[fix] engine: core.ac.uk - don't split a string into a tag list (#6689)

docs/dev/engines/online/europepmc.rst

@@ -0,0 +1,8 @@
+.. _europepmc engine:
+
+==========
+Europe PMC
+==========
+
+.. automodule:: searx.engines.europepmc
+   :members:

searx/engines/core.py

@@ -141,12 +141,13 @@ def response(resp: "SXNG_Response") -> EngineResults:
             if name:
                 authors.add(name)
 
+        tag = result.get("fieldOfStudy")
         res.add(
             res.types.Paper(
                 title=result.get("title"),
                 url=url,
                 content=result.get("fullText", "") or "",
-                tags=result.get("fieldOfStudy", []),
+                tags=[tag] if tag else [],
                 publishedDate=published_date,
                 type=result.get("documentType", "") or "",
                 authors=authors,

searx/engines/duckduckgo_web.py

@@ -14,11 +14,12 @@ can't build it ourselves and must scrape it from the HTML pages.
 """
 
 import typing as t
+import re
 
-from urllib.parse import quote_plus
+from urllib.parse import quote_plus, urljoin
 from lxml import html
 
-from searx.utils import html_to_text, gen_useragent, extract_text, eval_xpath
+from searx.utils import html_to_text, extract_text, eval_xpath
 from searx.result_types import EngineResults
 from searx.enginelib import EngineCache
 from searx.network import get
@@ -38,7 +39,6 @@ about = {
 # engine dependent config
 categories = ["general"]
 paging = True
-_HTTP_User_Agent: str = gen_useragent()
 
 base_url = "https://duckduckgo.com"
 
@@ -73,6 +73,8 @@ def _fetch_first_page_link(
     resp = get(
         url=f"{base_url}/?q={quote_plus(query)}&t=h_&ia=web",
         headers=headers,
+        impersonate="firefox",
+        default_headers=False,
         timeout=2,
     )
 
@@ -96,6 +98,43 @@ def _cache_key(query: str, pageno: int) -> str:
     return f"nextpage_url|{query}|{pageno}"
 
 
+def _solve_jsa(resp: "SXNG_Response") -> "SXNG_Response":
+    """Duckduckgo sometimes issues a challenge instead of json."""
+
+    # length that a real browser would report for where the broken snippet is
+    html_len = {
+        "<p><div></p><p></div": 32,
+        "<li><div></li><li></div": 29,
+        "<div><div></div><div></div": 33,
+        "<br><div></br><br></div": 23,
+    }
+
+    js = resp.text or ""
+    jsa_match = re.search(r"let jsa = (\d+);.*?DDG\.deep\.initialize\('([^']+)'", js, re.S)
+    if not jsa_match:
+        return resp
+
+    js_functions = dict(re.findall(r"let (\w+) = function\(num\) \{([^}]*)\};", js))
+    jsa = int(jsa_match.group(1))
+    try:
+        for name in re.findall(r"jsa = (\w+)\(jsa\);", js):
+            body = js_functions[name]
+            mul = re.search(r"num \* (\d+)", body)
+            jsa = jsa * int(mul.group(1)) if mul else jsa + html_len[re.search(r"`([^`]+)`", body).group(1)]
+    except (KeyError, AttributeError):
+        return resp
+
+    params = resp.search_params
+    follow = get(
+        urljoin("https://links.duckduckgo.com", jsa_match.group(2) + str(jsa)),
+        headers=params["headers"],
+        impersonate="firefox",
+        default_headers=False,
+    )
+    follow.search_params = params
+    return follow
+
+
 def request(query: str, params: "OnlineParams") -> None:
 
     if len(query) >= 500:
@@ -103,25 +142,15 @@ def request(query: str, params: "OnlineParams") -> None:
         params["url"] = None
         return
 
-    headers = params["headers"]
-
-    # The vqd value is generated from the query and the UA header.  To be able
-    # to reuse the vqd value, the UA header must be static.
-    headers["User-Agent"] = _HTTP_User_Agent
-    headers["Accept"] = "*/*"
-    headers["Referer"] = f"{base_url}/"
-    headers["Host"] = "duckduckgo.com"
-
-    # Sec-Fetch headers are required to not get blocked when sending a Firefox user agent
-    headers["Sec-Fetch-Dest"] = "script"
-    headers["Sec-Fetch-Mode"] = "no-cors"
-    headers["Sec-Fetch-Site"] = "same-site"
+    # firefox TLS only
+    params["impersonate"] = "firefox"
+    params["default_headers"] = False
 
     api_url = ""
     if params["pageno"] > 1:
         api_url = CACHE.get(_cache_key(query, params["pageno"]))
     else:
-        api_url = _fetch_first_page_link(query, headers)
+        api_url = _fetch_first_page_link(query, params["headers"])
 
     if not api_url:
         params["url"] = None
@@ -129,14 +158,27 @@ def request(query: str, params: "OnlineParams") -> None:
 
     params["url"] = api_url.replace("/d.js?", "/d.js?o=json&")
 
