tediparse 2.0.0

checksums.yaml

@@ -0,0 +1,7 @@
+---
+SHA256:
+  metadata.gz: 8a9edcb495d00d92eb5add48fa6d6706db419e3a78c0f01b96519fb22034c501
+  data.tar.gz: a862a70cdd1bbfb4aeb925cc9203a8794d7c4a3113c1d92527920629d4d710d9
+SHA512:
+  metadata.gz: fc2fd8ef0383d83016259c68757c5c1a7434c4c77f4357566f43c432676e08e5ba9884ee7dc010d415c2d253f140349612a7b8b0d715642ef6be4b6609f9874f
+  data.tar.gz: 305d907ad0b62760ccc75556df795512d46cc4f85d3c897a6fe9a06d4b89a58f76ba6121545af1b2f1676b555c248046a5c597c1ba0fb6ccbbc7f52b9e3a3cdf

data/CHANGELOG.md

@@ -0,0 +1,319 @@
+v 2.0.0 - Aug 27, 2026
+
+  First release of tediparse. The version continues the `stupidedi`
+  numbering (upstream's last release was 1.4.5) and steps the major to
+  mark the breaking removal of X12 grammar content described below.
+
+  Tediparse is now a maintained fork of `stupidedi` (Kyle Putnam, upstream
+  maintained by Isi Robayna). The internal `Stupidedi` module name is
+  preserved for source compatibility — `require "tediparse"` and `require
+  "stupidedi"` both work.
+
+  **Breaking: X12 grammar content removed**
+
+  Tediparse ships the engine, not the grammars. Roughly 856 files of X12
+  IP have been removed from `lib/`: every per-era version namespace
+  (`Versions::TwoThousandOne` through `FiftyTen`), every transaction-set
+  definition (standards and HIPAA implementations under
+  `transaction_sets/<version>/`), every interchange envelope
+  (`interchanges/{00200,00300,00400,00401,00501}/`), and the
+  `editor/` / `contrib/` / `guides/` trees.
+
+  Before relying on tediparse you must register your own grammar against
+  `Stupidedi::Config`. See `spec/support/synthetic/` in the source tree
+  for a worked authoring example.
+
+  * **Recovering the upstream corpus.** The full content tree and the
+    ~146 `.edi` fixtures live at the `pre-x12-removal` git tag. Use
+    `git worktree add ../conformance pre-x12-removal` to drive a private
+    conformance suite against the engine while keeping X12 IP out of
+    the published gem.
+  * `Validation::Ambiguity.build` now requires an `InterchangeDef` as
+    a third positional argument — pass your synthetic envelope (see
+    `spec/support/synthetic/interchange_def.rb`). The previous
+    nil-default would crash inside `mkconfig` when the validator
+    actually ran.
+
+  **Helpful failure surface**
+
+  * Add `Stupidedi::Exceptions::MissingGrammarError` (subclass of
+    `StupidediError`). Const lookups against the removed namespaces —
+    `Stupidedi::Editor`, `Stupidedi::Contrib`, `Stupidedi::Guides`,
+    the per-era `Stupidedi::Versions::FiftyTen` /
+    `Stupidedi::TransactionSets::FortyTen` /
+    `Stupidedi::Interchanges::FiveOhOne` style references, plus the
+    long-deprecated `Versions::FunctionalGroups` /
+    `Versions::Interchanges` aliases — raise it with a message that
+    names the requested constant and points users at the authoring
+    reference. Legitimate typos still surface as `NameError`.
+  * Parser-driven lookups against an empty config now push a
+    `FailureState` whose reason is the same helpful message instead of
+    the generic "unknown … version" string. Applies to all three
+    parser states (`InterchangeState`, `FunctionalGroupState`,
+    `TransactionSetState`).
+  * `Config::InterchangeConfig`, `FunctionalGroupConfig`, and
+    `TransactionSetConfig` gain an `empty?` predicate.
