Tarek Ziade | 1231a4e | 2011-05-19 13:07:25 +0200 | [diff] [blame] | 1 | """Spider using the XML-RPC PyPI API. |
| 2 | |
| 3 | This module contains the class Client, a spider that can be used to find |
| 4 | and retrieve distributions from a project index (like the Python Package |
| 5 | Index), using its XML-RPC API (see documentation of the reference |
| 6 | implementation at http://wiki.python.org/moin/PyPiXmlRpc). |
| 7 | """ |
| 8 | |
| 9 | import xmlrpc.client |
| 10 | |
| 11 | from packaging import logger |
| 12 | from packaging.errors import IrrationalVersionError |
| 13 | from packaging.version import get_version_predicate |
| 14 | from packaging.pypi.base import BaseClient |
| 15 | from packaging.pypi.errors import (ProjectNotFound, InvalidSearchField, |
| 16 | ReleaseNotFound) |
| 17 | from packaging.pypi.dist import ReleaseInfo |
| 18 | |
| 19 | __all__ = ['Client', 'DEFAULT_XMLRPC_INDEX_URL'] |
| 20 | |
| 21 | DEFAULT_XMLRPC_INDEX_URL = 'http://python.org/pypi' |
| 22 | |
| 23 | _SEARCH_FIELDS = ['name', 'version', 'author', 'author_email', 'maintainer', |
| 24 | 'maintainer_email', 'home_page', 'license', 'summary', |
| 25 | 'description', 'keywords', 'platform', 'download_url'] |
| 26 | |
| 27 | |
| 28 | class Client(BaseClient): |
| 29 | """Client to query indexes using XML-RPC method calls. |
| 30 | |
| 31 | If no server_url is specified, use the default PyPI XML-RPC URL, |
| 32 | defined in the DEFAULT_XMLRPC_INDEX_URL constant:: |
| 33 | |
Éric Araujo | 25d5737 | 2011-06-01 14:41:11 +0200 | [diff] [blame] | 34 | >>> client = Client() |
Tarek Ziade | 1231a4e | 2011-05-19 13:07:25 +0200 | [diff] [blame] | 35 | >>> client.server_url == DEFAULT_XMLRPC_INDEX_URL |
| 36 | True |
| 37 | |
Éric Araujo | 25d5737 | 2011-06-01 14:41:11 +0200 | [diff] [blame] | 38 | >>> client = Client("http://someurl/") |
Tarek Ziade | 1231a4e | 2011-05-19 13:07:25 +0200 | [diff] [blame] | 39 | >>> client.server_url |
| 40 | 'http://someurl/' |
| 41 | """ |
| 42 | |
| 43 | def __init__(self, server_url=DEFAULT_XMLRPC_INDEX_URL, prefer_final=False, |
| 44 | prefer_source=True): |
| 45 | super(Client, self).__init__(prefer_final, prefer_source) |
| 46 | self.server_url = server_url |
| 47 | self._projects = {} |
| 48 | |
| 49 | def get_release(self, requirements, prefer_final=False): |
| 50 | """Return a release with all complete metadata and distribution |
| 51 | related informations. |
| 52 | """ |
| 53 | prefer_final = self._get_prefer_final(prefer_final) |
| 54 | predicate = get_version_predicate(requirements) |
| 55 | releases = self.get_releases(predicate.name) |
| 56 | release = releases.get_last(predicate, prefer_final) |
| 57 | self.get_metadata(release.name, str(release.version)) |
| 58 | self.get_distributions(release.name, str(release.version)) |
| 59 | return release |
| 60 | |
| 61 | def get_releases(self, requirements, prefer_final=None, show_hidden=True, |
| 62 | force_update=False): |
| 63 | """Return the list of existing releases for a specific project. |
| 64 | |
| 65 | Cache the results from one call to another. |
| 66 | |
| 67 | If show_hidden is True, return the hidden releases too. |
| 68 | If force_update is True, reprocess the index to update the |
| 69 | informations (eg. make a new XML-RPC call). |
| 70 | :: |
| 71 | |
Éric Araujo | 25d5737 | 2011-06-01 14:41:11 +0200 | [diff] [blame] | 72 | >>> client = Client() |
Tarek Ziade | 1231a4e | 2011-05-19 13:07:25 +0200 | [diff] [blame] | 73 | >>> client.get_releases('Foo') |
| 74 | ['1.1', '1.2', '1.3'] |
| 75 | |
| 76 | If no such project exists, raise a ProjectNotFound exception:: |
| 77 | |
| 78 | >>> client.get_project_versions('UnexistingProject') |
| 79 | ProjectNotFound: UnexistingProject |
| 80 | |
| 81 | """ |
| 82 | def get_versions(project_name, show_hidden): |
| 83 | return self.proxy.package_releases(project_name, show_hidden) |
| 84 | |
| 85 | predicate = get_version_predicate(requirements) |
| 86 | prefer_final = self._get_prefer_final(prefer_final) |
| 87 | project_name = predicate.name |
| 88 | if not force_update and (project_name.lower() in self._projects): |
| 89 | project = self._projects[project_name.lower()] |
| 90 | if not project.contains_hidden and show_hidden: |
| 91 | # if hidden releases are requested, and have an existing |
| 92 | # list of releases that does not contains hidden ones |
| 93 | all_versions = get_versions(project_name, show_hidden) |
| 94 | existing_versions = project.get_versions() |
