@graphy/content.nq.read - npm Package Compare versions

Comparing version 3.0.6 to 3.1.0

main.js

 
 
 
-const FN_TO_STRING_OBJECT = Object.prototype.toString;
-
-
-
-// try something else
-
-
-
 const stream = require('@graphy/core.iso.stream');
@@ -18,2 +10,11 @@ const factory = require('@graphy/core.data.factory').raw;
 
+const H_ESCAPES_JSON = 	{
+	'\t': '\\t',
+	'\u0008': '\\b',
+	'\n': '\\n',
+	'\r': '\\r',
+	'\f': '\\f',
+	'"': '\\"',
+};
+
 const R_ESCAPES = /(\\[\\])|\\([^tbnrfuU\\])/g;
@@ -26,10 +27,12 @@ const R_UNICODE_8 = /\\U([0-9A-Fa-f]{8})/g;
 const R_CLEAN = /\s*(?:#[^\n]*\n\s*)*\s*/y;
+const R_CLEAN_COMMENTS = /\s*(#[^\n]*\n\s*)*\s*/y;
 const R_LITERAL_ESCAPELESS = /^"([^\\"]*)"(?:\^\^<([^\\>]*)>|@([^ \t.]+)|)?$/;
 const R_LITERAL = /^"(.*)"(?:\^\^<(.*)>|@([^ \t.]+)|)?$/;
 const RT_HAS_ESCAPES = /[\\]/;
+const R_EOL = /[^\n]+\n/y;
 
 const RT_ABSOLUTE_IRI_VALID = /^[a-z][a-z0-9+\-.]*:([^\0-\x20<>"{}|^`\\]|\\u[A-Fa-f0-9]{4}|\\U[A-Fa-f0-9]{8})*$/;
 const RT_BLANK_NODE_LABEL_VALID = /^(?:[A-Za-z\xc0-\xd6\xd8-\xf6\xf8-\u{02ff}\u{0370}-\u{037d}\u{037f}-\u{1fff}\u{200c}-\u{200d}\u{2070}-\u{218f}\u{2c00}-\u{2fef}\u{3001}-\u{d7ff}\u{f900}-\u{fdcf}\u{fdf0}-\u{fffd}\u{10000}-\u{effff}_0-9])(?:(?:[A-Za-z\xc0-\xd6\xd8-\xf6\xf8-\u{02ff}\u{0370}-\u{037d}\u{037f}-\u{1fff}\u{200c}-\u{200d}\u{2070}-\u{218f}\u{2c00}-\u{2fef}\u{3001}-\u{d7ff}\u{f900}-\u{fdcf}\u{fdf0}-\u{fffd}\u{10000}-\u{effff}_\-0-9\xb7\u{0300}-\u{036f}\u{203f}-\u{2040}.])*[A-Za-z\xc0-\xd6\xd8-\xf6\xf8-\u{02ff}\u{0370}-\u{037d}\u{037f}-\u{1fff}\u{200c}-\u{200d}\u{2070}-\u{218f}\u{2c00}-\u{2fef}\u{3001}-\u{d7ff}\u{f900}-\u{fdcf}\u{fdf0}-\u{fffd}\u{10000}-\u{effff}_\-0-9\xb7\u{0300}-\u{036f}\u{203f}-\u{2040}])?$/u;
-const RT_ESCAPES_INVALID = /\\[^"tbnrfuU\\_~.!$&'()*+,;=/?#@%-]|\\u[^A-Fa-f0-9]{4}|\\U[^A-Fa-f0-9]{8}/;
-const RT_LITERAL_CONTENTS_VALID = /^([^"\\\n\r]|\\[tbnrf"'\\]|\\u[A-Fa-f0-9]{4}|\\U[A-Fa-f0-9]{8})*$/;
+const RT_ESCAPES_INVALID = /(?:(?:^|[^\\])(?:\\(?:\\\\)*[^"tbnrfuU\\_~.!$&'()*+,;=/?#@%-])(?:[^\\]|$))|\\u[^A-Fa-f0-9]{4}|\\U[^A-Fa-f0-9]{8}/;
+const RT_LITERAL_CONTENTS_VALID = /^(?:[^\\\n\r]|\\[tbnrf"'\\]|\\u[A-Fa-f0-9]{4}|\\U[A-Fa-f0-9]{8})*$/;
 const RT_LANGUAGE_VALID = /^[a-z]+(-[a-z0-9]+)*$/;
@@ -39,7 +42,63 @@
 