+  * `InterchangeConfig#register`, `FunctionalGroupConfig#register`, and
+    `TransactionSetConfig#register` now raise `ArgumentError` when
+    called with neither a definition nor a block, instead of silently
+    storing `nil` under the key. Catches the `register("DEMO01")` typo
+    that would otherwise produce a confusing `NoMethodError` deep in
+    the parser.
+
+  **Removed**
+
+  * `bin/edi-pp`, `bin/edi-ed`, `bin/edi-obfuscate` — the gem no longer
+    ships executables.
+  * `lib/stupidedi/editor/` and `editor.rb` (the entire TA1 / 999 /
+    277CA acknowledgement subsystem).
+  * `lib/stupidedi/contrib/`, `contrib.rb`, `guides.rb`.
+  * Per-era autoloads from `lib/stupidedi.rb`, `interchanges.rb`,
+    `versions.rb`, `transaction_sets.rb`.
+  * `Config.default` / `Config.hipaa` / `Config.contrib` factory bodies
+    — the methods are preserved as documented no-ops returning a fresh
+    empty `Config`, for source compatibility.
+  * `Stupidedi::Config::EditorConfig` and the `Config#editor` reader —
+    the editor registry only had callers inside the now-deleted
+    `lib/stupidedi/editor/` tree.
+  * Top-level `notes/` scratch directory — historical maintainer
+    examples (X12 generator, recovery script, JSON writer demo) that
+    all referenced the deleted X12 grammar tree. Still recoverable
+    from the `pre-x12-removal` tag.
+  * Top-level `examples/` directory (`generate.rb`, `tokenizer.rb`) —
+    both scripts were hard-wired to `Config.hipaa` and crashed at
+    runtime against the now-empty config. `spec/support/synthetic/` is
+    the canonical worked example; the old scripts are still recoverable
+    from the `pre-x12-removal` tag.
+
+  **Added**
+
+  * `spec/support/synthetic/` — a small non-X12 grammar harness
+    (interchange + functional group + transaction set + adversarial
+    variants) used to exercise the engine in tests. Doubles as the
+    canonical grammar-authoring reference.
+
+  * Grammar generation supports release `003060` (module `ThirtySixty`,
+    interchange version `00306`). Its ASC X12 Table Data is pure ASCII,
+    and is declared as Windows-1252 in
+    `FlatFileReader::SOURCE_ENCODINGS` alongside the other pre-008010
+    releases rather than being left to the UTF-8 default.
+
+  **Bug fixes**
+
+  * `Schema::Generation::FlatFileReader` now rejects two shapes of
+    defective SETDETL row instead of carrying them into the generated
+    grammar. Both are column shifts in the source distribution — a value
+    landing one field left of where it belongs — which the neighbouring
+    release carries correctly:
+
+    - A blank or zero loop repeat, or a blank maximum use, produced
+      `RepeatCount.bounded(0)`. That is not a repeat count the engine can
+      build, so the emitted grammar raised when the consumer loaded it,
+      naming neither the release nor the transaction set nor the row.
+    - `"0"` in the loop-id column (whatever the repeat says) opened a
+      loop identified as `"0"`. Nothing objects to that: it builds, loads
+      and parses, quietly reparenting every row that follows it — in one
+      real case the rest of the heading table, which then could not be
+      parsed past that point.
+
+    The reader now raises at generation time, naming the file,
+    transaction set, area, sequence and segment, so the row can be
+    corrected in the table data.
+
+  * `Schema::RepeatCount.bounded` raised
+    `Exception::InvalidSchemaError`, which resolves to `::Exception` and
+    has no such constant, so a non-positive count surfaced as
+    `NameError: uninitialized constant Exception::InvalidSchemaError`
+    rather than the intended `InvalidSchemaError, "n must be positive"`.
