evennia/src/utils/batchprocessors.py

448 lines
16 KiB
Python

"""
This file contains the core methods for the Batch-command- and
Batch-code-processors respectively. In short, these are two different
ways to build a game world using a normal text-editor without having
to do so 'on the fly' in-game. They also serve as an automatic backup
so you can quickly recreate a world also after a server reset. The
functions in this module is meant to form the backbone of a system
called and accessed through game commands.
The Batch-command processor is the simplest. It simply runs a list of
in-game commands in sequence by reading them from a text file. The
advantage of this is that the builder only need to remember the normal
in-game commands. They are also executing with full permission checks
etc, making it relatively safe for builders to use. The drawback is
that in-game there is really a builder-character walking around
building things, and it can be important to create rooms and objects
in the right order, so the character can move between them. Also
objects that affects players (such as mobs, dark rooms etc) will
affect the building character too, requiring extra care to turn
off/on.
The Batch-code processor is a more advanced system that accepts full
Python code, executing in chunks. The advantage of this is much more
power; practically anything imaginable can be coded and handled using
the batch-code processor. There is no in-game character that moves and
that can be affected by what is being built - the database is
populated on the fly. The drawback is safety and entry threshold - the
code is executed as would any server code, without mud-specific
permission checks and you have full access to modifying objects
etc. You also need to know Python and Evennia's API. Hence it's
recommended that the batch-code processor is limited only to
superusers or highly trusted staff.
Batch-command processor file syntax
The batch-command processor accepts 'batchcommand files' e.g 'batch.ev',
containing a sequence of valid evennia commands in a simple
format. The engine runs each command in sequence, as if they had been
run at the game prompt.
This way entire game worlds can be created and planned offline; it is
especially useful in order to create long room descriptions where a
real offline text editor is often much better than any online text
editor or prompt.
Example of batch.ev file:
----------------------------
# batch file
# all lines starting with # are comments; they also indicate
# that a command definition is over.
@create box
# this comment ends the @create command.
@set box/desc = A large box.
Inside are some scattered piles of clothing.
It seems the bottom of the box is a bit loose.
# Again, this comment indicates the @set command is over. Note how
# the description could be freely added. Excess whitespace on a line
# is ignored. An empty line in the command definition is parsed as a \n
# (so two empty lines becomes a new paragraph).
@teleport #221
# (Assuming #221 is a warehouse or something.)
# (remember, this comment ends the @teleport command! Don'f forget it)
@drop box
# Done, the box is in the warehouse! (this last comment is not necessary to
# close the @drop command since it's the end of the file)
-------------------------
An example batch file is game/gamesrc/commands/examples/batch_example.ev.
Batch-code processor file syntax
The Batch-code processor accepts full python modules (e.g. "batch.py") that
looks identical to normal Python files with a few exceptions that allows them
to the executed in blocks. This way of working assures a sequential execution
of the file and allows for features like stepping from block to block
(without executing those coming before), as well as automatic deletion
of created objects etc. You can however also run a batch-code python file
directly using Python (and can also be de).
Code blocks are separated by python comments starting with special code words.
#HEADER - this denotes commands global to the entire file, such as
import statements and global variables. They will
automatically be pasted at the top of all code blocks. Observe
that changes to these variables made in one block is not
preserved between blocks!
#CODE [objname, objname, ...] - This designates a code block that will be executed like a
stand-alone piece of code together with any #HEADER
defined. <objname>s mark the (variable-)names of objects created in the code,
and which may be auto-deleted by the processor if desired (such as when
debugging the script). E.g., if the code contains the command
myobj = create.create_object(...), you could put 'myobj' in the #CODE header
regardless of what the created object is actually called in-game.
The following variables are automatically made available for the script:
caller - the object executing the script
Example batch.py file
-----------------------------------
#HEADER
import traceback
from django.config import settings
from src.utils import create
from game.gamesrc.typeclasses import basetypes
GOLD = 10
#CODE obj, obj2
obj = create.create_object(basetypes.Object)
obj2 = create.create_object(basetypes.Object)
obj.location = caller.location
obj.db.gold = GOLD
caller.msg("The object was created!")
