parser.py

__author__ = "Johannes Köster"
__copyright__ = "Copyright 2021, Johannes Köster"
__email__ = "johannes.koester@uni-due.de"
__license__ = "MIT"

import tokenize
import textwrap
import os
from urllib.error import HTTPError, URLError, ContentTooShortError
import urllib.request
from io import TextIOWrapper

from snakemake.exceptions import WorkflowError
from snakemake import common

dd = textwrap.dedent

INDENT = "\t"


def is_newline(token, newline_tokens=set((tokenize.NEWLINE, tokenize.NL))):
    return token.type in newline_tokens


def is_indent(token):
    return token.type == tokenize.INDENT


def is_dedent(token):
    return token.type == tokenize.DEDENT


def is_op(token):
    return token.type == tokenize.OP


def is_greater(token):
    return is_op(token) and token.string == ">"


def is_comma(token):
    return is_op(token) and token.string == ","


def is_name(token):
    return token.type == tokenize.NAME


def is_colon(token):
    return is_op(token) and token.string == ":"


def is_comment(token):
    return token.type == tokenize.COMMENT


def is_string(token):
    return token.type == tokenize.STRING


def is_eof(token):
    return token.type == tokenize.ENDMARKER


def lineno(token):
    return token.start[0]


class StopAutomaton(Exception):
    def __init__(self, token):
        self.token = token


class TokenAutomaton:

    subautomata = dict()

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        self.root = root
        self.snakefile = snakefile
        self.state = None
        self.base_indent = base_indent
        self.line = 0
        self.indent = 0
        self.was_indented = False
        self.lasttoken = None
        self._dedent = dedent

    @property
    def dedent(self):
        return self._dedent

    @property
    def effective_indent(self):
        return self.base_indent + self.indent - self.dedent

    def indentation(self, token):
        if is_indent(token) or is_dedent(token):
            self.indent = token.end[1] - self.base_indent
            self.was_indented |= self.indent > 0

    def consume(self):
        for token in self.snakefile:
            self.indentation(token)
            try:
                for t, orig in self.state(token):
                    if self.lasttoken == "\n" and not t.isspace():
                        yield INDENT * self.effective_indent, orig
                    yield t, orig
                    self.lasttoken = t
            except tokenize.TokenError as e:
                self.error(
                    str(e).split(",")[0].strip("()''"), token
                )  # TODO the inferred line number seems to be wrong sometimes

    def error(self, msg, token):
        raise SyntaxError(msg, (self.snakefile.path, lineno(token), None, None))

    def subautomaton(self, automaton, *args, **kwargs):
        return self.subautomata[automaton](
            self.snakefile,
            *args,
            base_indent=self.base_indent + self.indent,
            dedent=self.dedent,
            root=False,
            **kwargs,
        )


class KeywordState(TokenAutomaton):

    prefix = ""

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.line = 0
        self.state = self.colon

    @property
    def keyword(self):
        return self.__class__.__name__.lower()[len(self.prefix) :]

    def end(self):
        yield ")"

    def decorate_end(self, token):
        for t in self.end():
            if isinstance(t, tuple):
                yield t
            else:
                yield t, token

    def colon(self, token):
        if is_colon(token):
            self.state = self.block
            for t in self.start():
                yield t, token
        else:
            self.error("Colon expected after keyword {}.".format(self.keyword), token)

    def is_block_end(self, token):
        return (self.line and self.indent <= 0) or is_eof(token)

    def block(self, token):
        if self.lasttoken == "\n" and is_comment(token):
            # ignore lines containing only comments
            self.line -= 1
        if self.is_block_end(token):
            yield from self.decorate_end(token)
            yield "\n", token
            raise StopAutomaton(token)

        if is_newline(token):
            self.line += 1
            yield token.string, token
        elif not (is_indent(token) or is_dedent(token)):
            if is_comment(token):
                yield token.string, token
            else:
                yield from self.block_content(token)

