cloro-haystack 0.1.0__tar.gz → 0.3.0__tar.gz

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
@@ -1,7 +1,7 @@
1
1
  Metadata-Version: 2.5
2
2
  Name: cloro-haystack
3
- Version: 0.1.0
4
- Summary: Haystack components for cloro: Google search results with the AI Overview, and ChatGPT, Google AI Mode and Gemini answers with sources.
3
+ Version: 0.3.0
4
+ Summary: Haystack components for cloro: Google and Bing search results with the AI Overview, and ChatGPT, Google AI Mode, Gemini and Bing AI Mode answers with sources.
5
5
  Project-URL: Homepage, https://cloro.dev/?utm_source=cloro-haystack&utm_medium=byok
6
6
  Project-URL: Documentation, https://github.com/cloro-dev/cloro-haystack#readme
7
7
  Project-URL: Issues, https://github.com/cloro-dev/cloro-haystack/issues
@@ -9,7 +9,7 @@ Project-URL: Source, https://github.com/cloro-dev/cloro-haystack
9
9
  Author-email: cloro <support@cloro.dev>
10
10
  License-Expression: MIT
11
11
  License-File: LICENSE
12
- Keywords: AI Mode,AI Overview,ChatGPT,Gemini,Google,Haystack,Web Search,cloro
12
+ Keywords: AI Mode,AI Overview,Bing,ChatGPT,Gemini,Google,Haystack,Web Search,cloro
13
13
  Classifier: Development Status :: 4 - Beta
14
14
  Classifier: License :: OSI Approved :: MIT License
15
15
  Classifier: Programming Language :: Python
@@ -28,8 +28,8 @@ Description-Content-Type: text/markdown
28
28
 
29
29
  | Component | cloro endpoint | Output | Credits per run |
30
30
  |---|---|---|--:|
31
- | `CloroWebSearch` | `POST /v1/monitor/google` | `documents` and `links` from Google organic results. With `include_ai_overview=True`, the Google AI Overview is the first Document. | 5, or 7 with the AI Overview |
32
- | `CloroAIAnswer` | `POST /v1/monitor/chatgpt`, `/aimode` or `/gemini` | `replies` (the answer as Markdown), `documents` and `links` (the cited sources) | ChatGPT 7, AI Mode 6, Gemini 6 |
31
+ | `CloroWebSearch` | `POST /v1/monitor/google` or `/bing` | `documents` and `links` from Google or Bing organic results. With `include_ai_overview=True`, the AI Overview is the first Document. | 5, or 7 with the AI Overview |
32
+ | `CloroAIAnswer` | `POST /v1/monitor/chatgpt`, `/aimode`, `/gemini` or `/bing/aimode` | `replies` (the answer as Markdown), `documents` and `links` (the cited sources) | ChatGPT 7, Google AI Mode 6, Gemini 6, Bing AI Mode 6 |
33
33
 
34
34
  ## Installation
35
35
 
@@ -55,9 +55,10 @@ for doc in result["documents"]:
55
55
  print(doc.meta.get("title"), doc.meta.get("link"))
56
56
  ```
57
57
 
58
- Each organic result is a `Document`. Its `content` is the Google snippet and its `meta` has `title`,
59
- `link` and `position`. The AI Overview Document has `meta["source"] == "google_ai_overview"` and the
60
- URLs it cites in `meta["links"]`. Results with an empty link or a Google redirect link are skipped.
58
+ Each organic result is a `Document`. Its `content` is the result snippet and its `meta` has `title`,
59
+ `link` and `position`. The AI Overview Document has `meta["source"] == "google_ai_overview"` (or
60
+ `"bing_ai_overview"`) and the URLs it cites in `meta["links"]`. Results with an empty link or a Google
61
+ redirect link are skipped.
61
62
 
62
63
  Parameters:
63
64
 
@@ -66,16 +67,26 @@ Parameters:
66
67
  results (about 10). Set `search_params={"pages": 2}` for more (2 credits for each extra page).
67
68
  - `country`: ISO 3166-1 alpha-2 country code. Defaults to `"US"`.
68
69
  - `allowed_domains`: domains to limit the search to, sent as `site:` filters.
69
- - `include_ai_overview`: request the Google AI Overview. Defaults to `False`.
70
- - `search_params`: other request fields, for example `device`, `location`, `hl` or `pages`.
70
+ - `include_ai_overview`: request the AI Overview. Defaults to `False`.
71
+ - `search_params`: other request fields, for example `device`, `location`, `hl` or `pages` for Google.
71
72
  - `timeout`: request timeout in seconds. Defaults to 60.
73
+ - `engine`: `"google"` (Google Search) or `"bing"` (Bing Search). Defaults to `"google"`.
74
+
75
+ With `engine="bing"`, the component searches Bing. The country sets the Bing market. Bing Search accepts
76
+ all countries except CN, KR and RU. It has no `device`, `location`, `hl` or `pages` fields and returns
77
+ one page of results. The query can have a maximum of 3,500 characters after URL encoding, and fewer for
78
+ non-Latin scripts.
79
+
80
+ ```python
81
+ websearch = CloroWebSearch(engine="bing", country="GB", include_ai_overview=True)
82
+ ```
72
83
 
73
84
  ### CloroAIAnswer
74
85
 
75
86
  ```python
76
87
  from haystack_integrations.components.websearch.cloro import CloroAIAnswer
77
88
 
