airbyte-source-google-search-console 1.7.0__py3-none-any.whl → 1.8.0__py3-none-any.whl

source_google_search_console/source.py

@@ -3,14 +3,14 @@
 #
 
 import json
-from typing import Any, List, Mapping, Optional, Tuple, Union
+from typing import Any, List, Mapping, Optional, Union
 from urllib.parse import urlparse
 
 import jsonschema
 import pendulum
 import requests
 
-from airbyte_cdk.models import ConfiguredAirbyteCatalog, FailureType, SyncMode
+from airbyte_cdk.models import ConfiguredAirbyteCatalog, FailureType
 from airbyte_cdk.sources.declarative.yaml_declarative_source import YamlDeclarativeSource
 from airbyte_cdk.sources.source import TState
 from airbyte_cdk.sources.streams import Stream
@@ -23,15 +23,7 @@ from source_google_search_console.exceptions import (
 )
 from source_google_search_console.service_account_authenticator import ServiceAccountAuthenticator
 from source_google_search_console.streams import (
-    SearchAnalyticsAllFields,
     SearchAnalyticsByCustomDimensions,
-    SearchAnalyticsByDate,
-    SearchAnalyticsByDevice,
-    SearchAnalyticsByPage,
-    SearchAnalyticsByQuery,
-    SearchAnalyticsPageReport,
-    SearchAnalyticsSiteReportByPage,
-    SearchAnalyticsSiteReportBySite,
 )
 
 
@@ -147,18 +139,6 @@ class SourceGoogleSearchConsole(YamlDeclarativeSource):
 
         streams = super().streams(config=config)
 
-        streams.extend(
-            [
-                SearchAnalyticsByDevice(**stream_config),
-                SearchAnalyticsByDate(**stream_config),
-                SearchAnalyticsByQuery(**stream_config),
-                SearchAnalyticsByPage(**stream_config),
-                SearchAnalyticsAllFields(**stream_config),
-                SearchAnalyticsPageReport(**stream_config),
-                SearchAnalyticsSiteReportBySite(**stream_config),
-                SearchAnalyticsSiteReportByPage(**stream_config),
-            ]
-        )
         streams = streams + self.get_custom_reports(config=config, stream_config=stream_config)
 
         return streams

source_google_search_console/spec.json

@@ -158,6 +158,15 @@
         "default": "final",
         "order": 6
       },
+      "num_workers": {
+        "type": "integer",
+        "title": "Number of concurrent workers",
+        "minimum": 2,
+        "maximum": 100,
+        "default": 40,
+        "examples": [30, 40, 50],
+        "description": "The number of worker threads to use for the sync. For more details on Google Search Console rate limits, refer to the <a href=\"https://developers.google.com/webmaster-tools/limits\">docs</a>."
+      },
       "always_use_aggregation_type_auto": {
         "type": "boolean",
         "title": "Always Use Aggregation Type Auto",

source_google_search_console/streams.py

@@ -294,53 +294,6 @@ class SearchAnalytics(GoogleSearchConsole, CheckpointMixin, ABC):
             yield record
 
 
-class SearchAnalyticsByDate(SearchAnalytics):
-    primary_key = ["site_url", "date", "search_type"]
-    search_types = ["web", "news", "image", "video", "discover", "googleNews"]
-    dimensions = ["date"]
-
-
-class SearchAnalyticsByDevice(SearchAnalytics):
-    primary_key = ["site_url", "date", "device", "search_type"]
-    search_types = ["web", "news", "image", "video", "googleNews"]
-    dimensions = ["date", "device"]
-
-
-class SearchAnalyticsByPage(SearchAnalytics):
-    primary_key = ["site_url", "date", "page", "search_type"]
-    search_types = ["web", "news", "image", "video", "discover", "googleNews"]
-    dimensions = ["date", "page"]
-
-
-class SearchAnalyticsByQuery(SearchAnalytics):
-    primary_key = ["site_url", "date", "query", "search_type"]
-    dimensions = ["date", "query"]
-
-
-class SearchAnalyticsAllFields(SearchAnalytics):
-    primary_key = ["site_url", "date", "country", "device", "query", "page", "search_type"]
-    dimensions = ["date", "country", "device", "page", "query"]
-
-
-class SearchAnalyticsSiteReportBySite(SearchAnalytics):
-    primary_key = ["site_url", "date", "country", "device", "search_type"]
-    dimensions = ["date", "country", "device"]
-    aggregation_type = QueryAggregationType.by_property
-
-
-class SearchAnalyticsSiteReportByPage(SearchAnalytics):
-    primary_key = ["site_url", "date", "country", "device", "search_type"]
-    search_types = ["web", "news", "image", "video", "googleNews"]
-    dimensions = ["date", "country", "device"]
-    aggregation_type = QueryAggregationType.by_page
-
-
-class SearchAnalyticsPageReport(SearchAnalytics):
-    primary_key = ["site_url", "date", "country", "device", "search_type", "page"]
-    search_types = ["web", "news", "image", "video", "googleNews"]
-    dimensions = ["date", "country", "device", "page"]
-
-
 class SearchAnalyticsByCustomDimensions(SearchAnalytics):
     # `date` is a cursor field therefore should be mandatory
     DEFAULT_DIMENSIONS = ["date"]