#CODE
script = create.create_script()
"""
import re
import codecs
from traceback import format_exc
from django.conf import settings
from django.core.management import setup_environ
from src.utils import logger
from src.utils import utils
from game import settings as settings_module
ENCODINGS = settings.ENCODINGS
#------------------------------------------------------------
# Helper function
#------------------------------------------------------------
def read_batchfile(pythonpath, file_ending='.py'):
"""
This reads the contents of a batch-file.
Filename is considered to be the name of the batch file
relative the directory specified in settings.py.
file_ending specify which batchfile ending should be
assumed (.ev or .py).
"""
# open the file
if pythonpath and not (pythonpath.startswith('src.') or pythonpath.startswith('game.')
or pythonpath.startswith('contrib.')):
abspaths = []
for basepath in settings.BASE_BATCHPROCESS_PATHS:
abspaths.append(utils.pypath_to_realpath("%s.%s" % (basepath, pythonpath), file_ending))
else:
abspaths = [utils.pypath_to_realpath(pythonpath, file_ending)]
fobj, lines, err = None, [], None
for file_encoding in ENCODINGS:
# try different encodings, in order
load_errors = []
for abspath in abspaths:
# try different paths, until we get a match
try:
# we read the file directly into unicode.
fobj = codecs.open(abspath, 'r', encoding=file_encoding)
except IOError:
load_errors.append("Could not open batchfile '%s'." % abspath)
continue
break
if not fobj:
continue
load_errors = []
err =None
# We have successfully found and opened the file. Now actually
# try to decode it using the given protocol.
try:
lines = fobj.readlines()
except UnicodeDecodeError:
# give the line of failure
fobj.seek(0)
try:
lnum = 0
for lnum, line in enumerate(fobj):
pass
except UnicodeDecodeError, err:
# lnum starts from 0, so we add +1 line,
# besides the faulty line is never read
# so we add another 1 (thus +2) to get
# the actual line number seen in an editor.
err.linenum = lnum + 2
fobj.close()
# possibly try another encoding
continue
# if we get here, the encoding worked. Stop iteration.
break
if load_errors:
logger.log_errmsg("\n".join(load_errors))
if err:
return err
else:
return lines
#------------------------------------------------------------
#
# Batch-command processor
#
#------------------------------------------------------------
class BatchCommandProcessor(object):
"""
This class implements a batch-command processor.
"""
def parse_file(self, pythonpath):
"""
This parses the lines of a batchfile according to the following
rules:
1) # at the beginning of a line marks the end of the command before it.
It is also a comment and any number of # can exist on subsequent
lines (but not inside comments).
2) Commands are placed alone at the beginning of a line and their
arguments are considered to be everything following (on any
number of lines) until the next comment line beginning with #.
3) Newlines are ignored in command definitions
4) A completely empty line in a command line definition is condered
a newline (so two empty lines is a paragraph).
5) Excess spaces and indents inside arguments are stripped.
"""
#helper function
def identify_line(line):
"""
Identifies the line type (comment, commanddef or empty)
"""
try:
if line.strip()[0] == '#':
return "comment"
else:
return "commanddef"
except IndexError:
return "empty"
#read the indata, if possible.
lines = read_batchfile(pythonpath, file_ending='.ev')
#line = utils.to_unicode(line)
if not lines:
return None
commands = []
curr_cmd = ""
#purge all superfluous whitespace and newlines from lines
reg1 = re.compile(r"\s+")
lines = [reg1.sub(" ", l) for l in lines]
#parse all command definitions into a list.
for line in lines:
typ = identify_line(line)
if typ == "commanddef":
curr_cmd += line
elif typ == "empty" and curr_cmd:
curr_cmd += "\r\n"
else: #comment
if curr_cmd:
commands.append(curr_cmd.strip())
curr_cmd = ""
if curr_cmd:
commands.append(curr_cmd.strip())
#second round to clean up now merged line edges etc.