78
- ai_answer = CloroAIAnswer(engine="aimode") # "chatgpt", "aimode" or "gemini"
89
+ ai_answer = CloroAIAnswer(engine="aimode") # "chatgpt", "aimode", "gemini" or "bing_aimode"
79
90
  result = ai_answer.run(query="What is retrieval-augmented generation?")
80
91
 
81
92
  print(result["replies"][0])
@@ -86,6 +97,10 @@ print(result["links"])
86
97
  the answer cites, with `title`, `link` and `position` in `meta`. An answer can take 15 to 60 seconds,
87
98
  so the default `timeout` is 180 seconds.
88
99
 
100
+ With `engine="bing_aimode"`, the query can have a maximum of 3,500 characters after URL encoding,
101
+ and fewer for non-Latin scripts. To send the query from a US state, set
102
+ `search_params={"state": "CA"}` with `country="US"` (2 more credits).
103
+
89
104
  ### In a pipeline
90
105
 
91
106
  The output names of `CloroAIAnswer` match `AnswerBuilder`:
@@ -5,8 +5,8 @@
5
5
 
6
6
  | Component | cloro endpoint | Output | Credits per run |
7
7
  |---|---|---|--:|
8
- | `CloroWebSearch` | `POST /v1/monitor/google` | `documents` and `links` from Google organic results. With `include_ai_overview=True`, the Google AI Overview is the first Document. | 5, or 7 with the AI Overview |
9
- | `CloroAIAnswer` | `POST /v1/monitor/chatgpt`, `/aimode` or `/gemini` | `replies` (the answer as Markdown), `documents` and `links` (the cited sources) | ChatGPT 7, AI Mode 6, Gemini 6 |
8
+ | `CloroWebSearch` | `POST /v1/monitor/google` or `/bing` | `documents` and `links` from Google or Bing organic results. With `include_ai_overview=True`, the AI Overview is the first Document. | 5, or 7 with the AI Overview |
9
+ | `CloroAIAnswer` | `POST /v1/monitor/chatgpt`, `/aimode`, `/gemini` or `/bing/aimode` | `replies` (the answer as Markdown), `documents` and `links` (the cited sources) | ChatGPT 7, Google AI Mode 6, Gemini 6, Bing AI Mode 6 |
10
10
 
