richter 00/05/08 11:16:29
Modified: . Tag: Embperl2 embperl.h epdom.c epparse.c
test/html Tag: Embperl2 input.htm
Added: . Tag: Embperl2 epcomp.c epdom.h
Log:
Embperl 2 - DOM
Revision Changes Path
No revision
No revision
1.17.2.3 +0 -25 embperl/embperl.h
Index: embperl.h
===================================================================
RCS file: /home/cvs/embperl/embperl.h,v
retrieving revision 1.17.2.2
retrieving revision 1.17.2.3
diff -u -r1.17.2.2 -r1.17.2.3
--- embperl.h 2000/05/03 14:03:49 1.17.2.2
+++ embperl.h 2000/05/08 18:16:27 1.17.2.3
@@ -189,31 +189,6 @@
extern pid_t nPid ;
-/*
- Node Types
-*/
-
-enum tNodeType
- {
- ntypCDATA = 1,
- ntypTag = 2,
- ntypStartTag = 3,
- ntypEndTag = 4,
- ntypAttr = 5,
- ntypAttrValue = 6,
- } ;
-
-
-
-struct tNode
- {
- enum tNodeType nType ;
- const char * sText ;
- struct tNode * pFirstChild ;
- struct tNode * pLastChild ;
- struct tNode * pSibling ;
- struct tNode * pParent ;
- } ;
1.1.2.4 +348 -34 embperl/Attic/epdom.c
Index: epdom.c
===================================================================
RCS file: /home/cvs/embperl/Attic/epdom.c,v
retrieving revision 1.1.2.3
retrieving revision 1.1.2.4
diff -u -r1.1.2.3 -r1.1.2.4
--- epdom.c 2000/05/04 05:42:02 1.1.2.3
+++ epdom.c 2000/05/08 18:16:27 1.1.2.4
@@ -15,8 +15,178 @@
#include "ep.h"
#include "epmacro.h"
+#include "epdom.h"
+HV * pStringTableHash ; /* Hash to translate strings to index number */
+const char * * pStringTableArray ; /* Array with pointers to strings */
+
+struct tNodeData * * pNodeLookup ;
+struct tNodePad * * pPadLookup ;
+
+int nInitialNodePadSize = 64 ;
+
+int nMemUsage = 0 ;
+int numNodes = 0 ;
+int numAttr = 0 ;
+int numStr = 0 ;
+int numPads = 0 ;
+
+#define dom_malloc(s) (nMemUsage += s, malloc(s))
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* ArrayNew */
+/* */
+/* Create a new dynamic array */
+/* */
+/* ------------------------------------------------------------------------ */
+
+int ArrayNew (/*in*/ const tArray * pArray,
+ /*in*/ int nAdd,
+ /*in*/ int nElementSize)
+
+
+ {
+ struct tArrayCtrl * pNew ;
+
+ if ((pNew = dom_malloc (nAdd * nElementSize + sizeof (struct tArrayCtrl))) ==
NULL)
+ return 0 ;
+
+ *(void * *)pArray = (struct tArray *)(pNew + 1) ;
+ pNew -> nMax = nAdd ;
+ pNew -> nAdd = nAdd ;
+ pNew -> nFill = 0 ;
+ pNew -> nElementSize = nElementSize ;
+
+ return ok ;
+ }
+
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* ArrayAdd */
+/* */
+/* Make space for numElements in Array and return index of first one */
+/* */
+/* ------------------------------------------------------------------------ */
+
+
+int ArrayAdd (/*in*/ const tArray * pArray,
+ /*in*/ int numElements)
+
+ {
+ struct tArrayCtrl * pCtrl = ((struct tArrayCtrl *)(*(void * *)pArray)) - 1 ;
+ int nNdx ;
+
+
