parse-comment.js

const nlp = require("compromise");

// Types that are valid (multi words must all be lower case)
const validContributionTypes = [
  "a11y",
  "audio",
  "blog",
  "bug",
  "business",
  "code",
  "content",
  "data",
  "design",
  "doc",
  "eventorganizing",
  "example",
  "financial",
  "fundingfinding",
  "ideas",
  "infra",
  "maintenance",
  "mentoring",
  "platform",
  "plugin",
  "projectManagement",
  "promotion",
  "question",
  "research",
  "review",
  "security",
  "talk",
  "test",
  "tool",
  "translation",
  "tutorial",
  "usertesting",
  "video",
];

// Types that are valid multi words, that we need to re map back to their camelCase parts
const validMultiContributionTypesMapping = {
  eventorganizing: "eventOrganizing",
  fundingfinding: "fundingFinding",
  usertesting: "userTesting",
  projectmanagement: "projectManagement",
};

// Additional terms to match to types (plurals, aliases etc)
const contributionTypeMappings = {
  accessibility: "a11y",
  advertisement: "promotion",
  blogs: "blog",
  blogging: "blog",
  bugs: "bug",
  codes: "code",
  coding: "code",
  dataset: "data",
  datasets: "data",
  designing: "design",
  designs: "design",
  doc: "doc",
  docs: "doc",
  documenting: "doc",
  documentation: "doc",
  examples: "example",
  finance: "financial",
  financials: "financial",
  funds: "fundingfinding",
  idea: "ideas",
  infras: "infra",
  infrastructure: "infra",
  maintaining: "maintenance",
  management: "projectManagement",
  managing: "projectManagement",
  mentor: "mentoring",
  music: "audio",
  platforms: "platform",
  plugins: "plugin",
  project: "projectManagement",
  projectManaging: "projectManagement",
  questions: "question",
  reviews: "review",
  securing: "security",
  sound: "audio",
  talks: "talk",
  tests: "test",
  testing: "test",
  tools: "tool",
  tooling: "tool",
  translator: "translation",
  translating: "translation",
  translations: "translation",
  tutorials: "tutorial",
  videos: "video",
};

// Additional terms to match to types (plurals, aliases etc) that are multi word
const contributionTypeMultiWordMapping = {
  "audio production": "audio",
  "audio recording": "audio",
  "music production": "audio",
  "data collection": "data",
  "data collections": "data",
  "data set": "data",
  "data sets": "data",
  "event organising": "eventOrganizing",
  "event organizing": "eventOrganizing",
  "fund finding": "fundingFinding",
  "funding finding": "fundingFinding",
  "user testing": "userTesting",
  "business development": "business",
  "project management": "projectManagement",
  "social media": "promotion",
};

const Contributions = {};

validContributionTypes.forEach((type) => {
  Contributions[type] = "Contribution";
});

Object.keys(contributionTypeMappings).forEach((type) => {
  Contributions[`${type}`] = "Contribution";
});

const plugin = {
  words: {
    ...Contributions,
    add: "Action",
  },
};

nlp.plugin(plugin);

function findWho(message, action) {
  function findWhoSafe(match) {
    message = message.replace(/\-/g, "#/#"); // workaround (https://github.com/spencermountain/compromise/issues/726)
    const whoNormalizeSettings = {
      whitespace: true, // remove hyphens, newlines, and force one space between words
      case: false, // keep only first-word, and 'entity' titlecasing
      numbers: false, // turn 'seven' to '7'
      punctuation: true, // remove commas, semicolons - but keep sentence-ending punctuation
      unicode: false, // visually romanize/anglicize 'Björk' into 'Bjork'.
      contractions: false, // turn "isn't" to "is not"
      acronyms: false, //remove periods from acronyms, like 'F.B.I.'
      parentheses: false, //remove words inside brackets (like these)
      possessives: false, // turn "Google's tax return" to "Google tax return"
      plurals: false, // turn "batmobiles" into "batmobile"
      verbs: false, // turn all verbs into Infinitive form - "I walked" → "I walk"
      honorifics: false, //turn 'Vice Admiral John Smith' to 'John Smith'
    };
    const matchedSet = nlp(message)
      .match(match)
      .normalize(whoNormalizeSettings)
      .data();

    if (matchedSet.length > 0) {
      matchedText = matchedSet[0].text;
      matchedText = matchedText.replace(/#\/#/g, "-");

      return matchedText;
    }
  }

  const whoMatchedMention = findWhoSafe(`@[.]`);
  // TODO: Unused
  // if (whoMatchedMention) {
  //   return whoMatchedMention;
  // }

  const whoMatchedByAction = findWhoSafe(`${action} [.]`);
  if (whoMatchedByAction) {
    return whoMatchedByAction;
  }

  const whoMatchedByFor = findWhoSafe(`[.] for`);
  if (whoMatchedByFor) {
    return whoMatchedByFor;
  }
}

function parseAddSentence(message, action) {
  message = message
    .split(",")
    .map((e) => e.trim())
    .join(", ");

  const whoMatched = findWho(message, action);
  if (!whoMatched) {
    return {
      who: undefined,
    };
  }

  const who = whoMatched.startsWith("@") ? whoMatched.substr(1) : whoMatched;

  // Contributions
  const doc = nlp(message).toLowerCase();
  // This is to support multi word 'matches' (altho the compromise docs say it supports this *confused*)
  Object.entries(contributionTypeMultiWordMapping).forEach(
    ([multiWordType, singleWordType]) => {
      doc.replace(multiWordType, singleWordType);
    }
  );

  const rawContributions = doc
    .match("#Contribution")
    .data()
    .map((data) => {
      // This removes whitespace, commas etc
      return data.normal;
    });

  const contributions = doc
    .match("#Contribution")
    .data()
    .map((data) => {
      // This removes whitespace, commas etc
      return data.normal;
    })
    .filter((element, index) => {
      return rawContributions.indexOf(element) === index;
    })
    .map((type) => {
      if (contributionTypeMappings[type]) return contributionTypeMappings[type];
      return type;
    })
    .map((type) => {
      // Convert usertesting to userTesting for the node api
      if (validMultiContributionTypesMapping[type])
        return validMultiContributionTypesMapping[type];
      return type;
    });

  return {
    who,
    contributions,
  };
}

function parseAddComment(message, action) {
  const contributors = {};

  const sentences = nlp(message).sentences();
  sentences.forEach(function (sentence) {
    const sentenceRaw = sentence.data()[0].text;
    const { who, contributions } = parseAddSentence(sentenceRaw, action);

    if (who) {
      contributors[who] = contributions;
    }
  });

  return {
    action: "add",
    contributors,
  };
}

function parseComment(message) {
  const doc = nlp(message);

  const action = doc.toLowerCase().match("#Action").normalize().out("string");

  if (action.match("add")) {
    return parseAddComment(message, action);
  }
  
  return {
    action: false,
  };
}

module.exports = parseComment;