From 22c6c24457fe45cf2fa6bd8f53721bf00de48cba Mon Sep 17 00:00:00 2001
From: Joakim Wennergren <jokedst@gmail.com>
Date: Fri, 26 Jan 2018 18:37:14 +0100
Subject: [PATCH] More streams. Less OutOfMemoryException

---
 CsvQuery/Csv/CsvAnalyzer.cs   |  7 ++-----
 CsvQuery/Csv/CsvSettings.cs   | 15 ++++++++++++++-
 CsvQuery/Forms/QueryWindow.cs | 32 +++++++++++++++++++++-----------
 3 files changed, 37 insertions(+), 17 deletions(-)

diff --git a/CsvQuery/Csv/CsvAnalyzer.cs b/CsvQuery/Csv/CsvAnalyzer.cs
index 0a722fb..b4d4ebc 100644
--- a/CsvQuery/Csv/CsvAnalyzer.cs
+++ b/CsvQuery/Csv/CsvAnalyzer.cs
@@ -22,11 +22,6 @@ private class Stat
         /// <returns></returns>
         public static CsvSettings Analyze(string csvString)
         {
-            // TODO: strings with quoted values (e.g. 'hej,san')
-            // Not sure how to detect this, but we could just run the variance analysis
-            // 3 times, one for none, one for ' and one for " and see which has best variances
-            // That wouldn't detect escape chars though, or odd variants like [this]
-
             var result = DetectW3C(csvString);
             if (result != null)
                 return result;
@@ -120,6 +115,8 @@ public static CsvSettings Analyze(string csvString)
             var separatorQuoted = GetSeparatorFromVariance(variancesQuoted, occurrencesQuoted, linesQuoted, out var uncertancyQuoted);
             if (uncertancyQuoted < uncertancy)
                 result.Separator = separatorQuoted;
+            else if (uncertancy < uncertancyQuoted) // It was better ignoring quotes!
+                result.TextQualifier = '\0';
 
             if (result.Separator != default(char)) return result;
 
diff --git a/CsvQuery/Csv/CsvSettings.cs b/CsvQuery/Csv/CsvSettings.cs
index 9857a88..a399a63 100644
--- a/CsvQuery/Csv/CsvSettings.cs
+++ b/CsvQuery/Csv/CsvSettings.cs
@@ -44,15 +44,28 @@ public CsvSettings(char separator, char quoteEscapeChar, char commentChar, bool?
             this.HasHeader = hasHeader;
         }
 
+
         /// <summary>
         /// Parses a big text blob into rows and columns, using the settings
         /// </summary>
         /// <param name="text">Big blob of text</param>
         /// <returns>Parsed data</returns>
         public List<string[]> Parse(string text)
+        {
+            using (var reader = new StringReader(text))
+            {
+                return Parse(reader);
+            }
+        }
+
+        /// <summary>
+        /// Parses a big text blob into rows and columns, using the settings
+        /// </summary>
+        /// <param name="reader">Big blob of text</param>
+        /// <returns>Parsed data</returns>
+        public List<string[]> Parse(TextReader reader)
         {
             // The actual _parsing_ .NET can handle. Well, VisualBasic anyway...
-            using(var reader = new StringReader(text))
             using (var parser = new Microsoft.VisualBasic.FileIO.TextFieldParser(reader))
             {
                 var errors = new StringBuilder();
diff --git a/CsvQuery/Forms/QueryWindow.cs b/CsvQuery/Forms/QueryWindow.cs
index 6d8b5cb..7258b28 100644
--- a/CsvQuery/Forms/QueryWindow.cs
+++ b/CsvQuery/Forms/QueryWindow.cs
@@ -173,13 +173,14 @@ private void Analyze(bool silent)
             var watch = new DiagnosticTimer();
             var bufferId = NotepadPPGateway.GetCurrentBufferId();
 
-            string text;
+            var textLength = PluginBase.CurrentScintillaGateway.GetTextLength();
+            var text = PluginBase.CurrentScintillaGateway.GetTextRange(0, Math.Min(100000, textLength));
            // var text = PluginBase.CurrentScintillaGateway.GetAllText();
 
-            using (var sr = new StreamReader(ScintillaStreams.StreamAllText(), Encoding.UTF8))
-            {
-                text = sr.ReadToEnd();
-            }
+            //using (var sr = new StreamReader(ScintillaStreams.StreamAllText(), Encoding.UTF8))
+            //{
+            //    text = sr.ReadToEnd();
+            //}
             
 
             watch.Checkpoint("GetText");
@@ -197,10 +198,13 @@ private void Analyze(bool silent)
             }
             watch.Checkpoint("Analyze");
 
-            Parse(csvSettings, watch, text, bufferId);
+            using (var sr = new StreamReader(ScintillaStreams.StreamAllText(), Encoding.UTF8))
+            {
+                Parse(csvSettings, watch, sr, bufferId);
+            }
         }
 
-        private void Parse(CsvSettings csvSettings, DiagnosticTimer watch, string text, IntPtr bufferId)
+        private void Parse(CsvSettings csvSettings, DiagnosticTimer watch, TextReader text, IntPtr bufferId)
         {
             var data = csvSettings.Parse(text);
             watch.Checkpoint("Parse");
@@ -227,10 +231,16 @@ private void Parse(CsvSettings csvSettings, DiagnosticTimer watch, string text,
 
         public void StartParse(CsvSettings settings)
         {
-            StartSomething(() => Parse(settings,
-                new DiagnosticTimer(),
-                PluginBase.CurrentScintillaGateway.GetAllText(),
-                NotepadPPGateway.GetCurrentBufferId()));
+            StartSomething(() =>
+            {
+                using (var sr = new StreamReader(ScintillaStreams.StreamAllText(), Encoding.UTF8))
+                {
+                    Parse(settings,
+                        new DiagnosticTimer(),
+                        sr,
+                        NotepadPPGateway.GetCurrentBufferId());
+                }
+            });
         }
 
         private void Execute(IntPtr bufferId, DiagnosticTimer watch)