+ if (pCtrl -> nFill + numElements > pCtrl -> nMax)
+ {
+ struct tArrayCtrl * pNew ;
+ int nNewMax = pCtrl -> nFill + numElements + pCtrl -> nAdd ;
+
+ if ((pNew = realloc (pCtrl, nNewMax * pCtrl -> nElementSize + sizeof (struct
tArrayCtrl))) == NULL)
+ return 0 ;
+
+ *(void * *)pArray = (struct tArray *)(pNew + 1) ;
+ pNew -> nMax = nNewMax ;
+ pCtrl = pNew ;
+ }
+
+ nNdx = pCtrl -> nFill ;
+ pCtrl -> nFill += numElements ;
+ return nNdx ;
+ }
+
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* String2Ndx */
+/* */
+/* Convert String to an unique index */
+/* */
+/* ------------------------------------------------------------------------ */
+
+
+int String2Ndx (/*in*/ const char * sText,
+ /*in*/ int nLen)
+
+ {
+ SV * * ppSV ;
+ SV * pSVKey ;
+ SV * pSVNdx ;
+ HE * pHEKey ;
+ int nNdx ;
+
+ if ((ppSV = hv_fetch (pStringTableHash, (char *)sText, nLen, 1)) == NULL)
+ return 0 ;
+
+ if (*ppSV != NULL && SvTYPE (*ppSV) == SVt_IV)
+ return SvIV (*ppSV) ;
+
+ /* new string */
+
+ nNdx = ArrayAdd (&pStringTableArray, 1) ;
+
+ pSVNdx = newSViv (nNdx) ;
+ SvREFCNT_inc (pSVNdx) ;
+ pSVKey = newSVpvn ((char *)sText, nLen) ;
+ pHEKey = hv_store_ent (pStringTableHash, pSVKey, pSVNdx, 0) ;
+
+ pStringTableArray[nNdx] = HeKEY (pHEKey) ;
+
+ numStr++ ;
+
+ return nNdx ;
+ }
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* Ndx2String */
+/* */
+/* Get String from index */
+/* */
+/* ------------------------------------------------------------------------ */
+
+const char * Ndx2String (/*in*/ int nNdx)
+
+ {
+ return pStringTableArray[nNdx] ;
+ }
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* DomInit */
+/* */
+/* */
+/* */
+/* ------------------------------------------------------------------------ */
+
+
+int DomInit (void)
+
+ {
+ pStringTableHash = newHV () ;
+
+ ArrayNew (&pStringTableArray, 128, sizeof (char *)) ;
+ String2Ndx ("", 0) ;
+
+ ArrayNew (&pNodeLookup, 4096, sizeof (struct tNodeData *)) ;
+ ArrayAdd (&pNodeLookup, 1) ;
+ pNodeLookup[0] = NULL ;
+ ArrayNew (&pPadLookup, 256, sizeof (struct tNodePad *)) ;
+ ArrayAdd (&pPadLookup, 1) ;
+ pPadLookup[0] = NULL ;
+ }
+
+
+int mydie (char * msg)
+ {
+ puts (msg) ;
+ exit (1) ;
+ }
+
+
/* ------------------------------------------------------------------------
interface Node {
@@ -63,56 +233,200 @@
*/
+/* ------------------------------------------------------------------------ */
+/* */
+/* NewPad */
+/* */
+/* Create a new pad for storing nodes */
+/* */
+/* ------------------------------------------------------------------------ */
+
+static struct tNodePad * NewPad (/*in*/ tIndex xParent)
+
+ {
+ tIndex xNdx = ArrayAdd (&pPadLookup, 1) ;
+ struct tNodePad * pChilds ;
+
+ if ((pPadLookup [xNdx] = pChilds = dom_malloc (nInitialNodePadSize)) == NULL)
+ return NULL ;
+
+ pChilds -> nFill = sizeof (struct tNodePad) ;
