@ezs/basics 1.23.4 → 1.24.0

package/CHANGELOG.md

@@ -3,6 +3,28 @@
 All notable changes to this project will be documented in this file.
 See [Conventional Commits](https://conventionalcommits.org) for commit guidelines.
 
+# [1.24.0](https://github.com/Inist-CNRS/ezs/compare/@ezs/basics@1.23.5...@ezs/basics@1.24.0) (2023-03-28)
+
+
+### Features
+
+* **basics:** Add TXTSentences ([c66abb5](https://github.com/Inist-CNRS/ezs/commit/c66abb5f242afcff57f98a7fa8eff918a7d60098))
+
+
+
+
+
+## [1.23.5](https://github.com/Inist-CNRS/ezs/compare/@ezs/basics@1.23.4...@ezs/basics@1.23.5) (2023-03-24)
+
+
+### Bug Fixes
+
+* 🐛 multi bytes in txtconcat ([7c705fa](https://github.com/Inist-CNRS/ezs/commit/7c705facc1378fae709bbae6a2b416b059e81731))
+
+
+
+
+
 ## [1.23.4](https://github.com/Inist-CNRS/ezs/compare/@ezs/basics@1.23.3...@ezs/basics@1.23.4) (2023-03-24)
 
 

package/README.md

@@ -30,6 +30,7 @@ npm install @ezs/basics
 -   [TXTConcat](#txtconcat)
 -   [TXTObject](#txtobject)
 -   [TXTParse](#txtparse)
+-   [TXTSentences](#txtsentences)
 -   [TXTZip](#txtzip)
 -   [URLConnect](#urlconnect)
 -   [URLFetch](#urlfetch)
@@ -601,6 +602,24 @@ Output:
 
 Returns **[String](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)** 
 
+### TXTSentences
+
+Take a `String` and split it into an array of sentences.
+
+Input:
+
+```json
+"First sentence? Second sentence. My name is Bond, J. Bond."
+```
+
+Output:
+
+```json
+["First sentence?", "Second sentence.", "My name is Bond, J. Bond."]
+```
+
+Returns **[String](https://developer.mozilla.org/docs/Web/JavaScript/Reference/Global_Objects/String)** 
+
 ### TXTZip
 
 Take a `String` and zip it.

package/lib/index.js

@@ -21,6 +21,8 @@ var _txtObject = _interopRequireDefault(require("./txt-object"));
 
 var _txtParse = _interopRequireDefault(require("./txt-parse"));
 
+var _txtSentences = _interopRequireDefault(require("./txt-sentences"));
+
 var _xmlParse = _interopRequireDefault(require("./xml-parse"));
 
 var _xmlString = _interopRequireDefault(require("./xml-string"));
@@ -72,6 +74,7 @@ const funcs = {
   TXTParse: _txtParse.default,
   TXTObject: _txtObject.default,
   TXTConcat: _txtConcat.default,
+  TXTSentences: _txtSentences.default,
   XMLParse: _xmlParse.default,
   XMLString: _xmlString.default,
   XMLConvert: _xmlConvert.default,

package/lib/txt-concat.js

@@ -5,16 +5,23 @@ Object.defineProperty(exports, "__esModule", {
 });
 exports.default = void 0;
 
+var _string_decoder = require("string_decoder");
+
 function TXTConcat(data, feed) {
+  if (!this.decoder) {
+    this.decoder = new _string_decoder.StringDecoder('utf8');
+  }
+
   if (this.buffer === undefined) {
     this.buffer = '';
   }
 
   if (this.isLast()) {
+    this.decoder.end();
     feed.send(this.buffer);
     feed.close();
   } else {
-    this.buffer = this.buffer.concat(data);
+    this.buffer = this.buffer.concat(Buffer.isBuffer(data) ? this.decoder.write(data) : data);
     feed.end();
   }
 }