-const R_QUAD_ESCAPELESS_SP = /(?:<([^\\>]*)>|_:([^\x20\t<]+))[\x20\t]*<([^\\>]*)>[\x20\t]*(?:(<[^\\>]*)>|_:([^\x20\t<]+)|"([^"\\]*)"(?:\^\^<([^\\>]*)>|@([^\x20\t.]+)|))[\x20\t]*(?:<([^\\>]*)>|_:([^\x20\t<]+)|)[\x20\t]*\.\s*(?:#[^\n]*\n\s*|\n\s*)+/y;
-const R_QUAD = /(?:<([^>]*)>|_:([^\x20\t<]+))[\x20\t]*<([^>]*)>[\x20\t]*(?:(<[^>]*)>|_:([^\x20\t<]+)|"((?:[^"\\]|\\.)*)"(?:\^\^<([^>]*)>|@([^\x20\t.]+)|))[\x20\t]*(?:<([^>]*)>|_:([^\x20\t<]+)|)[\x20\t]*\.\s*(?:#[^\n]*\n\s*|\n\s*)+/y;
+const R_QUAD_ESCAPELESS_SP = /(?:<([^\\>]*)>|_:([^\x20\t<]+))[\x20\t]*<([^\\>]*)>[\x20\t]*(?:(<[^\\>]*)>|_:([^\x20\t<]+)|"([^"\\]*)"(?:\^\^<([^\\>]*)>|@([^\x20\t.]+)|))[\x20\t]*(?:<([^\\>]*)>|_:([^\x20\t<]+)|)[\x20\t]*\.\s*(#[^\n]*\n\s*|\n\s*)+/y;
+const R_QUAD = /(?:<([^>]*)>|_:([^\x20\t<]+))[\x20\t]*<([^>]*)>[\x20\t]*(?:(<[^>]*)>|_:([^\x20\t<]+)|"((?:[^"\\]|\\.)*)"(?:\^\^<([^>]*)>|@([^\x20\t.]+)|))[\x20\t]*(?:<([^>]*)>|_:([^\x20\t<]+)|)[\x20\t]*\.\s*(#[^\n]*\n\s*|\n\s*)+/y;
 
 
 
+class NQuads_Reader extends stream.Transform {
+	constructor(g_impls) {
+		super({
+			// do not decode strings into buffers
+			decodeStrings: false,
+
+			// accept strings as input on writable side
+			writableObjectMode: false,
+
+			// output quad objects on readable side
+			readableObjectMode: true,
+
+			// implementations
+			flush: g_impls.flush,
+			transform: g_impls.transform,
+		});
+
+		// when the writable side is piped into
+		this.on('pipe', (ds_input) => {
+			// input stream has encoding option; ensure stream encoding is utf8
+			if('function' === typeof ds_input.setEncoding) {
+				ds_input.setEncoding('utf8');
+			}
+		});
+	}
+
+	// intercept pipe
+	pipe(ds_out) {
+		let ds_dst = ds_out;
+
+		// non-object mode
+		if(!ds_dst._writableState.objectMode) {
+			// transform to JSON
+			ds_out = stream.quads_to_json();
+		}
+		// yet object mode and graphy writable
+		else if(ds_out.isGraphyWritable) {
+			// transform to quad-stream
+			ds_out = stream.quads_to_writable();
+		}
+
+		// interim stream created
+		if(ds_out !== ds_dst) {
+			// forward output to super
+			super.pipe(ds_out);
+
+			// pipe outpu to destination
+			return ds_out.pipe(ds_dst);
+		}
+		// forward as-is to super
+		else {
+			return super.pipe(ds_dst);
+		}
+	}
+}
+
 class reader {
@@ -69,4 +128,9 @@ constructor(g_config) {
 			validate: b_validate,
+
+			destroyed: false,
 		});
 
+		// clean regex
+		let r_clean = R_CLEAN;
+
 		// validation
@@ -92,5 +156,4 @@ Object.assign(this, b_validate
 		.replace(R_ESCAPES, '$1$2') // no need to escape anything other than reserved characters
-
-		.replace(/"/g, '\\"') // escape all quotes ;)
-					+'"'): s_literal);
+		.replace(/[\t"\f\u0008]/g, s => H_ESCAPES_JSON[s])
+				+'"'): s_literal);
 					}
@@ -111,5 +174,4 @@ catch(e_parse) {
 		.replace(R_ESCAPES, '$1$2') // no need to escape anything other than reserved characters
-
-		.replace(/"/g, '\\"') // escape all quotes ;)
-					+'"'): s_literal),
+		.replace(/[\t"\f\u0008]/g, s => H_ESCAPES_JSON[s])
+				+'"'): s_literal),
 			});
@@ -124,9 +186,46 @@
 		// create transform
-		ds_transform = this.transform = new stream.Transform({
-			// do not decode strings into buffers
-			decodeStrings: false,
+		ds_transform = this.transform = new NQuads_Reader({
+			// on data event
+			transform: (s_chunk, s_encoding, fk_chunk) => {
+				// first transform
+				if(!b_init) {
+					// notify that data will begin
+					ds_transform.emit('ready');
 