+ pChilds -> nMax = nInitialNodePadSize ;
+ pChilds -> numChilds = 0 ;
+ pChilds -> xParent = xParent ;
+ pChilds -> xNdx = xNdx ;
+ pChilds -> xFirst = xNdx ;
+ pChilds -> xNext = 0 ;
+ pChilds -> xPrev = 0 ;
+ pChilds -> xLast = 0 ;
+
+
+ numPads++ ;
+ return pChilds ;
+ }
+
+
/* ------------------------------------------------------------------------ */
/* */
-/* CreateNode */
+/* Node_appendChild */
/* */
-/* */
+/* Append a child node to a parent node */
/* */
/* ------------------------------------------------------------------------ */
+
-int CreateNode (/*i/o*/ register req * r,
- /*in*/ enum tNodeType nType,
- /*in*/ const char * sText,
- /*in*/ int nTextLen,
- /*in*/ struct tNode * pParent,
- /*in*/ int nLevel,
- /*out*/ struct tNode * * pNewNode)
+tNode Node_appendChild (/*i/o*/ register req * r,
+ /*in*/ tNodeType nType,
+ /*in*/ const char * sText,
+ /*in*/ int nTextLen,
+ /*in*/ tNode xParent,
+ /*in*/ int nLevel)
{
- struct tNode * pNode ;
+ struct tNodeData * pParent = pNodeLookup [xParent] ;
+
+ if (r -> bDebug & dbgParse)
+ lprintf (r, "[%d]PARSE: AddNode: +%02d parent=%d %*s type=%d text=%*.*s\n",
r -> nPid, nLevel, xParent, nLevel * 2, "", nType, nTextLen, nTextLen, sText) ;
- if ((pNode = _malloc (r, sizeof (struct tNode))) == NULL)
- return rcOutOfMemory ;
+ if (nType == ntypAttr)
+ {
+ struct tNodePad * pPad = (struct tNodePad * )(((tUInt8 *)pParent) -
pParent -> nPadNdx) ;
+ struct tAttrData * pNew = ((struct tAttrData * )(pParent + 1)) + pParent ->
numAttr ;
+ if (((tUInt8 *)pNew) - ((tUInt8 *)pPad) != pPad -> nFill)
+ { /* not last child in pad -> make room */
+ mydie ("not last child in pad -> make room") ;
- pNode -> nType = nType ;
- pNode -> sText = _ep_memdup (r, sText, nTextLen) ;
- pNode -> pParent = pParent ;
- pNode -> pSibling = NULL ;
- pNode -> pFirstChild = NULL ;
- pNode -> pLastChild = NULL ;
- if (pParent)
+ }
+ if (((tUInt8 *)pNew) - ((tUInt8 *)pPad) + sizeof (struct tAttrData) > pPad ->
nMax)
+ { /* pad full -> move into new pad */
+ struct tNodePad * pNext = NewPad (pPad -> xParent) ;
+ int nNodeSize = ((tUInt8 *)pNew) - ((tUInt8 *)pParent) ;
+ struct tNodeData * pNewParent = (struct tNodeData * )(pNext + 1) ;
+ lprintf (r, "[%d]PARSE: NewPad Extent Attr: %d -> %d\n", r -> nPid, pPad
-> xParent, pNext -> xNdx) ;
+
+ pNext -> xPrev = pPad -> xNdx ;
+ pPad -> xNext = pNext -> xNdx ;
+ pPadLookup[pPad->xFirst] -> xLast = pNext -> xNdx ;
+ pNext -> xFirst = pPad->xFirst ;
+
+ memcpy (pNewParent, pParent, nNodeSize) ;
+ pNewParent -> nPadNdx = pNext -> nFill ;
+ pNodeLookup [xParent] = pParent = pNewParent ;
+ pNew = ((struct tAttrData * )(pParent + 1)) + pParent -> numAttr ;
+ pPad -> nFill -= nNodeSize ;
+ pNext -> nFill += nNodeSize ;
+ pPad = pNext ;
+ }
+
+ pNew -> nName = String2Ndx (sText, nTextLen) ;
