airbyte-source-google-search-console 1.5.16.dev202503252224__py3-none-any.whl → 1.6.0.dev202503282247__py3-none-any.whl

{airbyte_source_google_search_console-1.5.16.dev202503252224.dist-info → airbyte_source_google_search_console-1.6.0.dev202503282247.dist-info}/METADATA

@@ -1,6 +1,6 @@
 Metadata-Version: 2.3
 Name: airbyte-source-google-search-console
-Version: 1.5.16.dev202503252224
+Version: 1.6.0.dev202503282247
 Summary: Source implementation for Google Search Console.
 License: Elv2
 Author: Airbyte

{airbyte_source_google_search_console-1.5.16.dev202503252224.dist-info → airbyte_source_google_search_console-1.6.0.dev202503282247.dist-info}/RECORD

@@ -9,18 +9,21 @@ source_google_search_console/schemas/search_analytics_by_device.json,sha256=VtoF
 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_keyword_page_report.json,sha256=hCQZbpW9UNiCOrnUMkj4oQdzY0k8Hqoh1XkuKHvyvZw,1681
+source_google_search_console/schemas/search_analytics_keyword_page_report_minimal_dimensions.json,sha256=hCQZbpW9UNiCOrnUMkj4oQdzY0k8Hqoh1XkuKHvyvZw,1681
 source_google_search_console/schemas/search_analytics_keyword_site_report_by_page.json,sha256=lFpbShmTQhFyySc_JCdIOqMP9RfOIA15fpQN9xNXtTI,1626
+source_google_search_console/schemas/search_analytics_keyword_site_report_by_page_minimal_dimensions.json,sha256=lFpbShmTQhFyySc_JCdIOqMP9RfOIA15fpQN9xNXtTI,1626
 source_google_search_console/schemas/search_analytics_keyword_site_report_by_site.json,sha256=AVUl5x9jLL0FcStbTR-FMy-_A7uM-VlrJ-kdCBXWz_g,1755
+source_google_search_console/schemas/search_analytics_keyword_site_report_by_site_minimal_dimensions.json,sha256=AVUl5x9jLL0FcStbTR-FMy-_A7uM-VlrJ-kdCBXWz_g,1755
 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/schemas/sitemaps.json,sha256=coyPSZCAfzMheybfRp4WPAZCp5JF2KGRF2rWK8oC080,1775
 source_google_search_console/schemas/sites.json,sha256=WNiCRuStPL1YkJiFa8FEbNJmqaERAOf9Yow6ygIumvo,383
 source_google_search_console/service_account_authenticator.py,sha256=gjUxt0xFxj82uviCQNTsA1Jlee__UDhYNjE7bRO1G0U,1227
-source_google_search_console/source.py,sha256=8n10_agSa2rvPvEyBvOfOpmpEardbEhi3H0vlK2A4_g,10002
+source_google_search_console/source.py,sha256=4UJW4KUVGo5WxH9ewWBVSemXH17Iqp-eaNRtUfLNET8,10431
 source_google_search_console/spec.json,sha256=WYtFvaSqWYGm1Dt2yV9G92U78Q94rh9oarbxJe3H7xo,8470
-source_google_search_console/streams.py,sha256=7C9E8gYpzU4VdQGsrS2AqjQTdMjeQ3mS3iAfvywADek,20436
-airbyte_source_google_search_console-1.5.16.dev202503252224.dist-info/METADATA,sha256=llHA_zmljhs2_GK5_vKBp_TJj7mgizzaB_kFodADb_A,5646
-airbyte_source_google_search_console-1.5.16.dev202503252224.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
-airbyte_source_google_search_console-1.5.16.dev202503252224.dist-info/entry_points.txt,sha256=DMcgc9bCX-Vt6hm_68pa77qS3eGdeMhg-UdlFc-XKUM,85
-airbyte_source_google_search_console-1.5.16.dev202503252224.dist-info/RECORD,,
+source_google_search_console/streams.py,sha256=1rgtlcVUyAP3oKRNQ34qIh6rwTzvlr636GiEy0RPQLs,20508
+airbyte_source_google_search_console-1.6.0.dev202503282247.dist-info/METADATA,sha256=YyTR_gsHe3OcxaiSStl2wtO8Wa5Vsfm16QJ6miZG8JA,5645
+airbyte_source_google_search_console-1.6.0.dev202503282247.dist-info/WHEEL,sha256=XbeZDeTWKc1w7CSIyre5aMDU_-PohRwTQceYnisIYYY,88
+airbyte_source_google_search_console-1.6.0.dev202503282247.dist-info/entry_points.txt,sha256=DMcgc9bCX-Vt6hm_68pa77qS3eGdeMhg-UdlFc-XKUM,85
+airbyte_source_google_search_console-1.6.0.dev202503282247.dist-info/RECORD,,

source_google_search_console/schemas/search_analytics_keyword_page_report_minimal_dimensions.json

