-#include "basics.h"\r
+#include "common.h"\r
#include "XMLTree.h"\r
#include "timings.h"\r
#include <errno.h>\r
using std::min;\r
using std::string;\r
\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
+// 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
// Current implementation corresponds to balanced-parentheses representation for\r
// the tree, and storing 2 tags per tree node (opening and closing tags).\r
\r
-// tag position -> tree node\r
-static treeNode tagpos2node(int t) \r
- {\r
- return (treeNode) t;\r
- }\r
\r
static int bits8 (int t ) {\r
int r = bits(t);\r
return 8;\r
else if (r <= 16)\r
return 16;\r
- else \r
+ else\r
return r;\r
}\r
\r
-// tree node -> tag position\r
-static int node2tagpos(treeNode x) \r
-{\r
- return (int)x;\r
-}\r
-\r
-static int fast_find_close(bp *b,int s)\r
-{\r
- return fwd_excess(b,s,-1);\r
-}\r
-\r
-static int fast_inspect(bp* Par,treeNode i)\r
-{\r
- int j,l;\r
- j = i >> logD;\r
- l = i & (D-1);\r
- return (Par->B[j] >> (D-1-l)) & 1;\r
-}\r
-\r
-static treeNode fast_first_child(bp *Par, treeNode x)\r
-{\r
- x = x+1;\r
- return (fast_inspect(Par,x) == OP) ? x : NULLT;\r
-}\r
-\r
-static treeNode fast_next_sibling(bp* Par,treeNode x)\r
-{\r
- x = fast_find_close(Par,x)+1;\r
- return (fast_inspect(Par,x) == OP) ? x : NULLT;\r
-}\r
\r
\r
static treeNode fast_sibling(bp* Par,treeNode x,TagType tag){\r
\r
if (tag == PCDATA_TAG_ID){\r
x = x+2;\r
- return fast_inspect(Par,x)==OP ? x : NULLT;\r
- } else return fast_next_sibling(Par,x);\r
+ return bp_inspect(Par,x)==OP ? x : NULLT;\r
+ } else return bp_next_sibling(Par,x);\r
\r
}\r
\r
-static bool fast_isleaf(bp* Par,treeNode x){\r
- return (fast_inspect(Par,x+1) == CP ? true : false);\r
-}\r
+\r
\r
\r
inline uint get_field_no_power(uint *A, uint len, uint index) {\r
- \r
+\r
register uint i=index*len/W, j=index*len-W*i;\r
return (j+len <= W) ? (A[i] << (W-j-len)) >> (W-len) : (A[i] >> j) | (A[i+1] << (WW-j-len)) >> (W-len);\r
\r
\r
}\r
\r
-inline bool fast_is_ancestor(bp * Par,treeNode x,treeNode y){\r
- if (x > y) \r
- return false;\r
- else\r
- return (x==0) || (y <= fast_find_close(Par,x));\r
-}\r
\r
\r
-XMLTree::XMLTree( pb * const par, uint npar, vector<string> * const TN, TagIdMap * const tim, \r
+\r
+XMLTree::XMLTree( pb * const par, uint npar, vector<string> * const TN, TagIdMap * const tim,\r
uint *empty_texts_bmp, TagType *tags,\r
- TextCollection * const TC, bool dis_tc)\r
+ TextCollection * const TC, bool dis_tc,\r
+ TextCollectionBuilder::index_type_t _index_type )\r
{\r
buffer = 0;\r
- // creates the data structure for the tree topology\r
- Par = (bp *)umalloc(sizeof(bp));\r
- STARTTIMER();\r
- bp_construct(Par, npar, (pb*) par, OPT_DEGREE|0);\r
- STOPTIMER(Building);\r
- PRINTTIME("Building parenthesis struct", Building);\r
- STARTTIMER();\r
+ print_stack = 0;\r
+ // creates the data structure for the tree topology\r
+ STARTTIMER();\r
+ Par = bp_construct(npar, (pb*) par, OPT_FAST_PREORDER_SELECT | OPT_DEGREE|0);\r
+ STOPTIMER(Building);\r
+ PRINTTIME("Building parenthesis struct", Building);\r
+ STARTTIMER();\r
+\r
\r
- \r
// creates structure for tags\r
\r
TagName = (vector<string>*)TN;\r
tIdMap = (TagIdMap *) tim;\r
- \r
+\r
uint max_tag = TN->size() - 1;\r
- \r
- \r
+\r
+\r
static_bitsequence_builder *bmb = new static_bitsequence_builder_sdarray();\r
alphabet_mapper *am = new alphabet_mapper_none();\r
Tags = new static_sequence_bs((uint*)tags,npar,am,bmb);\r
- \r
+\r
//cout << "Tags test: " << Tags->test((uint*)tags,npar) << endl;\r
\r
//Ensures that for small tag numbers, we are on an 8bit boundary.\r
tags_fix = new uint[uint_len(tags_blen,tags_len)];\r
for(uint i=0;i<(uint)npar;i++)\r
set_field(tags_fix,tags_blen,i,tags[i]);\r
- delete bmb; \r
+ delete bmb;\r
free(tags);\r
tags = NULL;\r
\r
STOPTIMER(Building);\r
PRINTTIME("Building Tag Structure", Building);\r
- \r
+\r
Text = (TextCollection*) TC;\r
\r
\r
free(empty_texts_bmp);\r
empty_texts_bmp = NULL;\r
\r
- \r
+\r
disable_tc = dis_tc;\r
- stream = NULL;\r
- stream_fd = 0;\r
+ text_index_type = _index_type;\r
std::cerr << "Number of distinct tags " << TagName->size() << "\n";\r
//std::cerr.flush();\r