+ pNew -> nValue = 0 ;
+ pParent -> numAttr++ ;
+ pPad -> nFill += sizeof (struct tAttrData) ;
+ numAttr++ ;
+ return 1 ;
+ }
+ else if (nType == ntypAttrValue)
+ {
+ struct tAttrData * pNew = ((struct tAttrData * )pParent + 1) + (pParent ->
numAttr - 1);
+ pNew -> nValue = String2Ndx (sText, nTextLen) ;
+ return 1 ;
+ }
+ else
{
- if (pParent -> pLastChild)
+ struct tNodePad * pChilds ;
+ struct tNodeData * pNew ;
+ int xNdx ;
+
+ if (!pParent || !pParent -> xChilds)
{
- pParent -> pLastChild -> pSibling = pNode ;
- pParent -> pLastChild = pNode ;
+ pChilds = NewPad (xParent) ;
+ lprintf (r, "[%d]PARSE: NewPad NewChild: %d -> %d\n", r -> nPid, xParent,
pChilds -> xNdx) ;
+
+ if (pParent)
+ pParent -> xChilds = pChilds -> xNdx ;
}
- else
+ else
{
- pParent -> pFirstChild = pNode ;
- pParent -> pLastChild = pNode ;
+ struct tNodePad * pFirst = pChilds = pPadLookup [pParent -> xChilds] ;
+ if (pChilds -> xLast)
+ pChilds = pPadLookup [pChilds -> xLast] ;
+
+ if (pChilds -> nFill + sizeof (struct tNodeData) > pChilds -> nMax)
+ { /* pad full -> make room */
+ struct tNodePad * pNext = NewPad (xParent) ;
+ lprintf (r, "[%d]PARSE: NewPad Extent Childs: %d -> %d\n", r -> nPid,
xParent, pNext -> xNdx) ;
+
+ pNext -> xPrev = pChilds -> xNdx ;
+ pChilds -> xNext = pNext -> xNdx ;
+ pFirst -> xLast = pNext -> xNdx ;
+ pNext -> xFirst = pFirst -> xFirst ;
+
+ pChilds = pNext ;
+ }
}
-
+
+ pNew = (struct tNodeData *)(((tUInt8 *)pChilds) + pChilds -> nFill) ;
+
+ xNdx = ArrayAdd (&pNodeLookup, 1) ;
+ pNodeLookup[xNdx] = pNew ;
+
+ pNew -> nText = String2Ndx (sText, nTextLen) ;
+ pNew -> nType = nType ;
+ pNew -> numAttr = 0 ;
+ pNew -> xNdx = xNdx ;
+ pNew -> xChilds = 0 ;
+ pNew -> nPadNdx = pChilds -> nFill ;
+
+ pChilds -> numChilds++ ;
+ pChilds -> nFill += sizeof (struct tNodeData) ;
+ numNodes++ ;
+ return xNdx ;
}
-
- if (r -> bDebug & dbgParse)
- lprintf (r, "[%d]PARSE: AddNode: +%02d %*s type=%d text=%s\n", r -> nPid,
nLevel, nLevel * 2, "", nType, pNode -> sText) ;
-
- *pNewNode = pNode ;
-
- return ok ;
- }
\ No newline at end of file
+ }
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* Node_parentNode */
+/* */
+/* Get parent node */
+/* */
+/* ------------------------------------------------------------------------ */
+
+
+tNode Node_parentNode (/*in*/ tNode xNode)
+
+ {
+ struct tNodeData * pNode = pNodeLookup [xNode] ;
+ struct tNodePad * pPad = (struct tNodePad * )(((tUInt8 *)pNode) -
pNode -> nPadNdx) ;
+ lprintf (pCurrReq, "[%d]PARSE: parentNode: %d -> %d\n", pCurrReq -> nPid,
xNode, pPad -> xParent) ;
+ return pPad -> xParent ;
+ }
+
+
+/* ------------------------------------------------------------------------ */
+/* */
+/* Node_nodeName */
+/* */
+/* Get name of node */
+/* */
+/* ------------------------------------------------------------------------ */