@@ -0,0 +1,54 @@
+{
+  "$schema": "https://json-schema.org/draft-07/schema#",
+  "type": "object",
+  "additionalProperties": true,
+  "properties": {
+    "site_url": {
+      "description": "The URL of the website being monitored.",
+      "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 data collected.",
+      "type": ["null", "string"],
+      "format": "date"
+    },
+    "country": {
+      "description": "The country where the search is made.",
+      "type": ["null", "string"]
+    },
+    "device": {
+      "description": "The device type used for the search (e.g., desktop, mobile).",
+      "type": ["null", "string"]
+    },
+    "page": {
+      "description": "The page URL on which the keyword appears in search results.",
+      "type": ["null", "string"]
+    },
+    "query": {
+      "description": "The search query used to find the site.",
+      "type": ["null", "string"]
+    },
+    "clicks": {
+      "description": "The number of clicks for the keyword on a specific page.",
+      "type": ["null", "integer"]
+    },
+    "impressions": {
+      "description": "The number of times the keyword appeared in search results.",
+      "type": ["null", "integer"]
+    },
+    "ctr": {
+      "description": "Click-through rate which is the percentage of clicks divided by impressions.",
+      "type": ["null", "number"],
+      "multipleOf": 1e-25
+    },
+    "position": {
+      "description": "The average position of the keyword on search results pages.",
+      "type": ["null", "number"],
+      "multipleOf": 1e-25
+    }
+  }
+}

source_google_search_console/schemas/search_analytics_keyword_site_report_by_page_minimal_dimensions.json

@@ -0,0 +1,50 @@
+{
+  "$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 retrieved.",
+      "type": ["null", "string"]
+    },
+    "search_type": {
+      "description": "The type of search conducted (e.g., web, image, video).",
+      "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 query originated.",
+      "type": ["null", "string"]
+    },
+    "device": {
+      "description": "The device type used for the search query (e.g., desktop, mobile).",
+      "type": ["null", "string"]
+    },
+    "query": {
+      "description": "The search query used by the user.",
+      "type": ["null", "string"]
+    },
+    "clicks": {
+      "description": "The number of times users clicked on your website link in search results.",
+      "type": ["null", "integer"]
+    },
+    "impressions": {
+      "description": "The number of times your website link appeared in search results.",
+      "type": ["null", "integer"]
+    },
+    "ctr": {
+      "description": "Click-through rate: Number of clicks divided by the number of impressions.",
+      "type": ["null", "number"],
+      "multipleOf": 1e-25
+    },
+    "position": {
+      "description": "The average position of your website link in search results.",
+      "type": ["null", "number"],
+      "multipleOf": 1e-25
+    }
+  }
+}

source_google_search_console/schemas/search_analytics_keyword_site_report_by_site_minimal_dimensions.json

@@ -0,0 +1,50 @@
+{
+  "$schema": "https://json-schema.org/draft-07/schema#",
+  "type": "object",
+  "additionalProperties": true,
+  "properties": {
+    "site_url": {
+      "description": "The URL of the site for which the search analytics data is recorded.",
+      "type": ["null", "string"]
+    },
+    "search_type": {
+      "description": "The type of search (e.g., web search, image search) that generated the analytics data.",
+      "type": ["null", "string"]
+    },
+    "date": {
+      "description": "The date for which the search analytics data is recorded.",
+      "type": ["null", "string"],
+      "format": "date"
+    },
+    "country": {
+      "description": "The country from which the search originated.",
+      "type": ["null", "string"]
+    },
+    "device": {
+      "description": "The type of device used by the user during the search (e.g., desktop, mobile).",
+      "type": ["null", "string"]
+    },
+    "query": {
+      "description": "The search query used by the user to find the site in search results.",
+      "type": ["null", "string"]
+    },
+    "clicks": {
+      "description": "The number of times users clicked on the search result linking to the site.",
+      "type": ["null", "integer"]
+    },
+    "impressions": {
+      "description": "The number of times the site was shown in search results to users.",
+      "type": ["null", "integer"]
+    },
+    "ctr": {
+      "description": "Click-through rate represents the percentage of users who clicked on the site's link after seeing it in search results.",
+      "type": ["null", "number"],
+      "multipleOf": 1e-25
+    },
+    "position": {
+      "description": "The average ranking position of the site in search results.",
+      "type": ["null", "number"],
+      "multipleOf": 1e-25
+    }
+  }
+}

