summaryrefslogtreecommitdiff
path: root/lib/ansible/plugins/inventory/ini.py
blob: fa8fdc4b09b0918bc8d445f930ef358a3ffc6e4a (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
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
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
# Copyright (c) 2017 Ansible Project
# GNU General Public License v3.0+ (see COPYING or https://www.gnu.org/licenses/gpl-3.0.txt)
from __future__ import (absolute_import, division, print_function)
__metaclass__ = type

DOCUMENTATION = '''
    name: ini
    version_added: "2.4"
    short_description: Uses an Ansible INI file as inventory source.
    description:
        - INI file based inventory, sections are groups or group related with special C(:modifiers).
        - Entries in sections C([group_1]) are hosts, members of the group.
        - Hosts can have variables defined inline as key/value pairs separated by C(=).
        - The C(children) modifier indicates that the section contains groups.
        - The C(vars) modifier indicates that the section contains variables assigned to members of the group.
        - Anything found outside a section is considered an 'ungrouped' host.
        - Values passed in the INI format using the C(key=value) syntax are interpreted differently depending on where they are declared within your inventory.
        - When declared inline with the host, INI values are processed by Python's ast.literal_eval function
          (U(https://docs.python.org/3/library/ast.html#ast.literal_eval)) and interpreted as Python literal structures
          (strings, numbers, tuples, lists, dicts, booleans, None). Host lines accept multiple C(key=value) parameters per line.
          Therefore they need a way to indicate that a space is part of a value rather than a separator.
        - When declared in a C(:vars) section, INI values are interpreted as strings. For example C(var=FALSE) would create a string equal to C(FALSE).
          Unlike host lines, C(:vars) sections accept only a single entry per line, so everything after the C(=) must be the value for the entry.
        - Do not rely on types set during definition, always make sure you specify type with a filter when needed when consuming the variable.
        - See the Examples for proper quoting to prevent changes to variable type.
    notes:
        - Whitelisted in configuration by default.
        - Consider switching to YAML format for inventory sources to avoid confusion on the actual type of a variable.
          The YAML inventory plugin processes variable values consistently and correctly.
'''

EXAMPLES = '''# fmt: ini
# Example 1
[web]
host1
host2 ansible_port=222 # defined inline, interpreted as an integer

[web:vars]
http_port=8080 # all members of 'web' will inherit these
myvar=23 # defined in a :vars section, interpreted as a string

[web:children] # child groups will automatically add their hosts to parent group
apache
nginx

[apache]
tomcat1
tomcat2 myvar=34 # host specific vars override group vars
tomcat3 mysecret="'03#pa33w0rd'" # proper quoting to prevent value changes

[nginx]
jenkins1

[nginx:vars]
has_java = True # vars in child groups override same in parent

[all:vars]
has_java = False # 'all' is 'top' parent

# Example 2
host1 # this is 'ungrouped'

# both hosts have same IP but diff ports, also 'ungrouped'
host2 ansible_host=127.0.0.1 ansible_port=44
host3 ansible_host=127.0.0.1 ansible_port=45

[g1]
host4

[g2]
host4 # same host as above, but member of 2 groups, will inherit vars from both
      # inventory hostnames are unique
'''

import ast
import re

from ansible.inventory.group import to_safe_group_name
from ansible.plugins.inventory import BaseFileInventoryPlugin

from ansible.errors import AnsibleError, AnsibleParserError
from ansible.module_utils._text import to_bytes, to_text
from ansible.utils.shlex import shlex_split


class InventoryModule(BaseFileInventoryPlugin):
    """
    Takes an INI-format inventory file and builds a list of groups and subgroups
    with their associated hosts and variable settings.
    """
    NAME = 'ini'
    _COMMENT_MARKERS = frozenset((u';', u'#'))
    b_COMMENT_MARKERS = frozenset((b';', b'#'))

    def __init__(self):

        super(InventoryModule, self).__init__()

        self.patterns = {}
        self._filename = None

    def parse(self, inventory, loader, path, cache=True):

        super(InventoryModule, self).parse(inventory, loader, path)

        self._filename = path

        try:
            # Read in the hosts, groups, and variables defined in the inventory file.
            if self.loader:
                (b_data, private) = self.loader._get_file_contents(path)
            else:
                b_path = to_bytes(path, errors='surrogate_or_strict')
                with open(b_path, 'rb') as fh:
                    b_data = fh.read()

            try:
                # Faster to do to_text once on a long string than many
                # times on smaller strings
                data = to_text(b_data, errors='surrogate_or_strict').splitlines()
            except UnicodeError:
                # Handle non-utf8 in comment lines: https://github.com/ansible/ansible/issues/17593
                data = []
                for line in b_data.splitlines():
                    if line and line[0] in self.b_COMMENT_MARKERS:
                        # Replace is okay for comment lines
                        # data.append(to_text(line, errors='surrogate_then_replace'))
                        # Currently we only need these lines for accurate lineno in errors
                        data.append(u'')
                    else:
                        # Non-comment lines still have to be valid uf-8
                        data.append(to_text(line, errors='surrogate_or_strict'))

            self._parse(path, data)
        except Exception as e:
            raise AnsibleParserError(e)

    def _raise_error(self, message):
        raise AnsibleError("%s:%d: " % (self._filename, self.lineno) + message)

    def _parse(self, path, lines):
        '''
        Populates self.groups from the given array of lines. Raises an error on
        any parse failure.
        '''

        self._compile_patterns()

