JSON Exchange Validation

JSON specifies a way to encode data objects for web services, particularly when used with JavaScript. JSON does not describe a way to stream a set of data objects or to perform reliable data exchange, as such protocols are external to JSON’s purpose.

This article details the WARES components and methods which implement reliable data exchange with JSON structured document files.

Directory Sharing

Directory sharing services are the Internet’s answer for data transfer between partners. These services replace EDI’s proprietary Value-Added Networks (VANs), dial-up storage systems, FTP sites, and email attachments. DropBox provides a shared file system which, when mounted, appears as a native directory structure on any computer or server using an Apple, Linux, or Microsoft operating system.

When a directory is shared for data exchange, within the directory should be a folder for each document type to be exchanged, and corresponding sub-folders to archive files which have been picked up. To exchange warehouse orders, for example, there would be an ORDERS folder, an ORDERS/HISTORY sub-folder, an ORDERS/STREAMS sub-folder, a SHIPPERS folder, and a SHIPPERS/HISTORY sub-folder, as shown in JSON-integrate-view.

DropBox uses file system extensions to make the exchange fileshare appear as a part of the local file system. This allows WARES to perform file operations directly on the shared files during JSON import, unlike using FTP sites or EDI VANs which require a separate communications program for file exchange.

Transaction Numbering

Traditional EDI uses sequential counters to track documents, groups, and exchanges. The WARES implementation of EDI maintains a perpetual document history, and if a document number of a specific type is received twice, the second receipt is detected as a duplicate by the EDI software. By comparison, JSON does not have streaming protocols to which the concept of sequenced transmission documents would apply.

Note

The EDI numbering model is applied externally to the exchanged documents, so the retransmission of an exchange is detected this way but a duplication due to resubmitting a document, which produces a new sequence number, is not.

Detecting Duplicate Documents

When WARES imports an order, combinations of the internal consignee identifier and the document reference numbers are indexed to look for duplications. When a reference number is repeated the order is marked as a duplicate, and a human operator must resolve this conflict. This methodology detects duplications whether a file is retransmitted or a document is resubmitted. (Order Status 0 is used to indicate orders received with errors, including order duplicates.)

Using JSON, when a warehouse client writes individual order documents into a file share using a reference number as the filename, resubmitting an order will just replace the file – up to the point when the warehouse picks up the files. There is no such thing as a “transmission duplication,” and WARES duplicate detection catches resubmissions that occur once an original order is streamed.

Warning

This process has a consequence: if a JSON order is shipped from the warehouse, and then it is resubmitted by the client, the resubmitted order will overwrite the original in the fileshare history.

Note

EDI transaction numbering is sequential, so that it is possible to detect a missing or faulty document by number and request resubmission automatically. No one does this actually, as automatic resubmission of failed documents would be pointless when a document is faulty to begin with.

Batch Numbering

WARES processes streams, not individual order files. When WARES selects a group of JSON orders for processing, the group is compiled into a stream file and written to a file processing stream history, separate from the exchange share. The stream filename is a datetime stamp of when the files were selected.

Note

Stream files are never overwritten by the client, so resubmitting a document after it has been processed saves the original and its resubmission in separate streams, and no information is lost in stream history.

Document Counts and Lengths

Measuring the reliability of document exchange is an issue.

EDI transmission standards include counts of groups and document counts within groups. Further, the footer of each document has a count of the included data segments. Nothing in JSON or any other structured data presentation corresponds to the EDI concept of data segments. Similarly, character counts which are used in transmission are not a reliable measure of structured textual content in either EDI or JSON, as reformatting can either add or eliminate white space.

So, we are left with counting documents while processing. WARES does this quite well, reporting the number of processed documents at the end of import or export. This should be compared with the number of records written to the WARES Orders file for example, and to the directory of documents which were processed and moved to the ORDERS/HISTORY fileshare. To this end, WARES writes both the stream file and a directory listing of processed document files to the ORDERS/STREAMS folder. The directory listing has file extension ”.ls”, while ”.JSN” is the stream extension.

Document Acknowledgments

EDI specifies the form of Acknowledgments (set 997) which may return counts for documents accepted without errors, accepted with errors, and rejected. Classic EDI is a standalone application, and so documents are checked for syntactical compliance with standards, but semantic implications of documents cannot be reviewed. A document which is accepted without errors may contain entirely useless information, as for instance ordering products which are not a part of inventory, or requesting an expired delivery date.

Because JSON exchange includes importing actual documents, it may be possible to add semantic checks as part of acknowledgment, as is mentioned previously with duplicate detection. However, such semantic checks will likely be limited to comparing values to datatype and formatting specifications, which is just what EDI does.

THE NATURE OF SUCH JSON ACKNOWLEDGMENTS IS YET TO BE DETERMINED.

Stream DateTime Decoding

WARES limits stream filenames to no more than eight characters. An encoding scheme is employed to include date and time information in this name, and to arrange streams in chronological order when sorting (left-aligned). Following is the decoding to determine date and time of a stream from its filename.

Year, Month, Day, and Seconds since midnight are presented in 8 characters: YYMDDTTT.

  • YY Year: Add century digits 20 in front of the year to complete the year.
  • M Month: Letters [A,B,C,D,E,F,G,H,I,J,K,L] represent consecutive months January through December.
  • DD Day: The day of month is displayed in two right-aligned, zero-filled digits.
  • TTT Time: 36 symbols: [0,1,...,9,A,B,...,Z] are used to encode time in base 36. Convert from base 36 to base 10 and then multiply by 3 to get the time in seconds after midnight.

Note

The five-character date portion of stream names will always sort correctly. The chronology of times within a date may be out of sequence if the operating system encodes letters before numbers in character sequence.

Exchange Validation TLDR;

JSON exchange through file sharing bypasses the concept of data transmission, so there is no sequence numbering for transmissions as in EDI. Similarly each document type being exchanged uses a different folder in the fileshare, replacing the EDI concept of document groups. Also, documents are placed into streams by the recipient, not the sender, so group sequence numbering is replaced by process timestamps on stream files. Reference numbers extracted from document data are used as filenames for individual documents, replacing document transmission sequence numbers.

File sharing allows programs to manipulate exchanged datafiles directly, which eliminates unreliable interplays between programming and system functions. Each recipient is responsible for reporting documents received and processed to the sender, so that the sender can verify the success of data exchange. All data are available for review, providing transparency in exchange. The customer can guarantee the confidentiality and security of data by owning the DropBox session used for data exchange.