| 95 | hidden_versions = set(all_versions) - set(existing_versions) |
| 96 | for version in hidden_versions: |
| 97 | project.add_release(release=ReleaseInfo(project_name, |
| 98 | version, index=self._index)) |
| 99 | else: |
| 100 | versions = get_versions(project_name, show_hidden) |
| 101 | if not versions: |
| 102 | raise ProjectNotFound(project_name) |
| 103 | project = self._get_project(project_name) |
| 104 | project.add_releases([ReleaseInfo(project_name, version, |
| 105 | index=self._index) |
| 106 | for version in versions]) |
| 107 | project = project.filter(predicate) |
| 108 | if len(project) == 0: |
| 109 | raise ReleaseNotFound("%s" % predicate) |
| 110 | project.sort_releases(prefer_final) |
| 111 | return project |
| 112 | |
| 113 | |
| 114 | def get_distributions(self, project_name, version): |
| 115 | """Grab informations about distributions from XML-RPC. |
| 116 | |
| 117 | Return a ReleaseInfo object, with distribution-related informations |
| 118 | filled in. |
| 119 | """ |
| 120 | url_infos = self.proxy.release_urls(project_name, version) |
| 121 | project = self._get_project(project_name) |
| 122 | if version not in project.get_versions(): |
| 123 | project.add_release(release=ReleaseInfo(project_name, version, |
| 124 | index=self._index)) |
| 125 | release = project.get_release(version) |
| 126 | for info in url_infos: |
| 127 | packagetype = info['packagetype'] |
| 128 | dist_infos = {'url': info['url'], |
| 129 | 'hashval': info['md5_digest'], |
| 130 | 'hashname': 'md5', |
| 131 | 'is_external': False, |
| 132 | 'python_version': info['python_version']} |
| 133 | release.add_distribution(packagetype, **dist_infos) |
| 134 | return release |
| 135 | |
| 136 | def get_metadata(self, project_name, version): |
| 137 | """Retrieve project metadata. |
| 138 | |
| 139 | Return a ReleaseInfo object, with metadata informations filled in. |
| 140 | """ |
| 141 | # to be case-insensitive, get the informations from the XMLRPC API |
| 142 | projects = [d['name'] for d in |
| 143 | self.proxy.search({'name': project_name}) |
| 144 | if d['name'].lower() == project_name] |
| 145 | if len(projects) > 0: |
| 146 | project_name = projects[0] |
| 147 | |
| 148 | metadata = self.proxy.release_data(project_name, version) |
| 149 | project = self._get_project(project_name) |
| 150 | if version not in project.get_versions(): |
| 151 | project.add_release(release=ReleaseInfo(project_name, version, |
| 152 | index=self._index)) |
| 153 | release = project.get_release(version) |
| 154 | release.set_metadata(metadata) |
| 155 | return release |
| 156 | |
| 157 | def search_projects(self, name=None, operator="or", **kwargs): |
| 158 | """Find using the keys provided in kwargs. |
| 159 | |
| 160 | You can set operator to "and" or "or". |
| 161 | """ |
| 162 | for key in kwargs: |
| 163 | if key not in _SEARCH_FIELDS: |
| 164 | raise InvalidSearchField(key) |
| 165 | if name: |
| 166 | kwargs["name"] = name |
| 167 | projects = self.proxy.search(kwargs, operator) |
| 168 | for p in projects: |
| 169 | project = self._get_project(p['name']) |
| 170 | try: |
| 171 | project.add_release(release=ReleaseInfo(p['name'], |
| 172 | p['version'], metadata={'summary': p['summary']}, |
| 173 | index=self._index)) |
| 174 | except IrrationalVersionError as e: |
| 175 | logger.warning("Irrational version error found: %s", e) |
| 176 | return [self._projects[p['name'].lower()] for p in projects] |
| 177 | |
| 178 | def get_all_projects(self): |
| 179 | """Return the list of all projects registered in the package index""" |
| 180 | projects = self.proxy.list_packages() |
| 181 | for name in projects: |
| 182 | self.get_releases(name, show_hidden=True) |
| 183 | |
| 184 | return [self._projects[name.lower()] for name in set(projects)] |
| 185 | |
| 186 | @property |
| 187 | def proxy(self): |
| 188 | """Property used to return the XMLRPC server proxy. |
| 189 | |
| 190 | If no server proxy is defined yet, creates a new one:: |
| 191 | |
Éric Araujo | 25d5737 | 2011-06-01 14:41:11 +0200 | [diff] [blame] | 192 | >>> client = Client() |
Tarek Ziade | 1231a4e | 2011-05-19 13:07:25 +0200 | [diff] [blame] | 193 | >>> client.proxy() |
| 194 | <ServerProxy for python.org/pypi> |
| 195 | |
| 196 | """ |
| 197 | if not hasattr(self, '_server_proxy'): |
| 198 | self._server_proxy = xmlrpc.client.ServerProxy(self.server_url) |
| 199 | |
| 200 | return self._server_proxy |