lib/couchdb/client.py

# -*- coding: utf-8 -*-
#
# Copyright (C) 2007-2009 Christopher Lenz
# All rights reserved.
#
# This software is licensed as described in the file COPYING, which
# you should have received as part of this distribution.

"""Python client API for CouchDB.

>>> server = Server()
>>> db = server.create('python-tests')
>>> doc_id, doc_rev = db.save({'type': 'Person', 'name': 'John Doe'})
>>> doc = db[doc_id]
>>> doc['type']
'Person'
>>> doc['name']
'John Doe'
>>> del db[doc.id]
>>> doc.id in db
False

>>> del server['python-tests']
"""

import mimetypes
import os
from types import FunctionType
from inspect import getsource
from textwrap import dedent
import re
import warnings

from couchdb import http, json

__all__ = ['Server', 'Database', 'Document', 'ViewResults', 'Row']
__docformat__ = 'restructuredtext en'


DEFAULT_BASE_URL = os.environ.get('COUCHDB_URL', 'http://localhost:5984/')


class Server(object):
    """Representation of a CouchDB server.

    >>> server = Server()

    This class behaves like a dictionary of databases. For example, to get a
    list of database names on the server, you can simply iterate over the
    server object.

    New databases can be created using the `create` method:

    >>> db = server.create('python-tests')
    >>> db
    <Database 'python-tests'>

    You can access existing databases using item access, specifying the database
    name as the key:

    >>> db = server['python-tests']
    >>> db.name
    'python-tests'

    Databases can be deleted using a ``del`` statement:

    >>> del server['python-tests']
    """

    def __init__(self, url=DEFAULT_BASE_URL, full_commit=True, session=None):
        """Initialize the server object.

        :param url: the URI of the server (for example
                    ``http://localhost:5984/``)
        :param full_commit: turn on the X-Couch-Full-Commit header
        :param session: an http.Session instance or None for a default session
        """
        if isinstance(url, basestring):
            self.resource = http.Resource(url, session or http.Session())
        else:
            self.resource = url # treat as a Resource object
        if not full_commit:
            self.resource.headers['X-Couch-Full-Commit'] = 'false'

    def __contains__(self, name):
        """Return whether the server contains a database with the specified
        name.

        :param name: the database name
        :return: `True` if a database with the name exists, `False` otherwise
        """
        try:
            self.resource.head(validate_dbname(name))
            return True
        except http.ResourceNotFound:
            return False

    def __iter__(self):
        """Iterate over the names of all databases."""
        status, headers, data = self.resource.get_json('_all_dbs')
        return iter(data)

    def __len__(self):
        """Return the number of databases."""
        status, headers, data = self.resource.get_json('_all_dbs')
        return len(data)

    def __nonzero__(self):
        """Return whether the server is available."""
        try:
            self.resource.head()
            return True
        except:
            return False

    def __repr__(self):
        return '<%s %r>' % (type(self).__name__, self.resource.url)

    def __delitem__(self, name):
        """Remove the database with the specified name.

        :param name: the name of the database
        :raise ResourceNotFound: if no database with that name exists
        """
        self.resource.delete_json(validate_dbname(name))

    def __getitem__(self, name):
        """Return a `Database` object representing the database with the
        specified name.

        :param name: the name of the database
        :return: a `Database` object representing the database
        :rtype: `Database`
        :raise ResourceNotFound: if no database with that name exists
        """
        db = Database(self.resource(name), validate_dbname(name))
        db.resource.head() # actually make a request to the database
        return db

    def config(self):
        """The configuration of the CouchDB server.

        The configuration is represented as a nested dictionary of sections and
        options from the configuration files of the server, or the default
        values for options that are not explicitly configured.

        :rtype: `dict`
        """
        status, headers, data = self.resource.get_json('_config')
        return data

    def version(self):
        """The version string of the CouchDB server.

        Note that this results in a request being made, and can also be used
        to check for the availability of the server.

        :rtype: `unicode`"""
        status, headers, data = self.resource.get_json()
        return data['version']

    def stats(self, name=None):
        """Server statistics.

