ws-bom-robot-app 0.0.47__py3-none-any.whl → 0.0.48__py3-none-any.whl
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.
- ws_bom_robot_app/llm/vector_store/integration/sitemap.py +4 -1
- {ws_bom_robot_app-0.0.47.dist-info → ws_bom_robot_app-0.0.48.dist-info}/METADATA +1 -1
- {ws_bom_robot_app-0.0.47.dist-info → ws_bom_robot_app-0.0.48.dist-info}/RECORD +5 -5
- {ws_bom_robot_app-0.0.47.dist-info → ws_bom_robot_app-0.0.48.dist-info}/WHEEL +1 -1
- {ws_bom_robot_app-0.0.47.dist-info → ws_bom_robot_app-0.0.48.dist-info}/top_level.txt +0 -0
|
@@ -20,6 +20,7 @@ class Sitemap(IntegrationStrategy):
|
|
|
20
20
|
data["excludeTag"] (list[str]): default to ["script", "noscript", "style", "head", "header","nav","footer", "iframe"]
|
|
21
21
|
data["excludeClass"] (list[str]): ["class1", "class2"]
|
|
22
22
|
data["excludeId"] (list[str]): ["id1", "id2"]
|
|
23
|
+
data["restrictDomain"] (bool): if True, only urls from the same domain will be loaded, default to True
|
|
23
24
|
"""
|
|
24
25
|
def __init__(self, knowledgebase_path: str, data: dict[str, Any]):
|
|
25
26
|
super().__init__(knowledgebase_path, data)
|
|
@@ -30,6 +31,7 @@ class Sitemap(IntegrationStrategy):
|
|
|
30
31
|
self.__exclude_tag: list[str] = self.data.get("excludeTag",[]) # type: ignore
|
|
31
32
|
self.__exclude_class: list[str] = self.data.get("excludeClass",[]) # type: ignore
|
|
32
33
|
self.__exclude_id: list[str] = self.data.get("excludeId",[]) # type: ignore
|
|
34
|
+
self.__restrict_to_same_domain: bool = self.data.get("restrictDomain", True) # type: ignore
|
|
33
35
|
def working_subdirectory(self) -> str:
|
|
34
36
|
return ""
|
|
35
37
|
def _extract(self, tag: Tag) -> str:
|
|
@@ -75,7 +77,8 @@ class Sitemap(IntegrationStrategy):
|
|
|
75
77
|
web_path=self._remap_if_local(self.__sitemap_url),
|
|
76
78
|
filter_urls=self.__filter_urls,
|
|
77
79
|
parsing_function=self._parse,
|
|
78
|
-
is_local=self._is_local(self.__sitemap_url)
|
|
80
|
+
is_local=self._is_local(self.__sitemap_url),
|
|
81
|
+
restrict_to_same_domain=self.__restrict_to_same_domain
|
|
79
82
|
)
|
|
80
83
|
_docs = self._output([document async for document in self.alazy_load(_loader)])
|
|
81
84
|
if self._is_local(self.__sitemap_url):
|
|
@@ -54,13 +54,13 @@ ws_bom_robot_app/llm/vector_store/integration/manager.py,sha256=5Fl3XML6f1wmgrai
|
|
|
54
54
|
ws_bom_robot_app/llm/vector_store/integration/s3.py,sha256=3kh-VmH84IW7DdSLvOk6td1VBJ9aohlVJsk5F3cYj0U,3320
|
|
55
55
|
ws_bom_robot_app/llm/vector_store/integration/sftp.py,sha256=WNzjjS1EUykgFB-8e7QkecSa1r1jTJqKyGzR25uJCtM,2848
|
|
56
56
|
ws_bom_robot_app/llm/vector_store/integration/sharepoint.py,sha256=zqqn-6qPrK50Phch4nZHJTgaPyPkGe7W2InGL_Ru6vE,5376
|
|
57
|
-
ws_bom_robot_app/llm/vector_store/integration/sitemap.py,sha256=
|
|
57
|
+
ws_bom_robot_app/llm/vector_store/integration/sitemap.py,sha256=g0TIRZ2qIpEuVCZ9Bn0MvvxYZtU3wptnTEjoKNZyBAg,5019
|
|
58
58
|
ws_bom_robot_app/llm/vector_store/integration/slack.py,sha256=FMjESXm2QetFXI6i8epze7Kbbu22fV8CVaxb71AHnJ8,2572
|
|
59
59
|
ws_bom_robot_app/llm/vector_store/loader/__init__.py,sha256=47DEQpj8HBSa-_TImW-5JCeuQeRkm5NMpJWZG3hSuFU,0
|
|
60
60
|
ws_bom_robot_app/llm/vector_store/loader/base.py,sha256=L_ugekNuAq0N9O-24wtlHSNHkqSeD-KsJrfGt_FX9Oc,5340
|
|
61
61
|
ws_bom_robot_app/llm/vector_store/loader/docling.py,sha256=yP0zgXLeFAlByaYuj-6cYariuknckrFds0dxdRcnVz8,3456
|
|
62
62
|
ws_bom_robot_app/llm/vector_store/loader/json_loader.py,sha256=LDppW0ZATo4_1hh-KlsAM3TLawBvwBxva_a7k5Oz1sc,858
|
|
63
|
-
ws_bom_robot_app-0.0.
|
|
64
|
-
ws_bom_robot_app-0.0.
|
|
65
|
-
ws_bom_robot_app-0.0.
|
|
66
|
-
ws_bom_robot_app-0.0.
|
|
63
|
+
ws_bom_robot_app-0.0.48.dist-info/METADATA,sha256=pzuC_inHOf1_cwPufxB6ntSFPEbERRZ3VStx2cI6xuY,8348
|
|
64
|
+
ws_bom_robot_app-0.0.48.dist-info/WHEEL,sha256=CmyFI0kx5cdEMTLiONQRbGQwjIoR1aIYB7eCAQ4KPJ0,91
|
|
65
|
+
ws_bom_robot_app-0.0.48.dist-info/top_level.txt,sha256=Yl0akyHVbynsBX_N7wx3H3ZTkcMLjYyLJs5zBMDAKcM,17
|
|
66
|
+
ws_bom_robot_app-0.0.48.dist-info/RECORD,,
|
|
File without changes
|