git.r.bdr.sh - rbdr/prompt/blob - bin/wordnet_to_json.js

   1 #!/usr/bin/env node
   2
   3 import { basename, join } from 'path';
   4 import { mkdir, readFile, writeFile } from 'fs/promises';
   5 import { debuglog } from 'util';
   6
   7 const internals = {
   8   kParseRegex: /[0-9]+-(?<type>[anrv])\s+lemma\s+(?<term>.*)/,
   9   kTermMapping: {
  10     a: 'adjectives',
  11     n: 'nouns',
  12     v: 'verbs',
  13     r: 'adverbs'
  14   },
  15
  16   debug: debuglog('wordnet_to_json'),
  17
  18   // Validates that inputs are present
  19
  20   validateInputs() {
  21     if (process.argv.length < 4) {
  22       internals.printUsage();
  23       throw new Error(`Insufficient arguments, expected 2, found ${process.argv.length - 2}`);
  24     }
  25   },
  26
  27   // Prints the usage of the program
  28
  29   printUsage() {
  30
  31     console.error('Usage:');
  32     console.error(`${basename(process.argv[1])} <path/to/wordnet_file.tab> <path/to/data/dir>`);
  33   },
  34
  35   // Loads the contents of a wordnet tab file.
  36
  37   async load(pathToTab) {
  38
  39     try {
  40       return await readFile(pathToTab, { encoding: 'utf8' });
  41     }
  42     catch (error) {
  43       internals.debug(error.stack);
  44       throw new Error(`Could not read tab file at ${pathToTab}`);
  45     }
  46   },
  47
  48   // Parses a wordner tab file and turns it into a JSON structure
  49
  50   async parse(wordnetTab) {
  51
  52     try {
  53       return wordnetTab.split('\n').reduce((parsedTerms, currentTerm) => {
  54         const matches = currentTerm.match(internals.kParseRegex);
  55         if (matches) {
  56           parsedTerms[internals.kTermMapping[matches.groups.type]].push(matches.groups.term);
  57         }
  58         return parsedTerms;
  59       }, {
  60         adjectives: [],
  61         nouns: [],
  62         verbs: [],
  63         adverbs: []
  64       });
  65     }
  66     catch (error) {
  67       internals.debug(error.stack);
  68       throw new Error('Could not parse wordnet data.');
  69     }
  70   },
  71
  72   async write(parsedTerms, dataDirectory) {
  73
  74     await internals.createDataDirectory(dataDirectory);
  75     try {
  76       for (const [type, terms] of Object.entries(parsedTerms)) {
  77         const targetFile = join(dataDirectory, `${type}.json`);
  78         await writeFile(targetFile, JSON.stringify(terms, null, 2));
  79       }
  80     }
  81     catch (error) {
  82       internals.debug(error.stack);
  83       throw new Error(`Could not write wordnet data to ${dataDirectory}.`);
  84     }
  85   },
  86
  87   async createDataDirectory(dataDirectory) {
  88
  89     try {
  90       await mkdir(dataDirectory, { recursive: true });
  91     }
  92     catch (error) {
  93       internals.debug(error.stack);
  94       throw new Error(`Could not create data directory at ${dataDirectory}`);
  95     }
  96   },
  97
  98   async run() {
  99
 100     internals.validateInputs();
 101     const wordnetTab = await internals.load(process.argv[2]);
 102     const parsedTerms = await internals.parse(wordnetTab);
 103     await internals.write(parsedTerms, process.argv[3]);
 104   }
 105 };
 106
 107 internals.run()
 108   .then(() => process.exit(0))
 109   .catch((err) => {
 110
 111     console.error(err.message || err);
 112     process.exit(1);
 113   });