Mercurial > libervia-web
diff libervia/server/server.py @ 1483:595e7fef41f3
merge bookmark @
author | Goffi <goffi@goffi.org> |
---|---|
date | Fri, 12 Nov 2021 17:48:30 +0100 |
parents | fc91b78b71db 83cd4862b134 |
children | 6643855770a5 |
line wrap: on
line diff
--- a/libervia/server/server.py Tue Sep 28 18:18:37 2021 +0200 +++ b/libervia/server/server.py Fri Nov 12 17:48:30 2021 +0100 @@ -23,6 +23,7 @@ import urllib.request, urllib.error import time import copy +from typing import Optional from pathlib import Path from twisted.application import service from twisted.internet import reactor, defer, inotify @@ -179,6 +180,7 @@ def getChildForRequest(self, request): return super().getChildForRequest(request) + class LiberviaRootResource(ProtectedFile): """Specialized resource for Libervia root @@ -234,8 +236,8 @@ self.pages_redirects = {} self.cached_urls = {} self.main_menu = None - # map SàT application names => data - self.sat_apps = {} + # map Libervia application names => data + self.libervia_apps = {} self.build_path = host.getBuildPath(site_name) self.build_path.mkdir(parents=True, exist_ok=True) self.dev_build_path = host.getBuildPath(site_name, dev=True) @@ -295,7 +297,7 @@ await self.host.bridgeCall( "applicationStart", app_name, data_format.serialise(extra) ) - app_data = self.sat_apps[app_name] = data_format.deserialise( + app_data = self.libervia_apps[app_name] = data_format.deserialise( await self.host.bridgeCall( "applicationExposedGet", app_name, "", "")) @@ -337,168 +339,184 @@ self.redirections = {} self.inv_redirections = {} # new URL to old URL map - for old, new_data in url_redirections.items(): - # new_data can be a dictionary or a unicode url - if isinstance(new_data, dict): - # new_data dict must contain either "url", "page" or "path" key - # (exclusive) - # if "path" is used, a file url is constructed with it - if len({"path", "url", "page"}.intersection(list(new_data.keys()))) != 1: - raise ValueError( - 'You must have one and only one of "url", "page" or "path" key ' - 'in your url_redirections_dict data') - if "url" in new_data: - new = new_data["url"] - elif "page" in new_data: - new = new_data - new["type"] = "page" - new.setdefault("path_args", []) - if not isinstance(new["path_args"], list): - log.error( - _('"path_args" in redirection of {old} must be a list. ' - 'Ignoring the redirection'.format(old=old))) - continue - new.setdefault("query_args", {}) - if not isinstance(new["query_args"], dict): - log.error( - _( - '"query_args" in redirection of {old} must be a ' - 'dictionary. Ignoring the redirection'.format(old=old))) - continue - new["path_args"] = [quote(a) for a in new["path_args"]] - # we keep an inversed dict of page redirection - # (page/path_args => redirecting URL) - # so getURL can return the redirecting URL if the same arguments - # are used # making the URL consistent - args_hash = tuple(new["path_args"]) - self.pages_redirects.setdefault(new_data["page"], {})[ - args_hash - ] = old + for old, new_data_list in url_redirections.items(): + # several redirections can be used for one path by using a list. + # The redirection will be done using first item of the list, and all items + # will be used for inverse redirection. + # e.g. if a => [b, c], a will redirect to c, and b and c will both be + # equivalent to a + if not isinstance(new_data_list, list): + new_data_list = [new_data_list] + for new_data in new_data_list: + # new_data can be a dictionary or a unicode url + if isinstance(new_data, dict): + # new_data dict must contain either "url", "page" or "path" key + # (exclusive) + # if "path" is used, a file url is constructed with it + if (( + len( + {"path", "url", "page"}.intersection(list(new_data.keys())) + ) != 1 + )): + raise ValueError( + 'You must have one and only one of "url", "page" or "path" ' + 'key in your url_redirections_dict data' + ) + if "url" in new_data: + new = new_data["url"] + elif "page" in new_data: + new = new_data + new["type"] = "page" + new.setdefault("path_args", []) + if not isinstance(new["path_args"], list): + log.error( + _('"path_args" in redirection of {old} must be a list. ' + 'Ignoring the redirection'.format(old=old))) + continue + new.setdefault("query_args", {}) + if not isinstance(new["query_args"], dict): + log.error( + _( + '"query_args" in redirection of {old} must be a ' + 'dictionary. Ignoring the redirection' + ).format(old=old) + ) + continue + new["path_args"] = [quote(a) for a in new["path_args"]] + # we keep an inversed dict of page redirection + # (page/path_args => redirecting URL) + # so getURL can return the redirecting URL if the same arguments + # are used # making the URL consistent + args_hash = tuple(new["path_args"]) + self.pages_redirects.setdefault(new_data["page"], {}).setdefault( + args_hash, + old + ) - # we need lists in query_args because it will be used - # as it in request.path_args - for k, v in new["query_args"].items(): - if isinstance(v, str): - new["query_args"][k] = [v] - elif "path" in new_data: - new = "file:{}".format(urllib.parse.quote(new_data["path"])) - elif isinstance(new_data, str): - new = new_data - new_data = {} - else: - log.error( - _("ignoring invalid redirection value: {new_data}").format( - new_data=new_data + # we need lists in query_args because it will be used + # as it in request.path_args + for k, v in new["query_args"].items(): + if isinstance(v, str): + new["query_args"][k] = [v] + elif "path" in new_data: + new = "file:{}".format(urllib.parse.quote(new_data["path"])) + elif isinstance(new_data, str): + new = new_data + new_data = {} + else: + log.error( + _("ignoring invalid redirection value: {new_data}").format( + new_data=new_data + ) ) - ) - continue + continue - # some normalization - if not old.strip(): - # root URL special case - old = "" - elif not old.startswith("/"): - log.error(_("redirected url must start with '/', got {value}. Ignoring") - .format(value=old)) - continue - else: - old = self._normalizeURL(old) + # some normalization + if not old.strip(): + # root URL special case + old = "" + elif not old.startswith("/"): + log.error( + _("redirected url must start with '/', got {value}. Ignoring") + .format(value=old) + ) + continue + else: + old = self._normalizeURL(old) - if isinstance(new, dict): - # dict are handled differently, they contain data - # which ared use dynamically when the request is done - self.redirections[old] = new - if not old: - if new["type"] == "page": - log.info( - _("Root URL redirected to page {name}").format( - name=new["page"] + if isinstance(new, dict): + # dict are handled differently, they contain data + # which ared use dynamically when the request is done + self.redirections.setdefault(old, new) + if not old: + if new["type"] == "page": + log.info( + _("Root URL redirected to page {name}").format( + name=new["page"] + ) ) - ) - else: - if new["type"] == "page": - page = self.getPageByName(new["page"]) - url = page.getURL(*new.get("path_args", [])) - self.inv_redirections[url] = old - continue + else: + if new["type"] == "page": + page = self.getPageByName(new["page"]) + url = page.getURL(*new.get("path_args", [])) + self.inv_redirections[url] = old + continue - # at this point we have a redirection URL in new, we can parse it - new_url = urllib.parse.urlsplit(new) + # at this point we have a redirection URL in new, we can parse it + new_url = urllib.parse.urlsplit(new) - # we handle the known URL schemes - if new_url.scheme == "xmpp": - location = self.getPagePathFromURI(new) - if location is None: - log.warning( - _("ignoring redirection, no page found to handle this URI: " - "{uri}").format(uri=new)) - continue - request_data = self._getRequestData(location) - if old: + # we handle the known URL schemes + if new_url.scheme == "xmpp": + location = self.getPagePathFromURI(new) + if location is None: + log.warning( + _("ignoring redirection, no page found to handle this URI: " + "{uri}").format(uri=new)) + continue + request_data = self._getRequestData(location) self.inv_redirections[location] = old - elif new_url.scheme in ("", "http", "https"): - # direct redirection - if new_url.netloc: - raise NotImplementedError( - "netloc ({netloc}) is not implemented yet for " - "url_redirections_dict, it is not possible to redirect to an " - "external website".format(netloc=new_url.netloc)) - location = urllib.parse.urlunsplit( - ("", "", new_url.path, new_url.query, new_url.fragment) - ) - request_data = self._getRequestData(location) - if old: + elif new_url.scheme in ("", "http", "https"): + # direct redirection + if new_url.netloc: + raise NotImplementedError( + "netloc ({netloc}) is not implemented yet for " + "url_redirections_dict, it is not possible to redirect to an " + "external website".format(netloc=new_url.netloc)) + location = urllib.parse.urlunsplit( + ("", "", new_url.path, new_url.query, new_url.fragment) + ) + request_data = self._getRequestData(location) self.inv_redirections[location] = old - elif new_url.scheme == "file": - # file or directory - if new_url.netloc: - raise NotImplementedError( - "netloc ({netloc}) is not implemented for url redirection to " - "file system, it is not possible to redirect to an external " - "host".format( - netloc=new_url.netloc)) - path = urllib.parse.unquote(new_url.path) - if not os.path.isabs(path): - raise ValueError( - "file redirection must have an absolute path: e.g. " - "file:/path/to/my/file") - # for file redirection, we directly put child here - resource_class = ( - ProtectedFile if new_data.get("protected", True) else static.File - ) - res = resource_class(path, defaultType="application/octet-stream") - self.addResourceToPath(old, res) - log.info("[{host_name}] Added redirection from /{old} to file system " - "path {path}".format(host_name=self.host_name, - old=old, - path=path)) + elif new_url.scheme == "file": + # file or directory + if new_url.netloc: + raise NotImplementedError( + "netloc ({netloc}) is not implemented for url redirection to " + "file system, it is not possible to redirect to an external " + "host".format( + netloc=new_url.netloc)) + path = urllib.parse.unquote(new_url.path) + if not os.path.isabs(path): + raise ValueError( + "file redirection must have an absolute path: e.g. " + "file:/path/to/my/file") + # for file redirection, we directly put child here + resource_class = ( + ProtectedFile if new_data.get("protected", True) else static.File + ) + res = resource_class(path, defaultType="application/octet-stream") + self.addResourceToPath(old, res) + log.info("[{host_name}] Added redirection from /{old} to file system " + "path {path}".format(host_name=self.host_name, + old=old, + path=path)) - # we don't want to use redirection system, so we continue here - continue - - elif new_url.scheme == "sat-app": - # a SàT application - - app_name = urllib.parse.unquote(new_url.path).lower().strip() - extra = {"url_prefix": f"/{old}"} - try: - await self._startApp(app_name, extra) - except Exception as e: - log.warning(_( - "Can't launch {app_name!r} for path /{old}: {e}").format( - app_name=app_name, old=old, e=e)) + # we don't want to use redirection system, so we continue here continue - log.info("[{host_name}] Added redirection from /{old} to application " - "{app_name}".format( - host_name=self.host_name, - old=old, - app_name=app_name)) + elif new_url.scheme == "libervia-app": + # a Libervia application - # normal redirection system is not used here - continue + app_name = urllib.parse.unquote(new_url.path).lower().strip() + extra = {"url_prefix": f"/{old}"} + try: + await self._startApp(app_name, extra) + except Exception as e: + log.warning(_( + "Can't launch {app_name!r} for path /{old}: {e}").format( + app_name=app_name, old=old, e=e)) + continue + + log.info("[{host_name}] Added redirection from /{old} to application " + "{app_name}".format( + host_name=self.host_name, + old=old, + app_name=app_name)) + + # normal redirection system is not used here + continue elif new_url.scheme == "proxy": # a reverse proxy host, port = new_url.hostname, new_url.port @@ -521,18 +539,18 @@ # normal redirection system is not used here continue - else: - raise NotImplementedError( - "{scheme}: scheme is not managed for url_redirections_dict".format( - scheme=new_url.scheme + else: + raise NotImplementedError( + "{scheme}: scheme is not managed for url_redirections_dict".format( + scheme=new_url.scheme + ) ) - ) - self.redirections[old] = request_data - if not old: - log.info(_("[{host_name}] Root URL redirected to {uri}") - .format(host_name=self.host_name, - uri=request_data[1])) + self.redirections.setdefault(old, request_data) + if not old: + log.info(_("[{host_name}] Root URL redirected to {uri}") + .format(host_name=self.host_name, + uri=request_data[1])) # the default root URL, if not redirected if not "" in self.redirections: @@ -554,8 +572,8 @@ log.error(msg) raise ValueError(msg) page_name, url = menu - elif menu.startswith("sat-app:"): - app_name = menu[8:].strip().lower() + elif menu.startswith("libervia-app:"): + app_name = menu[13:].strip().lower() app_data = await self._startApp(app_name) front_url = app_data['front_url'] options = self.host.options @@ -746,7 +764,7 @@ # if nothing was found, we try our luck with redirections # XXX: we