        :param name: name of single statistic, e.g. httpd/requests
                     (None -- return all statistics)
        """
        if not name:
            resource = self.resource('_stats')
        else:
            resource = self.resource('_stats', *name.split('/'))
        status, headers, data = resource.get_json()
        return data

    def tasks(self):
        """A list of tasks currently active on the server."""
        status, headers, data = self.resource.get_json('_active_tasks')
        return data

    def uuids(self, count=None):
        """Retrieve a batch of uuids

        :param count: a number of uuids to fetch
                      (None -- get as many as the server sends)
        :return: a list of uuids
        """
        if count is None:
            _, _, data = self.resource.get_json('_uuids')
        else:
            _, _, data = self.resource.get_json('_uuids', count=count)
        return data['uuids']

    def create(self, name):
        """Create a new database with the given name.

        :param name: the name of the database
        :return: a `Database` object representing the created database
        :rtype: `Database`
        :raise PreconditionFailed: if a database with that name already exists
        """
        self.resource.put_json(validate_dbname(name))
        return self[name]

    def delete(self, name):
        """Delete the database with the specified name.

        :param name: the name of the database
        :raise ResourceNotFound: if a database with that name does not exist
        :since: 0.6
        """
        del self[name]

    def replicate(self, source, target, **options):
        """Replicate changes from the source database to the target database.

        :param source: URL of the source database
        :param target: URL of the target database
        :param options: optional replication args, e.g. continuous=True
        """
        data = {'source': source, 'target': target}
        data.update(options)
        status, headers, data = self.resource.post_json('_replicate', data)
        return data


class Database(object):
    """Representation of a database on a CouchDB server.

    >>> server = Server()
    >>> db = server.create('python-tests')

    New documents can be added to the database using the `save()` method:

    >>> doc_id, doc_rev = db.save({'type': 'Person', 'name': 'John Doe'})

    This class provides a dictionary-like interface to databases: documents are
    retrieved by their ID using item access

    >>> doc = db[doc_id]
    >>> doc                 #doctest: +ELLIPSIS
    <Document '...'@... {...}>

    Documents are represented as instances of the `Row` class, which is
    basically just a normal dictionary with the additional attributes ``id`` and
    ``rev``:

    >>> doc.id, doc.rev     #doctest: +ELLIPSIS
    ('...', ...)
    >>> doc['type']
    'Person'
    >>> doc['name']
    'John Doe'

    To update an existing document, you use item access, too:

    >>> doc['name'] = 'Mary Jane'
    >>> db[doc.id] = doc

    The `save()` method creates a document with a random ID generated by
    CouchDB (which is not recommended). If you want to explicitly specify the
    ID, you'd use item access just as with updating:

    >>> db['JohnDoe'] = {'type': 'person', 'name': 'John Doe'}

    >>> 'JohnDoe' in db
    True
    >>> len(db)
    2

    >>> del server['python-tests']
    """

    def __init__(self, url, name=None, session=None):
        if isinstance(url, basestring):
            if not url.startswith('http'):
                url = DEFAULT_BASE_URL + url
            self.resource = http.Resource(url, session)
        else:
            self.resource = url
        self._name = name

    def __repr__(self):
        return '<%s %r>' % (type(self).__name__, self.name)

    def __contains__(self, id):
        """Return whether the database contains a document with the specified
        ID.

        :param id: the document ID
        :return: `True` if a document with the ID exists, `False` otherwise
        """
        try:
            _doc_resource(self.resource, id).head()
            return True
        except http.ResourceNotFound:
            return False

    def __iter__(self):
        """Return the IDs of all documents in the database."""
        return iter([item.id for item in self.view('_all_docs')])

    def __len__(self):
        """Return the number of documents in the database."""
        _, _, data = self.resource.get_json()
        return data['doc_count']

    def __nonzero__(self):
        """Return whether the database is available."""
        try:
            self.resource.head()
            return True
        except:
            return False

    def __delitem__(self, id):
        """Remove the document with the specified ID from the database.

        :param id: the document ID
        """
        resource = _doc_resource(self.resource, id)
        status, headers, data = resource.head()
        resource.delete_json(rev=headers['etag'].strip('"'))

    def __getitem__(self, id):
        """Return the document with the specified ID.

        :param id: the document ID
        :return: a `Row` object representing the requested document
        :rtype: `Document`
        """
        _, _, data = _doc_resource(self.resource, id).get_json()
        return Document(data)

    def __setitem__(self, id, content):
        """Create or update a document with the specified ID.