+
+ int i ;
+ struct tNodePad * p ;
+ const char * s ;
+
+const char * Node_nodeName (/*in*/ tNode xNode)
+
+ {
+ i = pNodeLookup [xNode] -> nText ;
+ p = pPadLookup [pNodeLookup [xNode] -> xChilds] ;
+ s = pStringTableArray[pNodeLookup [xNode] -> nText] ;
+ return s ;
+ }
+
+
+
+
1.1.2.7 +38 -23 embperl/Attic/epparse.c
Index: epparse.c
===================================================================
RCS file: /home/cvs/embperl/Attic/epparse.c,v
retrieving revision 1.1.2.6
retrieving revision 1.1.2.7
diff -u -r1.1.2.6 -r1.1.2.7
--- epparse.c 2000/05/04 05:42:03 1.1.2.6
+++ epparse.c 2000/05/08 18:16:28 1.1.2.7
@@ -15,6 +15,7 @@
#include "ep.h"
#include "epmacro.h"
+#include "epdom.h"
typedef unsigned char tCharMap [256/(sizeof(unsigned char)*8)] ;
@@ -310,7 +311,7 @@
const char * pParentContains,
enum tNodeType nCDataType,
const char * pParentNodeName,
- struct tNode * pParentNode,
+ tNode xParentNode,
int level)
{
@@ -319,18 +320,18 @@
int nEndText = sEndText?strlen (sEndText):0 ;
char * pCurr = *ppCurr ;
char * pCurrStart = pCurr ;
- struct tNode *pNewNode ;
+ tNode xNewNode ;
int rc ;
while (pCurr < pEnd)
{
+ struct tToken * pToken = NULL ;
if (pStartChars [*pCurr >> 3] & 1 << (*pCurr & 7))
{
- struct tToken * pToken = NULL ;
- struct tTokenTable * pNextTokenTab = pTokenTable ;
- const char * pNodeName = NULL ;
- enum tNodeType nNodeType = 0 ;
- char * pCurrTokenStart = pCurr ;
+ struct tTokenTable * pNextTokenTab = pTokenTable ;
+ const char * pNodeName = NULL ;
+ enum tNodeType nNodeType = 0 ;
+ char * pCurrTokenStart = pCurr ;
do
@@ -389,8 +390,8 @@
if (pCurrStart < pCurrTokenStart)
{
if (nCDataType)
- if ((rc = CreateNode (r, nCDataType, pCurrStart,
pCurrTokenStart - pCurrStart, pParentNode, level, &pNewNode)) != ok)
- return rc ;
+ if (!(xNewNode = Node_appendChild (r, nCDataType, pCurrStart,
pCurrTokenStart - pCurrStart, xParentNode, level)))
+ return 1 ;
pCurrStart = pCurrTokenStart ;
}
@@ -403,42 +404,48 @@
if (pEndCurr)
pCurr = pEndCurr + strlen (pToken -> sEndText) ;
level-- ;
- pParentNode = pParentNode -> pParent ;
+ xParentNode = Node_parentNode (xParentNode) ;
}
else
{
- if ((rc = CreateNode (r, pToken -> nNodeType, pNodeName, strlen
(pNodeName), pParentNode, level, &pNewNode)) != ok)
+ if (!(xNewNode = Node_appendChild (r, pToken -> nNodeType,
pNodeName, strlen (pNodeName), xParentNode, level)))
return rc ;
if (pInside = pToken -> pInside)
{
- ParseTokens (r, &pCurr, pEnd, pInside, pToken -> sEndText,
pToken -> pContains, pToken -> nCDataType, pNodeName, pNewNode, level+1) ;
+ ParseTokens (r, &pCurr, pEnd, pInside, pToken -> sEndText,
pToken -> pContains, pToken -> nCDataType, pNodeName, xNewNode, level+1) ;
}
else
{
char * pEndCurr ;
unsigned char * pContains ;
+ int nSkip ;
if ((pContains = pToken -> pContains))
