master..1ea573d · searx

.github/workflows/data-update.yml+1−1

since splitsnapshots

unifiedsplit

@@ -29,7 +29,7 @@ jobs:

29 - name: Set up Python

30 uses: actions/setup-python@v2

31 with:

32− python-version: '3.10'

32+ python-version: '3.9'

33 architecture: 'x64'

34

35 - name: Install Python dependencies

.github/workflows/integration.yml+3−32

@@ -13,7 +13,7 @@ jobs:

13 strategy:

14 matrix:

15 os: [ubuntu-20.04]

16− python-version: [3.7, 3.8, 3.9, "3.10"]

16+ python-version: [3.6, 3.7, 3.8, 3.9]

17 steps:

18 - name: Checkout

19 uses: actions/checkout@v2

@@ -56,17 +56,6 @@ jobs:

56 uses: actions/checkout@v2

57 - name: Install Ubuntu packages

58 run: sudo ./utils/searx.sh install packages

59− - name: Set up Python

60− uses: actions/setup-python@v2

61− with:

62− python-version: '3.9'

63− architecture: 'x64'

64− - name: Cache Python dependencies

65− id: cache-python

66− uses: actions/cache@v2

67− with:

68− path: ./local

69− key: python-ubuntu-20.04-3.9-${{ hashFiles('requirements*.txt', 'setup.py') }}

70 - name: Install node dependencies

59 - name: Install node dependencies

71 run: make V=1 node.env

60 run: make V=1 node.env

72 - name: Build themes

61 - name: Build themes

@@ -79,21 +68,14 @@ jobs:

79 - name: Checkout

68 - name: Checkout

80 uses: actions/checkout@v2

69 uses: actions/checkout@v2

81 with:

70 with:

82− fetch-depth: '0'

83 persist-credentials: false

71 persist-credentials: false

84 - name: Install Ubuntu packages

72 - name: Install Ubuntu packages

85 run: sudo ./utils/searx.sh install buildhost

73 run: sudo ./utils/searx.sh install buildhost

86 - name: Set up Python

74 - name: Set up Python

87 uses: actions/setup-python@v2

75 uses: actions/setup-python@v2

88 with:

76 with:

89− python-version: '3.10'

77+ python-version: '3.9'

90 architecture: 'x64'

78 architecture: 'x64'

91− - name: Cache Python dependencies

92− id: cache-python

93− uses: actions/cache@v2

94− with:

95− path: ./local

96− key: python-ubuntu-20.04-3.9-${{ hashFiles('requirements*.txt', 'setup.py') }}

97 - name: Build documentation

79 - name: Build documentation

98 run: |

80 run: |

99 make V=1 docs.clean docs.html

81 make V=1 docs.clean docs.html

@@ -115,7 +97,7 @@ jobs:

115 - documentation

97 - documentation

116 env:

98 env:

117 DOCKERHUB_USERNAME: ${{ secrets.DOCKERHUB_USERNAME }}

99 DOCKERHUB_USERNAME: ${{ secrets.DOCKERHUB_USERNAME }}

118− runs-on: ubuntu-20.04

100+ runs-on: ubuntu-18.04

119 steps:

101 steps:

120 - name: Checkout

102 - name: Checkout

121 if: env.DOCKERHUB_USERNAME != null

103 if: env.DOCKERHUB_USERNAME != null

@@ -123,17 +105,6 @@ jobs:

123 with:

105 with:

124 # make sure "make docker.push" can get the git history

106 # make sure "make docker.push" can get the git history

125 fetch-depth: '0'

107 fetch-depth: '0'

126− - name: Set up Python

127− uses: actions/setup-python@v2

128− with:

129− python-version: '3.9'

130− architecture: 'x64'

131− - name: Cache Python dependencies

132− id: cache-python

133− uses: actions/cache@v2

134− with:

135− path: ./local

136− key: python-ubuntu-20.04-3.9-${{ hashFiles('requirements*.txt', 'setup.py') }}

137 - name: Set up QEMU

108 - name: Set up QEMU

138 if: env.DOCKERHUB_USERNAME != null

109 if: env.DOCKERHUB_USERNAME != null

139 uses: docker/setup-qemu-action@v1

110 uses: docker/setup-qemu-action@v1

.gitignore−2

@@ -26,5 +26,3 @@ dist/

26local/

27gh-pages/

28searx.egg-info/

29−.env

30−geckodriver.log

.pylintrc+1−1

@@ -59,7 +59,7 @@ confidence=

59# --enable=similarities". If you want to run only the classes checker, but have

60# no Warning level messages displayed, use"--disable=all --enable=classes

61# --disable=W"

62−disable=bad-whitespace, duplicate-code, consider-using-f-string

62+disable=bad-whitespace, duplicate-code

63

64# Enable the message, report, category or checker with the given id(s). You can