11
11
  ## Installation
12
12
 
@@ -32,9 +32,10 @@ for doc in result["documents"]:
32
32
  print(doc.meta.get("title"), doc.meta.get("link"))
33
33
  ```
34
34
 
35
- Each organic result is a `Document`. Its `content` is the Google snippet and its `meta` has `title`,
36
- `link` and `position`. The AI Overview Document has `meta["source"] == "google_ai_overview"` and the
37
- URLs it cites in `meta["links"]`. Results with an empty link or a Google redirect link are skipped.
35
+ Each organic result is a `Document`. Its `content` is the result snippet and its `meta` has `title`,
36
+ `link` and `position`. The AI Overview Document has `meta["source"] == "google_ai_overview"` (or
37
+ `"bing_ai_overview"`) and the URLs it cites in `meta["links"]`. Results with an empty link or a Google
38
+ redirect link are skipped.
38
39
 
39
40
  Parameters:
40
41
 
@@ -43,16 +44,26 @@ Parameters:
43
44
  results (about 10). Set `search_params={"pages": 2}` for more (2 credits for each extra page).
44
45
  - `country`: ISO 3166-1 alpha-2 country code. Defaults to `"US"`.
45
46
  - `allowed_domains`: domains to limit the search to, sent as `site:` filters.
46
- - `include_ai_overview`: request the Google AI Overview. Defaults to `False`.
47
- - `search_params`: other request fields, for example `device`, `location`, `hl` or `pages`.
47
+ - `include_ai_overview`: request the AI Overview. Defaults to `False`.
48
+ - `search_params`: other request fields, for example `device`, `location`, `hl` or `pages` for Google.
48
49
  - `timeout`: request timeout in seconds. Defaults to 60.
50
+ - `engine`: `"google"` (Google Search) or `"bing"` (Bing Search). Defaults to `"google"`.
51
+
52
+ With `engine="bing"`, the component searches Bing. The country sets the Bing market. Bing Search accepts
53
+ all countries except CN, KR and RU. It has no `device`, `location`, `hl` or `pages` fields and returns
54
+ one page of results. The query can have a maximum of 3,500 characters after URL encoding, and fewer for
55
+ non-Latin scripts.
56
+
57
+ ```python
58
+ websearch = CloroWebSearch(engine="bing", country="GB", include_ai_overview=True)
59
+ ```
49
60
 
50
61
  ### CloroAIAnswer
51
62
 
52
63
  ```python
53
64
  from haystack_integrations.components.websearch.cloro import CloroAIAnswer
54
65
 
55
- ai_answer = CloroAIAnswer(engine="aimode") # "chatgpt", "aimode" or "gemini"
66
+ ai_answer = CloroAIAnswer(engine="aimode") # "chatgpt", "aimode", "gemini" or "bing_aimode"
56
67
  result = ai_answer.run(query="What is retrieval-augmented generation?")
57
68
 
58
69
  print(result["replies"][0])
@@ -63,6 +74,10 @@ print(result["links"])
63
74
  the answer cites, with `title`, `link` and `position` in `meta`. An answer can take 15 to 60 seconds,
64
75
  so the default `timeout` is 180 seconds.
65
76
 
77
+ With `engine="bing_aimode"`, the query can have a maximum of 3,500 characters after URL encoding,
78
+ and fewer for non-Latin scripts. To send the query from a US state, set
79
+ `search_params={"state": "CA"}` with `country="US"` (2 more credits).
80
+
66
81
  ### In a pipeline
67
82
 
68
83
  The output names of `CloroAIAnswer` match `AnswerBuilder`:
@@ -4,13 +4,13 @@ build-backend = "hatchling.build"
4
4
 
5
5
  [project]
6
6
  name = "cloro-haystack"
7
- version = "0.1.0"
8
- description = "Haystack components for cloro: Google search results with the AI Overview, and ChatGPT, Google AI Mode and Gemini answers with sources."
7
+ version = "0.3.0"
8
+ description = "Haystack components for cloro: Google and Bing search results with the AI Overview, and ChatGPT, Google AI Mode, Gemini and Bing AI Mode answers with sources."
9
9
  readme = "README.md"
10
10
  requires-python = ">=3.11"
11
11
  license = "MIT"
12
12
  license-files = ["LICENSE"]
13
- keywords = ["Haystack", "cloro", "Web Search", "Google", "AI Overview", "AI Mode", "ChatGPT", "Gemini"]
13
+ keywords = ["Haystack", "cloro", "Web Search", "Google", "AI Overview", "AI Mode", "ChatGPT", "Gemini", "Bing"]
14
14
  authors = [{ name = "cloro", email = "support@cloro.dev" }]
15
15
  classifiers = [
16
16
  "Development Status :: 4 - Beta",
@@ -11,7 +11,15 @@ from haystack_integrations.components.websearch.cloro._api import is_direct_link
11
11
 
12
12
  logger = logging.getLogger(__name__)
13
13
 
14
- Engine = Literal["chatgpt", "aimode", "gemini"]
14
+ Engine = Literal["chatgpt", "aimode", "gemini", "bing_aimode"]
15
+
16
+ # The API path for each engine.
17
+ _PATHS: dict[str, str] = {
18
+ "chatgpt": "/v1/monitor/chatgpt",
19
+ "aimode": "/v1/monitor/aimode",
20
+ "gemini": "/v1/monitor/gemini",
21
+ "bing_aimode": "/v1/monitor/bing/aimode",
22
+ }
15
23
 
