From 31e7708fd4affa327a7ae701f0e07cd34b1e08e7 Mon Sep 17 00:00:00 2001
From: "Unknown (roddey)" <dev-null@apache.org>
Date: Thu, 6 Apr 2000 23:50:38 +0000
Subject: [PATCH] Now the low level formatter handles doing char refs for
unrepresentable chars (in addition to the replacement char style already
done.)
git-svn-id: https://svn.apache.org/repos/asf/xerces/c/trunk@172031 13f79535-47bb-0310-9956-ffa450edef68
---
src/framework/XMLFormatter.cpp | 97 ++++++++++++++++++++++++++++++++++
src/framework/XMLFormatter.hpp | 11 ++++
2 files changed, 108 insertions(+)
diff --git a/src/framework/XMLFormatter.cpp b/src/framework/XMLFormatter.cpp
index 9651f56ce..d509146e0 100644
--- a/src/framework/XMLFormatter.cpp
+++ b/src/framework/XMLFormatter.cpp
@@ -56,6 +56,11 @@
/**
* $Log$
+ * Revision 1.4 2000/04/06 23:50:38 roddey
+ * Now the low level formatter handles doing char refs for
+ * unrepresentable chars (in addition to the replacement char style
+ * already done.)
+ *
* Revision 1.3 2000/04/06 19:09:21 roddey
* Some more improvements to output formatting. Now it will correctly
* handle doing the 'replacement char' style of dealing with chars
@@ -269,6 +274,17 @@ XMLFormatter::formatBuf(const XMLCh* const toFormat
const UnRepFlags actualUnRep = (unrepFlags == DefaultUnRep)
? fUnRepFlags : unrepFlags;
+ //
+ // If the actual unrep action is that they want to provide char refs
+ // for unrepresentable chars, then this one is a much more difficult
+ // one to do cleanly, and we handle it separately.
+ //
+ if (actualUnRep == UnRep_CharRef)
+ {
+ specialFormat(toFormat, count, actualEsc);
+ return;
+ }
+
//
// Use that to figure out what I should pass to the transcoder. If we
// are doing character references or failing for unrepresentable chars,
@@ -428,6 +444,7 @@ XMLFormatter::formatBuf(const XMLCh* const toFormat
srcPtr++;
}
}
+
}
}
}
@@ -564,3 +581,83 @@ const XMLByte* XMLFormatter::getQuoteRef()
memcpy(fQuoteRef, fTmpBuf, outBytes + 1);
return fQuoteRef;
}
+
+
+void XMLFormatter::specialFormat(const XMLCh* const toFormat
+ , const unsigned int count
+ , const EscapeFlags escapeFlags)
+{
+ //
+ // We have to check each character and see if it could be represented.
+ // As long as it can, we just keep up with where we started and how
+ // many chars we've checked. When we hit an unrepresentable one, we
+ // stop, transcode everything we've collected, then start handling
+ // the unrepresentables via char refs. We repeat this until we get all
+ // the chars done.
+ //
+ const XMLCh* srcPtr = toFormat;
+ const XMLCh* endPtr = toFormat + count;
+
+ // Set up the common part of the buffer that we build char refs into
+ XMLCh tmpBuf[32];
+ tmpBuf[0] = chAmpersand;
+ tmpBuf[1] = chPound;
+ tmpBuf[2] = chLatin_x;
+
+ while (srcPtr < endPtr)
+ {
+ const XMLCh* tmpPtr = srcPtr;
+ while (tmpPtr < endPtr)
+ {
+ if (fXCoder->canTranscodeTo(*tmpPtr))
+ tmpPtr++;
+ else
+ break;
+ }
+
+ if (tmpPtr > srcPtr)
+ {
+ // We got at least some chars that can be done normally
+ formatBuf
+ (
+ srcPtr
+ , tmpPtr - srcPtr
+ , escapeFlags
+ , XMLFormatter::UnRep_Fail
+ );
+
+ // Update the source pointer to our new spot
+ srcPtr = tmpPtr;
+ }
+ else
+ {
+ //
+ // We hit something unrepresentable. So continue forward doing
+ // char refs until we hit something representable again or the
+ // end of input.
+ //
+ while (srcPtr < endPtr)
+ {
+ // Build a char ref for the current char
+ XMLString::binToText(*srcPtr, &tmpBuf[3], 8, 16);
+ const unsigned int bufLen = XMLString::stringLen(tmpBuf);
+ tmpBuf[bufLen] = chSemiColon;
+ tmpBuf[bufLen+1] = chNull;
+
+ // And now call recursively back to our caller to format this
+ formatBuf
+ (
+ tmpBuf
+ , bufLen + 1
+ , XMLFormatter::NoEscapes
+ , XMLFormatter::UnRep_Fail
+ );
+
+ // Move up the source pointer and break out if needed
+ srcPtr++;
+ if (fXCoder->canTranscodeTo(*srcPtr))
+ break;
+ }
+ }
+ }
+}
diff --git a/src/framework/XMLFormatter.hpp b/src/framework/XMLFormatter.hpp
index 6d7ab02fb..3e1722854 100644
--- a/src/framework/XMLFormatter.hpp
+++ b/src/framework/XMLFormatter.hpp
@@ -56,6 +56,11 @@
/*
* $Log$
+ * Revision 1.4 2000/04/06 23:50:38 roddey
+ * Now the low level formatter handles doing char refs for
+ * unrepresentable chars (in addition to the replacement char style
+ * already done.)
+ *
* Revision 1.3 2000/04/06 19:09:21 roddey
* Some more improvements to output formatting. Now it will correctly
* handle doing the 'replacement char' style of dealing with chars
@@ -216,6 +221,12 @@ private :
const XMLByte* getLTRef();
const XMLByte* getQuoteRef();
+ void specialFormat
+ (
+ const XMLCh* const toFormat
+ , const unsigned int count
+ , const EscapeFlags escapeFlags
+ );
// -----------------------------------------------------------------------
--
GitLab