+    # loads as a script
+    headers = params["headers"]
+    headers["Accept"] = "*/*"
+    headers["Sec-Fetch-Dest"] = "script"
+    headers["Sec-Fetch-Mode"] = "no-cors"
+    headers["Sec-Fetch-Site"] = "same-site"
+    headers["Referer"] = f"{base_url}/"
+
     # TODO: support safesearch, timerange and engine traits  # pylint:disable=fixme
 
 
 def response(resp: "SXNG_Response"):
     res = EngineResults()
-    res_json = resp.json()
 
-    for result in res_json["results"]:
+    # check if ddg returns a challenge
+    # e.g. 'site:github.com searxng'
+    if "let jsa =" in (resp.text or ""):
+        resp = _solve_jsa(resp)
+
+    results = resp.json()["results"]
+    for result in results:
         if "u" not in result:
             continue
 
@@ -144,8 +186,8 @@ def response(resp: "SXNG_Response"):
             res.types.MainResult(url=result["u"], title=html_to_text(result["t"]), content=html_to_text(result["a"]))
         )
 
-    # link to next page
-    next_page_path = res_json["results"][-1].get("n")
+    if results:
+        next_page_path = results[-1].get("n")
         if next_page_path:
             CACHE.set(
                 _cache_key(resp.search_params["query"], resp.search_params["pageno"] + 1),

searx/engines/europepmc.py

@@ -0,0 +1,150 @@
+# SPDX-License-Identifier: AGPL-3.0-or-later
+"""`Europe PMC`_ provides comprehensive access to life sciences literature from
+trusted sources.  With Europe PMC you can search and read millions of
+publications, preprints and other documents enriched with links to supporting
+data, reviews, protocols, and other relevant resources.
+
+.. _Europe PMC: https://europepmc.org/
+
+Configuration
+=============
+
+.. code:: yaml
+
+   - name: europepmc
+     engine: europepmc
+     shortcut: epmc
+
+Implementations
+===============
+
+"""
+
+import typing as t
+
+from datetime import datetime
+from urllib.parse import urlencode
+
+from dateutil.parser import isoparse
+
+from searx.enginelib import EngineCache
+from searx.result_types import EngineResults
+from searx.utils import html_to_text
+
+if t.TYPE_CHECKING:
+    from searx.extended_types import SXNG_Response
+    from searx.search.processors import OnlineParams
+
+
+about = {
+    "website": "https://europepmc.org/",
+    "wikidata_id": "Q5412157",
+    "official_api_documentation": "https://europepmc.org/RestfulWebService",
+    "use_official_api": True,
+    "require_api_key": False,
+    "results": "JSON",
+}
+
+categories = ["science", "scientific publications"]
+paging = True
+
+# engine dependent config
+search_url = "https://www.ebi.ac.uk/europepmc/webservices/rest/search"
+article_url = "https://europepmc.org/article/"
+
+page_size = 20
+
+CACHE: EngineCache
+"""Cache for storing the pagination cursor."""
+
+
+def setup(engine_settings: dict[str, t.Any]):
+    global CACHE  # pylint: disable=global-statement
+    CACHE = EngineCache(engine_settings["name"])
+
+
+def _cache_key(query: str, page: int) -> str:
+    return f"{query}|{page}"
+
+
+def request(query: str, params: "OnlineParams") -> None:
+    args = {
+        "query": query,
+        "format": "json",
+        "resultType": "core",
+        "pageSize": page_size,
+    }
+
+    if params["pageno"] > 1:
+        if cursor := CACHE.get(_cache_key(query, params["pageno"])):
+            args["cursorMark"] = cursor
+        else:
+            # no cached cursor for that page
+            params["url"] = None
+            return
+
+    params["url"] = f"{search_url}?{urlencode(args)}"
+
+
+def response(resp: "SXNG_Response") -> EngineResults:
+    res = EngineResults()
+
+    json_resp = resp.json()
+
+    # store pagination cursor for loading next pages in cache
+    if next_cursor := json_resp.get("nextCursorMark"):
+        next_page = resp.search_params["pageno"] + 1
+        query = resp.search_params["query"]
+        CACHE.set(_cache_key(query, next_page), next_cursor)
+
+    all_results = json_resp.get("resultList", {}).get("result", [])
+
+    for item in all_results:
+        source = item.get("source", "")
+        identifier = item.get("id", "")
+        url = f"{article_url}{source}/{identifier}" if source and identifier else ""
+
+        journal_info: dict[str, t.Any] = item.get("journalInfo", {})
+        journal: dict[str, t.Any] = journal_info.get("journal", {})
+
+        res.add(
+            res.types.Paper(
+                url=url,
+                title=html_to_text(item.get("title", "")),
+                content=html_to_text(item.get("abstractText", "")),
+                journal=journal.get("title", ""),
+                issn=[journal.get("issn", "")],
+                authors=_get_authors(item),
+                doi=item.get("doi", ""),
+                publishedDate=_get_published_date(item),
+                type=", ".join((item.get("pubTypeList", {})).get("pubType", [])),
+                pdf_url=_get_pdf_url(item),
+                html_url=url,
+            )
+        )
+
+    return res
+
+
+def _get_authors(item: dict[str, t.Any]) -> list:
+    """Extract the list of authors from the item."""
+    if authors := item.get("authorString", None):
+        authors = [author.strip().rstrip(".") for author in authors.split(",") if author.strip()]
+    else:
+        authors = []
+    return authors
+
+
+def _get_pdf_url(item: dict[str, t.Any]) -> str:
+    """Extract the PDF URL in case it is open access."""
+    for url_info in (item.get("fullTextUrlList", {})).get("fullTextUrl", []):
+        if url_info.get("documentStyle") == "pdf" and url_info.get("availabilityCode") == "OA":
+            return url_info.get("url", "")
+    return ""
+
+
+def _get_published_date(item: dict[str, t.Any]) -> datetime | None:
+    """Extract the published date from the item and convert it to a datetime object."""
+    if unformatted_date := item.get("firstPublicationDate"):
+        return isoparse(unformatted_date)
+    return None