airbyte_source_google_search_console-1.7.0.dist-info/RECORD

@@ -1,22 +0,0 @@
-source_google_search_console/__init__.py,sha256=HQCPu-CK7XmVDtP9rmTdB2XyraVCc6pv9pw38-O8y48,1191
-source_google_search_console/components.py,sha256=5o8kH2xwYUvk3yjnSd6okJVF2KBUqUIF4V97xRpPAyI,2803
-source_google_search_console/config_migrations.py,sha256=Cl4SUdJpAf6wMM_vVhqjjU89NfUq9LIGJ9zNrWiBk-A,4235
-source_google_search_console/exceptions.py,sha256=iD3jYC4WxVCEKGsqQ7Vaj1tbjhJZ4S5mnSDnwFJdsIQ,1097
-source_google_search_console/manifest.yaml,sha256=m6IhF4bRwF9nAivJev2ebB8QPvoKRbYixF-5ZKSBkms,25882
-source_google_search_console/run.py,sha256=TBkPlseTERarkj6wL8AMEKgm5Xsb2drnltPVH6257-M,2195
-source_google_search_console/schemas/search_analytics_all_fields.json,sha256=iQxRh_c_yz3uGofqpo1KX571TMmzYjKScb0PtI6SN_Q,1729
-source_google_search_console/schemas/search_analytics_by_date.json,sha256=meCbWDayc1y0q-Lu-CAdjQVnsM8xZBX3BdF129UC1P8,1388
-source_google_search_console/schemas/search_analytics_by_device.json,sha256=VtoFjmmv9rx-uhSFaRn0wm4LeSxRIaexrxg2Spvbneo,1525
-source_google_search_console/schemas/search_analytics_by_page.json,sha256=KyUojZc4Lv3hPswxIJzUL5QDNsbvSugGjl_uHGF7Am4,1473
-source_google_search_console/schemas/search_analytics_by_query.json,sha256=mYc1Fu1A5TWLZCApZSjjfs_urW5BRUCPQI3rw3yQjt4,1439
-source_google_search_console/schemas/search_analytics_page_report.json,sha256=-b0Y0LenTchS0q9A2aQ4hIjUjXkYF8erOtyrTMhf6MM,1776
-source_google_search_console/schemas/search_analytics_site_report_by_page.json,sha256=hWKHkm1reqGGu1dNcWBe6_XkZ5tK-UaiymrYRVgxRxI,1515
-source_google_search_console/schemas/search_analytics_site_report_by_site.json,sha256=rAh6LuNy7nCrrNM9MTd0qxAVc886ecQaqWRgV63OfyA,1408
-source_google_search_console/service_account_authenticator.py,sha256=pAWKAXfwfTY3xkXvQJH0EyFphFULdCIcC47YXYTO9X8,1307
-source_google_search_console/source.py,sha256=ddqYq3ZSNrVtGvuQQ5WnHgMLuf92AtumzdDKWGffwRo,8721
-source_google_search_console/spec.json,sha256=SensIrWvoie2aGwxmgoMYpq2jb3XP3F2myQWCVQ4db8,8980
-source_google_search_console/streams.py,sha256=yb78hGaKPDCJoFqPSun3lDhr8VhSxcLgXTLtX2ouVpY,15804
-airbyte_source_google_search_console-1.7.0.dist-info/METADATA,sha256=R-pJQZUfWv2duwzSFWta7vbn8V01TeESsaoRO5728mo,5621
-airbyte_source_google_search_console-1.7.0.dist-info/WHEEL,sha256=b4K_helf-jlQoXBBETfwnf4B04YC67LOev0jo4fX5m8,88
-airbyte_source_google_search_console-1.7.0.dist-info/entry_points.txt,sha256=DMcgc9bCX-Vt6hm_68pa77qS3eGdeMhg-UdlFc-XKUM,85
-airbyte_source_google_search_console-1.7.0.dist-info/RECORD,,

source_google_search_console/schemas/search_analytics_all_fields.json

@@ -1,53 +0,0 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "properties": {
-    "site_url": {
-      "description": "The URL of the site from which the data originates.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search (e.g., web, image, video) that triggered the search result.",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date when the search query occurred.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "country": {
-      "description": "The country from which the search query originated.",
-      "type": ["null", "string"]
-    },
-    "device": {
-      "description": "The type of device used by the user (e.g., desktop, mobile).",
-      "type": ["null", "string"]
-    },
-    "page": {
-      "description": "The page URL that appeared in the search results.",
-      "type": ["null", "string"]
-    },
-    "query": {
-      "description": "The search query entered by the user.",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The number of times users clicked on the search result for a specific query.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The number of times a search result appeared in response to a query.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "Click-through rate, calculated as clicks divided by impressions.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position of the search result on the search engine results page.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_by_date.json