        # We behave as though the first line of the inventory is '[ungrouped]',
        # and begin to look for host definitions. We make a single pass through
        # each line of the inventory, building up self.groups and adding hosts,
        # subgroups, and setting variables as we go.

        pending_declarations = {}
        groupname = 'ungrouped'
        state = 'hosts'
        self.lineno = 0
        for line in lines:
            self.lineno += 1

            line = line.strip()
            # Skip empty lines and comments
            if not line or line[0] in self._COMMENT_MARKERS:
                continue

            # Is this a [section] header? That tells us what group we're parsing
            # definitions for, and what kind of definitions to expect.

            m = self.patterns['section'].match(line)
            if m:
                (groupname, state) = m.groups()

                groupname = to_safe_group_name(groupname)

                state = state or 'hosts'
                if state not in ['hosts', 'children', 'vars']:
                    title = ":".join(m.groups())
                    self._raise_error("Section [%s] has unknown type: %s" % (title, state))

                # If we haven't seen this group before, we add a new Group.
                if groupname not in self.inventory.groups:
                    # Either [groupname] or [groupname:children] is sufficient to declare a group,
                    # but [groupname:vars] is allowed only if the # group is declared elsewhere.
                    # We add the group anyway, but make a note in pending_declarations to check at the end.
                    #
                    # It's possible that a group is previously pending due to being defined as a child
                    # group, in that case we simply pass so that the logic below to process pending
                    # declarations will take the appropriate action for a pending child group instead of
                    # incorrectly handling it as a var state pending declaration
                    if state == 'vars' and groupname not in pending_declarations:
                        pending_declarations[groupname] = dict(line=self.lineno, state=state, name=groupname)

                    self.inventory.add_group(groupname)

                # When we see a declaration that we've been waiting for, we process and delete.
                if groupname in pending_declarations and state != 'vars':
                    if pending_declarations[groupname]['state'] == 'children':
                        self._add_pending_children(groupname, pending_declarations)
                    elif pending_declarations[groupname]['state'] == 'vars':
                        del pending_declarations[groupname]

                continue
            elif line.startswith('[') and line.endswith(']'):
                self._raise_error("Invalid section entry: '%s'. Please make sure that there are no spaces" % line + " " +
                                  "in the section entry, and that there are no other invalid characters")

            # It's not a section, so the current state tells us what kind of
            # definition it must be. The individual parsers will raise an
            # error if we feed them something they can't digest.

            # [groupname] contains host definitions that must be added to
            # the current group.
            if state == 'hosts':
                hosts, port, variables = self._parse_host_definition(line)
                self._populate_host_vars(hosts, variables, groupname, port)

            # [groupname:vars] contains variable definitions that must be
            # applied to the current group.
            elif state == 'vars':
                (k, v) = self._parse_variable_definition(line)
                self.inventory.set_variable(groupname, k, v)

            # [groupname:children] contains subgroup names that must be
            # added as children of the current group. The subgroup names
            # must themselves be declared as groups, but as before, they
            # may only be declared later.
            elif state == 'children':
                child = self._parse_group_name(line)
                if child not in self.inventory.groups:
                    if child not in pending_declarations:
                        pending_declarations[child] = dict(line=self.lineno, state=state, name=child, parents=[groupname])
                    else:
                        pending_declarations[child]['parents'].append(groupname)
                else:
                    self.inventory.add_child(groupname, child)
            else:
                # This can happen only if the state checker accepts a state that isn't handled above.
                self._raise_error("Entered unhandled state: %s" % (state))

