This method is useful for clients who need a connection open past the expiration time of the access token, which expires after 60 minutes.
\nInstead of user and password, this method requires a valid refresh_token obtained by initially signing in with user and password.
\n","urlObject":{"host":["https://api.parsel.ai/graphql"],"query":[],"variable":[]}},"response":[{"id":"455ee1c6-f376-4ca7-b06d-a59b2d557404","name":"Successful SignIn with refresh token","originalRequest":{"method":"POST","header":[{"key":"x-api-key","value":"da2-g4vjqusfzzcrvlgligh5owb2rq"},{"key":"Content-Type","value":"application/json"}],"body":{"mode":"graphql","graphql":{"query":"mutation ($refreshToken: String!){\n signIn(input: { refreshToken: $refreshToken}){\n token\n refreshToken\n }\n}","variables":"{\n \"refreshToken\": \"<You can use this mutation to change your current user's password.
\nNote: We strongly recommend to use the Parsel web platform to change your password at https://parsel.ai/forgot-password. If you have a US account and wish to change your password with this mutation, please ensure that the new password meets these requirements:
\n- \n
- At least 8 characters \n
- At least one number \n
- At least one uppercase character \n
- At least one lowercase character \n
- At least one special character \n
Generate a presigned URL
\nA Presigned URL acts as a verified location to upload your input documen to. The URL generated by this request is used in the following step to Upload your file, and subsequently referenced when creating your dataset. Each dataset maps to a singular Presigned URL, thus for every document that you wish to process, a new Presigned URL should be generated.
NOTE: The Presigned URL has an expiry date.
Input Parameters
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<Upload File
\nUsing a Presigned URL from the previous step, submit a PUT request to upload your chosen input file to the Presigned URL location.
Please note that a successful request has an empty response body, but has status code 200 OK. As long as the status code is 200 OK, you can be sure that the file was successfully uploaded.
WARNING : The Postman request documentation does not include a default example file, due to file restrictions implemented by Postman. If you're using Postman, please include your own file using Postman's interactive file selector in the binary Body. If using the code examples, please set the binary content according to the programming language and placeholder present.
SignatureDoesNotMatch\n SignatureDoesNotMatch\n Create Dataset
\nOnce a file has been uploaded to a Presigned URL it can then be encapsulated in a dataset and marked to begin processing. This is done by creating a dataset.
What is a dataset?
\nA dataset in the Parsel API refers to the bundled execution around a document. This bundle includes the uploaded file location, associated execution metadata and, if available, the generated outputs.
\nDatasets do not expire, and are stored securely, all data contained within is fully encrypted at rest and in flight.
\nThe dataset metadata contains:
\n- \n
- Dataset unique ID \n
- Execution status denoting the current stage of execution \n
- Execution progress, which is a numeric representation of the progress through the current execution stage \n
- The metadata inputted in the
Graphql Variables\n
All manipulation of uploaded files, outputs and executions are accessed through dataset management, as is described in the Manage Dataset section of the documentation.
By calling the Create Dataset mutation you both create the dataset and start execution upon it automatically.
The various input parameters can be found below.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<This section explains the necessary steps to upload a desired file and begin processing it through dataset creation.
\nRestrictions
\nThese restrictions apply to both files being uploaded and the general execution flow via dataset creation.
\n","_postman_id":"e8a2ba1d-aec4-483a-a617-e5053901fc31","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<You can use this mutation both to retrieve URLs for output files, and to check current Dataset execution status.
\nWhen checking dataset status, it can be one of three possible values:
\n* IN_PROGRESS
* COMPLETED
* ERROR
The output items in the outputList are generated in batches, for each group of 25 pages in the uploaded document that contain at least 1 table (or form field if enabled). So if you have a document with <= 25 pages, you’ll always receive a single batch per output format (.json, .xlsx, etc.). If your document has > 25 pages that contain tables, we will generate multiple batches in increments of 25 pages. The batch of each output item in the outputList is denoted by a -{BATCH NUMBER} suffix to the file name, and a dedicated batchNumber attribute, which is null if batches are not applicable for that output format, like ZIP and PDF.
Available Output Formats
\nAs part of the dataset creation process we create multiple outputs of different formats. The majority of these formats represent the same data in different ways, either to make the set more machine-readable or more human-readable.
However there are some outputs which represent either completely separate data, or a subset of other outputs. Each of the outputs are detailed and explained below.
