__init__.py 9.33 KB
Newer Older
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
1 2
from future import standard_library
standard_library.install_aliases()
3

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
4
import os.path
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
5
import os
6
import pickle
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
7
import logging
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
8
import copy
9 10 11 12 13 14 15

import fnmatch
import inspect
import sys
import subprocess
import importlib
import yaml
16
import threading
17

18 19
from contextlib import contextmanager

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
20
from .. import models, utils
21
from ..api import API_PARAMS
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
22 23 24

logger = logging.getLogger(__name__)

25

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
26
class Plugin(models.Plugin):
27
    def __init__(self, info=None, data_folder=None):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
28 29 30 31
        """
        Provides a canonical name for plugins and serves as base for other
        kinds of plugins.
        """
32
        if not info:
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
33 34
            raise models.Error(message=("You need to provide configuration"
                                        "information for the plugin."))
35
        logger.debug("Initialising {}".format(info))
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
36
        id = 'plugins/{}_{}'.format(info['name'], info['version'])
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
37
        super(Plugin, self).__init__(id=id, **info)
38
        self.is_activated = False
39
        self._lock = threading.Lock()
40
        self.data_folder = data_folder or os.getcwd()
41

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
42 43 44
    def get_folder(self):
        return os.path.dirname(inspect.getfile(self.__class__))

45 46 47 48 49 50
    def activate(self):
        pass

    def deactivate(self):
        pass

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
51
    def test(self):
52 53 54 55
        if not hasattr(self, 'test_cases'):
            import inspect
            raise AttributeError(('Plugin {} [{}] does not have any defined '
                                  'test cases').format(self.id, inspect.getfile(self.__class__)))
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
56 57 58 59 60 61
        for case in self.test_cases:
            res = list(self.analyse_entry(models.Entry(case['entry']),
                                          case['params']))
            exp = case['expected']
            if not isinstance(exp, list):
                exp = [exp]
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
62
            utils.check_template(res, exp)
63 64
            for r in res:
                r.validate()
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
65

66 67 68 69 70 71 72
    @contextmanager
    def open(self, fpath, *args, **kwargs):
        if not os.path.isabs(fpath):
            fpath = os.path.join(self.data_folder, fpath)
        with open(fpath, *args, **kwargs) as f:
            yield f

73

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
74 75 76 77
SenpyPlugin = Plugin


class AnalysisPlugin(Plugin):
78

79
    def analyse(self, *args, **kwargs):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
80 81 82 83 84 85 86 87 88 89
        raise NotImplemented(
            'Your method should implement either analyse or analyse_entry')

    def analyse_entry(self, entry, parameters):
        """ An implemented plugin should override this method.
        This base method is here to adapt old style plugins which only
        implement the *analyse* function.
        Note that this method may yield an annotated entry or a list of
        entries (e.g. in a tokenizer)
        """
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
90
        text = entry['nif:isString']
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
91 92 93 94 95
        params = copy.copy(parameters)
        params['input'] = text
        results = self.analyse(**params)
        for i in results.entries:
            yield i
96

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
97 98 99 100 101 102
    def analyse_entries(self, entries, parameters):
        for entry in entries:
            logger.debug('Analysing entry with plugin {}: {}'.format(self, entry))
            for result in self.analyse_entry(entry, parameters):
                yield result

103

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
104
class ConversionPlugin(Plugin):
105
    pass
106

107

108
class SentimentPlugin(models.SentimentPlugin, AnalysisPlugin):
109 110 111 112
    def __init__(self, info, *args, **kwargs):
        super(SentimentPlugin, self).__init__(info, *args, **kwargs)
        self.minPolarityValue = float(info.get("minPolarityValue", 0))
        self.maxPolarityValue = float(info.get("maxPolarityValue", 1))
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
113

114

115
class EmotionPlugin(models.EmotionPlugin, AnalysisPlugin):
116
    def __init__(self, info, *args, **kwargs):
117
        super(EmotionPlugin, self).__init__(info, *args, **kwargs)
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
118 119 120 121
        self.minEmotionValue = float(info.get("minEmotionValue", -1))
        self.maxEmotionValue = float(info.get("maxEmotionValue", 1))


122 123
class EmotionConversionPlugin(models.EmotionConversionPlugin, ConversionPlugin):
    pass
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
124 125 126 127 128


class ShelfMixin(object):
    @property
    def sh(self):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
129
        if not hasattr(self, '_sh') or self._sh is None:
130 131
            self.__dict__['_sh'] = {}
            if os.path.isfile(self.shelf_file):
132
                try:
133 134
                    with self.open(self.shelf_file, 'rb') as p:
                        self.__dict__['_sh'] = pickle.load(p)
135
                except (IndexError, EOFError, pickle.UnpicklingError):
136 137 138
                    logger.warning('{} has a corrupted shelf file!'.format(self.id))
                    if not self.get('force_shelf', False):
                        raise
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
139 140 141 142
        return self._sh

    @sh.deleter
    def sh(self):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
143 144
        if os.path.isfile(self.shelf_file):
            os.remove(self.shelf_file)
145 146
            del self.__dict__['_sh']
        self.save()
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
147