    def yield_indent(self, token):
        return token.string, token

    def block_content(self, token):
        yield token.string, token


class GlobalKeywordState(KeywordState):
    def start(self):
        yield "workflow.{keyword}(".format(keyword=self.keyword)


class DecoratorKeywordState(KeywordState):
    decorator = None
    args = list()

    def start(self):
        yield "@workflow.{}".format(self.decorator)
        yield "\n"
        yield "def __{}({}):".format(self.decorator, ", ".join(self.args))

    def end(self):
        yield ""


class RuleKeywordState(KeywordState):
    def __init__(self, snakefile, base_indent=0, dedent=0, root=True, rulename=None):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.rulename = rulename

    def start(self):
        yield "\n"
        yield "@workflow.{keyword}(".format(keyword=self.keyword)


class SectionKeywordState(KeywordState):
    def start(self):
        yield ", {keyword}=".format(keyword=self.keyword)

    def end(self):
        # no end needed
        return list()


# Global keyword states


class Envvars(GlobalKeywordState):
    @property
    def keyword(self):
        return "register_envvars"


class Include(GlobalKeywordState):
    pass


class Workdir(GlobalKeywordState):
    pass


class Configfile(GlobalKeywordState):
    pass


# PEPs


class Pepfile(GlobalKeywordState):
    pass


class Pepschema(GlobalKeywordState):
    pass


class Report(GlobalKeywordState):
    pass


class Scattergather(GlobalKeywordState):
    pass


class Ruleorder(GlobalKeywordState):
    def block_content(self, token):
        if is_greater(token):
            yield ",", token
        elif is_name(token):
            yield repr(token.string), token
        else:
            self.error(
                "Expected a descending order of rule names, "
                "e.g. rule1 > rule2 > rule3 ...",
                token,
            )


class GlobalWildcardConstraints(GlobalKeywordState):
    @property
    def keyword(self):
        return "global_wildcard_constraints"


class GlobalSingularity(GlobalKeywordState):
    @property
    def keyword(self):
        return "global_container"


class GlobalContainer(GlobalKeywordState):
    @property
    def keyword(self):
        return "global_container"


class GlobalContainerized(GlobalKeywordState):
    @property
    def keyword(self):
        return "global_containerized"


# subworkflows


class SubworkflowKeywordState(SectionKeywordState):
    prefix = "Subworkflow"


class SubworkflowSnakefile(SubworkflowKeywordState):
    pass


class SubworkflowWorkdir(SubworkflowKeywordState):
    pass


class SubworkflowConfigfile(SubworkflowKeywordState):
    pass


class Subworkflow(GlobalKeywordState):

    subautomata = dict(
        snakefile=SubworkflowSnakefile,
        workdir=SubworkflowWorkdir,
        configfile=SubworkflowConfigfile,
    )

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.state = self.name
        self.has_snakefile = False
        self.has_workdir = False
        self.has_name = False
        self.primary_token = None

    def end(self):
        if not (self.has_snakefile or self.has_workdir):
            self.error(
                "A subworkflow needs either a path to a Snakefile or to a workdir.",
                self.primary_token,
            )
        yield ")"

    def name(self, token):
        if is_name(token):
            yield "workflow.subworkflow({name!r}".format(name=token.string), token
            self.has_name = True
        elif is_colon(token) and self.has_name:
            self.primary_token = token
            self.state = self.block
        else:
            self.error("Expected name after subworkflow keyword.", token)

    def block_content(self, token):
        if is_name(token):
            try:
                if token.string == "snakefile":
                    self.has_snakefile = True
                if token.string == "workdir":
                    self.has_workdir = True
                for t in self.subautomaton(token.string).consume():
                    yield t
            except KeyError:
                self.error(
                    "Unexpected keyword {} in "
                    "subworkflow definition".format(token.string),
                    token,
                )
            except StopAutomaton as e:
                self.indentation(e.token)
                for t in self.block(e.token):
                    yield t
        elif is_comment(token):
            yield "\n", token
            yield token.string, token
        elif is_string(token):
            # ignore docstring
            pass
        else:
            self.error(
                "Expecting subworkflow keyword, comment or docstrings "
                "inside a subworkflow definition.",
                token,
            )


class Localrules(GlobalKeywordState):
    def block_content(self, token):
        if is_comma(token):
            yield ",", token
        elif is_name(token):
            yield repr(token.string), token
        else:
            self.error(
                "Expected a comma separated list of rules that shall "
                "not be executed by the cluster command.",
                token,
            )