+
+  * `Schema::Generation::FlatFileReader` declared the wrong encoding for
+    every supported release. It read the ASC X12 Table Data as ISO-8859-1;
+    004060 through 007010 are Windows-1252 (so CP1252 smart punctuation at
+    0x80-0x9F decoded to C1 control characters) and 008010 is UTF-8 (so
+    every multi-byte character was double-encoded). Both transcoded
+    silently, and the damage reached the generated grammar. The encoding is
+    now declared per release in `FlatFileReader::SOURCE_ENCODINGS`; an
+    undeclared release is read as UTF-8, deliberately, so a wrong assumption
+    raises instead of writing mojibake. Decoded text has CP1252 smart
+    quotes and dashes normalized to ASCII (accented letters are left
+    alone), and a C1 control character surviving the decode raises with the
+    file, line and codepoint. **Consumers should regenerate their grammar
+    tree**: affected element and segment names change.
+
+  **Bug fixes** (carried over from prior fork work)
+
+  * Fix `too much non-determinism` error when a `LoopDef` contains repeated
+    structurally-identical `SegmentUse` slots (same id, no qualifier element)
+    and only the earlier slot is filled — e.g. an N3 in the partner-customised
+    4010 PO850 N1 loop. `ConstraintTable::ValueBased` now consults the active
+    parser state's parse tree to pick the slot whose preceding sibling has
+    actually been consumed: the earliest position past the highest-consumed
+    sibling in the matching parent loop. This requires plumbing the active
+    `AbstractState` through `InstructionTable#matches` and the `ConstraintTable`
+    subclasses' `matches`. Behaviour is unchanged for qualifier-disambiguated
+    paths (e.g. N1 by N101) and for genuinely ambiguous cases across different
+    parent loops. During `find` navigation the `mode == :insert` gate skips
+    disambiguation entirely, so the full candidate set is preserved.
+  * Build out 5010 SO317 as a test case for interleaved segments and child
+    loops within a single `LoopDef`.
+
+v 1.4.1
+
+  **Bug Fixes**
+
+  * Fix missing method delegations in SimpleElementUse, ComponentElementUse, and CompositeElementUse #185
+  * Fix regression in edi-obfuscate
+  * Fix copy-pasted StringVal -> IdentifierVal #187
+  * Fix crash in 4010 editor, wrong method called #188
+  * Fix bug in RepeatedElementVal#==, incorrect comparison #190
+  * Fix error message when required composite element is missing #194
+  * Fix various typos in comments and descriptor strings #230 and #233
+
+  **Added**
+
+  * Add DSL for defining X12 grammars TransactionSets::Builder::Dsl #200
+  * Add support for 5010 X12-HN277 Health Care Information Status Notification
+  * Add support for 5010 X220A1-BE834 BGN05 Time Code #205
+  * Add support for "02 - Birth" maintenance reason code #209
+
+v 1.4.0
+
+  **Bug fixes**
+
+  * Fix ambiguous grammars (mostly due to incorrect parentheses)
+  * Fix errors in `Standards::FortyTen::HC837`
+  * Fix errors in `Standards::FiftyTen::BE834`
+  * Fix errors in `Standards::FiftyTen::HB271`
+  * Fix errors in `Standards::FiftyTen::RA820`
+  * Fix parsing invalid numeric data in Ruby 2.4+. Previously `"10B"` would be read as `10.0`, and `"AB10"` would be read as `0.0` due to using bigdecimal/util's implementation of `String#to_d`
+  * Fix `TimeVal` issue in all versions (not only `005010`)
+
+  **Added**
+
+  * Add `Stupidedi::Parser.build` as a shortcut for `Stupidedi::Parser::StateMachine.build`
+  * Add many stub definitions of segments, just the segment name and no elements, which are referred to by `RA820` and others.