J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
148 149
    @property
    def shelf_file(self):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
150
        if 'shelf_file' not in self or not self['shelf_file']:
151
            self.shelf_file = os.path.join(self.data_folder, self.name + '.p')
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
152
        return self['shelf_file']
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
153

154
    def save(self):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
155
        logger.debug('saving pickle')
156
        if hasattr(self, '_sh') and self._sh is not None:
157
            with self.open(self.shelf_file, 'wb') as f:
158
                pickle.dump(self._sh, f)
159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195


default_plugin_type = API_PARAMS['plugin_type']['default']


def pfilter(plugins, **kwargs):
    """ Filter plugins by different criteria """
    if isinstance(plugins, models.Plugins):
        plugins = plugins.plugins
    elif isinstance(plugins, dict):
        plugins = plugins.values()
    ptype = kwargs.pop('plugin_type', default_plugin_type)
    logger.debug('#' * 100)
    logger.debug('ptype {}'.format(ptype))
    if ptype:
        try:
            ptype = ptype[0].upper() + ptype[1:]
            pclass = globals()[ptype]
            logger.debug('Class: {}'.format(pclass))
            candidates = filter(lambda x: isinstance(x, pclass),
                                plugins)
        except KeyError:
            raise models.Error('{} is not a valid type'.format(ptype))
    else:
        candidates = plugins

    logger.debug(candidates)

    def matches(plug):
        res = all(getattr(plug, k, None) == v for (k, v) in kwargs.items())
        logger.debug(
            "matching {} with {}: {}".format(plug.name, kwargs, res))
        return res

    if kwargs:
        candidates = filter(matches, candidates)
    return {p.name: p for p in candidates}
196 197 198 199 200 201


def validate_info(info):
    return all(x in info for x in ('name', 'module', 'description', 'version'))


202 203 204
def load_module(name, root=None):
    if root:
        sys.path.append(root)
205
    tmp = importlib.import_module(name)
206 207
    if root:
        sys.path.remove(root)
208 209 210 211 212 213 214 215 216 217 218
    return tmp


def log_subprocess_output(process):
    for line in iter(process.stdout.readline, b''):
        logger.info('%r', line)
    for line in iter(process.stderr.readline, b''):
        logger.error('%r', line)


def install_deps(*plugins):
219
    installed = False
220 221 222
    for info in plugins:
        requirements = info.get('requirements', [])
        if requirements:
223
            pip_args = [sys.executable, '-m', 'pip', 'install', '--use-wheel']
224 225 226 227 228 229 230 231
            for req in requirements:
                pip_args.append(req)
            logger.info('Installing requirements: ' + str(requirements))
            process = subprocess.Popen(pip_args,
                                       stdout=subprocess.PIPE,
                                       stderr=subprocess.PIPE)
            log_subprocess_output(process)
            exitcode = process.wait()
232
            installed = True
233 234
            if exitcode != 0:
                raise models.Error("Dependencies not properly installed")
235
    return installed
236 237


238
def load_plugin_from_info(info, root=None, validator=validate_info, install=True, *args, **kwargs):
239 240
    if not root and '_path' in info:
        root = os.path.dirname(info['_path'])
241
    if not validator(info):
242
        raise ValueError('Plugin info is not valid: {}'.format(info))
243 244
    module = info["module"]

245 246 247 248 249 250 251
    try:
        tmp = load_module(module, root)
    except ImportError:
        if not install:
            raise
        install_deps(info)
        tmp = load_module(module, root)
252 253 254 255 256 257 258 259 260 261
    candidate = None
    for _, obj in inspect.getmembers(tmp):
        if inspect.isclass(obj) and inspect.getmodule(obj) == tmp:
            logger.debug(("Found plugin class:"
                          " {}@{}").format(obj, inspect.getmodule(obj)))
            candidate = obj
            break
    if not candidate:
        logger.debug("No valid plugin for: {}".format(module))
        return
262
    module = candidate(info=info, *args, **kwargs)
263
    return module
264 265


266
def parse_plugin_info(fpath):
267 268 269
    logger.debug("Loading plugin: {}".format(fpath))
    with open(fpath, 'r') as f:
        info = yaml.load(f)
270 271 272 273 274
    info['_path'] = fpath
    name = info['name']
    return name, info


275
def load_plugin(fpath, *args, **kwargs):
276
    name, info = parse_plugin_info(fpath)
277
    logger.debug("Info: {}".format(info))
278
    plugin = load_plugin_from_info(info, *args, **kwargs)
279
    return name, plugin
280 281


282
def load_plugins(folders, loader=load_plugin, *args, **kwargs):
283 284 285
    plugins = {}
    for search_folder in folders:
        for root, dirnames, filenames in os.walk(search_folder):
J. Fernando Sánchez's avatar
J. Fernando Sánchez committed
286 287
            # Do not look for plugins in hidden or special folders
            dirnames[:] = [d for d in dirnames if d[0] not in ['.', '_']]
288
            for filename in fnmatch.filter(filenames, '*.senpy'):
289
                fpath = os.path.join(root, filename)
290
                name, plugin = loader(fpath, *args, **kwargs)
291 292 293
                if plugin and name:
                    plugins[name] = plugin
    return plugins