# Rule keyword states


class Name(RuleKeywordState):
    pass


class Input(RuleKeywordState):
    pass


class Output(RuleKeywordState):
    pass


class Params(RuleKeywordState):
    pass


class Threads(RuleKeywordState):
    pass


class Shadow(RuleKeywordState):
    pass


class Resources(RuleKeywordState):
    pass


class Priority(RuleKeywordState):
    pass


class Version(RuleKeywordState):
    pass


class Log(RuleKeywordState):
    pass


class Message(RuleKeywordState):
    pass


class Benchmark(RuleKeywordState):
    pass


class Conda(RuleKeywordState):
    pass


class Singularity(RuleKeywordState):
    @property
    def keyword(self):
        return "container"


class Container(RuleKeywordState):
    pass


class Containerized(RuleKeywordState):
    pass


class EnvModules(RuleKeywordState):
    pass


class Group(RuleKeywordState):
    pass


class Cache(RuleKeywordState):
    @property
    def keyword(self):
        return "cache_rule"


class Handover(RuleKeywordState):
    pass


class WildcardConstraints(RuleKeywordState):
    @property
    def keyword(self):
        return "wildcard_constraints"


class Run(RuleKeywordState):
    def __init__(self, snakefile, rulename, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.rulename = rulename
        self.content = 0

    def start(self):
        yield "@workflow.run"
        yield "\n"
        yield (
            "def __rule_{rulename}(input, output, params, wildcards, threads, "
            "resources, log, version, rule, conda_env, container_img, "
            "singularity_args, use_singularity, env_modules, bench_record, jobid, "
            "is_shell, bench_iteration, cleanup_scripts, shadow_dir, edit_notebook, "
            "conda_base_path, basedir, runtime_sourcecache_path, {rule_func_marker}=True):".format(
                rulename=self.rulename
                if self.rulename is not None
                else self.snakefile.rulecount,
                rule_func_marker=common.RULEFUNC_CONTEXT_MARKER,
            )
        )

    def end(self):
        yield ""

    def block_content(self, token):
        self.content += 1
        yield token.string, token

    def is_block_end(self, token):
        return (self.content and self.line and self.indent <= 0) or is_eof(token)


class AbstractCmd(Run):

    overwrite_cmd = None
    start_func = None
    end_func = None

    def __init__(self, snakefile, rulename, base_indent=0, dedent=0, root=True):
        super().__init__(
            snakefile, rulename, base_indent=base_indent, dedent=dedent, root=root
        )
        self.cmd = list()
        self.token = None
        if self.overwrite_cmd is not None:
            self.block_content = self.overwrite_block_content

    def is_block_end(self, token):
        return (self.line and self.indent <= 0) or is_eof(token)

    def start(self):
        if self.start_func is not None:
            yield self.start_func
            yield "("

    def args(self):
        yield from []

    def end(self):
        # the end is detected. So we can savely reset the indent to zero here
        self.indent = 0
        yield "\n"
        yield ")"
        yield "\n"
        for t in super().start():
            yield t
        yield "\n"
        yield INDENT * (self.effective_indent + 1)
        yield self.end_func
        yield "("
        yield "\n".join(self.cmd)
        yield from self.args()
        yield "\n"
        yield ")"
        for t in super().end():
            yield t

    def decorate_end(self, token):
        if self.token is None:
            # no block after shell keyword
            self.error(
                "Command must be given as string after the shell keyword.", token
            )
        for t in self.end():
            yield t, self.token

    def block_content(self, token):
        self.token = token
        self.cmd.append(token.string)
        yield token.string, token