        :param id: the document ID
        :param content: the document content; either a plain dictionary for
                        new documents, or a `Row` object for existing
                        documents
        """
        resource = _doc_resource(self.resource, id)
        status, headers, data = resource.put_json(body=content)
        content.update({'_id': data['id'], '_rev': data['rev']})

    @property
    def name(self):
        """The name of the database.

        Note that this may require a request to the server unless the name has
        already been cached by the `info()` method.

        :rtype: basestring
        """
        if self._name is None:
            self.info()
        return self._name

    def create(self, data):
        """Create a new document in the database with a random ID that is
        generated by the server.

        Note that it is generally better to avoid the `create()` method and
        instead generate document IDs on the client side. This is due to the
        fact that the underlying HTTP ``POST`` method is not idempotent, and
        an automatic retry due to a problem somewhere on the networking stack
        may cause multiple documents being created in the database.

        To avoid such problems you can generate a UUID on the client side.
        Python (since version 2.5) comes with a ``uuid`` module that can be
        used for this::

            from uuid import uuid4
            doc_id = uuid4().hex
            db[doc_id] = {'type': 'person', 'name': 'John Doe'}

        :param data: the data to store in the document
        :return: the ID of the created document
        :rtype: `unicode`
        """
        warnings.warn('Database.create is deprecated, please use Database.save instead [2010-04-13]',
                      DeprecationWarning, stacklevel=2)
        _, _, data = self.resource.post_json(body=data)
        return data['id']

    def multisave(self, docs, **options):
        """Perform a bulk insertion of the given documents.
        """
        responses = []
        for doc in docs:
            id, rev = self.save(doc, **options)
            responses.append((id, rev))

    def save(self, doc, **options):
        """Create a new document or update an existing document.

        If doc has no _id then the server will allocate a random ID and a new
        document will be created. Otherwise the doc's _id will be used to
        identity the document to create or update. Trying to update an existing
        document with an incorrect _rev will raise a ResourceConflict exception.

        Note that it is generally better to avoid saving documents with no _id
        and instead generate document IDs on the client side. This is due to
        the fact that the underlying HTTP ``POST`` method is not idempotent,
        and an automatic retry due to a problem somewhere on the networking
        stack may cause multiple documents being created in the database.

        To avoid such problems you can generate a UUID on the client side.
        Python (since version 2.5) comes with a ``uuid`` module that can be
        used for this::

            from uuid import uuid4
            doc = {'_id': uuid4().hex, 'type': 'person', 'name': 'John Doe'}
            db.save(doc)

        :param doc: the document to store
        :param options: optional args, e.g. batch='ok'
        :return: (id, rev) tuple of the save document
        :rtype: `tuple`
        """
        if '_id' in doc:
            func = _doc_resource(self.resource, doc['_id']).put_json
        else:
            func = self.resource.post_json
        _, _, data = func(body=doc, **options)
        id, rev = data['id'], data.get('rev')
        doc['_id'] = id
        if rev is not None: # Not present for batch='ok'
            doc['_rev'] = rev
        return id, rev

    def cleanup(self):
        """Clean up old design document indexes.

        Remove all unused index files from the database storage area.

        :return: a boolean to indicate successful cleanup initiation
        :rtype: `bool`
        """
        headers = {'Content-Type': 'application/json'}
        _, _, data = self.resource('_view_cleanup').post_json(headers=headers)
        return data['ok']

    def commit(self):
        """If the server is configured to delay commits, or previous requests
        used the special ``X-Couch-Full-Commit: false`` header to disable
        immediate commits, this method can be used to ensure that any
        non-committed changes are committed to physical storage.
        """
        _, _, data = self.resource.post_json(
            '_ensure_full_commit',
            headers={'Content-Type': 'application/json'})
        return data

    def compact(self, ddoc=None):
        """Compact the database or a design document's index.

        Without an argument, this will try to prune all old revisions from the
        database. With an argument, it will compact the index cache for all
        views in the design document specified.

        :return: a boolean to indicate whether the compaction was initiated
                 successfully
        :rtype: `bool`
        """
        if ddoc:
            resource = self.resource('_compact', ddoc)
        else:
            resource = self.resource('_compact')
        _, _, data = resource.post_json(
            headers={'Content-Type': 'application/json'})
        return data['ok']

    def copy(self, src, dest):
        """Copy the given document to create a new document.