+  * `edi-pp` can now print different formats with `--format html`, `--format x12`, and `--format tree` (default)
+
+  **Deprecation notices**
+
+  * Remove support for Ruby < 2.0
+  * Remove workarounds for broken JRuby refinements
+  * Remove `Symbol#call` and `Symbol#to_proc` refinements
+
+  **Renamed**
+
+  * `Stupidedi::Builder` is renamed to `Stupidedi::Parser`
+  * `Stupidedi::Guides::*::GuideBuilder` is renamed to `Stupidedi::TransactionSets::Builder`
+  * `Stupidedi::Versions::Interchanges` is renamed to `Stupidedi::Interchanges`
+  * `Stupidedi::Versions::FunctionalGroups` is renamed to `Stupidedi::Versions`
+    * Lots of common code among versions has been factored into `Stupidedi::Versions::Common`
+  * Rename Guides `HC837P` and `HC837I` to `HC837`
+  * Moved all grammars, including `Guides` and `Contrib`, to `Stupidedi::TransactionSets`
+    * Each version now has `::Standards` and `::Implementations`
+  * `Stupidedi::Schema::Auditor` is renamed to `Stupidedi::TransactionSets::Validation::Ambiguity`
+
+  Most of these renames are not breaking changes (yet), but using the old name will print a warning:
+
+  ```
+  Stupidedi::Contrib is deprecated, use Stupidedi::TransactionSets
+  Stupidedi::Guides is deprecated, use Stupidedi::TransactionSets::*::Implementations
+  Stupidedi::TransactionSets::FiftyTen::Implementations::X222::HC837P is deprecated, use HC837 instead
+  Stupidedi::TransactionSets::FiftyTen::Implementations::X222A1::HC837P is deprecated, use HC837 instead
+  Stupidedi::TransactionSets::FiftyTen::Implementations::X223::HC837I is deprecated, use HC837 instead
+  Stupidedi::Versions::Interchanges is deprecated, use Stupidedi::Interchanges instead
+  Stupidedi::Versions::FunctionalGroups is deprecated, use Stupidedi::TransactionSets::*::Standards instead
+  ```
+
+  **Specs**
+
+  * Grammar specs automatically created when a fixture is added to spec/fixtures/<version>/<name>/pass/*.x12
+  * Remove support for `rcov`. Use only `simplecov` now
+  * Update all specs to use `expect(value).to matcher` syntax, instead of `value.should matcher`
+  * New specs to ensure element names match their `id` (eg, `E123.id == :E123`)
+  * New specs to ensure segment names match their `id` (eg `ST.id == :ST`)
+  * New specs to ensure `Config.hipaa`, `Config.contrib`, and `Config.default` reference valid definitions
+  * New specs for `Stupidedi::TransactionSets::Validation::Ambiguity`
+  * Fix fixture files that used `\n` as a segment terminator but didn't have one after `IEA`
+
+  **Miscellaneous**
+
+  * Create new examples in `examples/` that demonstrate undocumented `IdentifierStack`, and more
+  * Made whitespace and other formatting more consistent
+  * `Stupidedi::TransactionSets::Builder.build` no longer requires a `TransactionSetDef` argument
+  * Fix Travis CI to build older versions of Ruby < 2.3
+  * Ignore large definition files in Code Climate
+
+v 1.3.24
+  - Fix repeatability test in Navigation#iterate
+  - Adds implementation of June 2014 005010X223A3 (837I)
+  - Fixes misplaced 2330H and 2330I loops. Fixes names for 2310 Occurrenc…
+
+v 1.3.23 - Jan 10, 2019
+  - Fix decimal values for TimeVal being coerced incorrectly https://github.com/irobayna/stupidedi/pull/151
+  - Detect ambiguous grammar automatically https://github.com/irobayna/stupidedi/pull/153
+
+v 1.3.22
+  - Re-enable EC Segment on FifyTen group
+
+v 1.3.21 - Dec 9, 2018 ** Breaking Changes **
+  - Throw exception from #iterate if segment is not repeatable (fix #126) https://github.com/irobayna/stupidedi/pull/146
+  - Configurable limit of non-determinism (fixes #129) https://github.com/irobayna/stupidedi/pull/145
+  - Add utility to obfuscate data in X12 files https://github.com/irobayna/stupidedi/pull/147
+  - Fix potential frozen string issues in pretty_print methods https://github.com/irobayna/stupidedi/pull/148