JSON
\nSuffix: .json
NOTE: The JSON output described here refers to the file marked .json, and is functionally distinct from the output marked .fields.json, which is described further down.
This output represents all of the tabular and field data that was extracted, with metadata at base level and a list of detected tables, and if enabled fields, found within the document. Each table or field has an attribute denoting the page it was discovered on.
\nJSON Structure
\nTable Dictionary
\nThe table dictionary represents a single found table within the document. All indexing attributes are 1-indexed.
\nField Dictionary
\nThe field dictionary represents a single form field identified in the document. Form fields are composites of a label and a value. The form field object contains metadata surrounding the form field at top-level, index on page, page number, etc. Again all indexing attributes are 1-indexed.
The value and label attributes contain, respectively, a string representation of their contents, and their own bounding box denoting their position on the page.
Bounding Box Dictionary
\nThe bounding box dictionary denotes a rectangular area on a page. All of the measurements are taken respective to the left or top of the page.
All measurements are represented as float percentages of either page width or height depending on which direction they represent. To get the pixel values for the bounding box, see the Formula outlined for each attribute.
NOTE: d.p. in the table descriptions refer to decimal places. 0.001 for example is accurate to 3d.p.
\nCSV
\nSuffix: .csv
NOTE: The CSV output described here refers to the file marked .json, and is functionally distinct from the output marked .fields.csv, which is described further down.
This output represents all of the tabular and field data that was extracted. The CSV representation is made up of equal length rows of comma-delimited values. Each of these values align with the header row at the start of the file. This header row is described below, in two parallel key-value pairs (Header - Description) in order; Top to bottom first pair of columns, top to bottom second pair of columns
All values containing commas are appropriately escaped with quotation marks to allow correct csv parsing.
Note: d.p. in the table descriptions refer to decimal places. 0.001 for example is accurate to 3d.p.
\nFields JSON
\nSuffix: .fields.json
The fields JSON represents all form field data extracted from the document. This output represents a subset of data available in the complete JSON output file, but can be useful if you are only interested in the form fields extracted for a particular document
Fields CSV
\nSuffix: .fields.csv
Similar to the fields JSON above, this contains only records of form fields detected in the document. This output represents a subset of data available in the complete JSON output file, but can be useful if you are only interested in the form fields extracted for a particular document
The CSV representation is made up of equal length rows of comma-delimited values. Each of these values align with the header row at the start of the file. This header row is described below, in two parallel key-value pairs (Header - Description) in order; Top to bottom first pair of columns, top to bottom second pair of columns
All values containing commas are appropriately escaped with quotation marks to allow correct CSV parsing.
Note: d.p. in the table descriptions refer to decimal places. 0.001 for example is accurate to 3d.p.
\nRaw Text JSON
\nSuffix: .raw_text.json
The raw text JSON contains all text elements extracted from the document.
Raw Text txt
\nSuffix: .raw_text.txt
Row by row value dump of all extracted text elements from thge document, ordered from top-left to bottom-right, e.g.applepearzucchini
This request returns paginated results, which allows for relay infinite loading pagination.
\nWarning: This operation is not allowed for the docs demo user.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<The processed outputs of your input files are stored on Amazon S3 storage. In order to get the URLs to download these files make use of the Get Dataset or Get Datasets mutations.\nThe results of either mutation will include a link per output file in the response.
Reruns document processing on a given dataset. If dataset processing repeatedly fails there is usually a file or system restriction preventing execution. In such cases either perform processing on a different input file or get in touch with our support team by emailing support@parsel.ai.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<Sets the referenced dataset's name to a new identifier. This is purely for prettification and does not effect the internal identifiers for datasets in any way. As the dataset's ID on creation is immutable, it can be used to explicitly reference a target dataset.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<Deletes the referenced dataset and all associated files, data and metadata permanently. As the dataset's ID on creation is immutable, it can be used to explicitly reference a target dataset.
\nNOTE: A dataset can be deleted at anytime, even during processing.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<Search datasets available to you by keyword.
\nNote: Results are pagineated, so can be loaded with relay infinite loading pagination.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<The following mutations allow for manipulating existing datasets. These will become useful if you have a large set of test documents, with results you wish to save.
\n","_postman_id":"8e6d5ea0-1f90-46d6-b097-62a0b8ad37cc","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<Please see the diagrams and steps to follow to process a document at the Process a Document header above.
\n","auth":{"type":"apikey","apikey":{"key":"Authorization","value":"<