        :param src: the ID of the document to copy, or a dictionary or
                    `Document` object representing the source document.
        :param dest: either the destination document ID as string, or a
                     dictionary or `Document` instance of the document that
                     should be overwritten.
        :return: the new revision of the destination document
        :rtype: `str`
        :since: 0.6
        """
        if not isinstance(src, basestring):
            if not isinstance(src, dict):
                if hasattr(src, 'items'):
                    src = dict(src.items())
                else:
                    raise TypeError('expected dict or string, got %s' %
                                    type(src))
            src = src['_id']

        if not isinstance(dest, basestring):
            if not isinstance(dest, dict):
                if hasattr(dest, 'items'):
                    dest = dict(dest.items())
                else:
                    raise TypeError('expected dict or string, got %s' %
                                    type(dest))
            if '_rev' in dest:
                dest = '%s?%s' % (http.quote(dest['_id']),
                                  http.urlencode({'rev': dest['_rev']}))
            else:
                dest = http.quote(dest['_id'])

        _, _, data = self.resource._request('COPY', src,
                                            headers={'Destination': dest})
        data = json.decode(data.read())
        return data['rev']

    def delete(self, doc):
        """Delete the given document from the database.

        Use this method in preference over ``__del__`` to ensure you're
        deleting the revision that you had previously retrieved. In the case
        the document has been updated since it was retrieved, this method will
        raise a `ResourceConflict` exception.

        >>> server = Server()
        >>> db = server.create('python-tests')

        >>> doc = dict(type='Person', name='John Doe')
        >>> db['johndoe'] = doc
        >>> doc2 = db['johndoe']
        >>> doc2['age'] = 42
        >>> db['johndoe'] = doc2
        >>> db.delete(doc)
        Traceback (most recent call last):
          ...
        ResourceConflict: ('conflict', 'Document update conflict.')

        >>> del server['python-tests']

        :param doc: a dictionary or `Document` object holding the document data
        :raise ResourceConflict: if the document was updated in the database
        :since: 0.4.1
        """
        if doc['_id'] is None:
            raise ValueError('document ID cannot be None')
        _doc_resource(self.resource, doc['_id']).delete_json(rev=doc['_rev'])

    def get(self, id, default=None, **options):
        """Return the document with the specified ID.

        :param id: the document ID
        :param default: the default value to return when the document is not
                        found
        :return: a `Row` object representing the requested document, or `None`
                 if no document with the ID was found
        :rtype: `Document`
        """
        try:
            _, _, data = _doc_resource(self.resource, id).get_json(**options)
        except http.ResourceNotFound:
            return default
        if hasattr(data, 'items'):
            return Document(data)
        else:
            return data

    def revisions(self, id, **options):
        """Return all available revisions of the given document.

        :param id: the document ID
        :return: an iterator over Document objects, each a different revision,
                 in reverse chronological order, if any were found
        """
        try:
            resource = _doc_resource(self.resource, id)
            status, headers, data = resource.get_json(revs=True)
        except http.ResourceNotFound:
            return

        startrev = data['_revisions']['start']
        for index, rev in enumerate(data['_revisions']['ids']):
            options['rev'] = '%d-%s' % (startrev - index, rev)
            revision = self.get(id, **options)
            if revision is None:
                return
            yield revision

    def info(self, ddoc=None):
        """Return information about the database or design document as a
        dictionary.

        Without an argument, returns database information. With an argument,
        return information for the given design document.

        The returned dictionary exactly corresponds to the JSON response to
        a ``GET`` request on the database or design document's info URI.

        :return: a dictionary of database properties
        :rtype: ``dict``
        :since: 0.4
        """
        if ddoc is not None:
            _, _, data = self.resource('_design', ddoc, '_info').get_json()
        else:
            _, _, data = self.resource.get_json()
            self._name = data['db_name']
        return data

    def delete_attachment(self, doc, filename):
        """Delete the specified attachment.

        Note that the provided `doc` is required to have a ``_rev`` field.
        Thus, if the `doc` is based on a view row, the view row would need to
        include the ``_rev`` field.

        :param doc: the dictionary or `Document` object representing the
                    document that the attachment belongs to
        :param filename: the name of the attachment file
        :since: 0.4.1
        """
        resource = _doc_resource(self.resource, doc['_id'])
        _, _, data = resource.delete_json(filename, rev=doc['_rev'])
        doc['_rev'] = data['rev']

    def get_attachment(self, id_or_doc, filename, default=None):
        """Return an attachment from the specified doc id and filename.