+  - Remove erroneous code mappings from element definitions
+
+v 1.2.20 - Oct 23, 2018
+  - Json Writer functionality - Traverse stupidedit internal tree to a ruby hash
+  - Ruby 2.5.3 support
+
+v 1.2.19 - Oct 8, 2018
+  - EDI 276 support - Health Care Claim Status Inquiry
+    X212-HR276
+
+v 1.2.18 -
+  - SH856 rework
+  - PR855 support (v4010)
+  - This change fixes this issue, as if decimal is an empty string, it will be changed to 0 before to_d is called on it
+    https://github.com/irobayna/stupidedi/pull/132
+    Update  lib/stupidedi/versions/functional_groups/004010/element_types/time_val.rb
+
+
+v 1.2.17 - Aug 4, 2018
+  - EDI 270 / 271 support - Health Care Eligibility Benefit Inquiry and Response
+    X279-HS270
+    X279-HB271
+    X279A1-HS270
+    X279A1-HB271
+
+v 1.2.16 - May 27, 2018
+  - Fix item #127 (https://github.com/irobayna/stupidedi/issues/127)
+  - Ruby 2.5.1 support
+
+v 1.2.15 - Dec 20, 2017
+  - Gemfile Updates (fix security vulnerability CVE-2017-17042)
+  - Add Ruby 2.5.0-preview1 support
+  - Add 2.4.3 & 2.5.0-preview1 ruby versions to Travis CI
+
+v 1.2.14 - June 19, 2017
+  - Gemfile Updates
+  - use BigDecimal string refinement only on ruby versions < 2.4
+  - remove rake from gemspec
+  - Add Ruby 2.4.x support
+
+v 1.2.12 - July 29, 2016
+  - Fix a few issues with immutable strings
+
+v 1.2.2 - July 8, 2014
+  - Remove definition of module Enumerable::blank? and present?
+  - Add blank?, present? to Array class
+
+v 1.2.1 - July 7, 2014
+  - Don't redefine 'blank?' for Enumerable, if an implementation already (rails support)
+
+v 1.2.0 - July 7, 2014
+  - Don't redefine try, if active_support provides an implementation already (better rails support)
+
+v 1.1.0 - June 24, 2014
+  - Rspec 3.x support
+  - Drop Ruby 1.8.7 support
+  - Add Ruby 2.1.2 support

data/LICENSE

@@ -0,0 +1,26 @@
+Copyright (c) 2013, Kyle Putnam
+Copyright (c) 2025, Adrian Duyzer / Tediware
+All rights reserved.
+
+Redistribution and use in source and binary forms, with or without
+modification, are permitted provided that the following conditions are met:
+
+ * Redistributions of source code must retain the above copyright notice, this
+   list of conditions and the following disclaimer.
+ * Redistributions in binary form must reproduce the above copyright notice,
+   this list of conditions and the following disclaimer in the documentation
+   and/or other materials provided with the distribution.
+ * Neither the name of stupidedi nor the names of its contributors
+   may be used to endorse or promote products derived from this software
+   without specific prior written permission.
+
+THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
+AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE
+DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE
+FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
+SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER
+CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
+OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
+OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.

data/README.md

@@ -0,0 +1,308 @@
+# Tediparse
+
+- [GitHub project](https://github.com/Tediware/tediparse)
+- [Human Documentation](doc/README.md)
+
+Tediparse is a library for **parsing, generating, and validating** ASC X12 EDI
+documents. You bring the X12 transaction-set grammar; the library walks it.
+Very roughly, it's jQuery for EDI — once you've supplied the schema.
+
+For those unfamiliar with ASC X12 EDI, it is a data format used to encode
+common business documents like purchase orders, delivery notices, and health
+care claims. It is similar to XML in some ways, but precedes it by about 15
+years; so if you think XML sucks, you will love to hate EDI.