65# either give multiple identifier separated by comma (,) or put this option

AUTHORS.rst−1

@@ -165,4 +165,3 @@ generally made searx better:

165- @jhigginbotham

166- @xenrox

167- @OliveiraHermogenes

168−- Paul Alcock @Guilvareux

Dockerfile+4−5

@@ -1,4 +1,4 @@

1−FROM alpine:3.15

1+FROM alpine:3.12

2ENTRYPOINT ["/sbin/tini","--","/usr/local/searx/dockerfiles/docker-entrypoint.sh"]

3EXPOSE 8080

4VOLUME /etc/searx

@@ -46,19 +46,18 @@ RUN apk upgrade --no-cache \

46 uwsgi \

47 uwsgi-python3 \

48 brotli \

49− && pip3 install --upgrade pip wheel setuptools \

49+ && pip3 install --upgrade pip \

50 && pip3 install --no-cache -r requirements.txt \

51 && apk del build-dependencies \

52 && rm -rf /root/.cache

53

54−COPY searx ./searx

54+COPY --chown=searx:searx . .

55−COPY dockerfiles ./dockerfiles

56

55

57ARG TIMESTAMP_SETTINGS=0

56ARG TIMESTAMP_SETTINGS=0

58ARG TIMESTAMP_UWSGI=0

57ARG TIMESTAMP_UWSGI=0

59ARG VERSION_GITCOMMIT=unknown

58ARG VERSION_GITCOMMIT=unknown

60

59

61−RUN /usr/bin/python3 -m compileall -q searx; \

60+RUN su searx -c "/usr/bin/python3 -m compileall -q searx"; \

62 touch -c --date=@${TIMESTAMP_SETTINGS} searx/settings.yml; \

61 touch -c --date=@${TIMESTAMP_SETTINGS} searx/settings.yml; \

63 touch -c --date=@${TIMESTAMP_UWSGI} dockerfiles/uwsgi.ini; \

62 touch -c --date=@${TIMESTAMP_UWSGI} dockerfiles/uwsgi.ini; \

64 if [ ! -z $VERSION_GITCOMMIT ]; then\

63 if [ ! -z $VERSION_GITCOMMIT ]; then\

README.rst+1−1

@@ -61,7 +61,7 @@ our homepage_.

61.. _homepage: https://searx.github.io/searx

62

63contact:

64− openhub_ // twitter_ // IRC: #searx @ Libera (irc.libera.chat)

64+ openhub_ // twitter_ // IRC: #searx @ freenode

65

66.. _openhub: https://www.openhub.net/p/searx

67.. _twitter: https://twitter.com/Searx_engine

docs/blog/index.rst+3−3

@@ -3,14 +3,14 @@ Blog

3====

4

5.. toctree::

6− :titlesonly:

6+ :maxdepth: 2

7− :reversed:

7+ :caption: Contents

8

9+ lxcdev-202006

9 python3

10 python3

10 admin

11 admin

11 intro-offline

12 intro-offline

12 private-engines

13 private-engines

13− lxcdev-202006

14 command-line-engines

15 search-indexer-engines

16 sql-engines

docs/conf.py+1−2

@@ -10,7 +10,7 @@ from searx.version import VERSION_STRING

10# Project --------------------------------------------------------------

11

12project = u'searx'

13−copyright = u'2015-2021, Adam Tauber, Noémi Ványi'

13+copyright = u'2015-2020, Adam Tauber, Noémi Ványi'

14author = u'Adam Tauber'

15release, version = VERSION_STRING, VERSION_STRING

16highlight_language = 'none'

@@ -101,7 +101,6 @@ imgmath_font_size = 14

101

102html_theme_options = {"index_sidebar_logo": True}

103html_context = {"project_links": [] }

104−html_context["project_links"].append(ProjectLink("Blog", "blog/index.html"))

105if brand.GIT_URL:

104if brand.GIT_URL:

106 html_context["project_links"].append(ProjectLink("Source", brand.GIT_URL))

105 html_context["project_links"].append(ProjectLink("Source", brand.GIT_URL))

107if brand.WIKI_URL:

106if brand.WIKI_URL:

docs/dev/engine_overview.rst+1−1

@@ -171,7 +171,7 @@ headers set HTTP header information

171data set HTTP data information

172cookies set HTTP cookies

173verify bool Performing SSL-Validity check

174−follow_redirects bool Follow redirects

174+allow_redirects bool Follow redirects

175max_redirects int maximum redirects, hard limit

176soft_max_redirects int maximum redirects, soft limit. Record an error but don't stop the engine

177raise_for_httperror bool True by default: raise an exception if the HTTP code of response is >= 300

manage+4−4

@@ -13,7 +13,7 @@ source_dot_config

13

14PY_SETUP_EXTRAS='[test]'