}\r
\r
\r
// ~XMLTree: frees memory of XML tree.\r
-XMLTree::~XMLTree() \r
- { \r
+XMLTree::~XMLTree()\r
+ {\r
int i;\r
\r
- destroyTree(Par);\r
- free(Par); // frees the memory of struct Par\r
- \r
+ bp_delete(Par);\r
+ Par = NULL;\r
+\r
delete tIdMap;\r
tIdMap = NULL;\r
- \r
+\r
delete TagName;\r
TagName = NULL;\r
- \r
+\r
delete Tags;\r
Tags = NULL;\r
\r
- delete Text; \r
+ delete Text;\r
Text = NULL;\r
\r
delete EBVector;\r
EBVector = NULL;\r
- if (stream != NULL){\r
- fclose(stream);\r
- stream = NULL;\r
- stream_fd = 0;\r
- };\r
\r
}\r
\r
\r
-void XMLTree::print_stats() \r
+void XMLTree::print_stats()\r
{\r
uint total_space = Tags->size()+sizeof(static_sequence*);\r
total_space += sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len));\r
<< " *total* " << total_space << endl;\r
}\r
\r
-// Save: saves XML tree data structure to file. \r
-void XMLTree::Save(int fd) \r
+// Save: saves XML tree data structure to file.\r
+void XMLTree::Save(int fd, char * name)\r
{\r
FILE *fp;\r
- char filenameaux[1024];\r
int i;\r
\r
fp = fdopen(fd, "wa");\r
ufwrite(&ntags, sizeof(int), 1, fp);\r
for (i = 0; i<ntags;i++)\r
fprintf(fp, "%s\n",TagName->at(i).c_str());\r
- \r
+\r
\r
// stores the tags\r
Tags->save(fp);\r
ufwrite(&tags_len,sizeof(uint),1,fp);\r
ufwrite(tags_fix,sizeof(uint),uint_len(tags_blen,tags_len),fp);\r
\r
- // flags \r
+ // flags\r
ufwrite(&disable_tc, sizeof(bool),1,fp);\r
- \r
+\r
//text positions\r
EBVector->save(fp);\r
- \r
- // stores the texts \r
+\r
+ // stores the texts\r
if (!disable_tc) {\r
- Text->Save(fp);\r
- };\r
\r
+ ufwrite(&text_index_type, sizeof(TextCollectionBuilder::index_type_t), 1, fp);\r
\r
- }\r
\r
+ string file(name);\r
+ switch (text_index_type){\r
+ case TextCollectionBuilder::index_type_default:\r
+ file.append(".default");\r
+ break;\r
+ case TextCollectionBuilder::index_type_swcsa:\r
+ file.append(".swcsa");\r
+ break;\r
+ case TextCollectionBuilder::index_type_rlcsa:\r
+ file.append(".rlcsa");\r
+ break;\r
+ };\r
+\r
+ Text->Save(fp, file.c_str());\r
+\r
+\r
+ }\r
+ }\r
\r
// Load: loads XML tree data structure from file. Returns\r
// a pointer to the loaded data structure\r
-XMLTree *XMLTree::Load(int fd, bool load_tc,int sample_factor) \r
+XMLTree *XMLTree::Load(int fd, bool load_tc,int sample_factor, char * name)\r
{\r
\r
FILE *fp;\r
XML_Tree = new XMLTree();\r
STARTTIMER();\r
// Load the tree structure\r
- XML_Tree->Par = (bp *)umalloc(sizeof(bp));\r
-\r
- loadTree(XML_Tree->Par, fp); \r
+ XML_Tree->Par = loadTree(fp);\r
STOPTIMER(Loading);\r
PRINTTIME("Loading parenthesis struct", Loading);\r
STARTTIMER();\r
XML_Tree->tIdMap = new std::unordered_map<std::string,int>();\r
std::string s;\r
int ntags;\r
- \r
+\r
// Load the tag names\r
ufread(&ntags, sizeof(int), 1, fp);\r
\r
throw "Cannot read tag list";\r
s = buffer;\r
// remove the trailing \n\r
- s.erase(s.size()-1); \r
- XML_Tree->TagName->push_back(s); \r
+ s.erase(s.size()-1);\r
+ XML_Tree->TagName->push_back(s);\r
XML_Tree->tIdMap->insert(std::make_pair(s,i));\r
- \r
+\r
};\r
STOPTIMER(Loading);\r
PRINTTIME("Loading tag names struct", Loading);\r
// loads the tag structure\r
XML_Tree->Tags = static_sequence::load(fp);\r
ufread(&XML_Tree->tags_blen,sizeof(uint),1,fp);\r
- std::cerr << "tags_blen is "<< XML_Tree->tags_blen <<"\n"; \r
+ std::cerr << "tags_blen is "<< XML_Tree->tags_blen <<"\n";\r
ufread(&XML_Tree->tags_len,sizeof(uint),1,fp);\r
XML_Tree->tags_fix = new uint[uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)];\r
ufread(XML_Tree->tags_fix,sizeof(uint),uint_len(XML_Tree->tags_blen,XML_Tree->tags_len),fp);\r
//XML_Tree->Tags->test(seq,XML_Tree->tags_len);\r
//delete [] seq;\r
/// End ugly tests\r
- \r
+\r
STOPTIMER(Loading);\r
std::cerr << (uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)*sizeof(uint))/(1024*1024) << " MB for tag sequence" << std::endl;\r
PRINTTIME("Loading tag struct", Loading);\r
STARTTIMER();\r
\r
// loads the flags\r
- \r
+\r
ufread(&(XML_Tree->disable_tc), sizeof(bool), 1, fp);\r
if (load_tc) {\r
- XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
- //XML_Tree->EBVector = static_bitsequence_sdarray::load(fp);\r