@@ -1,37 +0,0 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "properties": {
-    "site_url": {
-      "description": "The URL of the site for which the search analytics data is being reported.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search query (e.g., web, image, video) that generated the search analytics data.",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date for which the search analytics data is being reported.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "clicks": {
-      "description": "The total number of times users clicked on the search result for the site URL on the specific date.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The number of times the site URL was displayed in the search results to users on the specific date.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "The click-through rate (CTR) represents the percentage of total impressions that resulted in a click to the site URL.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position of the site URL in the search results pages for the specific date.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_by_device.json

@@ -1,41 +0,0 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "properties": {
-    "site_url": {
-      "description": "The URL of the site for which search analytics data is being provided.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search performed (e.g., web search, image search, video search).",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date for which the search analytics data is provided.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "device": {
-      "description": "The type of device used by the user for the search query (e.g., desktop, mobile).",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The total number of times a user clicked on a search result linking to the target site.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The total number of times a user saw a link to the target site in search results.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "Click-through rate represents the ratio of clicks to impressions, showing the effectiveness of your site in attracting clicks from search results.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position of the site's URLs in search results for the given query or queries.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_by_page.json

@@ -1,41 +0,0 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "properties": {
-    "site_url": {
-      "description": "The URL of the site for which the search analytics data is being reported.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search query that led to the page being displayed in search results.",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date for which the search analytics data is reported.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "page": {
-      "description": "The URL of the specific page being analyzed for search analytics data.",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The number of times a user clicked on the search result linking to the page.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The number of times a page from the site appeared in the search results viewed by users.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "Click-through rate (CTR) is the ratio of clicks to impressions, indicating the effectiveness of the page in generating clicks.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position at which the page appeared in search results.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_by_query.json

@@ -1,41 +0,0 @@
-{
-  "$schema": "http://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "properties": {
-    "site_url": {
-      "description": "The URL of the site for which the search analytics data is captured.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search result (e.g., web, image, video) for the specific query.",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date for which the search analytics data is recorded.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "query": {
-      "description": "The search query for which the search analytics data is recorded.",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The number of times users clicked on the search result for the specific query.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The number of times the search result was displayed for the specific query.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "The click-through rate (percentage) for the specific query, calculated as clicks divided by impressions.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position at which the search result appeared for the specific query.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_page_report.json

@@ -1,50 +0,0 @@
-{
-  "$schema": "https://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "additionalProperties": true,
-  "properties": {
-    "site_url": {
-      "description": "The URL of the website for which the search analytics data is being reported.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search (e.g., web, image, video) that led users to the website.",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date when the search data was recorded.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "country": {
-      "description": "The country from which the search originated.",
-      "type": ["null", "string"]
-    },
-    "page": {
-      "description": "The specific page URL within the website that appeared in search results.",
-      "type": ["null", "string"]
-    },
-    "device": {
-      "description": "The type of device used by the user for the search query (e.g., desktop, mobile).",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The total number of times users clicked on search results that led to the linked website.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The total number of times a search result from the linked website was shown to users.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "Click-through rate: The percentage of clicks out of the total impressions for a given search query.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position at which the website's search results appeared to users.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_site_report_by_page.json

@@ -1,46 +0,0 @@
-{
-  "$schema": "https://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "additionalProperties": true,
-  "properties": {
-    "site_url": {
-      "description": "The URL of the page on the site that is being reported.",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search query that led to the page being shown.",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date for which the data is being reported.",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "country": {
-      "description": "The country from which the search traffic originated.",
-      "type": ["null", "string"]
-    },
-    "device": {
-      "description": "The type of device used by the searcher (e.g., desktop, mobile).",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The total number of clicks received by the page from search results.",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The total number of times the page appeared in search results.",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "The click-through rate, i.e., the percentage of total impressions that resulted in clicks.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position at which the page appeared in search results.",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}

source_google_search_console/schemas/search_analytics_site_report_by_site.json

@@ -1,46 +0,0 @@
-{
-  "$schema": "https://json-schema.org/draft-07/schema#",
-  "type": "object",
-  "additionalProperties": true,
-  "properties": {
-    "site_url": {
-      "description": "The URL of the site being analyzed",
-      "type": ["null", "string"]
-    },
-    "search_type": {
-      "description": "The type of search (e.g., web, image, video)",
-      "type": ["null", "string"]
-    },
-    "date": {
-      "description": "The date of the search analytics data",
-      "type": ["null", "string"],
-      "format": "date"
-    },
-    "country": {
-      "description": "The country where the search took place",
-      "type": ["null", "string"]
-    },
-    "device": {
-      "description": "The type of device used for the search (e.g., mobile, desktop)",
-      "type": ["null", "string"]
-    },
-    "clicks": {
-      "description": "The number of times users clicked on a search result linking to the site",
-      "type": ["null", "integer"]
-    },
-    "impressions": {
-      "description": "The number of times the site appeared in search results",
-      "type": ["null", "integer"]
-    },
-    "ctr": {
-      "description": "Click-through rate calculated as clicks divided by impressions",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    },
-    "position": {
-      "description": "The average position of the site in search results",
-      "type": ["null", "number"],
-      "multipleOf": 1e-25
-    }
-  }
-}