From 98be9c8f22d4d22e46817841f24ed1074d5a71d0 Mon Sep 17 00:00:00 2001
From: Michael Elsdoerfer <michael@elsdoerfer.com>
Date: Sun, 17 Apr 2011 00:48:45 +0200
Subject: [PATCH] Make it possible to feed CvsReader pieces of data from
 memory. In a way, this is a workaround for the lack of a memory-based stream
 object in Node.

---
 README.md      | 22 +++++++++++++++++++++-
 lib/ya-csv.js  | 45 +++++++++++++++++++++++++++------------------
 test/stream.js | 45 +++++++++++++++++++++++++++++++++++++++++++++
 3 files changed, 93 insertions(+), 19 deletions(-)
 create mode 100644 test/stream.js

diff --git a/README.md b/README.md
index f0fa9b0..5253123 100644
--- a/README.md
+++ b/README.md
@@ -22,7 +22,8 @@ Current version requires at least Node.js v0.1.99 and it's tested with Node.js v
 
  - event based, suitable for processing big CSV streams
  - configurable separator, quote and escape characters (comma, double-quote and double-quote by default)
-- ignores lines starting with configurable comment character (off by default)
+ - ignores lines starting with configurable comment character (off by default)
+ - supports memory-only streaming
 
 ## More examples
 
@@ -70,3 +71,22 @@ Convert the `/etc/passwd` file to comma separated format, drop commented lines a
     reader.addListener('data', function(data) {
         writer.writeRecord(data);
     });
+
+Parsing an upload as the data comes in, using node-formidable:
+
+    upload_form.onPart = function(part) {
+        if (!part.filename) { upload_form.handlePart(part); return }
+
+        var reader = csv.createCsvFileReader({'comment': '#'});
+        reader.addListener('data', function(data) {
+            saveRecord(data);
+        });
+
+        part.on('data', function(buffer) {
+            // Pipe incoming data into the reader.
+            reader.parse(buffer);
+        });
+        part.on('end', function() {
+            reader.end()
+        }
+    }
diff --git a/lib/ya-csv.js b/lib/ya-csv.js
index 995fbff..fe894d3 100644
--- a/lib/ya-csv.js
+++ b/lib/ya-csv.js
@@ -21,23 +21,11 @@ var CsvReader = csv.CsvReader = function(readStream, options) {
         commentedLine: false
     };
 
-    readStream.addListener('data', function(data) { self.parse(data) });
-    readStream.addListener('error', function() { self.emit('error') });
-    readStream.addListener('end', function() {
-        var ps = self.parsingStatus;
-        if (ps.quotedField) {
-            self.emit('error', new Error('Input stream ended but closing quotes expected'));
-        } else {
-            // dump open record
-            if (ps.openField) {
-                self._addField();
-            }
-            if (ps.openRecord.length > 0) {
-                self._addRecord();
-            }
-            self.emit('end');
-        }
-    });
+    if (readStream) {
+        readStream.addListener('data', function(data) { self.parse(data) });
+        readStream.addListener('error', function() { self.emit('error') });
+        readStream.addListener('end', function() { self.end() });
+    }
 };
 sys.inherits(CsvReader, events.EventEmitter);
 
@@ -114,6 +102,22 @@ CsvReader.prototype.parse = function(data) {
     }
 };
 
+CsvReader.prototype.end = function() {
+    var ps = this.parsingStatus;
+    if (ps.quotedField) {
+        this.emit('error', new Error('Input stream ended but closing quotes expected'));
+    } else {
+        // dump open record
+        if (ps.openField) {
+            this._addField();
+        }
+        if (ps.openRecord.length > 0) {
+            this._addRecord();
+        }
+        this.emit('end');
+    }
+}
+
 CsvReader.prototype._isEscapable = function(c) {
     if ((c === this.escapechar) || (c === this.quotechar)) {
         return true;
@@ -169,8 +173,13 @@ csv.createCsvFileReader = function(path, options) {
 };
 
 csv.createCsvStreamReader = function(readStream, options) {
+    if (options === undefined && typeof readStream === 'object') {
+        options = readStream;
+        readStream = undefined;
+    }
     options = options || {};
-    readStream.setEncoding(options.encoding ? options.encoding : 'utf8');
+    if (readStream)
+       readStream.setEncoding(options.encoding ? options.encoding : 'utf8');
     return new CsvReader(readStream, options);
 };
 
diff --git a/test/stream.js b/test/stream.js
new file mode 100644
index 0000000..3537b52
--- /dev/null
+++ b/test/stream.js
@@ -0,0 +1,45 @@
+var csv  = require('../lib/ya-csv'),
+    sys  = require('sys'),
+    fs = require('fs');
+
+sys.debug('start');
+
+if (process.argv.length < 3) {
+    sys.error("Usage: node " + process.argv[1] + " <csv file>");
+    process.exit(1);
+}
+
+var csvIn = csv.createCsvStreamReader({
+    'separator': ',',
+    'quote':   '"',
+    'comment': '#'
+});
+
+var lines   = 0;
+var columns = 0;
+
+csvIn.addListener('end', function() {
+    sys.debug('end');
+    sys.debug(columns + ' columns, ' + lines + ' lines');
+});
+
+csvIn.addListener('error', function(e) {
+    sys.debug('error');
+    sys.debug(e);
+});
+
+csvIn.addListener('data', function(data) {
+    lines++;
+    columns += data.length;
+});
+
+var file = process.argv[2];
+var fileIn = fs.createReadStream(file, {flags: 'r', bufferSize: 10});
+fileIn.setEncoding('utf8');
+fileIn.on('data', function(data) {
+  sys.debug(data);
+  csvIn.parse(data);
+});
+fileIn.on('end', function(data) {
+  csvIn.end();
+});