package/lib/txt-sentences.js

@@ -0,0 +1,100 @@
+"use strict";
+
+Object.defineProperty(exports, "__esModule", {
+  value: true
+});
+exports.default = void 0;
+
+var _string_decoder = require("string_decoder");
+
+const UPPER_LETTERS = 'ABCDEFGHIJKLMNOPQRSTUVWXYZ';
+const SENTENCE_INIT = '  ';
+const SENTENCE_ENDING = '.?!';
+/*
+ * Segment sentences from `str` into an array
+ * @param {string} str
+ * @returns {string[]}
+ */
+
+const segmentSentences = str => {
+  const characters = str.split('');
+  const sentences = characters.reduce(
+  /*
+   * @param {string[]} prevSentences
+   * @param {string} character
+   * @return {string[]}
+   */
+  (prevSentences, character) => {
+    const currentSentence = prevSentences.slice(-1)[0];
+    const [char1, char2] = currentSentence.slice(-2);
+
+    if (SENTENCE_ENDING.includes(character)) {
+      if (character !== '.') {
+        return [...prevSentences.slice(0, -1), currentSentence + character, SENTENCE_INIT];
+      }
+
+      if (char1 !== ' ') {
+        return [...prevSentences.slice(0, -1), currentSentence + character, SENTENCE_INIT];
+      }
+
+      if (!UPPER_LETTERS.includes(char2)) {
+        return [...prevSentences.slice(0, -1), currentSentence + character, SENTENCE_INIT];
+      }
+    }
+
+    return [...prevSentences.slice(0, -1), currentSentence + character];
+  }, [SENTENCE_INIT]).map(sentence => sentence.trimStart());
+  return sentences;
+};
+
+const TXTSentences = (data, feed, ctx) => {
+  if (!ctx.decoder) {
+    ctx.decoder = new _string_decoder.StringDecoder('utf8');
+  }
+
+  if (ctx.isLast()) {
+    ctx.decoder.end();
+    return feed.end();
+  }
+
+  ctx.remainder = ctx.remainder ?? '';
+  let str;
+
+  if (Buffer.isBuffer(data)) {
+    str = ctx.decoder.write(data);
+  } else if (typeof data === 'string') {
+    str = data;
+  }
+
+  const lines = str ? segmentSentences(str) : [];
+  lines.unshift(ctx.remainder + lines.shift());
+  ctx.remainder = lines.pop();
+  lines.forEach(line => {
+    feed.write(line);
+  });
+  return feed.end();
+};
+/**
+ * Take a `String` and split it into an array of sentences.
+ *
+ * Input:
+ *
+ * ```json
+ * "First sentence? Second sentence. My name is Bond, J. Bond."
+ * ```
+ *
+ * Output:
+ *
+ * ```json
+ * ["First sentence?", "Second sentence.", "My name is Bond, J. Bond."]
+ * ```
+ *
+ * @name TXTSentences
+ * @returns {String}
+ */
+
+
+var _default = {
+  TXTSentences
+};
+exports.default = _default;

package/package.json

@@ -1,7 +1,7 @@
 {
   "name": "@ezs/basics",
   "description": "Basics statements for EZS",
-  "version": "1.23.4",
+  "version": "1.24.0",
   "author": "Nicolas Thouvenin <nthouvenin@gmail.com>",
   "bugs": "https://github.com/Inist-CNRS/ezs/issues",
   "dependencies": {
@@ -39,7 +39,7 @@
   "directories": {
     "test": "test"
   },
-  "gitHead": "8da84f8d6833a91a2ae87fdb635070a811c3f9ef",
+  "gitHead": "0f52795f9c5d1a452adbad48ab567298bb8a6bea",
   "homepage": "https://github.com/Inist-CNRS/ezs/tree/master/packages/basics#readme",
   "keywords": [
     "ezs"

package/lib/fetch.js

@@ -1,74 +0,0 @@
-"use strict";
-
-Object.defineProperty(exports, "__esModule", {
-  value: true
-});
-exports.default = fetch;
-
-var _crossFetch = _interopRequireDefault(require("cross-fetch"));
-
-var _proxyFromEnv = require("proxy-from-env");
-
-var _http = _interopRequireDefault(require("http"));
-
-var _https = _interopRequireDefault(require("https"));
-
-var _betterHttpsProxyAgent = require("better-https-proxy-agent");
-
-function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
-
-const DefaultOptions = {
-  keepAlive: true,
-  timeout: 1000,
-  keepAliveMsecs: 500,
-  maxSockets: 200,
-  maxFreeSockets: 5,
-  maxCachedSessions: 500
-};
-
-const chooseAgent = (parsedURL, options) => {
-  const proxyurl = (0, _proxyFromEnv.getProxyForUrl)(parsedURL.href);
-
-  if (proxyurl) {
-    const proxyRequestOptions = new URL(proxyurl);
-    return new _betterHttpsProxyAgent.Agent(options, proxyRequestOptions);
-  }
-
-  if (parsedURL.protocol === 'https:') {
-    return new _https.default.Agent(options);
-  }
-
-  return new _http.default.Agent(options);
-};
-
-function fetch(url, options) {
-  const opts = options || {};
-  const {
-    keepAlive,
-    timeout,
-    keepAliveMsecs,
-    maxSockets,
-    maxFreeSockets,
-    maxCachedSessions
-  } = { ...options,
-    ...DefaultOptions
-  };
-  let agent = chooseAgent(new URL(url), {
-    keepAlive,
-    timeout,
-    keepAliveMsecs,
-    maxSockets,
-    maxFreeSockets,
-    maxCachedSessions
-  });
-  opts.agent = agent;
-
-  if (opts.signal) {
-    opts.signal.addEventListener('abort', () => {
-      agent.destroy();
-      agent = null;
-    });
-  }
-
-  return (0, _crossFetch.default)(url, options);
-}

