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