        :param id_or_doc: either a document ID or a dictionary or `Document`
                          object representing the document that the attachment
                          belongs to
        :param filename: the name of the attachment file
        :param default: default value to return when the document or attachment
                        is not found
        :return: a file-like object with read and close methods, or the value
                 of the `default` argument if the attachment is not found
        :since: 0.4.1
        """
        if isinstance(id_or_doc, basestring):
            id = id_or_doc
        else:
            id = id_or_doc['_id']
        try:
            _, _, data = _doc_resource(self.resource, id).get(filename)
            return data
        except http.ResourceNotFound:
            return default

    def put_attachment(self, doc, content, filename=None, content_type=None):
        """Create or replace an attachment.

        Note that the provided `doc` is required to have a ``_rev`` field. Thus,
        if the `doc` is based on a view row, the view row would need to include
        the ``_rev`` field.

        :param doc: the dictionary or `Document` object representing the
                    document that the attachment should be added to
        :param content: the content to upload, either a file-like object or
                        a string
        :param filename: the name of the attachment file; if omitted, this
                         function tries to get the filename from the file-like
                         object passed as the `content` argument value
        :param content_type: content type of the attachment; if omitted, the
                             MIME type is guessed based on the file name
                             extension
        :since: 0.4.1
        """
        if filename is None:
            if hasattr(content, 'name'):
                filename = os.path.basename(content.name)
            else:
                raise ValueError('no filename specified for attachment')
        if content_type is None:
            content_type = ';'.join(
                filter(None, mimetypes.guess_type(filename))
            )

        resource = _doc_resource(self.resource, doc['_id'])
        status, headers, data = resource.put_json(filename, body=content, headers={
            'Content-Type': content_type
        }, rev=doc['_rev'])
        doc['_rev'] = data['rev']

    def query(self, map_fun, reduce_fun=None, language='javascript',
              wrapper=None, **options):
        """Execute an ad-hoc query (a "temp view") against the database.

        >>> server = Server()
        >>> db = server.create('python-tests')
        >>> db['johndoe'] = dict(type='Person', name='John Doe')
        >>> db['maryjane'] = dict(type='Person', name='Mary Jane')
        >>> db['gotham'] = dict(type='City', name='Gotham City')
        >>> map_fun = '''function(doc) {
        ...     if (doc.type == 'Person')
        ...         emit(doc.name, null);
        ... }'''
        >>> for row in db.query(map_fun):
        ...     print row.key
        John Doe
        Mary Jane

        >>> for row in db.query(map_fun, descending=True):
        ...     print row.key
        Mary Jane
        John Doe

        >>> for row in db.query(map_fun, key='John Doe'):
        ...     print row.key
        John Doe

        >>> del server['python-tests']

        :param map_fun: the code of the map function
        :param reduce_fun: the code of the reduce function (optional)
        :param language: the language of the functions, to determine which view
                         server to use
        :param wrapper: an optional callable that should be used to wrap the
                        result rows
        :param options: optional query string parameters
        :return: the view reults
        :rtype: `ViewResults`
        """
        return TemporaryView(self.resource('_temp_view'), map_fun,
                             reduce_fun, language=language,
                             wrapper=wrapper)(**options)

    def update(self, documents, **options):
        """Perform a bulk update or insertion of the given documents using a
        single HTTP request.

        >>> server = Server()
        >>> db = server.create('python-tests')
        >>> for doc in db.update([
        ...     Document(type='Person', name='John Doe'),
        ...     Document(type='Person', name='Mary Jane'),
        ...     Document(type='City', name='Gotham City')
        ... ]):
        ...     print repr(doc) #doctest: +ELLIPSIS
        (True, '...', '...')
        (True, '...', '...')
        (True, '...', '...')

        >>> del server['python-tests']

        The return value of this method is a list containing a tuple for every
        element in the `documents` sequence. Each tuple is of the form
        ``(success, docid, rev_or_exc)``, where ``success`` is a boolean
        indicating whether the update succeeded, ``docid`` is the ID of the
        document, and ``rev_or_exc`` is either the new document revision, or
        an exception instance (e.g. `ResourceConflict`) if the update failed.

        If an object in the documents list is not a dictionary, this method
        looks for an ``items()`` method that can be used to convert the object
        to a dictionary. Effectively this means you can also use this method
        with `mapping.Document` objects.