-\r
- STOPTIMER(Loading);\r
- PRINTTIME("Loading text bitvector struct", Loading);\r
- STARTTIMER();\r
-\r
- // Not used \r
- // loads the texts\r
- if (!XML_Tree->disable_tc){\r
- XML_Tree->Text = TextCollection::Load(fp,sample_factor);\r
- }\r
- else XML_Tree->Text = NULL;\r
- STOPTIMER(Loading);\r
- PRINTTIME("Loading TextCollection", Loading);\r
- STARTTIMER(); \r
+ XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);\r
+\r
+ STOPTIMER(Loading);\r
+ PRINTTIME("Loading text bitvector struct", Loading);\r
+ STARTTIMER();\r
+\r
+ // Not used\r
+ // loads the texts\r
+ if (!XML_Tree->disable_tc){\r
+ ufread(&(XML_Tree->text_index_type),\r
+ sizeof(TextCollectionBuilder::index_type_t), 1, fp);\r
+ string file(name);\r
+ switch (XML_Tree->text_index_type){\r
+ case TextCollectionBuilder::index_type_default:\r
+ file.append(".default");\r
+ break;\r
+ case TextCollectionBuilder::index_type_swcsa:\r
+ file.append(".swcsa");\r
+ break;\r
+ case TextCollectionBuilder::index_type_rlcsa:\r
+ file.append(".rlcsa");\r
+ break;\r
+ };\r
+ XML_Tree->Text = TextCollection::Load(fp, file.c_str(), TextCollection::index_mode_default, sample_factor);\r
+\r
+ }\r
+ else XML_Tree->Text = NULL;\r
+ STOPTIMER(Loading);\r
+ PRINTTIME("Loading TextCollection", Loading);\r
+ STARTTIMER();\r
}\r
else {\r
XML_Tree->EBVector = NULL;\r
XML_Tree->disable_tc = true;\r
};\r
\r
- XML_Tree->stream = NULL;\r
- XML_Tree->stream_fd = 0;\r
- \r
+\r
return XML_Tree;\r
}\r
\r
\r
\r
-// SubtreeSize(x): the number of nodes (and attributes) in the subtree of node x.\r
-int XMLTree::SubtreeSize(treeNode x) \r
- {\r
- return subtree_size(Par, x);\r
- }\r
\r
-// SubtreeTags(x,tag): the number of occurrences of tag within the subtree of node x.\r
-int XMLTree::SubtreeTags(treeNode x, TagType tag) \r
+int XMLTree::SubtreeElements(treeNode x)\r
{\r
- if (x == Root())\r
- x = fast_first_child(Par,x);\r
- \r
\r
- int s = x + 2*subtree_size(Par, x) - 1;\r
- \r
- return (Tags->rank(tag, s) - Tags->rank(tag, node2tagpos(x)-1))+1;\r
- }\r
-int XMLTree::SubtreeElements(treeNode x) \r
- {\r
- \r
- int size = subtree_size(Par,x);\r
+ int size = bp_subtree_size(Par, x);\r
if (x == Root()){\r
- x = fast_first_child(Par,x);\r
+ x = bp_first_child(Par,x);\r
size = size - 1;\r
};\r
\r
int s = x + 2*size - 1;\r
int ntext = Tags->rank(PCDATA_TAG_ID, s) - Tags->rank(PCDATA_TAG_ID, node2tagpos(x)-1);\r
size = size - ntext;\r
- treeNode fin = fast_find_close(Par,x);\r
+ treeNode fin = bp_find_close(Par,x);\r
treeNode y = Tags->select_next(ATTRIBUTE_TAG_ID,node2tagpos(x));\r
while (y != NULLT && y < fin){\r
size -= SubtreeSize(y);\r
- y = Tags->select_next(ATTRIBUTE_TAG_ID,node2tagpos(y));\r
+ y = Tags->select_next(ATTRIBUTE_TAG_ID, node2tagpos(y));\r
};\r
- return size; \r
+ return size;\r
}\r
\r
// IsLeaf(x): returns whether node x is leaf or not. In the succinct representation\r
// this is just a bit inspection.\r
-bool XMLTree::IsLeaf(treeNode x) \r
+bool XMLTree::IsLeaf(treeNode x)\r
{\r
NULLT_IF(x==NULLT);\r
- return fast_isleaf(Par, x);\r
- } \r
+ return bp_isleaf(Par, x);\r
+ }\r
\r
// IsAncestor(x,y): returns whether node x is ancestor of node y.\r
-bool XMLTree::IsAncestor(treeNode x, treeNode y) \r
+bool XMLTree::IsAncestor(treeNode x, treeNode y)\r
{\r
- return fast_is_ancestor(Par, x, y);\r
+ return bp_is_ancestor(Par, x, y);\r
}\r
\r
// IsChild(x,y): returns whether node x is parent of node y.\r
-bool XMLTree::IsChild(treeNode x, treeNode y) \r
- {\r
- if (!fast_is_ancestor(Par, x, y)) return false;\r
- return depth(Par, x) == (depth(Par, y) + 1);\r
- }\r
-\r
-// IsFirstChild(x): returns whether node x is the first child of its parent.\r
-bool XMLTree::IsFirstChild(treeNode x)\r
+bool XMLTree::IsChild(treeNode x, treeNode y)\r
{\r
- return ((x != NULLT)&&(x==Root() || prev_sibling(Par,x) == (treeNode)-1));\r
+ if (!bp_is_ancestor(Par, x, y)) return false;\r
+ return bp_depth(Par, x) == (bp_depth(Par, y) + 1);\r
}\r
\r
\r
// NumChildren(x): number of children of node x. Constant time with the data structure\r
// of Sadakane.\r
-int XMLTree::NumChildren(treeNode x) \r
+int XMLTree::NumChildren(treeNode x)\r
{\r