16
24
 
17
25
  @component
@@ -19,7 +27,7 @@ class CloroAIAnswer:
19
27
  """
20
28
  Gets the answer that an AI search engine gives to a query, with the sources it cites.
21
29
 
22
- The component sends the query to ChatGPT, Google AI Mode or Gemini through the
30
+ The component sends the query to ChatGPT, Google AI Mode, Gemini or Bing AI Mode through the
23
31
  [cloro](https://cloro.dev/?utm_source=cloro-haystack&utm_medium=byok) API.
24
32
  `replies` has the answer as Markdown. `documents` has one
25
33
  Document for each cited source, with `title`, `link` and `position` in its meta.
@@ -51,16 +59,18 @@ class CloroAIAnswer:
51
59
  :param api_key:
52
60
  cloro API key. Defaults to the `CLORO_API_KEY` environment variable.
53
61
  :param engine:
54
- The AI search engine: `"chatgpt"`, `"aimode"` (Google AI Mode) or `"gemini"`.
62
+ The AI search engine: `"chatgpt"`, `"aimode"` (Google AI Mode), `"gemini"` or
63
+ `"bing_aimode"` (Bing AI Mode).
55
64
  :param country:
56
65
  ISO 3166-1 alpha-2 country code for the engine location.
57
66
  :param search_params:
58
- Other fields for the `POST /v1/monitor/<engine>` request body. See the cloro API reference.
67
+ Other fields for the engine request body, for example `state` for ChatGPT, Gemini or
68
+ Bing AI Mode. See the cloro API reference.
59
69
  :param timeout:
60
70
  Request timeout in seconds. An answer can take 15 to 60 seconds.
61
71
  """
62
- if engine not in ("chatgpt", "aimode", "gemini"):
63
- msg = f"engine must be 'chatgpt', 'aimode' or 'gemini', not {engine!r}"
72
+ if engine not in _PATHS:
73
+ msg = f"engine must be 'chatgpt', 'aimode', 'gemini' or 'bing_aimode', not {engine!r}"
64
74
  raise ValueError(msg)
65
75
  self.api_key = api_key
66
76
  self.engine = engine
@@ -129,7 +139,7 @@ class CloroAIAnswer:
129
139
  return self._parse(result)
130
140
 
131
141
  def _path(self) -> str:
132
- return f"/v1/monitor/{self.engine}"
142
+ return _PATHS[self.engine]
133
143
 
134
144
  def _payload(self, query: str) -> dict[str, Any]:
135
145
  payload: dict[str, Any] = {"prompt": query, "country": self.country, **(self.search_params or {})}
@@ -2,7 +2,7 @@
2
2
  #
3
3
  # SPDX-License-Identifier: MIT
4
4
 
5
- from typing import Any
5
+ from typing import Any, Literal
6
6
 
7
7
  from haystack import Document, component, default_from_dict, default_to_dict, logging
8
8
  from haystack.utils import Secret, deserialize_secrets_inplace
@@ -11,19 +11,23 @@ from haystack_integrations.components.websearch.cloro._api import is_direct_link
11
11
 
12
12
  logger = logging.getLogger(__name__)
13
13
 
14
- GOOGLE_PATH = "/v1/monitor/google"
14
+ Engine = Literal["google", "bing"]
15
+
16
+ # The API path and the display name for each engine.
17
+ _PATHS: dict[str, str] = {"google": "/v1/monitor/google", "bing": "/v1/monitor/bing"}
18
+ _NAMES: dict[str, str] = {"google": "Google", "bing": "Bing"}
15
19
 
16
20
 
17
21
  @component
18
22
  class CloroWebSearch:
19
23
  """
20
- Searches Google with the cloro API and returns the results as Documents.
24
+ Searches Google or Bing with the cloro API and returns the results as Documents.
21
25
 
22
26
  cloro: https://cloro.dev/?utm_source=cloro-haystack&utm_medium=byok
23
27
 
24
- Each organic result becomes a Document. Its content is the Google snippet. Its meta has
25
- `title`, `link` and `position`. With `include_ai_overview=True`, the Google AI Overview
26
- (when Google shows one) is the first Document.
28
+ Each organic result becomes a Document. Its content is the result snippet. Its meta has
29
+ `title`, `link` and `position`. With `include_ai_overview=True`, the AI Overview of the
30
+ engine (when the engine shows one) is the first Document.
27
31
 
