Data type for internal representation of code128 encoded values #1

glfmn · 2017-06-22T17:18:20Z

Implement Code128 barcode representation as underlying type for our visual encoding system, allowing us to fall back to regular bar-code scanning if we have issues with our custom format.

Rationale

Code128, as described by Wikipedia, is a high-density barcode data encoding format, which uses patterns to represent symbols in 3 different symbologies which cover different ranges of the ASCII standard.

The Code128 type is essentially intended to encapsulate conversion between many different representations of Code128 barcodes, especially between bit patterns (such as {0,0,0,1,1,1,0,0,1}) or associated weight patterns (such as {3,3,2,1}) and strings.

ToDo

Checksum validation
Conversion from Code128 to string
Conversion from string to Code128
Conversion from bit pattern to Code128
Conversion from weight pattern to Code128
Robust error handling using Result<T,E> specialization
Stable external API for Code128 module

Working checksum digit calculation and validation with basic tests covering hard-coded cases, pending a more complete solution.

Change module-level documentation to block-comment style for easier line reading, and change `From<u8>` to use stop code as wild-card conversion to catch all unxpectd values. Add variant documentation for `Symbology` enum, and fix values; stop codes had associated start code values shifted by one.

Better seperation between tests and base module, allows easier definition of macros or functions used only in the tests.

Simply panics if it recieves an unexpected value, and does not implement shift-codes. Uses a rudimentary finite state machine to model the change in behavior for each symbology, pushing characters symbol-per-symbol to build the encoded string.

Preferable to use iterators and expression-based for loop to avoid carying arround an extra mutable value and manually manipulating indecies.

`to_string` implies an implementation of `Display` or `ToString` for the type, so to disambiguate the method is now called `decode` I am considering an `Into` implementation, but since the conversion may fail if the checksum digit fails, `TryInto` would be a better option. However, since `TryInto` is only in Nightly, that decision is pushed down the road.

Improves ability for error handling, as the possible values are now constrained more strictly for each symbol, making impossible states impossible to represent. Deriving PartialOrd and Ord make it easy to keep the exact same behaviour as before, replacing only the u8 literals with Pattern literals.

Rather than just matching on possible symbologies, also match on shift codes to perform shift-code specific behaviour.

glfmn · 2017-06-30T02:20:07Z

Consider using traits to encapsulate encode and decode behaviour, so that we can support both standard UFT-8 strings, C-Strings, and perhaps even other types like Vec<u8> or bitvector which might encode barcode data.

A sufficiently general conversion API should be able to convert directly from any one supported encoding to another.

Allows for module to be public for integration and benchmark testing while making it clear that the contained modules are not meant for normal module users.

May end up making things take a long time, consider removing in the future if it causes problems.

Generic over formats, allowing support for decoding a variety of barcode types to a variety of target types, and encoding to formats from a variety of types. The ultimate purpose is to allow conversions to rust strings and c strings in the future, and to support the direct conversion of arbitrary encoding types directly to a supported Decode type (for a given format).

Use `Encoding` trait to represent different encoding types, starting with `u8` and `Pattern`. Use u8 in doc tests where appropriate.

Benchmark testing allows me to check for performance regressions and make performance improvements over time. Essential for having real time capability.

Better description of what the module actually does, as it mainly encapsulates all of the barcode logic.

Since I was using u8 to perform the actual string conversions anyway, simply use u8 directly and remove the repr() method from Encoding trait. Saves about 100 ns/iter on benchmarks.

Gwen Lofman added 5 commits June 22, 2017 12:34

Implement preliminary Code128 type

797c7f7

Working checksum digit calculation and validation with basic tests covering hard-coded cases, pending a more complete solution.

Move Code128 tests to test module

5d8bc16

Better seperation between tests and base module, allows easier definition of macros or functions used only in the tests.

Implement basic conversion to string

956c2a5

Simply panics if it recieves an unexpected value, and does not implement shift-codes. Uses a rudimentary finite state machine to model the change in behavior for each symbology, pushing characters symbol-per-symbol to build the encoded string.

Convert while loop anti-pattern to for loop

db9f33c

Preferable to use iterators and expression-based for loop to avoid carying arround an extra mutable value and manually manipulating indecies.

glfmn added the enhancement label Jun 22, 2017

glfmn self-assigned this Jun 22, 2017

glfmn changed the title ~~Data code128 barcode~~ Data type for internal representation and validation of code128 barcode Jun 22, 2017

glfmn changed the title ~~Data type for internal representation and validation of code128 barcode~~ Data type for internal representation of code128 encoded data Jun 22, 2017

glfmn changed the title ~~Data type for internal representation of code128 encoded data~~ Data type for internal representation of code128 encoded values Jun 22, 2017

Gwen Lofman added 4 commits June 23, 2017 12:05

Add deny missing docs attribute and fix docs

b2f690e

Add shift-code handling with state machine enum

8a1dc91

Rather than just matching on possible symbologies, also match on shift codes to perform shift-code specific behaviour.

Gwen Lofman added 2 commits July 9, 2017 02:31

Create internals module to contain code128 logic

f6d6781

Allows for module to be public for integration and benchmark testing while making it clear that the contained modules are not meant for normal module users.

Add bench command to commit testing

6bb3bb9

May end up making things take a long time, consider removing in the future if it causes problems.

glfmn added this to the 0.1 milestone Jul 20, 2017

Gwen Lofman added 5 commits July 22, 2017 00:49

Use traits to have encoding independent decode

89076b9

Use `Encoding` trait to represent different encoding types, starting with `u8` and `Pattern`. Use u8 in doc tests where appropriate.

Add rudimentary benchmark tests

68703e2

Benchmark testing allows me to check for performance regressions and make performance improvements over time. Essential for having real time capability.

Rename internals module to barcode

299e888

Better description of what the module actually does, as it mainly encapsulates all of the barcode logic.

Remove uneccesary copies by using u8

1c5a760

Since I was using u8 to perform the actual string conversions anyway, simply use u8 directly and remove the repr() method from Encoding trait. Saves about 100 ns/iter on benchmarks.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Data type for internal representation of code128 encoded values #1

Data type for internal representation of code128 encoded values #1

glfmn commented Jun 22, 2017 •

edited

Loading

glfmn commented Jun 30, 2017

Data type for internal representation of code128 encoded values #1

Are you sure you want to change the base?

Data type for internal representation of code128 encoded values #1

Conversation

glfmn commented Jun 22, 2017 • edited Loading

Rationale

ToDo

glfmn commented Jun 30, 2017

glfmn commented Jun 22, 2017 •

edited

Loading