searx/engines/startpage.py

@@ -1,5 +1,8 @@
 # SPDX-License-Identifier: AGPL-3.0-or-later
-"""Startpage's language & region selectors are a mess ..
+"""Startpage requires solving an Anubis POW captcha (difficulty 4).
+Solving it requires a lot of CPU, so the engine is set inactive by default.
+
+Startpage's language & region selectors are a mess ..
 
 .. _startpage regions:
 
@@ -84,6 +87,7 @@ Startpage's category (for Web-search, News, Videos, ..) is set by
 """
 # pylint: disable=too-many-statements
 
+import hashlib
 import re
 import typing as t
 from collections import OrderedDict
@@ -99,7 +103,7 @@ from searx.enginelib import EngineCache
 from searx.enginelib.traits import EngineTraits
 from searx.exceptions import SearxEngineCaptchaException
 from searx.locales import region_tag
-from searx.network import get  # see https://github.com/searxng/searxng/issues/762
+from searx.network import get, post  # see https://github.com/searxng/searxng/issues/762
 from searx.utils import (
     eval_xpath,
     extr,
@@ -176,6 +180,45 @@ def setup(_: dict[str, t.Any]) -> bool | None:
 sc_code_cache_sec = 3600
 """Time in seconds the sc-code is cached in memory :py:obj:`get_sc_code`."""
 
+# startpage's anubis difficulty is set to 4
+max_difficulty = 4
+
+
+def _solve_anubis(resp) -> str:
+    """Anubis POW solver"""
+    payload = loads(extr(resp.text, '<script id="anubis_challenge" type="application/json">', "</script>"))
+    challenge = payload["challenge"]
+    difficulty = int(payload["rules"]["difficulty"])
+    if difficulty > max_difficulty:
+        raise SearxEngineCaptchaException(message="startpage: Anubis difficulty too high")
+    prefix = "0" * difficulty
+    blob = challenge["randomData"].encode()
+    for nonce in range(16**difficulty * 8):  # max search is 8x average search, e^-8 = 0.034% will fail
+        digest = hashlib.sha256(blob + str(nonce).encode()).hexdigest()
+        if digest.startswith(prefix):
+            break
+    else:
+        raise SearxEngineCaptchaException(message="startpage: Anubis failed")
+
+    pass_resp = get(
+        f"{base_url}/.within.website/x/cmd/anubis/api/pass-challenge",
+        params={
+            "id": challenge["id"],
+            "response": digest,
+            "nonce": nonce,
+            "redir": str(resp.url),
+            "elapsedTime": "1",
+        },
+        cookies=resp.cookies,
+        allow_redirects=False,
+    )
+    auth = pass_resp.cookies.get("spchal-auth")
+    if not auth:
+        raise SearxEngineCaptchaException(message="startpage: Anubis pass-challenge failed")
+    auth = str(auth)
+    CACHE.set("SPCHAL_AUTH", auth, expire=240)
+    return auth
+
 
 def get_sc_code(params):
     """Get an actual ``sc`` argument from Startpage's search form (HTML page).
@@ -201,6 +244,9 @@ def get_sc_code(params):
     logger.debug("get_sc_code: request headers: %s", headers)
     resp = get(get_sc_url, headers=headers)
 
