep the cursor in a known column
+        num_spaces = 66 - len(truncated_thing)

 
 

-    If `verbose` is false, returns two no-op functions.
-    """
+        print "%s...%s" % (truncated_thing, ' ' * num_spaces),
+        sys.stdout.flush()

 
 

-    if verbose:
-        import sys
-        def print_start(thing):
-            # Truncate to 66 characters, leaving 10 characters for a success
-            # or failure message
-            truncated_thing = thing[0:66]
+    if sys.stdout.isatty():
+        # stdout is a terminal; stupid backspace tricks are OK.
+        # Don't use print, because it always adds magical spaces, which
+        # makes backspace accounting harder

 
 

-            # Also, space-pad to keep the cursor in a known column
-            num_spaces = 66 - len(truncated_thing)
-
-            print "%s...%s" % (truncated_thing, ' ' * num_spaces),
+        backspaces = [0]
+        def print_status(msg):
+            # Overwrite any status text with spaces before printing
+            sys.stdout.write('\b' * backspaces[0])
+            sys.stdout.write(' ' * backspaces[0])
+            sys.stdout.write('\b' * backspaces[0])
+            sys.stdout.write(msg)

             sys.stdout.flush()
             sys.stdout.flush()

+            backspaces[0] = len(msg)

 
         def print_done(msg='ok'):
 
         def print_done(msg='ok'):

-            print msg
+            # Overwrite any status text with spaces before printing
+            sys.stdout.write('\b' * backspaces[0])
+            sys.stdout.write(' ' * backspaces[0])
+            sys.stdout.write('\b' * backspaces[0])
+            sys.stdout.write(msg + "\n")

             sys.stdout.flush()
             sys.stdout.flush()

+            backspaces[0] = 0

 
 

-        return print_start, print_done
+    else:
+        # stdout is a file (or something); don't bother with status at all
+        def print_status(msg):
+            pass

 
 

-    # Not verbose; return dummies
-    def dummy(*args, **kwargs):
-        pass
+        def print_done(msg='ok'):
+            print msg

 
 

-    return dummy, dummy
+    return print_start, print_status, print_done

 
 
 
 

-def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
+def load(session, tables=[], directory=None, drop_tables=False, verbose=False, safe=True):

     """Load data from CSV files into the given database session.
 
     Tables are created automatically.
     """Load data from CSV files into the given database session.
 
     Tables are created automatically.


@@ -98,34 +115,41 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
 
     `verbose`
         If set to True, status messages will be printed to stdout.
 
     `verbose`
         If set to True, status messages will be printed to stdout.

+
+    `safe`
+        If set to False, load can be faster, but can corrupt the database if
+        it crashes or is interrupted.

     """
 
     # First take care of verbosity
     """
 
     # First take care of verbosity

-    print_start, print_done = _get_verbose_prints(verbose)
+    print_start, print_status, print_done = _get_verbose_prints(verbose)

 
 
 
 

-    if not directory:
-        directory = pkg_resources.resource_filename('pokedex', 'data/csv')
-
-    if tables:
-        regex = _wildcards_to_regex(tables)
-        table_names = filter(regex.match, metadata.tables.keys())
-    else:
-        table_names = metadata.tables.keys()
+    if directory is None:
+        directory = get_default_csv_dir()

 
 

+    table_names = _get_table_names(metadata, tables)

     table_objs = [metadata.tables[name] for name in table_names]
     table_objs = sqlalchemy.sql.util.sort_tables(table_objs)
 
     table_objs = [metadata.tables[name] for name in table_names]
     table_objs = sqlalchemy.sql.util.sort_tables(table_objs)
 

+    # SQLite speed tweaks
+    if not safe and session.connection().dialect.name == 'sqlite':
+        session.connection().execute("PRAGMA synchronous=OFF")
+        session.connection().execute("PRAGMA journal_mode=OFF")

 
     # Drop all tables if requested
     if drop_tables:
         print_start('Dropping tables')
 
     # Drop all tables if requested
     if drop_tables:
         print_start('Dropping tables')

-        for table in reversed(table_objs):
+        for n, table in enumerate(reversed(table_objs)):

             table.drop(checkfirst=True)
             table.drop(checkfirst=True)

+            print_status('%s/%s' % (n, len(table_objs)))

         print_done()
 
         print_done()
 