package/lib/skos-object.js

@@ -1,73 +0,0 @@
-"use strict";
-
-Object.defineProperty(exports, "__esModule", {
-  value: true
-});
-exports.default = void 0;
-
-function Concept(data, feed) {
-  const obj = {};
-  Object.keys(data).forEach(key => {
-    const newkey = key.replace('skos$', '');
-
-    if (Array.isArray(data[key])) {
-      data[key].filter(x => x.xml$lang).forEach(item => {
-        const localkey = newkey.concat('@').concat(item.xml$lang);
-        obj[localkey] = item.$t;
-      });
-    } else if (data[key].rdf$resource && !obj[newkey]) {
-      obj[newkey] = data[key].rdf$resource;
-    } else if (data[key].rdf$resource && obj[newkey]) {
-      obj[newkey] = [obj[newkey], data[key].rdf$resource];
-    } else if (data[key].$t && data[key].xml$lang) {
-      const localkey = newkey.concat('@').concat(data[key].xml$lang);
-      obj[localkey] = data[key].$t;
-    } else if (data[key].$t && Array.isArray(obj[newkey])) {
-      obj[newkey].push(data[key].$t);
-    } else if (data[key].$t && obj[newkey]) {
-      obj[newkey] = [obj[newkey], data[key].$t];
-    } else if (data[key].$t && !obj[newkey]) {
-      obj[newkey] = data[key].$t;
-    } else if (typeof data[key] === 'object') {
-      obj[newkey] = (this.getIndex().toString(36) + Math.random().toString(36).substr(2, 5)).toUpperCase();
-      let counter = 0;
-      Object.keys(data[key]).forEach(key2 => {
-        if (typeof data[key][key2] === 'object') {
-          data[key][key2].rdf$about = obj[newkey];
-          Concept.call(this, data[key][key2], feed);
-          counter += 1;
-        }
-      });
-
-      if (counter === 0) {
-        delete obj[newkey];
-      }
-    } else {
-      obj[newkey] = data[key];
-    }
-  });
-  feed.write(obj);
-}
-
-function SKOSObject(data, feed) {
-  if (this.isLast()) {
-    feed.close();
-  } else {
-    Concept.call(this, data, feed);
-    feed.end();
-  }
-}
-/**
- * Take `Object` generated by XMLMapping & SKOS data and
- * create a new basic object with only keys & values
- *
- * @name SKOSObject
- * @param {undefined} none
- * @returns {Object}
- */
-
-
-var _default = {
-  SKOSObject
-};
-exports.default = _default;