- return degree(Par, x);\r
+ return bp_degree(Par, x);\r
}\r
\r
// ChildNumber(x): returns i if node x is the i-th children of its parent.\r
-int XMLTree::ChildNumber(treeNode x) \r
+int XMLTree::ChildNumber(treeNode x)\r
{\r
- return child_rank(Par, x);\r
+ return bp_child_rank(Par, x);\r
}\r
\r
// Depth(x): depth of node x, a simple binary rank on the parentheses sequence.\r
-int XMLTree::Depth(treeNode x) \r
+int XMLTree::Depth(treeNode x)\r
{\r
- return depth(Par, x);\r
+ return bp_depth(Par, x);\r
}\r
\r
// Preorder(x): returns the preorder number of node x, just counting the tree\r
// nodes (i.e., tags, it disregards the texts in the tree).\r
-int XMLTree::Preorder(treeNode x) \r
+int XMLTree::Preorder(treeNode x)\r
{\r
- return preorder_rank(Par, x);\r
+ return bp_preorder_rank(Par, x);\r
}\r
\r
// Postorder(x): returns the postorder number of node x, just counting the tree\r
// nodes (i.e., tags, it disregards the texts in the tree).\r
-int XMLTree::Postorder(treeNode x) \r
- {\r
- return postorder_rank(Par, x);\r
- }\r
-/*\r
-// Tag(x): returns the tag identifier of node x.\r
-TagType XMLTree::Tag(treeNode x) \r
+int XMLTree::Postorder(treeNode x)\r
{\r
- return fast_get_field(tags_fix,tags_blen,node2tagpos(x));\r
+ return bp_postorder_rank(Par, x);\r
}\r
-*/\r
+\r
// DocIds(x): returns the range of text identifiers that descend from node x.\r
// returns {NULLT, NULLT} when there are no texts descending from x.\r
-range XMLTree::DocIds(treeNode x) \r
+range XMLTree::DocIds(treeNode x)\r
{\r
range r;\r
if (x == NULLT) {\r
r.max = NULLT;\r
return r;\r
};\r
- int min = EBVector->rank1(x-1); \r
- int max = EBVector->rank1(x+2*subtree_size(Par, x)-2); \r
+ int min = EBVector->rank1(x-1);\r
+ int max = EBVector->rank1(x+2*bp_subtree_size(Par, x)-2);\r
if (min==max) { // range is empty, no texts within the subtree of x\r
r.min = NULLT;\r
r.max = NULLT;\r
return r;\r
}\r
\r
-// Parent(x): returns the parent node of node x.\r
-\r
-treeNode XMLTree::Parent(treeNode x) \r
- {\r
- if (x == Root())\r
- return NULLT;\r
- else\r
- return parent(Par, x);;\r
- }\r
\r
// Child(x,i): returns the i-th child of node x, assuming it exists.\r
-treeNode XMLTree::Child(treeNode x, int i) \r
+treeNode XMLTree::Child(treeNode x, int i)\r
{\r
- if (i <= OPTD) return naive_child(Par, x, i);\r
- else return child(Par, x, i);\r
+ if (i <= OPTD) return bp_naive_child(Par, x, i);\r
+ else return bp_child(Par, x, i);\r
}\r
\r
-// FirstChild(x): returns the first child of node x, assuming it exists. Very fast in BP.\r
-\r
-treeNode XMLTree::FirstChild(treeNode x) \r
- {\r
- NULLT_IF(x==NULLT);\r
- return fast_first_child(Par, x);\r
- }\r
-\r
-treeNode XMLTree::FirstElement(treeNode x) \r
- {\r
- NULLT_IF(x==NULLT);\r
- x = fast_first_child(Par, x);\r
- NULLT_IF(x == NULLT);\r
- switch (Tag(x)){\r
- \r
- case PCDATA_TAG_ID:\r
- x = x+2;\r
- return (fast_inspect(Par,x)==OP)? x : NULLT;\r
- \r
- case ATTRIBUTE_TAG_ID: \r
- x = fast_next_sibling(Par,x);\r
- if (x != NULLT && Tag(x) == PCDATA_TAG_ID){\r
- x = x+2;\r
- return (fast_inspect(Par,x)==OP)? x : NULLT;\r
- } \r
- else return x; \r
- default:\r
- return x;\r
- }\r
- }\r
-\r
-treeNode XMLTree::NextElement(treeNode x) \r
-{\r
- NULLT_IF(x==NULLT);\r
- x = fast_next_sibling(Par, x);\r
- NULLT_IF(x == NULLT); \r
- if (Tag(x) == PCDATA_TAG_ID){\r
- x = x+2;\r
- return (fast_inspect(Par,x)==OP)? x : NULLT;\r
- }\r
- else return x; \r
-}\r
-\r
-// LastChild(x): returns the last child of node x.\r
-treeNode XMLTree::LastChild(treeNode x)\r
- {\r
- NULLT_IF(x == NULLT || fast_isleaf(Par,x));\r
- return find_open(Par, fast_find_close(Par, x)-1);\r
- }\r
-\r
-// NextSibling(x): returns the next sibling of node x, assuming it exists.\r
-treeNode XMLTree::NextSibling(treeNode x) \r
- {\r
- NULLT_IF(x==NULLT || x == Root() );\r
- x = fast_find_close(Par,x)+1;\r
- return (fast_inspect(Par,x) == CP ? NULLT : x);\r
- }\r
-\r
-\r
-// PrevSibling(x): returns the previous sibling of node x, assuming it exists.\r
-treeNode XMLTree::PrevSibling(treeNode x) \r
- {\r
- NULLT_IF(x==NULLT);\r
- return prev_sibling(Par, x);\r
- }\r
-\r
