datamarket 0.6.0__py3-none-any.whl → 0.10.3__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.
Potentially problematic release.
This version of datamarket might be problematic. Click here for more details.
- datamarket/__init__.py +0 -1
- datamarket/exceptions/__init__.py +1 -0
- datamarket/exceptions/main.py +118 -0
- datamarket/interfaces/alchemy.py +1934 -25
- datamarket/interfaces/aws.py +81 -14
- datamarket/interfaces/azure.py +127 -0
- datamarket/interfaces/drive.py +60 -10
- datamarket/interfaces/ftp.py +37 -14
- datamarket/interfaces/llm.py +1220 -0
- datamarket/interfaces/nominatim.py +314 -42
- datamarket/interfaces/peerdb.py +272 -104
- datamarket/interfaces/proxy.py +354 -50
- datamarket/interfaces/tinybird.py +7 -15
- datamarket/params/nominatim.py +439 -0
- datamarket/utils/__init__.py +1 -1
- datamarket/utils/airflow.py +10 -7
- datamarket/utils/alchemy.py +2 -1
- datamarket/utils/logs.py +88 -0
- datamarket/utils/main.py +138 -10
- datamarket/utils/nominatim.py +201 -0
- datamarket/utils/playwright/__init__.py +0 -0
- datamarket/utils/playwright/async_api.py +274 -0
- datamarket/utils/playwright/sync_api.py +281 -0
- datamarket/utils/requests.py +655 -0
- datamarket/utils/selenium.py +6 -12
- datamarket/utils/strings/__init__.py +1 -0
- datamarket/utils/strings/normalization.py +217 -0
- datamarket/utils/strings/obfuscation.py +153 -0
- datamarket/utils/strings/standardization.py +40 -0
- datamarket/utils/typer.py +2 -1
- datamarket/utils/types.py +1 -0
- datamarket-0.10.3.dist-info/METADATA +172 -0
- datamarket-0.10.3.dist-info/RECORD +38 -0
- {datamarket-0.6.0.dist-info → datamarket-0.10.3.dist-info}/WHEEL +1 -2
- datamarket-0.6.0.dist-info/METADATA +0 -49
- datamarket-0.6.0.dist-info/RECORD +0 -24
- datamarket-0.6.0.dist-info/top_level.txt +0 -1
- {datamarket-0.6.0.dist-info → datamarket-0.10.3.dist-info/licenses}/LICENSE +0 -0
datamarket/__init__.py
CHANGED
|
@@ -1 +0,0 @@
|
|
|
1
|
-
# init file
|
|
@@ -0,0 +1 @@
|
|
|
1
|
+
from .main import * # noqa: F403
|
|
@@ -0,0 +1,118 @@
|
|
|
1
|
+
########################################################################################################################
|
|
2
|
+
# CLASSES
|
|
3
|
+
|
|
4
|
+
|
|
5
|
+
from typing import Optional
|
|
6
|
+
|
|
7
|
+
from requests import Request, Response
|
|
8
|
+
from requests.exceptions import HTTPError
|
|
9
|
+
|
|
10
|
+
|
|
11
|
+
class ManagedHTTPError(HTTPError):
|
|
12
|
+
"""Signal that this HTTP status was handled and should not be retried."""
|
|
13
|
+
|
|
14
|
+
def __init__(
|
|
15
|
+
self,
|
|
16
|
+
message: Optional[str] = None,
|
|
17
|
+
response: Optional[Response] = None,
|
|
18
|
+
request: Optional[Request] = None,
|
|
19
|
+
*args,
|
|
20
|
+
**kwargs,
|
|
21
|
+
):
|
|
22
|
+
self.response = response
|
|
23
|
+
self.request = request or getattr(response, "request", None)
|
|
24
|
+
|
|
25
|
+
# Build a safe default message
|
|
26
|
+
if not message:
|
|
27
|
+
status = getattr(self.response, "status_code", "unknown")
|
|
28
|
+
url = getattr(self.request, "url", "unknown")
|
|
29
|
+
message = f"HTTP {status} for {url}"
|
|
30
|
+
|
|
31
|
+
self.message = message
|
|
32
|
+
|
|
33
|
+
super().__init__(message, *args, response=response, **kwargs)
|
|
34
|
+
|
|
35
|
+
|
|
36
|
+
class IgnoredHTTPError(ManagedHTTPError):
|
|
37
|
+
"""Exception type that signals the error should be ignored by retry logic."""