source_google_search_console/source.py

@@ -32,8 +32,11 @@ from source_google_search_console.streams import (
     SearchAnalyticsByPage,
     SearchAnalyticsByQuery,
     SearchAnalyticsKeywordPageReport,
+    SearchAnalyticsKeywordPageReportMinimalDimensions,
     SearchAnalyticsKeywordSiteReportByPage,
+    SearchAnalyticsKeywordSiteReportByPageMinimalDimensions,
     SearchAnalyticsKeywordSiteReportBySite,
+    SearchAnalyticsKeywordSiteReportBySiteMinimalDimensions,
     SearchAnalyticsPageReport,
     SearchAnalyticsSiteReportByPage,
     SearchAnalyticsSiteReportBySite,
@@ -182,11 +185,14 @@ class SourceGoogleSearchConsole(AbstractSource):
             SearchAnalyticsByPage(**stream_config),
             SearchAnalyticsAllFields(**stream_config),
             SearchAnalyticsKeywordPageReport(**stream_config),
+            SearchAnalyticsKeywordPageReportMinimalDimensions(**stream_config),
             SearchAnalyticsPageReport(**stream_config),
             SearchAnalyticsSiteReportBySite(**stream_config),
             SearchAnalyticsSiteReportByPage(**stream_config),
             SearchAnalyticsKeywordSiteReportByPage(**stream_config),
+            SearchAnalyticsKeywordSiteReportByPageMinimalDimensions(**stream_config),
             SearchAnalyticsKeywordSiteReportBySite(**stream_config),
+            SearchAnalyticsKeywordSiteReportBySiteMinimalDimensions(**stream_config),
         ]
 
         streams = streams + self.get_custom_reports(config=config, stream_config=stream_config)

source_google_search_console/streams.py

@@ -390,19 +390,11 @@ class SearchByKeyword(SearchAnalytics):
             keywords_records = search_appearance_stream.read_records(
                 sync_mode=SyncMode.full_refresh, stream_state=stream_state, stream_slice=stream_slice
             )
-            # Safely extract keywords, handling cases where "searchAppearance" might be missing
-            keywords = {record["searchAppearance"] for record in keywords_records if "searchAppearance" in record}
-
-            if keywords:
-                # If keywords exist, yield a slice for each keyword with filters
-                for keyword in keywords:
-                    filters = {"dimension": "searchAppearance", "operator": "equals", "expression": keyword}
-                    # Create a copy to avoid modifying the original stream_slice
-                    stream_slice_with_filter = stream_slice.copy()
-                    stream_slice_with_filter["dimensionFilterGroups"] = [{"groupType": "and", "filters": [filters]}]
-                    yield stream_slice_with_filter
-            else:
-                # If no keywords are found, yield the base slice without filters
+            keywords = {record["searchAppearance"] for record in keywords_records}
+
+            for keyword in keywords:
+                filters = {"dimension": "searchAppearance", "operator": "equals", "expression": keyword}
+                stream_slice["dimensionFilterGroups"] = [{"groupType": "and", "filters": filters}]
                 yield stream_slice
 
     def request_body_json(
@@ -412,8 +404,7 @@ class SearchByKeyword(SearchAnalytics):
         next_page_token: Mapping[str, Any] = None,
     ) -> Optional[Union[Dict[str, Any], str]]:
         data = super().request_body_json(stream_state, stream_slice, next_page_token)
-        if "dimensionFilterGroups" in stream_slice:
-            data["dimensionFilterGroups"] = stream_slice["dimensionFilterGroups"]
+        data["dimensionFilterGroups"] = stream_slice["dimensionFilterGroups"]
         return data
 
 
@@ -422,18 +413,35 @@ class SearchAnalyticsKeywordPageReport(SearchByKeyword):
     dimensions = ["date", "country", "device", "query", "page"]
 
 
+class SearchAnalyticsKeywordPageReportMinimalDimensions(SearchByKeyword):
+    primary_key = ["site_url", "date", "country", "device", "query", "page", "search_type"]
+    dimensions = ["date"]
+
+
 class SearchAnalyticsKeywordSiteReportByPage(SearchByKeyword):
     primary_key = ["site_url", "date", "country", "device", "query", "search_type"]
     dimensions = ["date", "country", "device", "query"]
     aggregation_type = QueryAggregationType.by_page
 
 
+class SearchAnalyticsKeywordSiteReportByPageMinimalDimensions(SearchByKeyword):
+    primary_key = ["site_url", "date", "country", "device", "query", "search_type"]
+    dimensions = ["date"]
+    aggregation_type = QueryAggregationType.by_page
+
+
 class SearchAnalyticsKeywordSiteReportBySite(SearchByKeyword):
     primary_key = ["site_url", "date", "country", "device", "query", "search_type"]
     dimensions = ["date", "country", "device", "query"]
     aggregation_type = QueryAggregationType.by_property
 
 
+class SearchAnalyticsKeywordSiteReportBySiteMinimalDimensions(SearchByKeyword):
+    primary_key = ["site_url", "date", "country", "device", "query", "search_type"]
+    dimensions = ["date"]
+    aggregation_type = QueryAggregationType.by_property
+
+
 class SearchAnalyticsSiteReportBySite(SearchAnalytics):
     primary_key = ["site_url", "date", "country", "device", "search_type"]
     dimensions = ["date", "country", "device"]