#include <stack>\r
#include "XMLTree.h"\r
#include "timings.h"\r
-\r
+#include <errno.h>\r
// functions to convert tag positions to the corresponding tree node and viceversa. \r
// These are implemented in order to be able to change the tree and Tags representations, \r
// without affecting the code so much.\r
\r
static treeNode fast_next_sibling(bp* Par,treeNode x)\r
{\r
- x = fwd_excess(Par,x,0);\r
+ x = fast_find_close(Par,x)+1;\r
return (fast_inspect(Par,x) == OP) ? x : NULLT;\r
}\r
\r
//WARNING this uses directly the underlying implementation for plain text\r
\r
\r
-\r
void XMLTree::Print(int fd,treeNode x, bool no_text){\r
\r
int newfd = dup(fd);\r
stream = fdopen(newfd,"wa");\r
+ if (stream == 0){\r
+ perror(NULL);\r
+ return;\r
+ };\r
+\r
if (buffer == 0)\r
buffer = new string();\r
\r
uchar * tagstr;\r
range r = DocIds(x);\r
treeNode first_idx;\r
- treeNode first_text = (tag == PCDATA_TAG_ID ? x : TaggedDescendant(x,PCDATA_TAG_ID));\r
- treeNode first_att = NULLT;//TaggedDesc(x,ATTRIBUTE_DATA_TAG_ID);\r
+ treeNode first_text = (tag == PCDATA_TAG_ID ? x : ParentNode(r.min-1));\r
+ treeNode first_att = NULLT;\r
\r
if (first_att == NULLT)\r
first_idx = first_text;\r
uchar * current_text=NULL;\r
if (first_idx != NULLT)\r
current_text = GetText(MyText(first_idx));\r
- int read = 0;\r
+ size_t read = 0;\r
std::vector<uchar*> st;\r
while (n <= fin){\r
if (fast_inspect(Par,n)){\r
if (tag == PCDATA_TAG_ID ) { \r
- // myfputs((const char*) (GetText(MyTextUnsafe(n))),fp);\r
+\r
if (no_text)\r
myfputs("<$/>",fp);\r
else{\r
- read = fprintf(fp,"%s",(const char*) current_text);\r
- current_text += (read + 1);\r
- }\r
+ read = myfprintf((const char*) current_text, fp);\r
+ current_text += (read + 1);\r
+ };\r
n+=2; // skip closing $\r
tag = Tag(n);\r
\r
while (fast_inspect(Par,n)){\r
if (no_text) {\r
myfputc('<',fp);\r
- const uchar * tmp = &(GetTagNameByRef(Tag(n))[3]);\r
- myfputs((const char*) tmp,fp);\r
+ myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
myfputc('>',fp);\r
myfputs("<$@/></",fp);\r
- myfputs((const char*) tmp,fp);\r
+ myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
myfputc('>',fp);\r
n+= 4;\r
}\r
myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
n++;\r
myfputs("=\"",fp);\r
- read = fprintf(fp,"%s",(const char*) current_text);\r
+ read = myfprintf((const char*) current_text,fp);\r
current_text += (read + 1);\r
- //myfputs((const char*) GetText(MyTextUnsafe(n)),fp);\r
myfputc('"',fp);\r
n+=3;\r
}\r
}while (!fast_inspect(Par,n) && !st.empty());\r
tag=Tag(n);\r
};\r
- //myfputc('\n',fp);\r
+ myfputc('\n',fp);\r
mybufferflush(fp);\r
- fflush(fp);\r
+ //fflush(fp);\r
fclose(fp);\r
}\r
fputs(buffer->c_str(), fp);\r
buffer->clear();\r
}\r
+\r
+ size_t myfprintf(const char* s, FILE * fp){\r
+ if (s == NULL)\r
+ return 0;\r
+ size_t i = buffer->size();\r
+ buffer->append(s);\r
+ size_t j = buffer->size();\r
+ if (buffer->size() >= 100000){\r
+ fputs(buffer->c_str(),fp);\r
+ buffer->clear();\r
+ };\r
+ return (j-i);\r
+ }\r
+\r
+ void PrintNode(treeNode n, int fd);\r
/** Data structure constructors */\r
XMLTree(){ buffer = 0;};\r
\r
uchar* GetText(DocID d) {\r
\r
uchar * s = Text->GetText(d);\r
- return (s[0] == 1 ? (uchar*)"" : s);\r
+ return (s[0] == 1 ? (s+1) : s);\r
}\r
\r
/** GetText(i, j): returns the texts corresponding to documents with\r