    def overwrite_block_content(self, token):
        if self.token is None:
            self.token = token
            cmd = repr(self.overwrite_cmd)
            self.cmd.append(cmd)
            yield cmd, token


class Shell(AbstractCmd):
    start_func = "@workflow.shellcmd"
    end_func = "shell"

    def args(self):
        yield ", bench_record=bench_record, bench_iteration=bench_iteration"


class Script(AbstractCmd):
    start_func = "@workflow.script"
    end_func = "script"

    def args(self):
        yield (
            ", basedir, input, output, params, wildcards, threads, resources, log, "
            "config, rule, conda_env, conda_base_path, container_img, singularity_args, env_modules, "
            "bench_record, jobid, bench_iteration, cleanup_scripts, shadow_dir, runtime_sourcecache_path"
        )


class Notebook(Script):
    start_func = "@workflow.notebook"
    end_func = "notebook"

    def args(self):
        yield (
            ", basedir, input, output, params, wildcards, threads, resources, log, "
            "config, rule, conda_env, conda_base_path, container_img, singularity_args, env_modules, "
            "bench_record, jobid, bench_iteration, cleanup_scripts, shadow_dir, "
            "edit_notebook, runtime_sourcecache_path"
        )


class Wrapper(Script):
    start_func = "@workflow.wrapper"
    end_func = "wrapper"

    def args(self):
        yield (
            ", input, output, params, wildcards, threads, resources, log, "
            "config, rule, conda_env, conda_base_path, container_img, singularity_args, env_modules, "
            "bench_record, workflow.wrapper_prefix, jobid, bench_iteration, "
            "cleanup_scripts, shadow_dir, runtime_sourcecache_path"
        )


class CWL(Script):
    start_func = "@workflow.cwl"
    end_func = "cwl"

    def args(self):
        yield (
            ", basedir, input, output, params, wildcards, threads, resources, log, "
            "config, rule, use_singularity, bench_record, jobid, runtime_sourcecache_path"
        )


rule_property_subautomata = dict(
    name=Name,
    input=Input,
    output=Output,
    params=Params,
    threads=Threads,
    resources=Resources,
    priority=Priority,
    version=Version,
    log=Log,
    message=Message,
    benchmark=Benchmark,
    conda=Conda,
    singularity=Singularity,
    container=Container,
    containerized=Containerized,
    envmodules=EnvModules,
    wildcard_constraints=WildcardConstraints,
    shadow=Shadow,
    group=Group,
    cache=Cache,
    handover=Handover,
)


class Rule(GlobalKeywordState):
    subautomata = dict(
        run=Run,
        shell=Shell,
        script=Script,
        notebook=Notebook,
        wrapper=Wrapper,
        cwl=CWL,
        **rule_property_subautomata,
    )

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.state = self.name
        self.lineno = None
        self.rulename = None
        self.run = False
        self.snakefile.rulecount += 1

    def start(self, aux=""):
        yield (
            "@workflow.rule(name={rulename!r}, lineno={lineno}, "
            "snakefile={snakefile!r}{aux})".format(
                rulename=self.rulename,
                lineno=self.lineno,
                snakefile=self.snakefile.path,
                aux=aux,
            )
        )

    def end(self):
        if not self.run:
            yield "@workflow.norun()"
            yield "\n"
            for t in self.subautomaton("run", rulename=self.rulename).start():
                yield t
            # the end is detected.
            # So we can savely reset the indent to zero here
            self.indent = 0
            yield "\n"
            yield INDENT * (self.effective_indent + 1)
            yield "pass"

    def name(self, token):
        if is_name(token):
            self.rulename = token.string
        elif is_colon(token):
            self.lineno = self.snakefile.lines + 1
            self.state = self.block
            for t in self.start():
                yield t, token
        else:
            self.error(
                "Expected name or colon after " "rule or checkpoint keyword.", token
            )