+
+## Scope and licensing
+
+**Tediparse ships the engine, not the grammars.** The library does not bundle
+any X12 transaction-set definitions, code lists, envelopes, or ack/editor
+content — that material is X12 IP and is not ours to redistribute. To parse
+or generate real X12 documents you must supply (or license) your own grammar
+and register it against `Stupidedi::Config` before calling the parser.
+
+If you reach for one of the per-era namespaces that the upstream `stupidedi`
+gem shipped — `Stupidedi::Versions::FiftyTen`, `Stupidedi::TransactionSets::FortyTen`,
+`Stupidedi::Interchanges::FiveOhOne`, and so on — tediparse raises a typed
+`Stupidedi::Exceptions::MissingGrammarError` with guidance, rather than
+`NameError`. Parser-driven lookups against an empty config produce a
+`FailureState` whose reason carries the same message.
+
+**Engine scope.** Tediparse provides parsing and generation as a library. The
+upstream gem's editor / acknowledgement subsystem (TA1 / 999 / 277CA) and
+the `bin/edi-pp` / `edi-ed` / `edi-obfuscate` command-line tools are not
+part of tediparse.
+
+### Authoring a grammar
+
+The canonical reference for how to wire up your own grammar against the
+engine lives in the spec harness:
+
+- `spec/support/synthetic/demo.rb` — a small adversarial transaction set
+  built with `Stupidedi::TransactionSets::Builder`. Exercises composites,
+  code lists, interleaved segments and child loops, qualifier-discriminated
+  sibling loops, and the `P` syntax note.
+- `spec/support/synthetic/interchange_def.rb` — minimal ISA envelope.
+- `spec/support/synthetic/functional_group_def.rb` — minimal GS envelope.
+- `spec/support/synthetic/config.rb` — wires the three pieces into a usable
+  `Stupidedi::Config`.
+
+Mirror that shape in your own application and register against
+`Stupidedi::Config.new` (the legacy `Config.default` / `hipaa` / `contrib`
+factories are preserved for source compatibility but now return empty
+configs).
+
+### Generating a grammar from ASC X12 Table Data
+
+Hand-authoring is fine for a handful of segments, but a full transaction set
+is hundreds of definitions. With an **ASC X12 license** you can download the
+**Table Data** (the official `.TXT` distribution — `ELEHEAD`, `SEGDETL`,
+`SETDETL`, `FREEFORM`, etc.) from <https://ecommerce.x12.org/downloads>, and
+tediparse can generate the grammar definition files for you. The full
+walkthrough — input file formats, the output tree, loading, multi-release
+trees, and how it works — is in
+[doc/Generating-Grammars.md](doc/Generating-Grammars.md); the short version:
+
+```sh
+tediparse generate --release 005010 \
+  --table-data vendor/x12/table_data/005010 \
+  --out lib
+```
+
+or from Ruby:
+
+```ruby
+Stupidedi::Schema::Generation.run(
+  table_data: "vendor/x12/table_data/005010",
+  release:    "005010",
+  out:        "lib",
+  namespace:  "Edi" # root module for the emitted code (default: "Edi")
+)
+```
+
+This reads the flat files and writes a grammar tree under `<out>/<namespace>/`
+— per-version support modules, `element_defs.rb`, `segment_defs.rb`,
+`functional_group_def.rb`, one file per transaction set under `standards/`, an
+interchange envelope under `interchanges/`, and a `stupidedi_registration.rb`
+that wires everything onto a `Config`. Use `--dry-run` to preview without
+writing. Supported releases: `003060`, `004010`, `004060`, `005010`, `006010`,
+`007010`, `008010`.
+
+You can keep several releases in one output tree. Generating an additional
+release preserves the ones already present: `stupidedi_registration.rb` (and the
+master loader, if any) are whole-tree artifacts that are rebuilt to cover every
+release in `out`, not just the one you just generated. To rebuild those
+aggregation files on their own — e.g. after adding or removing a release by other
+means — run `tediparse register --out DIR` (or `Stupidedi::Schema::Generation.register(out:)`).
+
+By default the generated tree is loaded by your application's autoloader (e.g.