-// TaggedChild(x,tag): returns the first child of node x tagged tag, or NULLT if there is none.\r
-// Because of the balanced-parentheses representation of the tree, this operation is not supported\r
-// efficiently, just iterating among the children of node x until finding the desired child.\r
-treeNode XMLTree::TaggedChild(treeNode x, TagType tag) \r
- {\r
- \r
- NULLT_IF(x==NULLT || fast_isleaf(Par,x));\r
- treeNode child; \r
- child = fast_first_child(Par, x); // starts at first child of node x\r
- if (Tag(child) == tag)\r
- return child;\r
- else\r
- return TaggedFollowingSibling(child,tag);\r
- }\r
-\r
-// TaggedSibling(x,tag): returns the first sibling of node x tagged tag, or NULLT if there is none.\r
-treeNode XMLTree::TaggedFollowingSibling(treeNode x, TagType tag)\r
-{\r
- NULLT_IF(x==NULLT);\r
- treeNode sibling = fast_next_sibling(Par, x);\r
- TagType ctag;\r
- while (sibling != NULLT) {\r
- ctag = Tag(sibling);\r
- if (ctag == tag) // current sibling is labeled with tag of interest\r
- return sibling; \r
- sibling = fast_sibling(Par, sibling, ctag); // OK, let's try with the next sibling\r
- }\r
- return NULLT; // no such sibling was found \r
-}\r
\r
treeNode XMLTree::SelectChild(treeNode x, TagIdSet *tags)\r
{\r
- \r
- NULLT_IF(x==NULLT || fast_isleaf(Par,x));\r
+\r
+ NULLT_IF(x==NULLT || bp_isleaf(Par,x));\r
int i;\r
- treeNode child = fast_first_child(Par, x); \r
+ treeNode child = bp_first_child(Par, x);\r
TagType t;\r
while (child != NULLT) {\r
t = Tag(child);\r
if (tags->find(t) != tags->end()) return child;\r
child = fast_sibling(Par, child,t);\r
}\r
- return NULLT; \r
+ return NULLT;\r
}\r
\r
\r
NULLT_IF(x==NULLT);\r
int i;\r
TagType t;\r
- treeNode sibling = fast_next_sibling(Par, x);\r
+ treeNode sibling = bp_next_sibling(Par, x);\r
while (sibling != NULLT) {\r
t = Tag(sibling);\r
if (tags->find(t) != tags->end()) return sibling;\r
sibling = fast_sibling(Par, sibling,t);\r
}\r
- return NULLT; \r
- }\r
-\r
-\r
-// TaggedDescendant(x,tag): returns the first node tagged tag with larger preorder than x and within\r
-// the subtree of x. Returns NULLT if there is none.\r
-treeNode XMLTree::TaggedDescendant(treeNode x, TagType tag) \r
- {\r
- //NULLT_IF(x==NULLT || fast_isleaf(Par,x));\r
-\r
- int s = (int) Tags->select_next(tag,node2tagpos(x));\r
- NULLT_IF (s == -1);\r
-\r
- treeNode y = tagpos2node(s); // transforms the tag position into a node position\r
- \r
- return (fast_is_ancestor(Par,x,y) ? y : NULLT);\r
- }\r
-\r
-\r
-treeNode XMLTree::SelectDescendant(treeNode x, TagIdSet *tags)\r
- {\r
- NULLT_IF (x ==NULLT || fast_isleaf(Par,x));\r
- int i;\r
- treeNode min = NULLT;\r
- treeNode fc = fast_first_child(Par,x);\r
- treeNode aux;\r
- TagIdSet::const_iterator tagit;\r
- for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\r
- aux = TaggedDescendant(x, (TagType) *tagit);\r
- if (aux == fc) return fc;\r
- if (aux == NULLT) continue;\r
- if ((min == NULLT) || (aux < min)) min = aux;\r
- };\r
- return min;\r
+ return NULLT;\r
}\r
\r
\r
-\r
// TaggedPrec(x,tag): returns the first node tagged tag with smaller preorder than x and not an\r
// ancestor of x. Returns NULLT if there is none.\r
-treeNode XMLTree::TaggedPreceding(treeNode x, TagType tag) \r
- { \r
+treeNode XMLTree::TaggedPreceding(treeNode x, TagType tag)\r
+ {\r
int r, s;\r
treeNode node_s, root;\r
r = (int)Tags->rank(tag, node2tagpos(x)-1);\r
if (r==0) return NULLT; // there is no such node.\r
s = (int)Tags->select(tag, r);\r
- root = root_node(Par);\r
+ root = bp_root_node(Par);\r
node_s = tagpos2node(s);\r
- while (fast_is_ancestor(Par, node_s, x) && (node_s!=root)) { // the one that we found is an ancestor of x\r
+ while (bp_is_ancestor(Par, node_s, x) && (node_s!=root)) { // the one that we found is an ancestor of x\r
r--;\r
if (r==0) return NULLT; // there is no such node\r
s = (int)Tags->select(tag, r); // we should use select_prev instead when provided\r
node_s = tagpos2node(s);\r
}\r
- return NULLT; // there is no such node \r
+ return NULLT; // there is no such node\r
}\r
\r
\r
// the subtree of x. Returns NULLT if there is none.\r
treeNode XMLTree::TaggedFollowing(treeNode x, TagType tag)\r
