We-Eat/server/app/com/weEat/controllers/ParserController.scala

package com.weEat.controllers

import com.weEat.shared.models._
import javax.inject.{Inject,Singleton}
import play.api.libs.json._
import play.api.mvc._
import scala.concurrent.Future
import com.weEat.models.Authorization
import scalaoauth2.provider.{AuthInfoRequest,OAuth2ProviderActionBuilders}
import com.weEat.services.OAuth2Service
import net.ruippeixotog.scalascraper.browser.JsoupBrowser
import net.ruippeixotog.scalascraper.dsl.DSL._
import net.ruippeixotog.scalascraper.dsl.DSL.Extract._
//import net.ruippeixotog.scalascraper.dsl.DSL.Parse._
import net.ruippeixotog.scalascraper.model.Element
import net.ruippeixotog.scalascraper.scraper.HtmlExtractor
import scala.util._

@Singleton
class ParserController @Inject()(
  val controllerComponents: ControllerComponents,
  oauth: OAuth2Service,
  usdaController: USDAController,
  foodController: FoodController
) extends BaseController
    with OAuth2ProviderActionBuilders {
  implicit val ec = scala.concurrent.ExecutionContext.global

  private val _browser = JsoupBrowser()

  def parseURL() = AuthorizedAction[Authorization](oauth).async(parse.text)({ implicit request: AuthInfoRequest[String, Authorization] =>
    val url = request.body
    _findParser(url).fold(Future.successful(NotFound(s"No parser available for $url."))) { (parser) =>
      val doc = _browser.get(url)
      val title = doc >> parser.titleExtractor
      val servings = doc >> parser.servingExtractor
      val prepTime = parser.prepTimeExtractor.flatMap(doc >?> _)
      val cookTime = parser.cookTimeExtractor.flatMap(doc >?> _)
      val ingredients = doc >> parser.ingredientExtractor
      val instructions = doc >> parser.instructionExtractor

      Future.sequence(ingredients.map(_parseIngredient _))
        .map((ingredients) => Ok(Json.toJson(RecipeNodeNoId(
          title,
          servings.getOrElse(1.0f),
          1.0f,
          UnitType.NUMBER,
          ingredients.toSeq,
          /* tflucke@[2023-10-26]: Do not pss along the instructions since this
           * could be a violation of the Recipe Author's copyright. */
          Nil, //instructions.toSeq,
          None,
          None,
          Some(url)
        ))))
    }
  })

  private def _findParser(url: String): Option[Parser] = {
    val host = new java.net.URL(url).getAuthority()
    val hostNoWWW = if (host.startsWith("www.")) host.substring("www.".length) else host
    Map(
      ("epicurious.com" -> Parser.epicurious),
      ("mccormick.com" -> Parser.mccormick),
      ("recipetineats.com" -> Parser.recipeTinEats),
      ("mamalovestocook.com" -> Parser.recipeTinEats)
    ).get(hostNoWWW)
  }

  private def _parseIngredient(ingredientLine: String): Future[Ingredient] = {
    val numberPattern = raw"(\d+)[\d-_]*\s(\w+)\s+(.+)".r
    val fractionPattern = raw"(\d+)/(\d+)[\d-_]*\s(\w+)\s+(.+)".r

    ingredientLine match {
      case numberPattern(amount, unit, rest) =>
        _guessFoodFromStr(rest).map(Ingredient(
          _,
          amount.toFloat,
          MeasureUnit.guessUnit(unit).getOrElse(Count)
        ))
      case fractionPattern(numerator, denominator, unit, rest) =>
        _guessFoodFromStr(rest).map(Ingredient(
          _,
          numerator.toFloat/denominator.toFloat,
          MeasureUnit.guessUnit(unit).getOrElse(Count)
        ))
      case noUnitLine => _guessFoodFromStr(noUnitLine).map(Ingredient(_, 1, Count))
    }

  }

  private def _guessFoodFromStr(foodLine: String): Future[Ingredient.IngredientId] = {
    import gov.usda.nal.fdc.models.DataType._
    usdaController.fdc.getFoodsSearch(foodLine.filterNot(_ == '/'), Seq(
      Foundation, Survey, SRLegacy
    ), pageSize = Some(10))().flatMap({ (fdcResult) =>
      Future.sequence(
        fdcResult.foods.map((food) => foodController.getByFdcId(food.fdcId))
      ).map(_.flatten
        .headOption
        .fold[Ingredient.IngredientId](
          Ingredient.USDAId(fdcResult.foods.head.fdcId)
        )((foodNode) => Ingredient.FoodNodeId(foodNode._id))
      ).transform({
        case Success(x) => Success(x)
        case Failure(x) => println(foodLine);Failure(x)
      })
    })
  }
}

case class Parser(
  titleExtractor: HtmlExtractor[Element, String],
  servingExtractor: HtmlExtractor[Element, Option[Float]],
  prepTimeExtractor: Option[HtmlExtractor[Element, String]],
  cookTimeExtractor: Option[HtmlExtractor[Element, String]],
  ingredientExtractor: HtmlExtractor[Element, Iterable[String]],
  instructionExtractor: HtmlExtractor[Element, Iterable[String]],
)

object Parser {
  val mccormick = Parser(
    text("h1"),
    // TODO use extractors
    text(".main-title .count").map(_.toFloatOption),
    Some(text(".prep_time .first_content")),
    cookTimeExtractor = Some(text(".ingredients .first_content")),
    ingredientExtractor = texts(".recipe-about-list li"),
    texts(".instructions-main span.para")
  )
  val epicurious = Parser(
    text("h1"),
    text("""div[data-testid="IngredientList"] > p""")
      .map("Yield: \\D*(\\d+).*".r.findFirstMatchIn(_).map(_.group(1).toFloat)),
    None,
    None,
    texts("""div[data-testid="IngredientList"] > div > div"""),
    texts("""div[data-testid="InstructionsWrapper"] > ol > li > p""")
  )
  val recipeTinEats = Parser(
    text("h2.wprm-recipe-name"),
    text("span.wprm-recipe-servings").map(_.toFloatOption),
    Some(text("span.wprm-recipe-prep_time-minutes")),
    Some(text("span.wprm-recipe-cook_time-minutes")),
    texts("li.wprm-recipe-ingredient")
      .map(_.map(_
        .replaceAll("\u00BD", "1/2")
        .replaceAll("\u00BC", "1/4")
        .replaceAll("\u00BE", "3/4")
        .replaceAll("\u2150", "1/7")
        .replaceAll("\u2151", "1/9")
        .replaceAll("\u2152", "1/10")
        .replaceAll("\u2153", "1/3")
        .replaceAll("\u2154", "2/3")
        .replaceAll("\u2155", "1/5")
        .replaceAll("\u2156", "2/5")
        .replaceAll("\u2157", "3/5")
        .replaceAll("\u2158", "4/5")
        .replaceAll("\u2159", "1/6")
        .replaceAll("\u215A", "5/6")
        .replaceAll("\u215B", "1/8")
        .replaceAll("\u215C", "3/8")
        .replaceAll("\u215D", "5/8")
        .replaceAll("\u215E", "7/8")
        .replaceAll("\u215F", "1/")
        .filter(_ <= 0x7f)
        .trim
      )),
    texts("div.wprm-recipe-instruction-text")
  )
}