+Rails/Zeitwerk). If you are not using an autoloader, pass `--master-loader`
+(or `master_loader: true`) to also emit a single entry file (`<namespace>.rb`,
+e.g. `edi.rb`) that `require`s the whole tree in dependency order, so
+`require "edi"` is all you need.
+
+The X12 Table Data you supply is licensed X12 IP, and the generated grammar is
+a derivative you own — **neither ships with this gem.** The generator is the
+machine; you bring the material. The committed fixture under
+`spec/support/generation/table_data/` is a hand-written synthetic grammar in
+the same flat-file format, not real X12 content.
+
+### Conformance suite
+
+If you have private grammars and need to keep them tested against tediparse,
+the upstream X12 fixture corpus and grammar tree remain recoverable from the
+pre-removal commit:
+
+    git fetch --tags
+    git worktree add ../conformance pre-x12-removal
+
+Build the conformance suite against that worktree's fixtures, layer your
+licensed grammar on top, and run it out-of-tree.
+
+## Attribution
+
+This product includes software from `stupidedi` by Kyle Putnam, available at
+<https://github.com/kputnam/stupidedi>.
+
+### Fork relationship
+
+Tediparse is a fork of [stupidedi](https://github.com/kputnam/stupidedi),
+maintained by [Adrian Duyzer](https://github.com/adriand) at
+[Tediware](https://github.com/Tediware). The fork was created to accelerate
+development and support a broader set of X12 documents. The internal Ruby
+module name remains `Stupidedi` for backward compatibility; `require "tediparse"` and `require "stupidedi"` both work.
+
+### Credits
+
+- **Original author**: [Kyle Putnam](https://github.com/kputnam)
+- **Tediparse maintainer**: [Adrian Duyzer](https://github.com/adriand)
+
+## What problem does it solve?
+
+Transaction set specifications can be enormous, boring, and vague. Trading
+partners can demand strict adherence (often to their own unique
+interpretation of the specification) of the documents you generate. However,
+documents they generate themselves are often non-standard and require
+flexibility to parse them.
+
+Tediparse enables you to encode these transaction set specifications directly
+in Ruby. From these specifications, it will generate a parser to read
+incoming messages and a DSL to generate outgoing messages. This approach has
+a huge advantage over writing a parser from scratch, which can be error-prone
+and difficult to change.
+
+### Robust tokenization and parsing
+
+Delimiters, line breaks, and out-of-band data between interchanges are
+handled correctly. While many trading partners follow common conventions, it
+only takes one unexpected deviation, like swapping the ":" and "~"
+delimiters, to render a hand-written parser broken.
+
+Tediparse handles many edge cases that can only be anticipated by reading
+carefully between the lines of the X12 documentation.
+
+### Instant feedback on error conditions
+
+When generating EDI documents, validation is performed incrementally on each
+segment. This means the instant your client code violates the specification,
+an exception is thrown with a meaningful stack trace. Other libraries only
+perform validation after the entire document has been generated, while some
+don't perform validation at all.
+
+### Encourages readable client code
+
+Unlike other libraries, generating documents doesn't involve naming obscure
+identifiers from the specification (like C001, DE522 or LOOP2000), for
+elements of the grammar that don't actually appear in the output.
+
+Like HAML or Builder::XmlMarkup, the DSL for generating documents closely
+matches terminology from the problem domain. You can see in the example
+below that code looks very similar to an EDI document.
+
+### Efficient parsing and traversing
+
+The parser is designed using immutable data structures, making it thread-safe
+for runtimes that can utilize multiple cores. In some cases, immutability
+places higher demand on garbage collection; this has been somewhat mitigated
+with careful optimization.
+
+## Examples
+
+The examples below assume `config` is a `Stupidedi::Config` you've populated
+with your own grammar — see `spec/support/synthetic/config.rb` for the
+authoring pattern.
+
+### Generating, Writing
+
+#### X12 Writer
+
+```ruby
+require "tediparse"
+
+# You bring the grammar. See spec/support/synthetic for the authoring shape.