        :param documents: a sequence of dictionaries or `Document` objects, or
                          objects providing a ``items()`` method that can be
                          used to convert them to a dictionary
        :return: an iterable over the resulting documents
        :rtype: ``list``

        :since: version 0.2
        """
        docs = []
        for doc in documents:
            if isinstance(doc, dict):
                docs.append(doc)
            elif hasattr(doc, 'items'):
                docs.append(dict(doc.items()))
            else:
                raise TypeError('expected dict, got %s' % type(doc))

        content = options
        content.update(docs=docs)
        _, _, data = self.resource.post_json('_bulk_docs', body=content)

        results = []
        for idx, result in enumerate(data):
            if 'error' in result:
                if result['error'] == 'conflict':
                    exc_type = http.ResourceConflict
                else:
                    # XXX: Any other error types mappable to exceptions here?
                    exc_type = http.ServerError
                results.append((False, result['id'],
                                exc_type(result['reason'])))
            else:
                doc = documents[idx]
                if isinstance(doc, dict): # XXX: Is this a good idea??
                    doc.update({'_id': result['id'], '_rev': result['rev']})
                results.append((True, result['id'], result['rev']))

        return results

    def purge(self, docs):
        """Perform purging (complete removing) of the given documents.

        Uses a single HTTP request to purge all given documents. Purged
        documents do not leave any meta-data in the storage and are not
        replicated.
        """
        content = {}
        for doc in docs:
            if isinstance(doc, dict):
                content[doc['_id']] = [doc['_rev']]
            elif hasattr(doc, 'items'):
                doc = dict(doc.items())
                content[doc['_id']] = [doc['_rev']]
            else:
                raise TypeError('expected dict, got %s' % type(doc))
        _, _, data = self.resource.post_json('_purge', body=content)
        return data

    def view(self, name, wrapper=None, **options):
        """Execute a predefined view.

        >>> server = Server()
        >>> db = server.create('python-tests')
        >>> db['gotham'] = dict(type='City', name='Gotham City')

        >>> for row in db.view('_all_docs'):
        ...     print row.id
        gotham

        >>> del server['python-tests']

        :param name: the name of the view; for custom views, use the format
                     ``design_docid/viewname``, that is, the document ID of the
                     design document and the name of the view, separated by a
                     slash
        :param wrapper: an optional callable that should be used to wrap the
                        result rows
        :param options: optional query string parameters
        :return: the view results
        :rtype: `ViewResults`
        """
        path = _path_from_name(name, '_view')
        return PermanentView(self.resource(*path), '/'.join(path),
                             wrapper=wrapper)(**options)

    def show(self, name, docid=None, **options):
        """Call a 'show' function.

        :param name: the name of the show function in the format
                     ``designdoc/showname``
        :param docid: optional ID of a document to pass to the show function.
        :param options: optional query string parameters
        :return: (headers, body) tuple, where headers is a dict of headers
                 returned from the show function and body is a readable
                 file-like instance
        """
        path = _path_from_name(name, '_show')
        if docid:
            path.append(docid)
        status, headers, body = self.resource(*path).get(**options)
        return headers, body

    def list(self, name, view, **options):
        """Format a view using a 'list' function.

        :param name: the name of the list function in the format
                     ``designdoc/listname``
        :param view: the name of the view in the format ``designdoc/viewname``
        :param options: optional query string parameters
        :return: (headers, body) tuple, where headers is a dict of headers
                 returned from the list function and body is a readable
                 file-like instance
        """
        path = _path_from_name(name, '_list')
        path.extend(view.split('/', 1))
        _, headers, body = _call_viewlike(self.resource(*path), options)
        return headers, body

    def update_doc(self, name, docid=None, **options):
        """Calls server side update handler.

        :param name: the name of the update handler function in the format
                     ``designdoc/updatename``.
        :param docid: optional ID of a document to pass to the update handler.
        :param options: optional query string parameters.
        :return: (headers, body) tuple, where headers is a dict of headers
                 returned from the list function and body is a readable
                 file-like instance
        """
        path = _path_from_name(name, '_update')
        if docid is None:
            func = self.resource(*path).post
        else:
            path.append(docid)
            func = self.resource(*path).put
        _, headers, body = func(**options)
        return headers, body

    def _changes(self, **opts):
        _, _, data = self.resource.get('_changes', **opts)
        lines = iter(data)
        for ln in lines:
            if not ln: # skip heartbeats
                continue
            doc = json.decode(ln)
            if 'last_seq' in doc: # consume the rest of the response if this
                for ln in lines:  # was the last line, allows conn reuse
                    pass
            yield doc

    def changes(self, **opts):
        """Retrieve a changes feed from the database.

