PagedList
trendminer_interface._services.PagedList
Bases: Generic[T]
Paginated list of search results
has_next
property
has_next: bool
Whether there is a next page of results available
page_number
property
page_number: int
The current page number (0-indexed)
size
property
size: int
The requested number of items per page. Not necessarily the same as the actual number of items on the page
data
property
data: list[T]
The content of the current result page.
total_pages
property
total_pages: int | None
The total number of pages
Not available for user groups
total_elements
property
total_elements: int | None
The total number of elements across all pages
Not available for user groups
collect
collect(max_pages: int | None = 10) -> list[T]
Collects all pages of results into a single list
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
max_pages
|
int | None
|
Maximum number of pages to collect. Default is set to 10 to prevent accidental excessive API calls. Can be set to None to collect all pages regardless of number. Must be 1 (only the current page) or higher if set. |
10
|
Returns:
| Type | Description |
|---|---|
content
|
A list with the items of all collected pages. |
stream
stream(max_pages: int | None = 10) -> Iterator[T]
Streams items one by one, loading new pages as needed
Parameters:
| Name | Type | Description | Default |
|---|---|---|---|
max_pages
|
int | None
|
Maximum number of pages to stream. Default is set to 10 to prevent accidental excessive API calls. Can be set to None to stream all pages regardless of number. Must be 1 (only the current page) or higher if set. |
10
|
Yields:
| Name | Type | Description |
|---|---|---|
content |
T
|
Each individual item from the content of all streamed pages, one at a time. |