{\r
- NULLT_IF (x ==NULLT || x == Root()); \r
- return tagpos2node(Tags->select_next(tag,fast_find_close(Par, x)));\r
-\r
- } \r
-\r
-// TaggedFollBelow(x,tag,root): returns the first node tagged tag with larger preorder than x \r
-// and not in the subtree of x. Returns NULLT if there is none.\r
-treeNode XMLTree::TaggedFollowingBelow(treeNode x, TagType tag, treeNode ancestor)\r
-{\r
- // NULLT_IF (x == NULLT || x == Root() || x == ancestor); \r
+ NULLT_IF (x ==NULLT || x == Root());\r
+ return tagpos2node(Tags->select_next(tag, bp_find_close(Par, x)));\r
\r
- //Special optimisation, test for the following sibling first\r
- treeNode close = fast_find_close(Par, x);\r
- /*\r
- treeNode ns = close+1;\r
- if (fast_inspect(Par,ns) == OP) {\r
- TagType tagns = Tag(ns);\r
- // cout << GetTagNameByRef(tagns) << endl;\r
- //cout.flush();\r
- if (tagns == PCDATA_TAG_ID){\r
- close = ns+1;\r
- ns = ns+2;\r
- if (fast_inspect(Par,ns) != OP)\r
- goto after;\r
- tagns = Tag(ns); \r
- };\r
- if (tagns == tag)\r
- return ns;\r
- };\r
- after:\r
- */\r
- treeNode s = tagpos2node(Tags->select_next(tag, close));\r
- \r
- if (ancestor == Root() || s==NULLT || s < fast_find_close(Par,ancestor)) return s;\r
- else return NULLT;\r
-} \r
-\r
-treeNode XMLTree::TaggedFollowingBefore(treeNode x, TagType tag, treeNode closing)\r
-{\r
+ }\r
\r
- NULLT_IF (x == NULLT || x == Root());\r
- \r
- treeNode s = tagpos2node(Tags->select_next(tag, fast_find_close(Par, x))); \r
- NULLT_IF (s == NULLT || s >= closing);\r
- \r
- return s;\r
-} \r
\r
/* Here we inline TaggedFoll to find the min globally, and only at the end\r
we check if the min is below the context node */\r
\r
NULLT_IF(x==NULLT || x==Root());\r
\r
- treeNode close = fast_find_close(Par,x);\r
+ treeNode close = bp_find_close(Par,x);\r
treeNode ns = close+1;\r
- if ( (fast_inspect(Par,ns) == OP) && (tags->find(Tag(ns)) != tags->end()))\r
+ if ( (bp_inspect(Par,ns) == OP) && (tags->find(Tag(ns)) != tags->end()))\r
return ns;\r
\r
int i;\r
treeNode min = NULLT;\r
treeNode aux;\r
- \r
\r
- TagIdSet::const_iterator tagit;\r
- for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\r
\r
- aux = tagpos2node(Tags->select_next(*tagit, close));\r
- \r
- // The next sibling of x is guaranteed to be below ctx\r
- // and is the node with lowest preorder which is after ctx.\r
- // if we find it, we return early; \r
- if (aux == NULLT) continue;\r
- if ((min == NULLT) || (aux < min)) min = aux;\r
- };\r
- \r
- // found the smallest node in preorder which is after x.\r
- // if ctx is the root node, just return what we found.\r
-\r
- if (ancestor == Root()) return min;\r
- // else check whether if is in below the ctx node\r
-\r
- NULLT_IF (min == NULLT || min >= fast_find_close(Par, ancestor));\r
- \r
- return min;\r
- \r
- }\r
-treeNode XMLTree::SelectFollowingBefore(treeNode x, TagIdSet *tags, treeNode closing)\r
- {\r
-\r
- NULLT_IF(x==NULLT || x==Root());\r
- int i;\r
- treeNode min = NULLT;\r
- treeNode ns = fast_next_sibling(Par, x);\r
- treeNode close = ns - 1;\r
- treeNode aux;\r
TagIdSet::const_iterator tagit;\r
- for (tagit = tags->begin(); tagit != tags->end(); tagit++) {\r
+ for (tagit = tags->begin(); tagit != tags->end(); ++tagit) {\r
\r
aux = tagpos2node(Tags->select_next(*tagit, close));\r
- \r
- // The next sibling of x is guaranteed to be below ctx\r
- // and is the node with lowest preorder which is after ctx.\r
- // if we find it, we return early;\r
- \r
- if (aux == ns ) return ns;\r
if (aux == NULLT) continue;\r
if ((min == NULLT) || (aux < min)) min = aux;\r
};\r
- \r
+\r
// found the smallest node in preorder which is after x.\r
// if ctx is the root node, just return what we found.\r
\r
- NULLT_IF (min == NULLT || min >= closing);\r
- \r
- return min;\r
- \r
- }\r
-\r
+ if (ancestor == Root() || min == NULLT || min < bp_find_close(Par, ancestor)) return min;\r
+ else return NULLT;\r
\r
+ }\r
// TaggedAncestor(x, tag): returns the closest ancestor of x tagged tag. Return\r
// NULLT is there is none.\r
treeNode XMLTree::TaggedAncestor(treeNode x, TagType tag)\r
- { \r
+ {\r
if (x == NULLT || x == Root())\r
return NULLT;\r
- \r