28
32
  ### Usage example
29
33
 
@@ -46,6 +50,7 @@ class CloroWebSearch:
46
50
  include_ai_overview: bool = False,
47
51
  search_params: dict[str, Any] | None = None,
48
52
  timeout: float = 60.0,
53
+ engine: Engine = "google",
49
54
  ) -> None:
50
55
  """
51
56
  Initialize the CloroWebSearch component.
@@ -53,22 +58,28 @@ class CloroWebSearch:
53
58
  :param api_key:
54
59
  cloro API key. Defaults to the `CLORO_API_KEY` environment variable.
55
60
  :param top_k:
56
- Maximum number of organic results to return. One request returns one page of Google
57
- results (about 10). Set `search_params={"pages": 2}` or more to get more results.
61
+ Maximum number of organic results to return. One request returns one page of results
62
+ (about 10). For Google, set `search_params={"pages": 2}` or more to get more results.
58
63
  :param country:
59
- ISO 3166-1 alpha-2 country code for the Google location.
64
+ ISO 3166-1 alpha-2 country code for the engine location.
60
65
  :param allowed_domains:
61
66
  Domains to limit the search to. They are added to the query as `site:` filters.
62
67
  :param include_ai_overview:
63
- If True, request the Google AI Overview and return it as the first Document.
68
+ If True, request the AI Overview and return it as the first Document.
64
69
  :param search_params:
65
- Other fields for the `POST /v1/monitor/google` request body, for example `device`,
66
- `location`, `hl` or `pages`. See the cloro API reference.
70
+ Other fields for the engine request body, for example `device`, `location`, `hl` or
71
+ `pages` for Google. See the cloro API reference.
67
72
  :param timeout:
68
73
  Request timeout in seconds.
74
+ :param engine:
75
+ The search engine: `"google"` (Google Search) or `"bing"` (Bing Search).
69
76
  """
77
+ if engine not in _PATHS:
78
+ msg = f"engine must be 'google' or 'bing', not {engine!r}"
79
+ raise ValueError(msg)
70
80
  self.api_key = api_key
71
81
  self.top_k = top_k
82
+ self.engine = engine
72
83
  self.country = country
73
84
  self.allowed_domains = allowed_domains
74
85
  self.include_ai_overview = include_ai_overview
@@ -90,6 +101,7 @@ class CloroWebSearch:
90
101
  include_ai_overview=self.include_ai_overview,
91
102
  search_params=self.search_params,
92
103
  timeout=self.timeout,
104
+ engine=self.engine,
93
105
  )
94
106
 
95
107
  @classmethod
@@ -106,7 +118,7 @@ class CloroWebSearch:
106
118
  @component.output_types(documents=list[Document], links=list[str])
107
119
  def run(self, query: str) -> dict[str, Any]:
108
120
  """
109
- Search Google with cloro.
121
+ Search Google or Bing with cloro.
110
122
 
111
123
  :param query: Search query.
112
124
  :returns: A dictionary with:
@@ -115,13 +127,13 @@ class CloroWebSearch:
115
127
  :raises TimeoutError: If the request times out.
116
128
  :raises CloroError: If the cloro API returns an error.
117
129
  """
118
- result = post(GOOGLE_PATH, self._payload(query), self.api_key, self.timeout)
130
+ result = post(_PATHS[self.engine], self._payload(query), self.api_key, self.timeout)
119
131
  return self._parse(result)
120
132
 
121
133
  @component.output_types(documents=list[Document], links=list[str])
122
134
  async def run_async(self, query: str) -> dict[str, Any]:
123
135
  """
124
- Search Google with cloro without blocking.
136
+ Search Google or Bing with cloro without blocking.
125
137
 
126
138
  This is the asynchronous version of `run`, with the same parameters and return values.
127
139
 
@@ -132,7 +144,7 @@ class CloroWebSearch:
132
144
  :raises TimeoutError: If the request times out.
133
145
  :raises CloroError: If the cloro API returns an error.
134
146
  """
135
- result = await post_async(GOOGLE_PATH, self._payload(query), self.api_key, self.timeout)
147
+ result = await post_async(_PATHS[self.engine], self._payload(query), self.api_key, self.timeout)
136
148
  return self._parse(result)
137
149
 
138
150
  def _payload(self, query: str) -> dict[str, Any]:
@@ -167,7 +179,11 @@ class CloroWebSearch:
167
179
  0,
168
180
  Document(
169
181
  content=overview["text"],
170
- meta={"title": "Google AI Overview", "source": "google_ai_overview", "links": sources},
182
+ meta={
183
+ "title": f"{_NAMES[self.engine]} AI Overview",
184
+ "source": f"{self.engine}_ai_overview",
185
+ "links": sources,
186
+ },
171
187
  ),
172
188
  )
173
189
 