    def block_content(self, token):
        if is_name(token):
            try:
                if (
                    token.string == "run"
                    or token.string == "shell"
                    or token.string == "script"
                    or token.string == "wrapper"
                    or token.string == "cwl"
                ):
                    if self.run:
                        raise self.error(
                            "Multiple run or shell keywords in rule {}.".format(
                                self.rulename
                            ),
                            token,
                        )
                    self.run = True
                elif self.run:
                    raise self.error(
                        "No rule keywords allowed after "
                        "run/shell/script/wrapper/cwl in "
                        "rule {}.".format(self.rulename),
                        token,
                    )
                for t in self.subautomaton(
                    token.string, rulename=self.rulename
                ).consume():
                    yield t
            except KeyError:
                self.error(
                    "Unexpected keyword {} in rule definition".format(token.string),
                    token,
                )
            except StopAutomaton as e:
                self.indentation(e.token)
                for t in self.block(e.token):
                    yield t
        elif is_comment(token):
            yield "\n", token
            yield token.string, token
        elif is_string(token):
            yield "\n", token
            yield "@workflow.docstring({})".format(token.string), token
        else:
            self.error(
                "Expecting rule keyword, comment or docstrings "
                "inside a rule definition.",
                token,
            )

    @property
    def dedent(self):
        return self.indent


class Checkpoint(Rule):
    def start(self):
        yield from super().start(aux=", checkpoint=True")


class OnSuccess(DecoratorKeywordState):
    decorator = "onsuccess"
    args = ["log"]


class OnError(DecoratorKeywordState):
    decorator = "onerror"
    args = ["log"]


class OnStart(DecoratorKeywordState):
    decorator = "onstart"
    args = ["log"]


# modules


class ModuleKeywordState(SectionKeywordState):
    prefix = "Module"


class ModuleSnakefile(ModuleKeywordState):
    pass


class ModuleMetaWrapper(ModuleKeywordState):
    @property
    def keyword(self):
        return "meta_wrapper"


class ModuleConfig(ModuleKeywordState):
    pass


class ModuleSkipValidation(ModuleKeywordState):
    @property
    def keyword(self):
        return "skip_validation"


class ModuleReplacePrefix(ModuleKeywordState):
    @property
    def keyword(self):
        return "replace_prefix"


class Module(GlobalKeywordState):
    subautomata = dict(
        snakefile=ModuleSnakefile,
        meta_wrapper=ModuleMetaWrapper,
        config=ModuleConfig,
        skip_validation=ModuleSkipValidation,
        replace_prefix=ModuleReplacePrefix,
    )

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.state = self.name
        self.has_snakefile = False
        self.has_meta_wrapper = False
        self.has_name = False
        self.primary_token = None

    def end(self):
        if not (self.has_snakefile or self.has_meta_wrapper):
            self.error(
                "A module needs either a path to a Snakefile or a meta wrapper URL.",
                self.primary_token,
            )
        yield ")"

    def name(self, token):
        if is_name(token):
            yield "workflow.module({name!r}".format(name=token.string), token
            self.has_name = True
        elif is_colon(token) and self.has_name:
            self.primary_token = token
            self.state = self.block
        else:
            self.error("Expected name after module keyword.", token)

    def block_content(self, token):
        if is_name(token):
            try:
                if token.string == "snakefile":
                    self.has_snakefile = True
                if token.string == "meta_wrapper":
                    self.has_meta_wrapper = True
                for t in self.subautomaton(token.string).consume():
                    yield t
            except KeyError:
                self.error(
                    "Unexpected keyword {} in "
                    "module definition".format(token.string),
                    token,
                )
            except StopAutomaton as e:
                self.indentation(e.token)
                for t in self.block(e.token):
                    yield t
        elif is_comment(token):
            yield "\n", token
            yield token.string, token
        elif is_string(token):
            # ignore docstring
            pass
        else:
            self.error(
                "Expecting module keyword, comment or docstrings "
                "inside a module definition.",
                token,
            )


class UseRule(GlobalKeywordState):
    subautomata = rule_property_subautomata

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.state = self.state_keyword_rule
        self.rules = []
        self.has_with = False
        self.name_modifier = []
        self.from_module = None
        self._with_block = []
        self.lineno = self.snakefile.lines + 1

    def end(self):
        name_modifier = "".join(self.name_modifier) if self.name_modifier else None
        yield "@workflow.userule(rules={!r}, from_module={!r}, name_modifier={!r}, lineno={})".format(
            self.rules, self.from_module, name_modifier, self.lineno
        )
        yield "\n"