want redirections to happen only if everything else failed path_elt = request.prepath + request.postpath - for idx in range(len(path_elt), 0, -1): + for idx in range(len(path_elt), -1, -1): test_url = b"/".join(path_elt[:idx]).decode('utf-8').lower() if test_url in self.redirections: request_data = self.redirections[test_url] @@ -1660,57 +1678,59 @@ """ ext_data = self.base_url_ext_data url_path = request.URLPath() - if not ext_data.scheme or not ext_data.netloc: - # ext_data is not specified, we check headers - if request.requestHeaders.hasHeader("x-forwarded-host"): - # we are behing a proxy - # we fill proxy_scheme and proxy_netloc value - proxy_host = request.requestHeaders.getRawHeaders("x-forwarded-host")[0] - try: - proxy_server = request.requestHeaders.getRawHeaders( - "x-forwarded-server" - )[0] - except TypeError: - # no x-forwarded-server found, we use proxy_host - proxy_netloc = proxy_host - else: - # if the proxy host has a port, we use it with server name - proxy_port = urllib.parse.urlsplit("//{}".format(proxy_host)).port - proxy_netloc = ( - "{}:{}".format(proxy_server, proxy_port) - if proxy_port is not None - else proxy_server - ) - try: - proxy_scheme = request.requestHeaders.getRawHeaders( - "x-forwarded-proto" - )[0] - except TypeError: - proxy_scheme = None - else: - proxy_scheme, proxy_netloc = None, None + + try: + forwarded = request.requestHeaders.getRawHeaders( + "forwarded" + )[0] + except TypeError: + # we try deprecated headers + try: + proxy_netloc = request.requestHeaders.getRawHeaders( + "x-forwarded-host" + )[0] + except TypeError: + proxy_netloc = None + try: + proxy_scheme = request.requestHeaders.getRawHeaders( + "x-forwarded-proto" + )[0] + except TypeError: + proxy_scheme = None else: - proxy_scheme, proxy_netloc = None, None + fwd_data = { + k.strip(): v.strip() + for k,v in (d.split("=") for d in forwarded.split(";")) + } + proxy_netloc = fwd_data.get("host") + proxy_scheme = fwd_data.get("proto") return urllib.parse.SplitResult( - ext_data.scheme or proxy_scheme or url_path.scheme.decode("utf-8"), - ext_data.netloc or proxy_netloc or url_path.netloc.decode("utf-8"), + ext_data.scheme or proxy_scheme or url_path.scheme.decode(), + ext_data.netloc or proxy_netloc or url_path.netloc.decode(), ext_data.path or "/", "", "", ) - def getExtBaseURL(self, request, path="", query="", fragment="", scheme=None): + def getExtBaseURL( + self, + request: server.Request, + path: str = "", + query: str = "", + fragment: str = "", + scheme: Optional[str] = None, + ) -> str: """Get external URL according to given elements external URL is the URL seen by external user - @param path(unicode): same as for urlsplit.urlsplit + @param path: same as for urlsplit.urlsplit path will be prefixed to follow found external URL if suitable - @param params(unicode): same as for urlsplit.urlsplit - @param query(unicode): same as for urlsplit.urlsplit - @param fragment(unicode): same as for urlsplit.urlsplit - @param scheme(unicode, None): if not None, will override scheme from base URL - @return (unicode): external URL + @param params: same as for urlsplit.urlsplit + @param query: same as for urlsplit.urlsplit + @param fragment: same as for urlsplit.urlsplit + @param scheme: if not None, will override scheme from base URL + @return: external URL """ split_result = self.getExtBaseURLData(request) return urllib.parse.urlunsplit( @@ -1723,23 +1743,23 @@ ) ) - def checkRedirection(self, vhost_root, url): + def checkRedirection(self, vhost_root: LiberviaRootResource, url_path: str) -> str: """check is a part of the URL prefix is redirected then replace it - @param vhost_root(web_resource.Resource): root of this virtual host - @param url(unicode): url to check - @return (unicode): possibly redirected URL which should link to the same location + @param vhost_root: root of this virtual host + @param url_path: path of the url to check + @return: possibly redirected URL which should link to the same location """ inv_redirections = vhost_root.inv_redirections - url_parts = url.strip("/").split("/") - for idx in range(len(url), 0, -1): + url_parts = url_path.strip("/").split("/") + for idx in range(len(url_parts), -1, -1): test_url = "/" + "/".join(url_parts[:idx]) if test_url in inv_redirections: rem_url = url_parts[idx:] return os.path.join( "/", "/".join([inv_redirections[test_url]] + rem_url) ) - return url + return url_path ## Sessions ##