-			// output quad objects on readable side
-			readableObjectMode: true,
+					// do not emit 'ready' event again
+					b_init = false;
+				}
 
+				// concatenate current chunk to previous chunk
+				let s = this.s += s_chunk;
+
+				// remove whitespace & comments from beginning
+				r_clean.lastIndex = 0;
+				let m_clean = r_clean.exec(s);
+				if(this.emit_comments) {
+					this.emit_comments(m_clean[1]);
+				}
+
+				// update index and prepare to match statement
+				this.i = r_clean.lastIndex;
+
+				// cache chunk length
+				this.n = s.length;
+
+				// resume parsing
+				try {
+					this.safe_parse(true);
+				}
+				// read error occurred; emit and destroy stream
+				catch(e_read) {
+					return ds_transform.destroy(e_read);
+				}
+
+				// emit progress event updates
+				ds_transform.emit('progress', s_chunk.length);
+
+				// done transforming this chunk
+				fk_chunk();
+			},
+
 			// once there's no more data to consume, invoke eof
@@ -140,11 +239,14 @@ flush: (fk_flush) => {
 					// parse safely
-					this.safe_parse();
+					try {
+						this.safe_parse();
+					}
+					// read error occurred; pass to flush errback and exit method
+					catch(e_read) {
+						// destroying during flush means overriding push
+						return ds_transform.demolish(e_read);
+					}
 
-					// still unparsed characters
+					// still unparsed characters; pass to flush errback and exit method
 					if(this.s.length) {
-						// throw parse error
-						fk_flush(new Error(`parsing error occurred in state: statement\n  ${this.s.substr(0, 50)}\n  ^ starting here`));
-
-						// exit method
-						return;
+						return ds_transform.demolish(new Error(`parsing error occurred in state: statement\n  ${this.s.substr(0, 50)}\n  ^ starting here`));
 					}
@@ -165,49 +267,2 @@ }
 			},
-
-			// on data event
-			transform: (s_chunk, s_encoding, fk_chunk) => {
-				// first transform
-				if(!b_init) {
-					// notify that data will begin
-					ds_transform.emit('ready');
-
-					// do not emit 'ready' event again
-					b_init = false;
-				}
-
-				// stream is paused
-				if(this.n < 0) {
-					return this.error('stream received new data while it was supposed to be paused!');
-				}
-
-				// concatenate current chunk to previous chunk
-				let s = this.s += s_chunk;
-
-				// remove whitespace & comments from beginning
-				R_CLEAN.lastIndex = 0;
-				R_CLEAN.exec(s);
-
-				// update index and prepare to match statement
-				this.i = R_CLEAN.lastIndex;
-
-				// cache chunk length
-				this.n = s.length;
-
-				// resume parsing
-				this.safe_parse(true);
-
-				// read error occurred
-				if(this.read_error) {
-					// destroy stream
-					ds_transform.destroy(this.read_error);
-				}
-				// no errors
-				else {
-					// emit progress event updates
-					ds_transform.emit('progress', s_chunk.length);
-
-					// done transforming this chunk
-					fk_chunk();
-				}
-			},
 		});
@@ -218,23 +273,39 @@
 
-		// bind events to transform stream
-		this.bind(g_config);
+		// new listener added
+		ds_transform.on('newListener', (s_event) => {
+			// comment
+			if('comment' === s_event) {
+				r_clean = R_CLEAN_COMMENTS;
+				this.emit_comments = (s_captured) => {
+					if(!s_captured) return;
+					let a_comments = s_captured.slice(1).replace(/\n\s+$/, '').split(/\n+\s*#/g);
 
-		// notify once and never again
-		ds_transform.once('pipe', (ds_input) => {
-			// input stream has encoding option
-			if(ds_input.setEncoding) {
-				// ensure stream is encoding in utf8
-				ds_input.setEncoding('utf8');
+					for(let s_comment of a_comments) {
+						ds_transform.emit('comment', s_comment);
+					}
+				};
 			}
 		});
 