- treeNode s = parent(Par, x), r = Root();\r
+\r
+ treeNode s = bp_parent(Par, x), r = Root();\r
while (s != r) {\r
if (Tag(s) == tag) return s;\r
- s = parent(Par, s);\r
+ s = bp_parent(Par, s);\r
}\r
return NULLT;\r
}\r
\r
\r
\r
-// MyText(x): returns the document identifier of the text below node x, \r
-// or NULLT if x is not a leaf node or the text is empty. Assumes Doc \r
+// MyText(x): returns the document identifier of the text below node x,\r
+// or NULLT if x is not a leaf node or the text is empty. Assumes Doc\r
// ids start from 0.\r
-DocID XMLTree::MyText(treeNode x) \r
+DocID XMLTree::MyText(treeNode x)\r
{\r
TagType tag = Tag(x);\r
// seems faster than testing EBVector->access(x);\r
\r
if (tag == PCDATA_TAG_ID || tag == ATTRIBUTE_DATA_TAG_ID)\r
- //if (EBVector->access(x))\r
- return (DocID) (EBVector->rank1(x)-1); //-1 because document ids start from 0\r
- else \r
+ return (DocID) (EBVector->rank1(x)-1);\r
+ else\r
return (DocID) NULLT;\r
- \r
+\r
}\r
-// MyText(x): returns the document identifier of the text below node x, \r
-// or NULLT if x is not a leaf node or the text is empty. Assumes Doc \r
+// MyText(x): returns the document identifier of the text below node x,\r
+// or NULLT if x is not a leaf node or the text is empty. Assumes Doc\r
// ids start from 0.\r
-DocID XMLTree::MyTextUnsafe(treeNode x) \r
+DocID XMLTree::MyTextUnsafe(treeNode x)\r
{\r
return (DocID) (EBVector->rank1(x)-1); //-1 because document ids start from 0\r
+\r
}\r
// TextXMLId(d): returns the preorder of document with identifier d in the tree consisting of\r
// all tree nodes and all text nodes. Assumes that the tree root has preorder 1.\r
-int XMLTree::TextXMLId(DocID d) \r
+int XMLTree::TextXMLId(DocID d)\r
{\r
NULLT_IF(d == NULLT);\r
int s = EBVector->select1(d+1);\r
- return rank_open(Par, s) + d + 1; // +1 because root has preorder 1\r
- \r
+ return bp_rank_open(Par, s) + d + 1; // +1 because root has preorder 1\r
+\r
}\r
\r
-// NodeXMLId(x): returns the preorder of node x in the tree consisting \r
+// NodeXMLId(x): returns the preorder of node x in the tree consisting\r
// of all tree nodes and all text nodes. Assumes that the tree root has\r
// preorder 0;\r
-int XMLTree::NodeXMLId(treeNode x) \r
+int XMLTree::NodeXMLId(treeNode x)\r
{\r
NULLT_IF(x == NULLT);\r
if (x == Root()) return 1; // root node has preorder 1\r
- return rank_open(Par, x) + EBVector->rank1(x-1);\r
+ return bp_rank_open(Par, x) + EBVector->rank1(x-1);\r
}\r
\r
// ParentNode(d): returns the parent node of document identifier d.\r
-treeNode XMLTree::ParentNode(DocID d) \r
- { \r
- NULLT_IF (d == NULLT); \r
- return (treeNode) EBVector->select1(d+1); \r
+treeNode XMLTree::ParentNode(DocID d)\r
+ {\r
+ NULLT_IF (d == NULLT);\r
+ return (treeNode) EBVector->select1(d+1);\r
}\r
\r
// GetTagId: returns the tag identifier corresponding to a given tag name.\r
// Returns NULLT in case that the tag name does not exists.\r
TagType XMLTree::GetTagId(unsigned char *tagname)\r
{\r
- \r
+\r
string s = (char *) tagname;\r
- TagIdMapIT it = tIdMap->find(s); \r
+ TagIdMapIT it = tIdMap->find(s);\r
return (TagType) ((it != tIdMap->end()) ? it->second : -1);\r
- \r
+\r
}\r
\r
\r
if ( tagid < 0 || tagid >= TagName->size())\r
return (unsigned char *) "<INVALID TAG>";\r
strcpy((char *)s, (*TagName)[tagid].c_str());\r
- \r
+\r
return (s == NULL ? (unsigned char*) "<INVALID TAG>" : s);\r
}\r
\r
unsigned char *s;\r
if ( tagid < 0 || tagid >= TagName->size())\r
return (unsigned char *) "<INVALID TAG>";\r
- \r
+\r
return (const unsigned char *) (*TagName)[tagid].c_str();\r
- \r
+\r
}\r
\r
\r
\r
TagType XMLTree::RegisterTag(unsigned char *tagname)\r
- { \r
+ {\r
TagType id = XMLTree::GetTagId(tagname);\r
if (id == NULLT) {\r
- string s = (char *) tagname; \r
- REGISTER_TAG(TagName,tIdMap,s); \r
+ string s = (char *) tagname;\r
+ REGISTER_TAG(TagName,tIdMap,s);\r
};\r
- \r
+\r
return id;\r
}\r
\r
\r
treeNode XMLTree::Closing(treeNode x) {\r
- return fast_find_close(Par,x); \r
+ return bp_find_close(Par,x);\r
}\r
-bool XMLTree::IsOpen(treeNode x) { return fast_inspect(Par,x); }\r
+bool XMLTree::IsOpen(treeNode x) { return bp_inspect(Par,x); }\r
