Browser PDF Text Extraction Prioritization
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
PDF document retrieval and searching in web browsers, which rely on single-thread processing, result in significant delays and poor user experience due to the need for extensive text extraction before search results can be provided, limiting user interaction during the process.
Innovation Solution
Implementing a JavaScript application in web browsers that prioritizes text extraction into small, manageable units (less than 200 ms) and focuses on extracting text from currently viewed pages first, allowing for rapid search results and minimizing interference with other operations like scrolling, by using available processing resources efficiently and storing extracted text in an indexed data structure.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If text extraction is performed on the entire PDF document before searching, then complete search results can be obtained, but the processing time becomes excessively long and the application appears unresponsive
Solution Approach 1:
The patent divides the PDF document into multiple pages and processes text extraction page by page rather than extracting all text at once. This segmentation allows the application to provide incremental search results as each page is processed, reducing the perceived waiting time while ultimately providing complete search results across all pages.
Solution Approach 2:
The patent performs text extraction on a page-by-page basis before the user initiates a search, preparing searchable text in advance. This preliminary action on individual pages allows search results to be generated quickly when the user submits a query, rather than waiting for complete document processing.
2Productivity
If text extraction is performed continuously in the single processing thread, then complete text can be extracted for searching, but other user operations like scrolling and page switching are blocked
Solution Approach 1:
The patent implements periodic text extraction where the single processing thread alternates between extracting text from PDF pages and yielding control to allow other user operations. This periodic approach maintains text extraction progress while periodically allowing scrolling, page switching, and other interactions to proceed, preventing the application from appearing frozen.
Solution Approach 2:
The patent dynamically adjusts the text extraction process by pausing and resuming based on user interactions. When user operations are detected, the extraction thread yields; when resources are available and no user input is pending, extraction resumes. This dynamic behavior balances productivity with user responsiveness.
3Measurement precision
If the application waits for complete text extraction before providing search results, then accurate search can be performed, but user experience deteriorates due to excessive waiting time
Solution Approach 1:
The patent provides search results based on partially extracted text from processed pages rather than waiting for complete document extraction. This partial action approach delivers usable search results quickly while continuing to extract text from remaining pages in the background, improving user experience without significantly compromising search accuracy for the portions that have been processed.
Data Source
AI summary
Methods and apparatus for retrieving PDF documents, performing text extraction operations on portions or all of a retrieved document and supporting search operations in a manner that allows search results to be quickly provided for at least portions of a PDF document being viewed are described. The methods and apparatus are particularly useful in applications, such as many applications executed by a browser, where the application is limited to a single processing thread and thus must perform all or many processing operations sequentially. By prioritizing document pages which are being viewed for text extraction even before a search is initiated and by performing text extraction in small periods of time and storing the results, in many cases a user can be provided with text search results for a page being viewed in relatively little time and without the program, e.g., java script browser application, appearing non-responsive.