@@ -0,0 +1,59 @@
1
+ {
2
+ "success": true,
3
+ "result": {
4
+ "organicResults": [
5
+ {
6
+ "position": 1,
7
+ "title": "How to Season a Wok",
8
+ "link": "https://www.wokandskillet.com/how-to-season-a-wok/",
9
+ "displayedLink": "https://www.wokandskillet.com › how-to-season-a-wok",
10
+ "snippet": "Wash the wok, heat it until the water evaporates, then coat it with a thin layer of oil.",
11
+ "page": 1
12
+ },
13
+ {
14
+ "position": 2,
15
+ "title": "How to Season a Wok - The Woks of Life",
16
+ "link": "https://thewoksoflife.com/how-to-season-a-wok/",
17
+ "displayedLink": "https://thewoksoflife.com › how-to-season-a-wok",
18
+ "snippet": "Seasoning a carbon steel wok gives it a natural nonstick surface.",
19
+ "page": 1
20
+ },
21
+ {
22
+ "position": 3,
23
+ "title": "3 Ways to Season a Wok - wikiHow",
24
+ "link": "https://www.wikihow.com/Season-a-Wok",
25
+ "displayedLink": "https://www.wikihow.com › Season-a-Wok",
26
+ "snippet": "Heat the wok over high heat, add oil and swirl it to coat the inside evenly.",
27
+ "page": 1,
28
+ "date": "Apr 5, 2024"
29
+ }
30
+ ],
31
+ "aioverview": {
32
+ "text": "Seasoning a carbon steel or cast iron wok means heating a thin layer of oil until it polymerizes.",
33
+ "sources": [
34
+ {
35
+ "position": 1,
36
+ "label": "Wok & Skillet",
37
+ "url": "https://www.wokandskillet.com/how-to-season-a-wok/",
38
+ "description": null
39
+ },
40
+ {
41
+ "position": 2,
42
+ "label": "Umamicart",
43
+ "url": "https://www.umamicart.com/how-to-season-a-wok/",
44
+ "description": null
45
+ }
46
+ ],
47
+ "citationPills": [
48
+ {
49
+ "position": 1,
50
+ "label": "Wok & Skillet",
51
+ "url": "https://www.wokandskillet.com/how-to-season-a-wok/",
52
+ "domain": "wokandskillet.com",
53
+ "citationPillId": 1
54
+ }
55
+ ],
56
+ "markdown": "Seasoning a carbon steel or cast iron wok means **heating a thin layer of oil until it polymerizes**.[Wok & Skillet](https://www.wokandskillet.com/how-to-season-a-wok/)"
57
+ }
58
+ }
59
+ }
@@ -0,0 +1,36 @@
1
+ {
2
+ "success": true,
3
+ "result": {
4
+ "text": "For most beginners, a neutral road shoe with soft cushioning is the safest start.",
5
+ "sources": [
6
+ {
7
+ "position": 1,
8
+ "label": "The Best Running Shoes for Beginners",
9
+ "url": "https://www.runnersworld.com/gear/a20865766/best-running-shoes-for-beginners/",
10
+ "description": "Our test team picked the best shoes for new runners."
11
+ },
12
+ {
13
+ "position": 2,
14
+ "label": "How to Choose Running Shoes",
15
+ "url": "https://www.rei.com/learn/expert-advice/running-shoes.html"
16
+ }
17
+ ],
18
+ "citationPills": [
19
+ {
20
+ "position": 1,
21
+ "label": "The Best Running Shoes for Beginners",
22
+ "url": "https://www.runnersworld.com/gear/a20865766/best-running-shoes-for-beginners/",
23
+ "domain": "runnersworld.com",
24
+ "citationPillId": 1
25
+ },
26
+ {
27
+ "position": 2,
28
+ "label": "How to Choose Running Shoes",
29
+ "url": "https://www.rei.com/learn/expert-advice/running-shoes.html",
30
+ "domain": "rei.com",
31
+ "citationPillId": 1
32
+ }
33
+ ],
34
+ "markdown": "For most beginners, a **neutral road shoe with soft cushioning** is the safest start.[The Best Running Shoes for Beginners](https://www.runnersworld.com/gear/a20865766/best-running-shoes-for-beginners/)[How to Choose Running Shoes](https://www.rei.com/learn/expert-advice/running-shoes.html)"
35
+ }
36
+ }
@@ -124,6 +124,27 @@ def test_websearch_top_k_domains_and_params(api):
124
124
  assert len(result["links"]) == 2
125
125
 