        # yield with block
        yield from self._with_block

        yield "@workflow.run"
        yield "\n"

        rulename = self.rules[0]
        if rulename == "*":
            rulename = "__allrules__"
        yield "def __userule_{}_{}():".format(self.from_module, rulename)
        # the end is detected.
        # So we can savely reset the indent to zero here
        self.indent = 0
        yield "\n"
        yield INDENT * (self.effective_indent + 1)
        yield "pass"

    def state_keyword_rule(self, token):
        if is_name(token) and token.string == "rule":
            self.state = self.state_rules_rule
            yield from ()
        else:
            self.error("Expecting keyword 'rule' after keyword 'use'", token)

    def state_rules_rule(self, token):
        if is_name(token):
            if token.string == "from" or token.string == "as" and not self.rules:
                self.error("Expecting rule names after 'use rule' statement.", token)

            self.rules.append(token.string)
            self.state = self.state_rules_comma_or_end
            yield from ()
        elif is_op(token):
            if token.string == "*":
                self.rules.append(token.string)
                self.state = self.state_rules_end
                yield from ()
            else:
                self.error(
                    "Expecting rule name or '*' after 'use rule' statement.", token
                )
        else:
            self.error(
                "Expecting rule listing (comma separated) after 'use rule' statement.",
                token,
            )
        # TODO newline and parentheses handling

    def state_rules_end(self, token):
        if is_name(token) and token.string == "from":
            self.state = self.state_from
            yield from ()
        else:
            self.error(
                "Expecting list of rules in 'use rule' statement to end with keyword 'from'.",
                token,
            )

    def state_rules_comma_or_end(self, token):
        if is_name(token):
            if token.string == "from" or token.string == "as":
                if not self.rules:
                    self.error(
                        "Expecting rule names after 'use rule' statement.", token
                    )
                if token.string == "from":
                    self.state = self.state_from
                else:
                    self.state = self.state_as
                yield from ()
            else:
                self.error(
                    "Expecting list of rules in 'use rule' statement to end with keyword 'from'.",
                    token,
                )
        elif is_comma(token):
            self.state = self.state_rules_rule
            yield from ()
        else:
            self.error(
                "Unexpected token in list of rules within 'use rule' statement.", token
            )

    def state_from(self, token):
        if is_name(token):
            self.state = self.state_modifier
            self.from_module = token.string
            yield from ()
        else:
            self.error(
                "Expecting module name after 'from' keyword in 'use rule' statement.",
                token,
            )

    def state_modifier(self, token):
        if is_name(token):
            if token.string == "as" and not self.name_modifier:
                self.state = self.state_as
                yield from ()
            elif token.string == "with":
                yield from self.handle_with(token)
            else:
                self.error(
                    "Expecting at most one 'as' or 'with' statement, or the end of the line.",
                    token,
                )
        elif is_newline(token) or is_comment(token) or is_eof(token):
            # end of the statement, close block manually
            yield from self.block(token)
        else:
            self.error(
                "Expecting either 'as', 'with' or end of line in 'use rule' statement.",
                token,
            )

    def handle_with(self, token):
        if "*" in self.rules:
            self.error(
                "Keyword 'with' in 'use rule' statement is not allowed in combination with rule pattern '*'.",
                token,
            )
        self.has_with = True
        self.state = self.state_with
        yield from ()