15NPM_PACKAGES="less@2.7 less-plugin-clean-css grunt-cli"

16−GECKODRIVER_VERSION="v0.30.0"

16+GECKODRIVER_VERSION="v0.28.0"

17# SPHINXOPTS=

18

19# These py files are linted by test.pylint(), all other files are linted by

@@ -224,7 +224,7 @@ docker.build() {

224

225 # build Docker image

226 build_msg DOCKER "Building image ${SEARX_IMAGE_NAME}:${SEARX_GIT_VERSION}"

227− docker build \

227+ sudo docker build \

228 --build-arg GIT_URL="${GIT_URL}" \

229 --build-arg SEARX_GIT_VERSION="${SEARX_GIT_VERSION}" \

230 --build-arg VERSION_GITCOMMIT="${VERSION_GITCOMMIT}" \

@@ -236,8 +236,8 @@ docker.build() {

236 -t "${SEARX_IMAGE_NAME}:latest" -t "${SEARX_IMAGE_NAME}:${SEARX_GIT_VERSION}" .

237

238 if [ "$1" = "push" ]; then

239− docker push "${SEARX_IMAGE_NAME}:latest"

239+ sudo docker push "${SEARX_IMAGE_NAME}:latest"

240− docker push "${SEARX_IMAGE_NAME}:${SEARX_GIT_VERSION}"

240+ sudo docker push "${SEARX_IMAGE_NAME}:${SEARX_GIT_VERSION}"

241 fi

242 )

243 dump_return $?

requirements-dev.txt+12−13

@@ -1,20 +1,19 @@

1mock==4.0.3

2nose2[coverage_plugin]==0.10.0

3cov-core==1.15.0

4−pycodestyle==2.8.0

4+pycodestyle==2.7.0

5−pylint==2.12.2

5+pylint==2.10.2

6−splinter==0.17.0

6+splinter==0.15.0

7−transifex-client==0.14.3; python_version < '3.10'

7+transifex-client==0.14.3

8−transifex-client==0.12.1; python_version == '3.10'

9−selenium==4.1.0

8+selenium==3.141.0

10−twine==3.7.1

9+twine==3.4.2

11−Pallets-Sphinx-Themes==2.0.2

10+Pallets-Sphinx-Themes==2.0.1

12docutils==0.16

11docutils==0.16

13−Sphinx==4.4.0

12+Sphinx==4.1.2

14−sphinx-issues==3.0.1

13+sphinx-issues==1.2.0

15−sphinx-jinja==1.4.0

14+sphinx-jinja==1.1.1

16sphinx-tabs==3.2.0

15sphinx-tabs==3.2.0

17sphinxcontrib-programoutput==0.17

16sphinxcontrib-programoutput==0.17

18sphinx-autobuild==2021.3.14

17sphinx-autobuild==2021.3.14

19−linuxdoc==20211220

18+linuxdoc==20210324

20−aiounittest==1.4.1

19+aiounittest==1.4.0

requirements.txt+7−7

@@ -1,16 +1,16 @@

1−certifi==2021.10.8

1+certifi==2021.5.30

2babel==2.9.1

3flask-babel==2.0.0

4−flask==2.0.2

4+flask==1.1.2

5−jinja2==3.0.3

5+jinja2==3.0.1

6−lxml==4.7.1

6+lxml==4.6.3

7pygments==2.8.0

8python-dateutil==2.8.2

9−pyyaml==6.0

9+pyyaml==5.4.1

10−httpx[http2]==0.21.3

10+httpx[http2]==0.19.0

11Brotli==1.0.9

12uvloop==0.16.0; python_version >= '3.7'

13uvloop==0.14.0; python_version < '3.7'

14−httpx-socks[asyncio]==0.7.2

14+httpx-socks[asyncio]==0.3.1

15langdetect==1.0.9

16setproctitle==1.2.2

searx/autocomplete.py+7−24

@@ -24,7 +24,6 @@ from httpx import HTTPError

24

25

26from searx import settings

27−from searx.data import ENGINES_LANGUAGES

28from searx.network import get as http_get

27from searx.network import get as http_get

29from searx.exceptions import SearxEngineResponseException

28from searx.exceptions import SearxEngineResponseException

30

29

@@ -36,22 +35,6 @@ def get(*args, **kwargs):

36 return http_get(*args, **kwargs)

35 return http_get(*args, **kwargs)

37

36

38

37

39−def brave(query, lang):

40− # brave search autocompleter

41− url = 'https://search.brave.com/api/suggest?{query}'

42−

43− resp = get(url.format(query=urlencode({'q': query})))

44−

45− results = []

46−

47− if resp.ok:

48− data = loads(resp.text)

49− for suggestion in data[1]:

50− results.append(suggestion)

51−

52− return results

53−

54−

55def dbpedia(query, lang):

38def dbpedia(query, lang):

56 # dbpedia autocompleter, no HTTPS

39 # dbpedia autocompleter, no HTTPS

57 autocomplete_url = 'https://lookup.dbpedia.org/api/search.asmx/KeywordSearch?'