+    if 'id="anubis_challenge"' in resp.text:
+        resp = get(get_sc_url, headers=headers, cookies={"spchal-auth": _solve_anubis(resp)})
+
     # ?? x = network.get('https://www.startpage.com/sp/cdn/images/filter-chevron.svg', headers=headers)
     # ?? https://www.startpage.com/sp/cdn/images/filter-chevron.svg
     # ?? ping-back URL: https://www.startpage.com/sp/pb?sc=TLsB0oITjZ8F21
@@ -239,8 +285,8 @@ def request(query, params):
     Additionally the arguments form Startpage's search form needs to be set in
     HTML POST data / compare ``<input>`` elements: :py:obj:`search_form_xpath`.
     """
-    engine_region = traits.get_region(params["searxng_locale"], "en-US")
-    engine_language = traits.get_language(params["searxng_locale"], "en")
+    engine_region = traits.get_region(params["searxng_locale"], "en_US")
+    engine_language = traits.get_language(params["searxng_locale"], "english")
 
     params["headers"]["Origin"] = base_url
     params["headers"]["Referer"] = base_url + "/"
@@ -262,9 +308,9 @@ def request(query, params):
         args["language"] = engine_language
         args["lui"] = engine_language
 
+    args["segment"] = "startpage.udog"
     if params["pageno"] > 1:
         args["page"] = params["pageno"]
-        args["segment"] = "startpage.udog"
 
     # Build cookie
     lang_homepage = "en"
@@ -289,6 +335,8 @@ def request(query, params):
         cookie["search_results_region"] = engine_region
 
     params["cookies"]["preferences"] = "N1N".join(["%sEEE%s" % x for x in cookie.items()])
+    if auth := CACHE.get("SPCHAL_AUTH"):
+        params["cookies"]["spchal-auth"] = auth
     logger.debug("cookie preferences: %s", params["cookies"]["preferences"])
 
     logger.debug("data: %s", args)
@@ -400,6 +448,18 @@ def _get_image_result(result) -> dict[str, t.Any] | None:
 
 
 def response(resp):
+    if 'id="anubis_challenge"' in resp.text:
+        params = resp.search_params
+        params["cookies"]["spchal-auth"] = _solve_anubis(resp)
+        resp = post(
+            params["url"] or search_url,
+            data=params["data"],
+            headers=params["headers"],
+            cookies=params["cookies"],
+        )
+        if 'id="anubis_challenge"' in resp.text:
+            raise SearxEngineCaptchaException()
+
     categ = startpage_categ.capitalize()
     results_raw = "{" + extr(resp.text, f"React.createElement(UIStartpage.AppSerp{categ}, {{", "}})") + "}}"
 

searx/search/processors/online.py

@@ -88,6 +88,9 @@ class HTTPParams(t.TypedDict):
     impersonate: t.NotRequired[str]
     """curl_cffi impersonate target. Default: ``chrome``."""
 
+    default_headers: t.NotRequired[bool]
+    """Ask curl_cffi to skip setting the impersonated browser's default headers."""
+
     curl_options: t.NotRequired[dict[int, t.Any]]
     """Any extra libcurl options for the request."""
 
@@ -169,8 +172,8 @@ class OnlineProcessor(EngineProcessor):
             "cookies": params["cookies"],
             "auth": params["auth"],
         }
-        for key in ("curl_options", "impersonate"):
-            if params.get(key):
+        for key in ("curl_options", "impersonate", "default_headers"):
+            if params.get(key) is not None:
                 request_args[key] = params[key]
 
         verify = params.get("verify")

searx/settings.yml

@@ -792,6 +792,10 @@ engines:
       require_api_key: false
       results: JSON
 
+  - name: europepmc
+    engine: europepmc
+    shortcut: epmc
+
   - name: erowid
     engine: xpath
     paging: true
@@ -2339,18 +2343,21 @@ engines:
     shortcut: sp
     startpage_categ: web
     categories: [general, web]
+    inactive: true  # uses a Proof Of Work captcha https://github.com/searxng/searxng/pull/6669
 
   - name: startpage news
     engine: startpage
     startpage_categ: news
     categories: [news, web]
     shortcut: spn
+    inactive: true  # uses a Proof Of Work captcha https://github.com/searxng/searxng/pull/6669
 
   - name: startpage images
     engine: startpage
     startpage_categ: images
     categories: [images, web]
     shortcut: spi
+    inactive: true  # uses a Proof Of Work captcha https://github.com/searxng/searxng/pull/6669
 
   - name: steam
     engine: steam