-    for table in table_objs:
+    print_start('Creating tables')
+    for n, table in enumerate(table_objs):

         table.create()
         table.create()

+        print_status('%s/%s' % (n, len(table_objs)))
+    print_done()

     connection = session.connection()
 
     # Okay, run through the tables and actually load the data now
     connection = session.connection()
 
     # Okay, run through the tables and actually load the data now


@@ -136,15 +160,44 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
         print_start(table_name)
 
         try:
         print_start(table_name)
 
         try:

-            csvfile = open("%s/%s.csv" % (directory, table_name), 'rb')
+            csvpath = "%s/%s.csv" % (directory, table_name)
+            csvfile = open(csvpath, 'rb')

         except IOError:
             # File doesn't exist; don't load anything!
             print_done('missing?')
             continue
 
         except IOError:
             # File doesn't exist; don't load anything!
             print_done('missing?')
             continue
 

+        csvsize = os.stat(csvpath).st_size
+

         reader = csv.reader(csvfile, lineterminator='\n')
         column_names = [unicode(column) for column in reader.next()]
 
         reader = csv.reader(csvfile, lineterminator='\n')
         column_names = [unicode(column) for column in reader.next()]
 

+        if not safe and session.connection().dialect.name == 'postgresql':
+            """
+            Postgres' CSV dialect works with our data, if we mark the not-null
+            columns with FORCE NOT NULL.
+            COPY is only allowed for DB superusers. If you're not one, use safe
+            loading (pokedex load -S).
+            """
+            session.commit()
+            not_null_cols = [c for c in column_names if not table_obj.c[c].nullable]
+            if not_null_cols:
+                force_not_null = 'FORCE NOT NULL ' + ','.join('"%s"' % c for c in not_null_cols)
+            else:
+                force_not_null = ''
+            command = "COPY {table_name} ({columns}) FROM '{csvpath}' CSV HEADER {force_not_null}"
+            session.connection().execute(
+                    command.format(
+                            table_name=table_name,
+                            csvpath=csvpath,
+                            columns=','.join('"%s"' % c for c in column_names),
+                            force_not_null=force_not_null,
+                        )
+                )
+            session.commit()
+            print_done()
+            continue
+

         # Self-referential tables may contain rows with foreign keys of other
         # rows in the same table that do not yet exist.  Pull these out and add
         # them to the session last
         # Self-referential tables may contain rows with foreign keys of other
         # rows in the same table that do not yet exist.  Pull these out and add
         # them to the session last


@@ -160,10 +213,15 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
 
         new_rows = []
         def insert_and_commit():
 
         new_rows = []
         def insert_and_commit():

+            if not new_rows:
+                return

             session.connection().execute(insert_stmt, new_rows)
             session.commit()
             new_rows[:] = []
 
             session.connection().execute(insert_stmt, new_rows)
             session.commit()
             new_rows[:] = []
 

+            progress = "%d%%" % (100 * csvfile.tell() // csvsize)
+            print_status(progress)
+

         for csvs in reader:
             row_data = {}
 
         for csvs in reader:
             row_data = {}
 


@@ -233,6 +291,10 @@ def load(session, tables=[], directory=None, drop_tables=False, verbose=False):
 
         print_done()
 
 
         print_done()
 

+    # SQLite check
+    if session.connection().dialect.name == 'sqlite':
+        session.connection().execute("PRAGMA integrity_check")
+

 
 
 def dump(session, tables=[], directory=None, verbose=False):
 
 
 def dump(session, tables=[], directory=None, verbose=False):


@@ -254,18 +316,13 @@ def dump(session, tables=[], directory=None, verbose=False):
     """
 
     # First take care of verbosity
     """
 
     # First take care of verbosity

-    print_start, print_done = _get_verbose_prints(verbose)
+    print_start, print_status, print_done = _get_verbose_prints(verbose)

 
 
     if not directory:
 
 
     if not directory:

-        directory = pkg_resources.resource_filename('pokedex', 'data/csv')
-
-    if tables:
-        regex = _wildcards_to_regex(tables)
-        table_names = filter(regex.match, metadata.tables.keys())
-    else:
-        table_names = metadata.tables.keys()
+        directory = get_default_csv_dir()

 
 

+    table_names = _get_table_names(metadata, tables)

     table_names.sort()
 
 
     table_names.sort()