diff options
| -rw-r--r-- | docs/dev/engines/online/alpinelinux.rst | 13 | ||||
| -rw-r--r-- | searx/engines/alpinelinux.py | 83 | ||||
| -rw-r--r-- | searx/settings.yml | 5 |
3 files changed, 101 insertions, 0 deletions
diff --git a/docs/dev/engines/online/alpinelinux.rst b/docs/dev/engines/online/alpinelinux.rst new file mode 100644 index 000000000..071a1985e --- /dev/null +++ b/docs/dev/engines/online/alpinelinux.rst @@ -0,0 +1,13 @@ +.. _alpinelinux engine: + +===================== +Alpine Linux Packages +===================== + +.. contents:: + :depth: 2 + :local: + :backlinks: entry + +.. automodule:: searx.engines.alpinelinux + :members: diff --git a/searx/engines/alpinelinux.py b/searx/engines/alpinelinux.py new file mode 100644 index 000000000..e5dcefed1 --- /dev/null +++ b/searx/engines/alpinelinux.py @@ -0,0 +1,83 @@ +# SPDX-License-Identifier: AGPL-3.0-or-later +"""`Alpine Linux binary packages`_. `Alpine Linux`_ is a Linux-based operation +system designed to be small, simple and secure. Contrary to many other Linux +distributions, it uses musl, BusyBox and OpenRC. Alpine is mostly used on +servers and for Docker images. + +.. _Alpine Linux binary packages: https://pkgs.alpinelinux.org +.. _Alpine Linux: https://www.alpinelinux.org + +""" + +import re + +from urllib.parse import urlencode +from lxml import html +from dateutil import parser + +from searx.utils import eval_xpath, eval_xpath_list, extract_text + +about = { + 'website': 'https://www.alpinelinux.org', + 'wikidata_id': 'Q4033826', + 'use_official_api': False, + 'official_api_documentation': None, + 'require_api_key': False, + 'results': 'HTML', +} +paging = True +categories = ['packages', 'it'] + +base_url = "https://pkgs.alpinelinux.org" +alpine_arch = 'x86_64' +"""Kernel architecture: ``x86_64``, ``x86``, ``aarch64``, ``armhf``, +``ppc64le``, ``s390x``, ``armv7`` or ``riscv64``""" + +ARCH_RE = re.compile("x86_64|x86|aarch64|armhf|ppc64le|s390x|armv7|riscv64") +"""Regular expression to match supported architectures in the query string.""" + + +def request(query, params): + query_arch = ARCH_RE.search(query) + if query_arch: + query_arch = query_arch.group(0) + query = query.replace(query_arch, '').strip() + + args = { + # use wildcards to match more than just packages with the exact same + # name as the query + 'name': f"*{query}*", + 'page': params['pageno'], + 'arch': query_arch or alpine_arch, + } + params['url'] = f"{base_url}/packages?{urlencode(args)}" + return params + + +def response(resp): + results = [] + + doc = html.fromstring(resp.text) + for result in eval_xpath_list(doc, "//table/tbody/tr"): + + if len(result.xpath("./td")) < 9: + # skip non valid entries in the result table + # e.g the "No item found..." message + continue + + results.append( + { + 'template': 'packages.html', + 'url': base_url + extract_text(eval_xpath(result, './td[contains(@class, "package")]/a/@href')), + 'title': extract_text(eval_xpath(result, './td[contains(@class, "package")]')), + 'package_name': extract_text(eval_xpath(result, './td[contains(@class, "package")]')), + 'publishedDate': parser.parse(extract_text(eval_xpath(result, './td[contains(@class, "bdate")]'))), + 'version': extract_text(eval_xpath(result, './td[contains(@class, "version")]')), + 'homepage': extract_text(eval_xpath(result, './td[contains(@class, "url")]/a/@href')), + 'maintainer': extract_text(eval_xpath(result, './td[contains(@class, "maintainer")]')), + 'license_name': extract_text(eval_xpath(result, './td[contains(@class, "license")]')), + 'tags': [extract_text(eval_xpath(result, './td[contains(@class, "repo")]'))], + } + ) + + return results diff --git a/searx/settings.yml b/searx/settings.yml index 0b804c542..1d506f8fc 100644 --- a/searx/settings.yml +++ b/searx/settings.yml @@ -325,6 +325,11 @@ engines: shortcut: 9g disabled: true + - name: alpine linux packages + engine: alpinelinux + disabled: true + shortcut: alp + - name: annas archive engine: annas_archive disabled: true |