        Takes since, feed, heartbeat and timeout options.
        """
        if opts.get('feed') == 'continuous':
            return self._changes(**opts)
        _, _, data = self.resource.get_json('_changes', **opts)
        return data


def _doc_resource(base, doc_id):
    """Return the resource for the given document id.
    """
    # Split an id that starts with a reserved segment, e.g. _design/foo, so
    # that the / that follows the 1st segment does not get escaped.
    if doc_id[:1] == '_':
        return base(*doc_id.split('/', 1))
    return base(doc_id)


def _path_from_name(name, type):
    """Expand a 'design/foo' style name to its full path as a list of
    segments.
    """
    if name.startswith('_'):
        return name.split('/')
    design, name = name.split('/', 1)
    return ['_design', design, type, name]


class Document(dict):
    """Representation of a document in the database.

    This is basically just a dictionary with the two additional properties
    `id` and `rev`, which contain the document ID and revision, respectively.
    """

    def __repr__(self):
        return '<%s %r@%r %r>' % (type(self).__name__, self.id, self.rev,
                                  dict([(k,v) for k,v in self.items()
                                        if k not in ('_id', '_rev')]))

    @property
    def id(self):
        """The document ID.

        :rtype: basestring
        """
        return self['_id']

    @property
    def rev(self):
        """The document revision.

        :rtype: basestring
        """
        return self['_rev']


class View(object):
    """Abstract representation of a view or query."""

    def __init__(self, url, wrapper=None, session=None):
        if isinstance(url, basestring):
            self.resource = http.Resource(url, session)
        else:
            self.resource = url
        self.wrapper = wrapper

    def __call__(self, **options):
        return ViewResults(self, options)

    def __iter__(self):
        return iter(self())

    def _exec(self, options):
        raise NotImplementedError


class PermanentView(View):
    """Representation of a permanent view on the server."""

    def __init__(self, uri, name, wrapper=None, session=None):
        View.__init__(self, uri, wrapper=wrapper, session=session)
        self.name = name

    def __repr__(self):
        return '<%s %r>' % (type(self).__name__, self.name)

    def _exec(self, options):
        _, _, data = _call_viewlike(self.resource, options)
        return data


class TemporaryView(View):
    """Representation of a temporary view."""

    def __init__(self, uri, map_fun, reduce_fun=None,
                 language='javascript', wrapper=None, session=None):
        View.__init__(self, uri, wrapper=wrapper, session=session)
        if isinstance(map_fun, FunctionType):
            map_fun = getsource(map_fun).rstrip('\n\r')
        self.map_fun = dedent(map_fun.lstrip('\n\r'))
        if isinstance(reduce_fun, FunctionType):
            reduce_fun = getsource(reduce_fun).rstrip('\n\r')
        if reduce_fun:
            reduce_fun = dedent(reduce_fun.lstrip('\n\r'))
        self.reduce_fun = reduce_fun
        self.language = language

    def __repr__(self):
        return '<%s %r %r>' % (type(self).__name__, self.map_fun,
                               self.reduce_fun)

    def _exec(self, options):
        body = {'map': self.map_fun, 'language': self.language}
        if self.reduce_fun:
            body['reduce'] = self.reduce_fun
        if 'keys' in options:
            options = options.copy()
            body['keys'] = options.pop('keys')
        content = json.encode(body).encode('utf-8')
        _, _, data = self.resource.post_json(body=content, headers={
            'Content-Type': 'application/json'
        }, **_encode_view_options(options))
        return data


def _encode_view_options(options):
    """Encode any items in the options dict that are sent as a JSON string to a
    view/list function.
    """
    retval = {}
    for name, value in options.items():
        if name in ('key', 'startkey', 'endkey') \
                or not isinstance(value, basestring):
            value = json.encode(value)
        retval[name] = value
    return retval


def _call_viewlike(resource, options):
    """Call a resource that takes view-like options.
    """
    if 'keys' in options:
        options = options.copy()
        keys = {'keys': options.pop('keys')}
        return resource.post_json(body=keys, **_encode_view_options(options))
    else:
        return resource.get_json(**_encode_view_options(options))


class ViewResults(object):
    """Representation of a parameterized view (either permanent or temporary)
    and the results it produces.