reg2 = re.compile(r"[ \t\f\v]+")
commands = [reg2.sub(" ", c) for c in commands]
#remove eventual newline at the end of commands
commands = [c.strip('\r\n') for c in commands]
return commands
#------------------------------------------------------------
#
# Batch-code processor
#
#------------------------------------------------------------
class BatchCodeProcessor(object):
"""
This implements a batch-code processor
"""
def parse_file(self, pythonpath):
"""
This parses the lines of a batchfile according to the following
rules:
1) Lines starting with #HEADER starts a header block (ends other blocks)
2) Lines starting with #CODE begins a code block (ends other blocks)
3) #CODE headers may be of the following form: #CODE (info) objname, objname2, ...
3) All lines outside blocks are stripped.
4) All excess whitespace beginning/ending a block is stripped.
"""
# helper function
def parse_line(line):
"""
Identifies the line type: block command, comment, empty or normal code.
"""
line = line.strip()
if line.startswith("#HEADER"):
return ("header", "", "")
elif line.startswith("#CODE"):
# parse code command
line = line.lstrip("#CODE").strip()
objs = []
info = ""
if line and '(' in line and ')' in line:
# a code description
lp = line.find('(')
rp = line.find(')')
info = line[lp:rp+1]
line = line[rp+1:]
if line:
objs = [obj.strip() for obj in line.split(',')]
return ("codeheader", info, objs)
elif line.startswith('#'):
return ('comment', "", "\n%s" % line)
else:
#normal line - return it with a line break.
return ('line', "", "\n%s" % line)
# read indata
lines = read_batchfile(pythonpath, file_ending='.py')
if not lines:
return None
# parse file into blocks
header = ""
codes = []
in_header = False
in_code = False
for line in lines:
# parse line
mode, info, line = parse_line(line)
# try:
# print "::", in_header, in_code, mode, line.strip()
# except:
# print "::", in_header, in_code, mode, line
if mode == 'header':
in_header = True
in_code = False
elif mode == 'codeheader':
in_header = False
in_code = True
# the line is a list of object variable names
# (or an empty list) at this point.
codedict = {'objs':line,
'info':info,
'code':""}
codes.append(codedict)
elif mode == 'comment' and in_header:
continue
else:
# another type of line (empty, comment or code)
if line and in_header:
header += line
elif line and in_code:
codes[-1]['code'] += line
else:
# not in a block (e.g. first in file). Ignore.
continue
# last, we merge the headers with all codes.
for codedict in codes:
codedict["code"] = "#CODE %s %s\n%s\n\n%s" % (codedict['info'],
", ".join(codedict["objs"]),
header.strip(),
codedict["code"].strip())
return codes
def code_exec(self, codedict, extra_environ=None, debug=False):
"""
Execute a single code block, including imports and appending global vars
extra_environ - dict with environment variables
"""
# define the execution environment
environ = "setup_environ(settings_module)"
environdict = {"setup_environ":setup_environ,
"settings_module":settings_module}
if extra_environ:
for key, value in extra_environ.items():
environdict[key] = value
# merge all into one block
code = "%s\n%s" % (environ, codedict['code'])
if debug:
# try to delete marked objects
for obj in codedict['objs']:
code += "\ntry: %s.delete()\nexcept: pass" % obj
# execute the block
try:
exec(code, environdict)
except Exception:
errlist = format_exc().split('\n')
if len(errlist) > 4:
errlist = errlist[4:]
err = "\n".join(" %s" % line for line in errlist if line)
if debug:
# try to delete objects again.
try:
for obj in codedict['objs']:
eval("%s.delete()" % obj, environdict)
except Exception:
pass
return err
return None
BATCHCMD = BatchCommandProcessor()
BATCHCODE = BatchCodeProcessor()