\r
//WARNING this uses directly the underlying implementation for plain text\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
+ if (buffer == 0) {\r
+ buffer = new string(BUFFER_ALLOC, 0);\r
+ buffer->clear();\r
+ print_stack = new std::vector<string *>();\r
+ print_stack->reserve(256);\r
+ };\r
\r
- FILE* fp = stream;\r
- treeNode fin = fast_find_close(Par,x);\r
+ treeNode fin = bp_find_close(Par,x);\r
treeNode n = x;\r
TagType tag = Tag(n);\r
- uchar * tagstr;\r
+\r
range r = DocIds(x);\r
treeNode first_idx;\r
treeNode first_text = (tag == PCDATA_TAG_ID ? x : ParentNode(r.min-1));\r
treeNode first_att = NULLT;\r
- \r
+\r
if (first_att == NULLT)\r
- first_idx = first_text;\r
+ first_idx = first_text;\r
else if (first_text == NULLT)\r
- first_idx = first_att;\r
+ first_idx = first_att;\r
else\r
- first_idx = min(first_att,first_text);\r
- \r
+ first_idx = min(first_att,first_text);\r
+\r
uchar * current_text=NULL;\r
+\r
if (first_idx != NULLT)\r
- current_text = GetText(MyText(first_idx));\r
+ current_text = GetText(MyTextUnsafe(first_idx));\r
+\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
-\r
- if (no_text)\r
- myfputs("<$/>",fp);\r
- else{\r
- read = myfprintf((const char*) current_text, fp);\r
- current_text += (read + 1);\r
- };\r
- n+=2; // skip closing $\r
- tag = Tag(n);\r
- \r
- }\r
- else {\r
- myfputc('<',fp);\r
- tagstr = (uchar*) GetTagNameByRef(tag);\r
- myfputs((const char*) tagstr ,fp);\r
- n++;\r
- if (fast_inspect(Par,n)) {\r
- st.push_back(tagstr);\r
+\r
+ while (n <= fin){\r
+ if (bp_inspect(Par,n)){\r
+ if (tag == PCDATA_TAG_ID) {\r
+\r
+ if (no_text)\r
+ _dputs("<$/>", fd);\r
+ else {\r
+ read = _dprintf((const char*) current_text, fd);\r
+ current_text += (read + 1);\r
+ };\r
+ n+=2; // skip closing $\r
tag = Tag(n);\r
- if (tag == ATTRIBUTE_TAG_ID){\r
- n++;\r
- if (no_text) myfputs("><@@>",fp);\r
- while (fast_inspect(Par,n)){\r
- if (no_text) {\r
- myfputc('<',fp);\r
- myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
- myfputc('>',fp);\r
- myfputs("<$@/></",fp);\r
- myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
- myfputc('>',fp);\r
- n+= 4;\r
- }\r
- else {\r
- myfputc(' ',fp);\r
- myfputs((const char*) &(GetTagNameByRef(Tag(n))[3]),fp);\r
- n++;\r
- myfputs("=\"",fp);\r
- read = myfprintf((const char*) current_text,fp);\r
- current_text += (read + 1);\r
- myfputc('"',fp);\r
- n+=3;\r
- }\r
- };\r
- if (no_text) \r
- myfputs("</@@>",fp);\r
- else myfputc('>',fp);\r
+\r
+ } else {\r
+\r
+ _dputc('<',fd);\r
+ _dput_str((*TagName)[tag], fd);\r
+ n++;\r
+ if (bp_inspect(Par,n)) {\r
+ print_stack->push_back(&((*TagName)[tag]));\r
+ tag = Tag(n);\r
+ if (tag == ATTRIBUTE_TAG_ID){\r
+ n++;\r
+ if (no_text) _dputs("><@@>",fd);\r
+\r
+ while (bp_inspect(Par,n)){\r
+ if (no_text) {\r
+ _dputc('<', fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ _dputc('>', fd);\r
+ _dputs("<$@/></", fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ _dputc('>', fd);\r
+ n+= 4;\r
+ } else {\r
+ _dputc(' ', fd);\r
+ _dputs((const char*) &(GetTagNameByRef(Tag(n))[3]), fd);\r
+ n++;\r
+ _dputs("=\"", fd);\r
+ read = _dprintf((const char*) current_text, fd);\r
+ current_text += (read + 1);\r
+ _dputc('"', fd);\r
+ n+=3;\r
+ }\r
+ };\r
+ if (no_text) _dputs("</@@>", fd);\r
+ else _dputc('>', fd);\r
+ n++;\r
+ tag=Tag(n);\r
+\r
+ } else\r
+ _dputc('>', fd);\r
+\r
+ } else {// <foo /> tag\r
+ _dputs("/>", fd);\r
n++;\r
tag=Tag(n);\r
- }\r
- else {\r
- myfputc('>',fp);\r
};\r
- }\r
- else {// <foo /> tag\r
- myfputs("/>",fp);\r
+ };\r
+ } else do {\r
+ _dputs("</", fd);\r
+ _dput_str(*(print_stack->back()), fd);\r
+ _dputc('>', fd);\r
+ print_stack->pop_back();\r
n++;\r
- tag=Tag(n); \r
- }; \r
- };\r
- }\r
- else\r
- do {\r
- myfputs("</",fp);\r
- myfputs((const char*)st.back(),fp);\r
- myfputc('>', fp);\r
- st.pop_back();\r
- n++;\r
- }while (!fast_inspect(Par,n) && !st.empty());\r
- tag=Tag(n);\r
- };\r
- myfputc('\n',fp);\r
- mybufferflush(fp);\r
- //fflush(fp);\r
- fclose(fp);\r
+ } while (!(bp_inspect(Par,n) || print_stack->empty()));\r
+ tag = Tag(n);\r
+ };\r
+ _dputc('\n', fd);\r
+ //_flush(fd);\r
}\r