- Settings file introduced

- Main module separated from everything else
- "show correct paragraphs" option removed
- custom dictionary removed

bin/cmd.js

@@ -0,0 +1,70 @@
+#!/usr/bin/env node
+
+//Helpers
+var fs = require('fs');
+var request = require('request');
+var program = require('commander');
+var mime = require('mime');
+var marked = require('marked');
+var Promise = require('promise');
+var config = require('../settings.json');
+var Proofreader = require('../lib/proofreader.js');
+
+program
+  .option('-u, --url [url]', 'URL to website that should be proofread.')
+  .option('-f, --file [path]', 'Path to HTML file that should be proofread.')
+  .option('-l, --file-list [path]', 'Path to a list of files that should be proofread.')
+  .option('-c, --config-file', 'Path to a configuration file')
+  .parse(process.argv);
+
+if(program.configFile) {
+  config = require(program.configFile);
+}
+
+var proofreader = new Proofreader(config);
+
+function toHTML(path, content) {
+  var mimeType = mime.lookup(path);
+
+  if(mimeType === 'text/x-markdown') {
+    return marked(content);
+  }
+
+  return content;
+}
+
+var resultPromise = null;
+
+if(program.url) {
+  request({uri: program.url}, function (err, response, body) {
+    if (err) {
+      throw err;
+    }
+
+    resultPromise = proofreader.proofread(toHTML(program.url, body));
+  });
+} else if(program.file) {
+  var content = fs.readFileSync(program.file).toString();
+
+  resultPromise = proofreader.proofread(toHTML(program.file, content));
+} else if(program.fileList) {
+  var listOfFiles = fs.readFileSync(program.fileList).toString().split("\n");
+  var promises = [];
+
+  listOfFiles.forEach(function(filePath) {
+    if(filePath) {
+      var content = fs.readFileSync(filePath).toString();
+      var promise = proofreader.proofread(toHTML(filePath, content));
+
+      promises.push(promise);
+    }
+  });
+
+  resultPromise = Promise.all(promises);
+}
+
+if(resultPromise) {
+  resultPromise.then(null, function() {
+    process.exit(1);
+  })
+}

lib/proofreader.js

@@ -0,0 +1,108 @@
+var path = require('path');
+var clc = require('cli-color');
+var Sync = require('sync');
+var cheerio = require('cheerio');
+var Promise = require('promise');
+var fs = require('fs');
+var writeGood = require('write-good');
+var nodehun = require('nodehun');
+var spellcheck = require('nodehun-sentences');
+
+module.exports = Proofreader;
+
+function Proofreader(settings) {
+  if(!settings) {
+    throw new Error('Settings object missing.');
+  } else if(!settings.dictionaries['build-in'] || !settings.dictionaries['build-in'].length) {
+    throw new Error('At least one build-in dictionary has to be set.');
+  } else if(!settings.selectors || !settings.selectors.whitelist) {
+    throw new Error('Whitelist has to be set.');
+  }
+
+  this._settings = settings;
+  this._setupDictionaries(settings.dictionaries);
+}
+
+Proofreader.prototype._setupDictionaries = function(dictionaries) {
+  var mainDictName = dictionaries['build-in'][0];
+  dictionaries['build-in'].shift();
+
+  //main dictionary
+  var dict = new nodehun(
+    fs.readFileSync(path.join(__dirname, '../dictionaries/' + mainDictName + '.aff')),
+    fs.readFileSync(path.join(__dirname, '../dictionaries/' + mainDictName + '.dic'))
+  );
+
+  //other build-in dictionaries
+  dictionaries['build-in'].forEach(function(dictName) {
+    dict.addDictionary(path.join(__dirname, '../dictionaries/' + dictName + '.dic'));
+  });
+
+  //other custom dictionaries
+  if(dictionaries['custom']) {
+    dictionaries['custom'].forEach(function(dictPath) {
+      dict.addDictionary(fs.readFileSync(dictPath));
+    });
+  }
+
+  this._dictionary = dict;
+};
+
+Proofreader.prototype.proofread = function(html) {
+  var $ = cheerio.load(html);
+  var dictionary = this._dictionary;
+  var whitelist = this._settings.selectors.whitelist;
+  var blacklist = this._settings.selectors.blacklist;
+
+  return new Promise(function(resolve, reject) {
+    Sync(function () {
+      var suggestionsCount = 0;
+
+      //Blacklisted elements are removed before text is processed
+      if(blacklist) {
+        $(blacklist).remove();
+      }
+
+      //Only whitelisted elements are processed
+      $(whitelist).each(function () {
+        var text = $(this).text();
+
+        //remove linebreaks from text
+        text = text.replace(/(\r\n|\n|\r)+/gm," ");
+
+        //replace ’ with '
+        text = text.replace(/’/g, "'");
+
+        if(text.trim().length) {
+          var writeGoodSuggestions = writeGood(text);
+          var spellingSuggestions = spellcheck.sync(null, dictionary, text);
+
+          //Printing output
+          if(writeGoodSuggestions.length || spellingSuggestions.length) {
+            console.log(clc.red(text));
+
+            writeGoodSuggestions.forEach(function(item) {
+              console.log(clc.blue.bold(' - ' + item.reason));
+            });
+
+            spellingSuggestions.forEach(function(item) {
+              console.log(clc.magenta.bold(' - "' + item.word + '" -> ' + item.suggestions));
+            });
+
+            console.log();
+          }
+
+          suggestionsCount += writeGoodSuggestions.length + spellingSuggestions.length;
+        }
+      });
+
+      return suggestionsCount;
+    }, function (err, result) {
+      if(err || result !== 0) {
+        reject();
+      } else {
+        resolve();
+      }
+    });
+  });
+};

package.json

@@ -1,7 +1,7 @@
 {
   "name": "proofreader",
   "description": "Simple text proofreader based on 'write-good' (hemingway-app-like suggestions) and 'nodehun' (spelling).",
-  "version": "0.4.0",
+  "version": "0.5.0",
   "repository": {
     "type": "git",
     "url": "git://github.com/kdzwinel/Proofreader.git"
@@ -16,7 +16,7 @@
     }
   ],
   "bin": {
-    "proofreader": "./bin/proofreader.js.sh"
+    "proofreader": "./bin/cmd.js"
   },
   "dependencies": {
     "cheerio": "^0.18.0",

settings.json

@@ -0,0 +1,10 @@
+{
+  "dictionaries": {
+    "build-in": ["en_US"],
+    "custom": []
+  },
+  "selectors": {
+    "whitelist": "p, li, h1, h2, h3, h4, th, td, dl, figcaption",
+    "blacklist": "pre, code"
+  }
+}