40 autocomplete_url = 'https://lookup.dbpedia.org/api/search.asmx/KeywordSearch?'

@@ -94,11 +77,12 @@ def google(query, lang):

94

77

95def startpage(query, lang):

78def startpage(query, lang):

96 # startpage autocompleter

79 # startpage autocompleter

97− lui = ENGINES_LANGUAGES['startpage'].get(lang, 'english')

98− url = 'https://startpage.com/suggestions?{query}'

80+ url = 'https://startpage.com/do/suggest?{query}'

81+

99− resp = get(url.format(query=urlencode({'q': query, 'segment': 'startpage.udog', 'lui': lui})))

82+ resp = get(url.format(query=urlencode({'query': query}))).text.split('\n')

100− data = resp.json()

83+ if len(resp) > 1:

101− return [e['text'] for e in data.get('suggestions', []) if 'text' in e]

84+ return resp

85+ return []

102

86

103

87

104def swisscows(query, lang):

88def swisscows(query, lang):

@@ -136,8 +120,7 @@ def wikipedia(query, lang):

136 return []

120 return []

137

121

138

122

139−backends = {'brave': brave,

140− 'dbpedia': dbpedia,

123+backends = {'dbpedia': dbpedia,

141 'duckduckgo': duckduckgo,

124 'duckduckgo': duckduckgo,

142 'google': google,

125 'google': google,

143 'startpage': startpage,

126 'startpage': startpage,

searx/data/__init__.py+14−39

@@ -1,50 +1,25 @@

1−# SPDX-License-Identifier: AGPL-3.0-or-later

2−# lint: pylint

3−"""This module holds the *data* created by::

4−

5− make data.all

6−

7−"""

8−

9−__all__ = [

10− 'ENGINES_LANGUAGES',

11− 'CURRENCIES',

12− 'USER_AGENTS',

13− 'EXTERNAL_URLS',

14− 'WIKIDATA_UNITS',

15− 'EXTERNAL_BANGS',

16− 'OSM_KEYS_TAGS',

17− 'ahmia_blacklist_loader',

18−]

19−

20import json

1import json

21from pathlib import Path

2from pathlib import Path

22

3

4+

5+__init__ = ['ENGINES_LANGUGAGES', 'CURRENCIES', 'USER_AGENTS', 'EXTERNAL_URLS', 'WIKIDATA_UNITS', 'EXTERNAL_BANGS',

6+ 'bangs_loader', 'ahmia_blacklist_loader']

23data_dir = Path(__file__).parent

7data_dir = Path(__file__).parent

24

8

25

9

26−def _load(filename):

10+def load(filename):

27− with open(data_dir / filename, encoding='utf-8') as f:

11+ with open(data_dir / filename, encoding='utf-8') as fd:

28− return json.load(f)

12+ return json.load(fd)

29

13

30

14

31def ahmia_blacklist_loader():

15def ahmia_blacklist_loader():

32− """Load data from `ahmia_blacklist.txt` and return a list of MD5 values of onion

33− names. The MD5 values are fetched by::

34−

35− searx_extra/update/update_ahmia_blacklist.py

36−

37− This function is used by :py:mod:`searx.plugins.ahmia_filter`.

38−

39− """

40− with open(str(data_dir / 'ahmia_blacklist.txt'), encoding='utf-8') as f:

16+ with open(str(data_dir / 'ahmia_blacklist.txt'), encoding='utf-8') as fd:

41− return f.read().split()

17+ return fd.read().split()

42

18

43

19

44−ENGINES_LANGUAGES = _load('engines_languages.json')

20+ENGINES_LANGUAGES = load('engines_languages.json')

45−CURRENCIES = _load('currencies.json')

21+CURRENCIES = load('currencies.json')

46−USER_AGENTS = _load('useragents.json')

22+USER_AGENTS = load('useragents.json')

47−EXTERNAL_URLS = _load('external_urls.json')

23+EXTERNAL_URLS = load('external_urls.json')

48−WIKIDATA_UNITS = _load('wikidata_units.json')

24+WIKIDATA_UNITS = load('wikidata_units.json')

49−EXTERNAL_BANGS = _load('external_bangs.json')

25+EXTERNAL_BANGS = load('external_bangs.json')

50−OSM_KEYS_TAGS = _load('osm_keys_tags.json')

searx/data/ahmia_blacklist.txt−8485