Implemented GetCharacterPointer feature.

8 files changed, 40 insertions, 0 deletions

diff --git a/doc/ScintillaDoc.html b/doc/ScintillaDoc.html
index f3aeec90b..8610dbe91 100644
--- a/doc/ScintillaDoc.html
+++ b/doc/ScintillaDoc.html
@@ -3978,6 +3978,7 @@ struct RangeToFormat {
     <h2 id="DirectAccess">Direct access</h2>
     <code><a class="message" href="#SCI_GETDIRECTFUNCTION">SCI_GETDIRECTFUNCTION</a><br />
      <a class="message" href="#SCI_GETDIRECTPOINTER">SCI_GETDIRECTPOINTER</a><br />
+     <a class="message" href="#SCI_GETCHARACTERPOINTER">SCI_GETCHARACTERPOINTER</a><br />
     </code>
 
     <p>On Windows, the message-passing scheme used to communicate between the container and
@@ -4022,6 +4023,25 @@ sptr_t CallScintilla(unsigned int iMessage, uptr_t wParam, sptr_t lParam){
     this once for each Scintilla window you create. When you call the direct function, you must
     pass in the direct pointer associated with the target window.</p>
 
+    <p><b id="SCI_GETCHARACTERPOINTER">SCI_GETCHARACTERPOINTER</b><br />
+     This moves the gap within Scintilla so that the text of the document is stored consecutively
+     and ensures there is a NUL character after the text, then returns a pointer to the first character.
+     Applications may then pass this to a function that accepts a character pointer such as a regular
+     expression search or a parser. The pointer should <em>not</em> be written to as that may desynchoronize
+     the internal state of Scintilla. </p>
+     <p>Since any action in Scintilla may change its internal state
+     this pointer becomes invalid after any call or by allowing user interface activity. The application
+     should reacquire the pointer after making any call or performing any user-interface calls such
+     as modifying a progress indicator.</p>
+     <p>This call takes similar time to inserting a character at the end of the document and this may
+     include moving the document contents. Specifically, all the characters after the document gap
+     are moved to before the gap. This compacted state should persist over calls and user interface
+     actions that do not change the document contents so reacquiring the pointer afterwards is very
+     quick. If this call is used to implement a global replace operation, then each replacement will
+     move the gap so if <code>SCI_GETCHARACTERPOINTER</code> is called after
+     each replacement then the operation will become O(n^2) rather than O(n). Instead, all
+     matches should be found and remembered, then all the replacements performed.</p>
+
     <h2 id="MultipleViews">Multiple views</h2>
 
     <p>A Scintilla window and the document that it displays are separate entities. When you create
diff --git a/include/Scintilla.h b/include/Scintilla.h
index ac32f40d8..90e7669ed 100644
--- a/include/Scintilla.h
+++ b/include/Scintilla.h
@@ -668,6 +668,7 @@ typedef sptr_t (*SciFnDirect)(sptr_t ptr, unsigned int iMessage, uptr_t wParam,
 #define SCI_SETPOSITIONCACHE 2514
 #define SCI_GETPOSITIONCACHE 2515
 #define SCI_COPYALLOWLINE 2519
+#define SCI_GETCHARACTERPOINTER 2520
 #define SCI_STARTRECORD 3001
 #define SCI_STOPRECORD 3002
 #define SCI_SETLEXER 4001
diff --git a/include/Scintilla.iface b/include/Scintilla.iface
index 946da891e..8641c6d91 100644
--- a/include/Scintilla.iface
+++ b/include/Scintilla.iface
@@ -1800,6 +1800,10 @@ get int GetPositionCache=2515(,)
 # Copy the selection, if selection empty copy the line with the caret
 fun void CopyAllowLine=2519(,)
 
+# Compact the document buffer and return a read-only pointer to the
+# characters in the document.
+get int GetCharacterPointer=2520(,)
+
 # Start notifying the container of all key presses and commands.
 fun void StartRecord=3001(,)
 
diff --git a/src/CellBuffer.cxx b/src/CellBuffer.cxx
index 85bf43552..0e9ae6950 100644
--- a/src/CellBuffer.cxx
+++ b/src/CellBuffer.cxx
@@ -587,6 +587,10 @@ char CellBuffer::StyleAt(int position) {
 	return style.ValueAt(position);
 }
 
+const char *CellBuffer::BufferPointer() { 
+	return substance.BufferPointer();
+}
+
 // The char* returned is to an allocation owned by the undo history
 const char *CellBuffer::InsertString(int position, const char *s, int insertLength, bool &startSequence) {
 	char *data = 0;
diff --git a/src/CellBuffer.h b/src/CellBuffer.h
index 4f654a8fd..4b83f48e0 100644
--- a/src/CellBuffer.h
+++ b/src/CellBuffer.h
@@ -171,6 +171,7 @@ public:
 	char CharAt(int position) const;
 	void GetCharRange(char *buffer, int position, int lengthRetrieve);
 	char StyleAt(int position);
+	const char *BufferPointer();
 
 	int Length() const;
 	void Allocate(int newSize);
diff --git a/src/Document.h b/src/Document.h
index a36c4aafe..04bbd5c74 100644
--- a/src/Document.h
+++ b/src/Document.h
@@ -159,6 +159,7 @@ public:
 	void EndUndoAction() { cb.EndUndoAction(); }
 	void SetSavePoint();
 	bool IsSavePoint() { return cb.IsSavePoint(); }
+	const char *BufferPointer() { return cb.BufferPointer(); }
 
 	int GetLineIndentation(int line);
 	void SetLineIndentation(int line, int indent);
diff --git a/src/Editor.cxx b/src/Editor.cxx
index af60624ff..3b6e7a2c3 100644
--- a/src/Editor.cxx
+++ b/src/Editor.cxx
@@ -7579,6 +7579,9 @@ sptr_t Editor::WndProc(unsigned int iMessage, uptr_t wParam, sptr_t lParam) {
 	case SCI_GETPASTECONVERTENDINGS:
 		return convertPastes ? 1 : 0;
 
+	case SCI_GETCHARACTERPOINTER:
+		return reinterpret_cast<sptr_t>(pdoc->BufferPointer());
+
 	default:
 		return DefWndProc(iMessage, wParam, lParam);
 	}
diff --git a/src/SplitVector.h b/src/SplitVector.h
index 9d62aef72..76f9a8f36 100644
--- a/src/SplitVector.h
+++ b/src/SplitVector.h
@@ -238,6 +238,12 @@ public:
 		DeleteRange(0, lengthBody);
 	}
 
+	T* BufferPointer() {
+		RoomFor(1);
+		GapTo(lengthBody);
+		body[lengthBody] = 0;
+		return body;
+	}
 };
 
 #endif