-		// input
+		// bind events to transform stream
+		this.bind(g_config);
+
+		// input given
 		if(g_input) {
+			// input is stream
 			if(g_input.stream) {
-				g_input.stream.pipe(ds_transform);
+				let ds_input = g_input.stream;
+
+				// go async so caller has chance to bind event listeners
+				setTimeout(() => {
+					ds_input.pipe(ds_transform);
+				}, 0);
 			}
 			// string
-			else if('string' in g_input) {
+			else if('string' === typeof g_input.string) {
+				let s_input = g_input.string;
+
+				// go async so caller has chance to bind event listeners
 				setTimeout(() => {
-					ds_transform.end(g_input.string, 'utf8');
+					ds_transform.end(s_input, 'utf8');
 				}, 0);
@@ -244,3 +315,3 @@ }
 			else {
-				throw new TypeError('invalid input: '+('object' === typeof g_input? JSON.stringify(g_input): g_input));
+				throw new TypeError(`Invalid argument for input parameter: ${'object' === typeof g_input? JSON.stringify(g_input): g_input}`);
 			}
@@ -251,3 +322,4 @@ }
 	_error(s_message) {
-		this.read_error = new Error(s_message);
+		this.destroyed = true;
+		throw new Error(s_message);
 	}
@@ -260,6 +332,8 @@
 		if(g_config.error) ds_transform.on('error', g_config.error);
+		if(g_config.comment) ds_transform.on('comment', g_config.comment);
 		if(g_config.read) ds_transform.once('read', g_config.read);
 		if(g_config.progress) ds_transform.on('progress', g_config.progress);
 		if(g_config.eof) ds_transform.once('eof', g_config.eof);
-		if(g_config.end) ds_transform.on('end', g_config.end);
+		if(g_config.end) ds_transform.once('end', g_config.end);
+		if(g_config.finish) ds_transform.once('finish', g_config.finish);
 		if(g_config.data) ds_transform.on('data', g_config.data);
@@ -331,2 +405,7 @@ }
 				);
+
+// comments
+				if(this.emit_comments) {
+					this.emit_comments(m_statement_e_sp[11]);
+				}
 			}
@@ -390,8 +469,23 @@ else {
 
-		// match counter: 1
+// comments
+					if(this.emit_comments) {
+						this.emit_comments(m_statement[11]);
+					}
 				}
 				else {
+	// prepare sticky regex index
+					R_EOL.lastIndex = i;
+
+					if(R_EOL.exec(s)) {
+	// advance index
+						this.i = R_EOL.lastIndex;
+						this._error(`invalid statement:\n${this.s.substr(0, 50)}\n  ^ starting here`);
+
+		// match counter: 2
+					}
+					else {
 	// break loop to retry on next chunk if eos
-					break;
-				}
+						break;
+					}
+				} // brace #2
 			} // brace #1
@@ -406,3 +500,3 @@ } // end of while
 
-const read = module.exports = function(...a_args) {
+module.exports = function(...a_args) {
 	let g_config = {};
@@ -427,3 +521,3 @@
 	// config struct
-		else if(z_arg_0 && 'object' === typeof z_arg_0 && '[object Object]' === FN_TO_STRING_OBJECT.call(z_arg_0)) {
+		else if(z_arg_0 && 'object' === typeof z_arg_0 && '[object Object]' === Object.prototype.toString.call(z_arg_0)) {
 			g_config = z_arg_0;
@@ -430,0 +524,0 @@

package.json

 {
   "name": "@graphy/content.nq.read",
-  "version": "3.0.6",
+  "version": "3.1.0",
   "description": "Single-threaded RDF N-Quads content reader",
@@ -26,6 +26,9 @@ "keywords": [
   "dependencies": {
-    "@graphy/core.data.factory": "^3.0.6",
-    "@graphy/core.iso.stream": "^3.0.6",
+    "@graphy/core.data.factory": "^3.1.0",
+    "@graphy/core.iso.stream": "^3.1.0",
     "uri-js": "^4.2.2"
+  },
+  "engines": {
+    "node": ">=8.4.0"
   }
-}
+}

No alert changes

Improved metrics

Total package byte prevSize

16466

increased by18.15%

Lines of code

445

increased by24.65%

Dependency changes

• Updated@graphy/core.data.factory@^3.1.0

• Updated@graphy/core.iso.stream@^3.1.0