package/lib/url-pager.js

@@ -1,179 +0,0 @@
-"use strict";
-
-Object.defineProperty(exports, "__esModule", {
-  value: true
-});
-exports.default = URLPager;
-
-var _debug = _interopRequireDefault(require("debug"));
-
-var _url = require("url");
-
-var _nodeAbortController = _interopRequireDefault(require("node-abort-controller"));
-
-var _lodash = _interopRequireDefault(require("lodash.get"));
-
-var _parseHeaders = _interopRequireDefault(require("parse-headers"));
-
-var _asyncRetry = _interopRequireDefault(require("async-retry"));
-
-var _request = _interopRequireDefault(require("./request"));
-
-function _interopRequireDefault(obj) { return obj && obj.__esModule ? obj : { default: obj }; }
-
-/**
- * Take `Object` as parameters of URL, throw each chunk from the result
- *
- *
- * Input:
- *
- * ```json
- * [{"q": "a"}]
- * ```
- *
- * Script:
- *
- * ```ini
- * [URLPager]
- * url = https://api.search.net
- * path = total
- * ```
- *
- * Output:
- *
- * ```json
- * [
- *      {
- *          "q": "a",
- *          "total": 22
- *          "offset": 0,
- *          "pageNumber": 1,
- *          "totalPages", 3,
- *          "maxPages": 1000,
- *          "limit": 10
- *      },
- *      {
- *          "q": "a",
- *          "total": 22
- *          "offset": 10,
- *          "pageNumber": 2,
- *          "totalPages", 3,
- *          "maxPages": 1000,
- *          "limit": 10
- *      },
- *      {
- *          "q": "a",
- *          "total": 22
- *          "offset": 20,
- *          "pageNumber": 3,
- *          "totalPages", 3,
- *          "maxPages": 1000,
- *          "limit": 10
- *      }
- *  ]
- * ```
- *
- * #### Example with URLs
- *
- * Input:
- *
- * ```json
- * [
- *   "https://httpbin.org/get?a=a",
- *   "https://httpbin.org/get?a=b",
- *   "https://httpbin.org/get?a=c"
- * ]
- * ```
- *
- * Script:
- *
- * ```ini
- * [URLPager]
- * path = .args
- * ```
- *
- * Output:
- *
- * ```json
- * [{"a": "a"}, {"a": "b"}, {"a": "c" }]
- * ```
- *
- * @name URLPager
- * @param {String} [url] URL to fetch (by default input string is taken)
- * @param {String} [path=total] choose the path to find the number of result
- * @param {Number} [timeout=1000] Timeout in milliseconds
- * @param {Boolean} [noerror=false] Ignore all errors, the target field will remain undefined
- * @param {Number} [retries=5] The maximum amount of times to retry the connection
- * @returns {Object}
- */
-async function URLPager(data, feed) {
-  if (this.isLast()) {
-    return feed.close();
-  }
-
-  const url = this.getParam('url');
-  const path = this.getParam('path', 'total');
-  const limit = Number(this.getParam('limit', 10));
-  const maxPages = Number(this.getParam('maxPages', 1000));
-  const retries = Number(this.getParam('retries', 5));
-  const noerror = Boolean(this.getParam('noerror', false));
-  const timeout = Number(this.getParam('timeout')) || 1000;
-  const headers = (0, _parseHeaders.default)([].concat(this.getParam('header')).filter(Boolean).join('\n'));
-  const cURL = new _url.URL(url || data);
-  const controller = new _nodeAbortController.default();
-  const parameters = {
-    timeout,
-    headers,
-    signal: controller.signal
-  };
-  const options = {
-    retries
-  };
-  cURL.search = new _url.URLSearchParams(data);
-
-  const onError = e => {
-    controller.abort();
-
-    if (noerror) {
-      (0, _debug.default)('ezs')(`Ignore item #${this.getIndex()} [URLPager] <${e}>`);
-      return feed.send(data);
-    }
-
-    (0, _debug.default)('ezs')(`Break item #${this.getIndex()} [URLPager] <${e}>`);
-    return feed.send(e);
-  };
-
-  try {
-    const response = await (0, _asyncRetry.default)((0, _request.default)(cURL.href, parameters), options);
-    const json = await response.json();
-    const total = (0, _lodash.default)(json, path);
-
-    if (total === 0) {
-      return onError(new Error('No result.'));
-    }
-
-    if (total === undefined) {
-      return onError(new Error('Unexpected response.'));
-    }
-
-    let totalPages = Math.ceil(json.total / limit);
-
-    if (totalPages > maxPages) {
-      totalPages = maxPages;
-    }
-
-    for (let pageNumber = 1; pageNumber <= totalPages; pageNumber += 1) {
-      feed.write({ ...data,
-        offset: (pageNumber - 1) * limit,
-        pageNumber,
-        totalPages,
-        maxPages,
-        limit
-      });
-    }
-
-    feed.end();
-  } catch (e) {
-    onError(e);
-  }
-}

package/lib/utils.js

@@ -1,18 +0,0 @@
-"use strict";
-
-Object.defineProperty(exports, "__esModule", {
-  value: true
-});
-exports.default = exports.writeTo = void 0;
-
-const writeTo = function writeTo(stream, data, cb) {
-  if (!stream.write(data)) {
-    stream.once('drain', cb);
-  } else {
-    process.nextTick(cb);
-  }
-};
-
-exports.writeTo = writeTo;
-var _default = writeTo;
-exports.default = _default;