|
|
38
|
+
|
|
39
|
+
pass
|
|
40
|
+
|
|
41
|
+
|
|
42
|
+
class NotFoundError(ManagedHTTPError):
|
|
43
|
+
def __init__(
|
|
44
|
+
self,
|
|
45
|
+
message: Optional[str] = None,
|
|
46
|
+
response: Optional[Response] = None,
|
|
47
|
+
request: Optional[Request] = None,
|
|
48
|
+
*args,
|
|
49
|
+
**kwargs,
|
|
50
|
+
):
|
|
51
|
+
if not message:
|
|
52
|
+
status = getattr(response, "status_code", 404)
|
|
53
|
+
req = request or getattr(response, "request", None)
|
|
54
|
+
url = getattr(req, "url", "unknown")
|
|
55
|
+
message = f"HTTP {status} for {url}"
|
|
56
|
+
super().__init__(message, response, request, *args, **kwargs)
|
|
57
|
+
|
|
58
|
+
|
|
59
|
+
class BadRequestError(ManagedHTTPError):
|
|
60
|
+
def __init__(
|
|
61
|
+
self,
|
|
62
|
+
message: Optional[str] = None,
|
|
63
|
+
response: Optional[Response] = None,
|
|
64
|
+
request: Optional[Request] = None,
|
|
65
|
+
*args,
|
|
66
|
+
**kwargs,
|
|
67
|
+
):
|
|
68
|
+
if not message:
|
|
69
|
+
status = getattr(response, "status_code", 400)
|
|
70
|
+
req = request or getattr(response, "request", None)
|
|
71
|
+
url = getattr(req, "url", "unknown")
|
|
72
|
+
message = f"HTTP {status} for {url}"
|
|
73
|
+
super().__init__(message, response, request, *args, **kwargs)
|
|
74
|
+
|
|
75
|
+
|
|
76
|
+
class EmptyResponseError(ManagedHTTPError):
|
|
77
|
+
def __init__(
|
|
78
|
+
self,
|
|
79
|
+
message: Optional[str] = None,
|
|
80
|
+
response: Optional[Response] = None,
|
|
81
|
+
request: Optional[Request] = None,
|
|
82
|
+
*args,
|
|
83
|
+
**kwargs,
|
|
84
|
+
):
|
|
85
|
+
if not message:
|
|
86
|
+
req = request or getattr(response, "request", None)
|
|
87
|
+
url = getattr(req, "url", "unknown")
|
|
88
|
+
message = f"Empty response for {url}"
|
|
89
|
+
super().__init__(message, response, request, *args, **kwargs)
|
|
90
|
+
|
|
91
|
+
|
|
92
|
+
class RedirectionDetectedError(ManagedHTTPError):
|
|
93
|
+
def __init__(
|
|
94
|
+
self,
|
|
95
|
+
message: Optional[str] = None,
|
|
96
|
+
response: Optional[Response] = None,
|
|
97
|
+
request: Optional[Request] = None,
|
|
98
|
+
*args,
|
|
99
|
+
**kwargs,
|
|
100
|
+
):
|
|
101
|
+
if not message:
|
|
102
|
+
status = getattr(response, "status_code", 300)
|
|
103
|
+
req = request or getattr(response, "request", None)
|
|
104
|
+
url = getattr(req, "url", "unknown")
|
|
105
|
+
message = f"HTTP {status} for {url}"
|
|
106
|
+
super().__init__(message, response, request, *args, **kwargs)
|
|
107
|
+
|
|
108
|
+
|
|
109
|
+
class NoWorkingProxiesError(Exception):
|
|
110
|
+
def __init__(self, message="No working proxies available"):
|
|
111
|
+
self.message = message
|
|
112
|
+
super().__init__(self.message)
|
|
113
|
+
|
|
114
|
+
|
|
115
|
+
class EnsureNewIPTimeoutError(Exception):
|
|
116
|
+
def __init__(self, message="Timed out waiting for new IP"):
|
|
117
|
+
self.message = message
|
|
118
|
+
super().__init__(self.message)
|