+config = MyApp::EDI.config
+
+b = Stupidedi::Parser::BuilderDsl.build(config)
+
+# These methods perform error checking: number of elements, element types, min/max
+# length requirements, conditionally required elements, valid segments, number of
+# segment occurrences, number of loop occurrences, etc.
+b.ISA "00", nil, "00", nil,
+      "ZZ", "SUBMITTER ID",
+      "ZZ", "RECEIVER ID",
+      "990531", "1230", nil, "00501", "123456789", "1", "T", nil
+
+# The API tracks the current position in the specification (e.g., the current loop,
+# table, etc) to ensure well-formedness as each segment is generated.
+b.GS "HC", "SENDER ID", "RECEIVER ID", "19990531", "1230", "1", "X", "005010X222"
+
+b.ST "837", "1234", b.default
+  b.BHT "0019", "00", "X"*30, "19990531", Time.now.utc, "CH"
+  b.NM1 b.default, "1", "PREMIER BILLING SERVICE", nil, nil, nil, nil, "46", "12EEER000TY"
+  # ...
+
+b.machine.zipper.tap do |z|
+  separators =
+    Stupidedi::Reader::Separators.build :segment    => "~\n",
+                                        :element    => "*",
+                                        :component  => ":",
+                                        :repetition => "^"
+
+  w = Stupidedi::Writer::Default.new(z.root, separators)
+  print w.write()
+end
+```
+
+#### HTML writer
+
+`Stupidedi::Writer::Default` outputs plain X12; `Stupidedi::Writer::Claredi`
+outputs a formatted HTML string.
+
+```ruby
+b.machine.zipper.tap do |z|
+  w = Stupidedi::Writer::Claredi.new(z.root)
+  File.open('output.html', 'w') { |f| f.write w.write }
+end
+```
+
+### Reading, Traversing
+
+```ruby
+require "tediparse"
+
+config = MyApp::EDI.config
+parser = Stupidedi::Parser.build(config)
+
+input  = File.open("path/to/your.edi", :encoding => "ISO-8859-1")
+
+# Reader.build accepts IO (File), String, and DelegateInput
+parser, result = parser.read(Stupidedi::Reader.build(input))
+
+# Report fatal tokenizer failures
+if result.fatal?
+  result.explain{|reason| raise reason + " at #{result.position.inspect}" }
+end
+
+# Helper function: fetch an element from the current segment
+def el(m, *ns, &block)
+  if Stupidedi::Either === m
+    m.tap{|m| el(m, *ns, &block) }
+  else
+    yield(*ns.map{|n| m.elementn(n).map(&:value).fetch })
+  end
+end
+
+parser.first
+  .flatmap{|m| m.find(:GS) }
+  .flatmap{|m| m.find(:ST) }
+  .tap do |m|
+    el(m.find(:N1, "PR"), 2){|e| puts "Payer: #{e}" }
+    el(m.find(:N1, "PE"), 2){|e| puts "Payee: #{e}" }
+  end
+  .flatmap{|m| m.find(:LX) }
+  .flatmap{|m| m.find(:CLP) }
+  .flatmap{|m| m.find(:NM1, "QC") }
+  .tap{|m| el(m, 3, 4){|l,f| puts "Patient: #{l}, #{f}" }}
+```
+
+### Testing
+
+```
+bundle exec rake spec
+```
+
+## What doesn't it solve?
+
+It isn't a translator. It doesn't have bells and whistles, like the
+commercial EDI translators have, so it...
+
+- Doesn't convert to/from XML, CSV, etc
+- Doesn't transmit or receive files
+- Doesn't do encryption
+- Doesn't connect to your database
+- Doesn't queue messages for delivery or receipt
+- Doesn't generate acknowledgements
+- Doesn't have a graphical interface
+
+These features are orthogonal to the problem tediparse aims to solve, but
+they can certainly be implemented with other code taking advantage of
+tediparse.