- Add nytToCooklang function to Roux.NYTimes
- buildMetadata: maps title, URL, description, time, servings
- buildSections: creates sections for ingredient groups then step groups
- ingredientGroupToSection: converts NYT ingredient items to
Cooklang step items with @ingredient references
- stepGroupToSection: converts method steps into Cooklang steps
- Add parseDuration, parseServings, parseNYTQuantity helpers
- Add extractIngredientName heuristic for core ingredient extraction
- Add 2 conversion tests verifying both example recipes
- All 32 tests passing, hlint clean
- Create src/Data/CookLang.hs with all recipe types (Recipe, Section,
StepItem, Ingredient, Cookware, Timer, RecipeRef, Quantity, Duration,
Metadata, emptyMetadata)
- Data.CookLang has zero Roux imports — can be extracted to its own
package
- Roux.Types now just re-exports from Data.CookLang
- Update all Roux.* modules and tests to import Data.CookLang directly
- All 30 tests passing, hlint clean
- Add Roux.NYTimes module with:
- extractNextData: finds <script id="__NEXT_DATA__"> tag in HTML
and parses its JSON content via aeson
- parseNYTRecipe: maps NYTimes recipe JSON structure into a
typed intermediate representation (NYTRecipe, NYTIngredientGroup,
NYTIngredientItem, NYTStepGroup, NYTStepItem)
- Supports all NYTimes Cooking data: title, URL, total time, yield,
topnote, ingredient groups (with section names), step groups
- Add aeson to dependencies
- Add 4 tests: extraction + parsing for both fried-rice and
carrot-risotto example HTML files
- Wire NYTimesSpec into test runner
Three categories of fixes:
1. Multi-line steps: removed \n from pText exclusion list so newlines
within a step are consumed as regular text
2. Range timers (e.g. ~{14-18%minutes}): handle hyphen-separated
ranges in parseSimpleRational by taking the lower value
3. & in ingredient names: added & to pSingleNameChar and
pMultiNameChar char sets
4. Markdown-style # headings: added pAnyChar fallback so # followed
by invalid cookware name falls through to text consumption
5. Note detection: tightened > check to require > (with space)
or > followed by non->, preventing >> metadata from being
treated as notes
- Add splitAmountUnit to parse quantities with unit attached without
% separator (e.g., @water{1,1/2cups}, @gelatine{3tsp})
- Add isNumericChar helper for detecting numeric quantity characters
- Add compound quantity test (1,1/2 cups = 3/2)
- Add no-separator unit test (3tsp)
- All 26 tests passing, all 4 example files parse correctly
- Fix pRecipeRef to include ./ prefix in stored path
- Add recipe reference test
- Add parse-validation tests for all 4 example .cook files
(EasyPancakes, FriedRice, CoffeeSouffle, OlivierSalad)
- All 24 tests passing, all examples parse without error
- Add pEndComment parser for -- end-of-line comments
- Add pInlineComment parser for [- ... -] block comments
- Update pText to stop before -- and [- sequences using notFollowedBy
- Add tests for both comment types (2 new tests, 19 total)
- Update Roux.Types: add SectionBodyItem (SecStep/SecComment/SecNote),
StepEndComment, remove StepNote (notes are now section-level)
- Rewrite Roux.Parser using Parsec for inline elements, text-splitting
for file-level structure
- Handle full-line -- comments, > notes, = Name = section headers
- Handle = Name (w/ and w/o closing =), == Name == variations
- Handle end-of-line backslash line breaks within steps
- Update tests to match new model + 3 EasyPancakes integration tests
- Add expectTextStep helper at module level
- Create Roux.Parser with parseCookFile (splits paragraphs, treats
each as a plain-text step, no metadata parsing yet)
- Add emptyMetadata to Roux.Types for convenience
- Add test dependencies: hspec, tasty-golden, tasty-hspec
- Add ParserSpec with 4 tests covering single step, multiple steps,
empty input, and whitespace trimming
- Wire ParserSpec into test runner via tasty-hspec
jbrechtel