    This class allows the specification of ``key``, ``startkey``, and
    ``endkey`` options using Python slice notation.

    >>> server = Server()
    >>> db = server.create('python-tests')
    >>> db['johndoe'] = dict(type='Person', name='John Doe')
    >>> db['maryjane'] = dict(type='Person', name='Mary Jane')
    >>> db['gotham'] = dict(type='City', name='Gotham City')
    >>> map_fun = '''function(doc) {
    ...     emit([doc.type, doc.name], doc.name);
    ... }'''
    >>> results = db.query(map_fun)

    At this point, the view has not actually been accessed yet. It is accessed
    as soon as it is iterated over, its length is requested, or one of its
    `rows`, `total_rows`, or `offset` properties are accessed:

    >>> len(results)
    3

    You can use slices to apply ``startkey`` and/or ``endkey`` options to the
    view:

    >>> people = results[['Person']:['Person','ZZZZ']]
    >>> for person in people:
    ...     print person.value
    John Doe
    Mary Jane
    >>> people.total_rows, people.offset
    (3, 1)

    Use plain indexed notation (without a slice) to apply the ``key`` option.
    Note that as CouchDB makes no claim that keys are unique in a view, this
    can still return multiple rows:

    >>> list(results[['City', 'Gotham City']])
    [<Row id='gotham', key=['City', 'Gotham City'], value='Gotham City'>]

    >>> del server['python-tests']
    """

    def __init__(self, view, options):
        self.view = view
        self.options = options
        self._rows = self._total_rows = self._offset = None

    def __repr__(self):
        return '<%s %r %r>' % (type(self).__name__, self.view, self.options)

    def __getitem__(self, key):
        options = self.options.copy()
        if type(key) is slice:
            if key.start is not None:
                options['startkey'] = key.start
            if key.stop is not None:
                options['endkey'] = key.stop
            return ViewResults(self.view, options)
        else:
            options['key'] = key
            return ViewResults(self.view, options)

    def __iter__(self):
        return iter(self.rows)

    def __len__(self):
        return len(self.rows)

    def _fetch(self):
        data = self.view._exec(self.options)
        wrapper = self.view.wrapper or Row
        self._rows = [wrapper(row) for row in data['rows']]
        self._total_rows = data.get('total_rows')
        self._offset = data.get('offset', 0)

    @property
    def rows(self):
        """The list of rows returned by the view.

        :rtype: `list`
        """
        if self._rows is None:
            self._fetch()
        return self._rows

    @property
    def total_rows(self):
        """The total number of rows in this view.

        This value is `None` for reduce views.

        :rtype: `int` or ``NoneType`` for reduce views
        """
        if self._rows is None:
            self._fetch()
        return self._total_rows

    @property
    def offset(self):
        """The offset of the results from the first row in the view.

        This value is 0 for reduce views.

        :rtype: `int`
        """
        if self._rows is None:
            self._fetch()
        return self._offset


class Row(dict):
    """Representation of a row as returned by database views."""

    def __repr__(self):
        keys = 'id', 'key', 'error', 'value'
        items = ['%s=%r' % (k, self[k]) for k in keys if k in self]
        return '<%s %s>' % (type(self).__name__, ', '.join(items))

    @property
    def id(self):
        """The associated Document ID if it exists. Returns `None` when it
        doesn't (reduce results).
        """
        return self.get('id')

    @property
    def key(self):
        return self['key']

    @property
    def value(self):
        return self.get('value')

    @property
    def error(self):
        return self.get('error')

    @property
    def doc(self):
        """The associated document for the row. This is only present when the
        view was accessed with ``include_docs=True`` as a query parameter,
        otherwise this property will be `None`.
        """
        doc = self.get('doc')
        if doc:
            return Document(doc)


SPECIAL_DB_NAMES = set(['_users'])
VALID_DB_NAME = re.compile(r'^[a-z][a-z0-9_$()+-/]*$')
def validate_dbname(name):
    if name in SPECIAL_DB_NAMES:
        return name
    if not VALID_DB_NAME.match(name):
        raise ValueError('Invalid database name')
    return name