        # Any entries in pending_declarations not removed by a group declaration above mean that there was an unresolved reference.
        # We report only the first such error here.
        for g in pending_declarations:
            decl = pending_declarations[g]
            if decl['state'] == 'vars':
                raise AnsibleError("%s:%d: Section [%s:vars] not valid for undefined group: %s" % (path, decl['line'], decl['name'], decl['name']))
            elif decl['state'] == 'children':
                raise AnsibleError("%s:%d: Section [%s:children] includes undefined group: %s" % (path, decl['line'], decl['parents'].pop(), decl['name']))

    def _add_pending_children(self, group, pending):
        for parent in pending[group]['parents']:
            self.inventory.add_child(parent, group)
            if parent in pending and pending[parent]['state'] == 'children':
                self._add_pending_children(parent, pending)
        del pending[group]

    def _parse_group_name(self, line):
        '''
        Takes a single line and tries to parse it as a group name. Returns the
        group name if successful, or raises an error.
        '''

        m = self.patterns['groupname'].match(line)
        if m:
            return m.group(1)

        self._raise_error("Expected group name, got: %s" % (line))

    def _parse_variable_definition(self, line):
        '''
        Takes a string and tries to parse it as a variable definition. Returns
        the key and value if successful, or raises an error.
        '''

        # TODO: We parse variable assignments as a key (anything to the left of
        # an '='"), an '=', and a value (anything left) and leave the value to
        # _parse_value to sort out. We should be more systematic here about
        # defining what is acceptable, how quotes work, and so on.

        if '=' in line:
            (k, v) = [e.strip() for e in line.split("=", 1)]
            return (k, self._parse_value(v))

        self._raise_error("Expected key=value, got: %s" % (line))

    def _parse_host_definition(self, line):
        '''
        Takes a single line and tries to parse it as a host definition. Returns
        a list of Hosts if successful, or raises an error.
        '''

        # A host definition comprises (1) a non-whitespace hostname or range,
        # optionally followed by (2) a series of key="some value" assignments.
        # We ignore any trailing whitespace and/or comments. For example, here
        # are a series of host definitions in a group:
        #
        # [groupname]
        # alpha
        # beta:2345 user=admin      # we'll tell shlex
        # gamma sudo=True user=root # to ignore comments

        try:
            tokens = shlex_split(line, comments=True)
        except ValueError as e:
            self._raise_error("Error parsing host definition '%s': %s" % (line, e))

        (hostnames, port) = self._expand_hostpattern(tokens[0])

        # Try to process anything remaining as a series of key=value pairs.
        variables = {}
        for t in tokens[1:]:
            if '=' not in t:
                self._raise_error("Expected key=value host variable assignment, got: %s" % (t))
            (k, v) = t.split('=', 1)
            variables[k] = self._parse_value(v)

        return hostnames, port, variables

    def _expand_hostpattern(self, hostpattern):
        '''
        do some extra checks over normal processing
        '''
        # specification?

        hostnames, port = super(InventoryModule, self)._expand_hostpattern(hostpattern)

        if hostpattern.strip().endswith(':') and port is None:
            raise AnsibleParserError("Invalid host pattern '%s' supplied, ending in ':' is not allowed, this character is reserved to provide a port." %
                                     hostpattern)
        for pattern in hostnames:
            # some YAML parsing prevention checks
            if pattern.strip() == '---':
                raise AnsibleParserError("Invalid host pattern '%s' supplied, '---' is normally a sign this is a YAML file." % hostpattern)

        return (hostnames, port)

    @staticmethod
    def _parse_value(v):
        '''
        Attempt to transform the string value from an ini file into a basic python object
        (int, dict, list, unicode string, etc).
        '''
        try:
            v = ast.literal_eval(v)
        # Using explicit exceptions.
        # Likely a string that literal_eval does not like. We wil then just set it.
        except ValueError:
            # For some reason this was thought to be malformed.
            pass
        except SyntaxError:
            # Is this a hash with an equals at the end?
            pass
        return to_text(v, nonstring='passthru', errors='surrogate_or_strict')

    def _compile_patterns(self):
        '''
        Compiles the regular expressions required to parse the inventory and
        stores them in self.patterns.
        '''

        # Section names are square-bracketed expressions at the beginning of a
        # line, comprising (1) a group name optionally followed by (2) a tag
        # that specifies the contents of the section. We ignore any trailing
        # whitespace and/or comments. For example:
        #
        # [groupname]
        # [somegroup:vars]
        # [naughty:children] # only get coal in their stockings

        self.patterns['section'] = re.compile(
            to_text(r'''^\[
                    ([^:\]\s]+)             # group name (see groupname below)
                    (?::(\w+))?             # optional : and tag name
                \]
                \s*                         # ignore trailing whitespace
                (?:\#.*)?                   # and/or a comment till the
                $                           # end of the line
            ''', errors='surrogate_or_strict'), re.X
        )

        # FIXME: What are the real restrictions on group names, or rather, what
        # should they be? At the moment, they must be non-empty sequences of non
        # whitespace characters excluding ':' and ']', but we should define more
        # precise rules in order to support better diagnostics.

        self.patterns['groupname'] = re.compile(
            to_text(r'''^
                ([^:\]\s]+)
                \s*                         # ignore trailing whitespace
                (?:\#.*)?                   # and/or a comment till the
                $                           # end of the line
            ''', errors='surrogate_or_strict'), re.X
        )