{
pEndCurr = pCurr ;
while (pContains [*pEndCurr >> 3] & (1 << (*pEndCurr & 7)))
pEndCurr++ ;
+ nSkip = 0 ;
}
else
+ {
pEndCurr = strstr (pCurr, pToken -> sEndText) ;
+ nSkip = strlen (pToken -> sEndText) ;
+ }
+
if (pEndCurr)
{
if (pEndCurr - pCurr && pToken -> nCDataType)
- if ((rc = CreateNode (r, pToken -> nCDataType, pCurr,
pEndCurr - pCurr, pNewNode, level+1, &pNewNode)) != ok)
- return rc ;
- pCurr = pEndCurr + strlen (pToken -> sEndText) ;
+ if (!(xNewNode = Node_appendChild (r, pToken ->
nCDataType, pCurr, pEndCurr - pCurr, xNewNode, level+1)))
+ return 1 ;
+ pCurr = pEndCurr + nSkip ;
}
}
if (pToken -> nNodeType == ntypStartTag)
{
level++ ;
- pParentNode = pNewNode ;
+ xParentNode = xNewNode ;
}
}
pCurrStart = pCurr ;
@@ -447,20 +454,20 @@
if (pParentContains && ((pParentContains [*pCurr >> 3] & 1 << (*pCurr & 7))
== 0) )
{
if (pCurr - pCurrStart && nCDataType)
- if ((rc = CreateNode (r, nCDataType, pCurrStart, pCurr - pCurrStart,
pParentNode, level, &pNewNode)) != ok)
- return rc ;
+ if (!(xNewNode = Node_appendChild (r, nCDataType, pCurrStart, pCurr -
pCurrStart, xParentNode, level)))
+ return 1 ;
break ;
}
else if (sEndText == NULL ||
(*pCurr == *sEndText && strncmp (pCurr, sEndText, nEndText) == 0))
{
if (pCurr - pCurrStart && nCDataType)
- if ((rc = CreateNode (r, nCDataType, pCurrStart, pCurr - pCurrStart,
pParentNode, level, &pNewNode)) != ok)
- return rc ;
+ if (!(xNewNode = Node_appendChild (r, nCDataType, pCurrStart, pCurr -
pCurrStart, xParentNode, level)))
+ return 1 ;
pCurr += nEndText ;
break ;
}
- else
+ else if (!pToken)
pCurr++ ;
}
@@ -489,8 +496,10 @@
char * sTokenHash = "HTML::Embperl::Tokens::Main" ;
HV * pTokenHash ;
int rc ;
-
+ tNode xDocNode ;
+ DomInit () ;
+
if ((pTokenHash = perl_get_hv ((char *)sTokenHash, TRUE)) == NULL)
{
return rcHashError ;
@@ -500,8 +509,14 @@
LogError (r, rc) ;
return rc ;
}
+
+ if (!(xDocNode = Node_appendChild (r, ntypTag, "attr", 3, 0, 0)))
+ return 1 ;
- return ParseTokens (r, &pStart, pEnd, &pTable, "", NULL, ntypCDATA, "root",
NULL, 0) ;
+ if (!(xDocNode = Node_appendChild (r, ntypTag, "doc", 3, 0, 0)))
+ return 1 ;
+
+ return ParseTokens (r, &pStart, pEnd, &pTable, "", NULL, ntypCDATA, "root",
xDocNode, 0) ;
}
No revision
No revision
1.1.2.1 +26 -0 embperl/Attic/epcomp.c
1.1.2.1 +100 -0 embperl/Attic/epdom.h
No revision
No revision
1.9.2.3 +1 -0 embperl/test/html/input.htm
Index: input.htm
===================================================================
RCS file: /home/cvs/embperl/test/html/input.htm,v
retrieving revision 1.9.2.2
retrieving revision 1.9.2.3
diff -u -r1.9.2.2 -r1.9.2.3
--- input.htm 2000/05/04 05:42:06 1.9.2.2
+++ input.htm 2000/05/08 18:16:28 1.9.2.3
@@ -1,3 +1,4 @@
+<table><table></table></table>
<html>
<head>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