126
126
 
127
+ def test_websearch_bing_request_and_ai_overview(api):
128
+ payload = _load("bing.json")
129
+ api.respond(httpx.Response(200, json=payload))
130
+ result = CloroWebSearch(engine="bing", country="GB", include_ai_overview=True).run(query="season a wok")
131
+ request = api.calls[0]
132
+ assert str(request.url) == "https://api.cloro.dev/v1/monitor/bing"
133
+ assert _body(request) == {"query": "season a wok", "country": "GB", "include": {"aioverview": True}}
134
+ overview = result["documents"][0]
135
+ assert overview.content == payload["result"]["aioverview"]["text"]
136
+ assert overview.meta["title"] == "Bing AI Overview"
137
+ assert overview.meta["source"] == "bing_ai_overview"
138
+ assert overview.meta["links"] == [s["url"] for s in payload["result"]["aioverview"]["sources"]]
139
+ assert result["links"] == [r["link"] for r in payload["result"]["organicResults"]]
140
+ assert len(result["documents"]) == 4
141
+
142
+
143
+ def test_websearch_rejects_other_engines():
144
+ with pytest.raises(ValueError, match="engine"):
145
+ CloroWebSearch(engine="other") # type: ignore[arg-type]
146
+
147
+
127
148
  def test_websearch_run_async(api):
128
149
  api.respond(httpx.Response(200, json=_load("google_aioverview.json")))
129
150
  result = asyncio.run(CloroWebSearch().run_async(query="q"))
@@ -133,12 +154,15 @@ def test_websearch_run_async(api):
133
154
  # CloroAIAnswer
134
155
 
135
156
 
136
- @pytest.mark.parametrize("engine", ["chatgpt", "aimode", "gemini"])
137
- def test_ai_answer_request(api, engine):
157
+ @pytest.mark.parametrize(
158
+ ("engine", "path"),
159
+ [("chatgpt", "chatgpt"), ("aimode", "aimode"), ("gemini", "gemini"), ("bing_aimode", "bing/aimode")],
160
+ )
161
+ def test_ai_answer_request(api, engine, path):
138
162
  api.respond(httpx.Response(200, json=_load("aimode.json")))
139
163
  CloroAIAnswer(engine=engine).run(query="rank tracker")
140
164
  request = api.calls[0]
141
- assert str(request.url) == f"https://api.cloro.dev/v1/monitor/{engine}"
165
+ assert str(request.url) == f"https://api.cloro.dev/v1/monitor/{path}"
142
166
  assert _body(request) == {"prompt": "rank tracker", "country": "US", "include": {"markdown": True}}
143
167
 
144
168
 
@@ -158,6 +182,16 @@ def test_ai_answer_output(api):
158
182
  assert result["documents"][0].content == sources[0]["description"]
159
183
 
160
184
 
185
+ def test_ai_answer_bing_aimode_output(api):
186
+ payload = _load("bing_aimode.json")
187
+ api.respond(httpx.Response(200, json=payload))
188
+ result = CloroAIAnswer(engine="bing_aimode", search_params={"state": "CA"}).run(query="running shoes")
189
+ assert _body(api.calls[0])["state"] == "CA"
190
+ assert result["replies"] == [payload["result"]["markdown"]]
191
+ assert result["links"] == [s["url"] for s in payload["result"]["sources"]]
192
+ assert result["documents"][1].content == payload["result"]["sources"][1]["label"]
193
+
194
+
161
195
  def test_ai_answer_chatgpt_without_markdown_or_description(api):
162
196
  payload = _load("chatgpt.json")
163
197
  api.respond(httpx.Response(200, json=payload))
@@ -228,6 +262,7 @@ def test_error_message_has_no_key(api):
228
262
  "component",
229
263
  [
230
264
  CloroWebSearch(top_k=3, country="GB", include_ai_overview=True, search_params={"device": "mobile"}),
265
+ CloroWebSearch(engine="bing", country="DE"),
231
266
  CloroAIAnswer(engine="gemini", country="FR"),
232
267
  ],
233
268
  )
File without changes