    def state_as(self, token):
        if is_name(token):
            if token.string != "with":
                self.name_modifier.append(token.string)
                yield from ()
            else:
                yield from self.handle_with(token)
        elif is_op(token) and token.string == "*":
            self.name_modifier.append(token.string)
            yield from ()
        elif is_newline(token) or is_comment(token) or is_eof(token):
            # end of the statement, close block manually
            yield from self.block(token)
        else:
            self.error(
                "Expecting rulename modifying pattern (e.g. modulename_*) after 'as' keyword.",
                token,
            )

    def state_with(self, token):
        if is_colon(token):
            self.state = self.block
            yield from ()
        else:
            self.error(
                "Expecting colon after 'with' keyword in 'use rule' statement.", token
            )

    def block_content(self, token):
        if is_comment(token):
            yield "\n", token
            yield token.string, token
        elif is_name(token):
            try:
                self._with_block.extend(self.subautomaton(token.string).consume())
                yield from ()
            except KeyError:
                self.error(
                    "Unexpected keyword {} in rule definition".format(token.string),
                    token,
                )
            except StopAutomaton as e:
                self.indentation(e.token)
                self.block(e.token)
        else:
            self.error(
                "Expecting a keyword or comment "
                "inside a 'use rule ... with:' statement.",
                token,
            )

    @property
    def dedent(self):
        return self.indent


class Python(TokenAutomaton):

    subautomata = dict(
        envvars=Envvars,
        include=Include,
        workdir=Workdir,
        configfile=Configfile,
        pepfile=Pepfile,
        pepschema=Pepschema,
        report=Report,
        ruleorder=Ruleorder,
        rule=Rule,
        checkpoint=Checkpoint,
        subworkflow=Subworkflow,
        localrules=Localrules,
        onsuccess=OnSuccess,
        onerror=OnError,
        onstart=OnStart,
        wildcard_constraints=GlobalWildcardConstraints,
        singularity=GlobalSingularity,
        container=GlobalContainer,
        containerized=GlobalContainerized,
        scattergather=Scattergather,
        module=Module,
        use=UseRule,
    )

    def __init__(self, snakefile, base_indent=0, dedent=0, root=True):
        super().__init__(snakefile, base_indent=base_indent, dedent=dedent, root=root)
        self.state = self.python

    def python(self, token):
        if not (is_indent(token) or is_dedent(token)):
            if self.lasttoken is None or self.lasttoken.isspace():
                try:
                    for t in self.subautomaton(token.string).consume():
                        yield t
                except KeyError:
                    yield token.string, token
                except StopAutomaton as e:
                    self.indentation(e.token)
                    for t in self.python(e.token):
                        yield t
            else:
                yield token.string, token


class Snakefile:
    def __init__(self, path, workflow, rulecount=0):
        self.path = path.get_path_or_uri()
        self.file = workflow.sourcecache.open(path)
        self.tokens = tokenize.generate_tokens(self.file.readline)
        self.rulecount = rulecount
        self.lines = 0

    def __next__(self):
        return next(self.tokens)

    def __iter__(self):
        return self

    def __enter__(self):
        return self

    def __exit__(self, *args):
        self.file.close()


def format_tokens(tokens):
    t_ = None
    for t in tokens:
        if t_ and not t.isspace() and not t_.isspace():
            yield " "
        yield t
        t_ = t


def parse(path, workflow, overwrite_shellcmd=None, rulecount=0):
    Shell.overwrite_cmd = overwrite_shellcmd
    with Snakefile(path, workflow, rulecount=rulecount) as snakefile:
        automaton = Python(snakefile)
        linemap = dict()
        compilation = list()
        for t, orig_token in automaton.consume():
            l = lineno(orig_token)
            linemap.update(
                dict(
                    (i, l)
                    for i in range(
                        snakefile.lines + 1, snakefile.lines + t.count("\n") + 1
                    )
                )
            )
            snakefile.lines += t.count("\n")
            compilation.append(t)
        compilation = "".join(format_tokens(compilation))
        if linemap:
            last = max(linemap)
            linemap[last + 1] = linemap[last]
        return compilation, linemap, snakefile.rulecount