X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=OCamlDriver.cpp;h=78d3e15de8ceb4e53bcc9069a394f44ac16fd570;hb=25dd7fcc77c2188732d96d5ff98d759bb81737cb;hp=636a95280c79e6170ea21290e05d0d10b100c4f2;hpb=24fdea81b5506233d139bd7d72364a190bef35b8;p=SXSI%2Fxpathcomp.git diff --git a/OCamlDriver.cpp b/OCamlDriver.cpp index 636a952..78d3e15 100644 --- a/OCamlDriver.cpp +++ b/OCamlDriver.cpp @@ -15,30 +15,74 @@ extern "C" { #include #include #include -} //extern C +#include -#include "TextCollection/TextCollection.h" + +#include +#include +#include +#include + + struct tms t1; + struct tms t2; + double ticks = (double) sysconf(_SC_CLK_TCK)/1000; + + void start_clock() { + times (&t1); + } + + + double stop_clock() { + times (&t2); + return (t2.tms_utime-t1.tms_utime)/ticks; + } +} //extern C + + +//#include "TextCollection/TextCollection.h" #include "XMLDocShredder.h" #include "XMLTree.h" #include "Utils.h" #define CAMLRAISECPP(e) (caml_failwith( ((e).what()))) #define NOT_IMPLEMENTED(s) (caml_failwith(s)) -#define XMLTREE(x) ((XMLTree *)(x)) +#define XMLTREE(x) ((XMLTree *)(* (XMLTree**) Data_custom_val(x))) #define TEXTCOLLECTION(x) #define TREENODEVAL(i) ((treeNode) (Int_val(i))) -extern "C" CAMLprim value caml_call_shredder_uri(value uri){ +extern "C" { + static struct custom_operations ops; + static bool initialized = false; +} +extern "C" void caml_xml_tree_finalize(value tree){ + delete XMLTREE(tree); + return; +} + +extern "C" void caml_init_ops () { + + if (initialized) + return; + ops.identifier = (char*) "XMLTree"; + ops.finalize = caml_xml_tree_finalize; + return; +} + + +extern "C" CAMLprim value caml_call_shredder_uri(value uri,value sf, value iet, value dtc){ CAMLparam1(uri); CAMLlocal1(doc); char *fn = String_val(uri); try { - XMLDocShredder shredder(fn); + XMLDocShredder shredder(fn,Int_val(sf),Bool_val(iet),Bool_val(dtc)); + XMLTree * tree; shredder.processStartDocument(fn); shredder.parse(); shredder.processEndDocument(); - doc = (value) shredder.storageIfc_->returnDocument(); - + caml_init_ops(); + doc = caml_alloc_custom(&ops,sizeof(XMLTree*),1,2); + tree = (XMLTree *) shredder.storageIfc_->returnDocument(); + memcpy(Data_custom_val(doc),&tree,sizeof(XMLTree*)); CAMLreturn(doc); } catch (const std::exception& e){ @@ -47,19 +91,22 @@ extern "C" CAMLprim value caml_call_shredder_uri(value uri){ } -extern "C" CAMLprim value caml_call_shredder_string(value data){ +extern "C" CAMLprim value caml_call_shredder_string(value data,value sf, value iet, value dtc){ CAMLparam1(data); CAMLlocal1(doc); unsigned int ln = string_length(data); unsigned char *fn = (unsigned char*) String_val(data); try { - XMLDocShredder shredder(fn,ln); + XMLDocShredder shredder(fn,ln,Int_val(sf),Bool_val(iet),Bool_val(dtc)); + XMLTree* tree; shredder.processStartDocument(""); shredder.parse(); shredder.processEndDocument(); - doc = (value) shredder.storageIfc_->returnDocument(); - + caml_init_ops(); + doc = caml_alloc_custom(&ops,sizeof(XMLTree*),1,2); + tree = (XMLTree *) shredder.storageIfc_->returnDocument(); + memcpy(Data_custom_val(doc),&tree,sizeof(XMLTree*)); CAMLreturn(doc); } catch (const std::exception& e) { @@ -67,11 +114,14 @@ extern "C" CAMLprim value caml_call_shredder_string(value data){ }; } + + + void traversal_rec(XMLTree* tree, treeNode id){ DocID tid; if (id == NULLT) return; - int tag = tree->Tag(id); + //int tag = tree->Tag(id); if (id) { tid = tree->PrevText(id); char * data = (char *) (tree->getTextCollection())->GetText(tid); @@ -97,50 +147,79 @@ extern "C" CAMLprim value caml_cpp_traversal(value tree){ CAMLreturn(Val_unit); } +extern "C" CAMLprim value caml_text_collection_get_text(value tree, value id){ + CAMLparam2(tree,id); + CAMLlocal1(str); + uchar* txt = XMLTREE(tree)->GetText((DocID) Int_val(id)); + str = caml_copy_string((const char*)txt); + delete (txt); + CAMLreturn (str); +} -extern "C" CAMLprim value caml_text_collection_get_text(value tc, value id){ - CAMLparam2(tc,id); - - const char* txt = (const char*) ((TextCollection*) tc)->GetText((DocID) Int_val(id)); - CAMLreturn (caml_copy_string(txt)); +extern "C" CAMLprim value caml_text_collection_get_cached_text(value tree, value id){ + CAMLparam2(tree,id); + CAMLlocal1(str); + char* txt = (char*) XMLTREE(tree)->GetCachedText((DocID) Int_val(id)); + str = caml_copy_string(txt); + free(txt); + CAMLreturn (str); } -extern "C" CAMLprim value caml_text_collection_empty_text(value tc,value id){ - CAMLparam2(tc,id); - CAMLreturn ( Val_int(((TextCollection*) tc)->EmptyText((DocID) Int_val(id)))); + + +extern "C" CAMLprim value caml_text_collection_empty_text(value tree,value id){ + CAMLparam2(tree,id); + CAMLreturn ( Val_int((XMLTREE(tree))->EmptyText((DocID) Int_val(id)))); } -extern "C" CAMLprim value caml_text_collection_is_contains(value tc,value str){ - CAMLparam2(tc,str); +extern "C" CAMLprim value caml_text_collection_is_contains(value tree,value str){ + CAMLparam2(tree,str); uchar * cstr = (uchar *) String_val(str); - CAMLreturn ( Val_bool((int) ((TextCollection*) tc)->IsContains(cstr))); + CAMLreturn ( Val_bool((int) XMLTREE(tree)->IsContains(cstr))); } -extern "C" CAMLprim value caml_text_collection_count_contains(value tc,value str){ - CAMLparam2(tc,str); +extern "C" CAMLprim value caml_text_collection_count_contains(value tree,value str){ + CAMLparam2(tree,str); uchar * cstr = (uchar *) String_val(str); - CAMLreturn ( Val_int(((TextCollection*) tc)->CountContains(cstr))); + CAMLreturn (Val_int((XMLTREE(tree)->CountContains(cstr)))); + +} +extern "C" CAMLprim value caml_text_collection_count(value tree,value str){ + CAMLparam2(tree,str); + uchar * cstr = (uchar *) String_val(str); + CAMLreturn (Val_int((XMLTREE(tree)->Count(cstr)))); + CAMLreturn (Val_unit); } -extern "C" CAMLprim value caml_text_collection_contains(value tc,value str){ - CAMLparam2(tc,str); +extern "C" CAMLprim value caml_text_collection_contains(value tree,value str){ + CAMLparam2(tree,str); CAMLlocal1(resarray); uchar * cstr = (uchar *) String_val(str); std::vector results; - results = ((TextCollection*) tc)->Contains(cstr); - + results = XMLTREE(tree)->Contains(cstr); + //free(cstr); resarray = caml_alloc_tuple(results.size()); - for (int i=0; i results; + start_clock(); + results = XMLTREE(tree)->Contains(cstr); + double d = stop_clock(); + std::cerr << "Internal timing " << d <<" ms\n"; + CAMLreturn (Val_unit); +} extern "C" CAMLprim value caml_xml_tree_root(value tree){ CAMLparam1(tree); - CAMLreturn (TREENODEVAL(XMLTREE(tree)->Root())); + CAMLreturn (Val_int(TREENODEVAL(XMLTREE(tree)->Root()))); } extern "C" CAMLprim value caml_xml_tree_text_collection(value tree){ CAMLparam1(tree); @@ -150,9 +229,19 @@ extern "C" CAMLprim value caml_xml_tree_parent(value tree, value id){ CAMLparam2(tree,id); CAMLreturn(Val_int (XMLTREE(tree)->Parent(TREENODEVAL(id)))); } +extern "C" CAMLprim value caml_xml_tree_prev_sibling(value tree, value id){ + CAMLparam2(tree,id); + CAMLreturn(Val_int (XMLTREE(tree)->PrevSibling(TREENODEVAL(id)))); +} + extern "C" CAMLprim value caml_xml_tree_parent_doc(value tree, value id){ CAMLparam2(tree,id); - CAMLreturn(Val_int (XMLTREE(tree)->ParentNode(TREENODEVAL(id)))); + CAMLreturn(Val_int (XMLTREE(tree)->ParentNode((DocID) Int_val(id)))); +} + +extern "C" CAMLprim value caml_xml_tree_prev_doc(value tree, value id){ + CAMLparam2(tree,id); + CAMLreturn(Val_int (XMLTREE(tree)->PrevNode((DocID) Int_val(id)))); } extern "C" CAMLprim value caml_xml_tree_is_ancestor(value tree,value id1, value id2) { @@ -172,6 +261,15 @@ extern "C" CAMLprim value caml_xml_tree_unserialize(value filename){ CAMLreturn(Val_unit); } +extern "C" CAMLprim value caml_xml_tree_last_child(value tree, value id){ + CAMLparam2(tree,id); + CAMLreturn(Val_int (XMLTREE(tree)->LastChild(TREENODEVAL(id)))); +} + +extern "C" CAMLprim value caml_xml_tree_is_first_child(value tree, value id){ + CAMLparam2(tree,id); + CAMLreturn(Val_bool (XMLTREE(tree)->IsFirstChild(TREENODEVAL(id)))); +} extern "C" CAMLprim value caml_xml_tree_first_child(value tree, value id){ CAMLparam2(tree,id); @@ -183,6 +281,22 @@ extern "C" CAMLprim value caml_xml_tree_is_leaf(value tree, value id){ CAMLreturn(Val_bool (XMLTREE(tree)->IsLeaf(TREENODEVAL(id)))); } +extern "C" CAMLprim value caml_xml_tree_tagged_desc(value tree, value id, value tag){ + CAMLparam3(tree,id,tag); + CAMLreturn(Val_int (XMLTREE(tree)->TaggedDesc(TREENODEVAL(id),(TagType) Int_val(tag)))); +} + + +extern "C" CAMLprim value caml_xml_tree_tagged_foll(value tree, value id, value tag){ + CAMLparam3(tree,id,tag); + CAMLreturn(Val_int (XMLTREE(tree)->TaggedFoll(TREENODEVAL(id),(TagType) Int_val(tag)))); +} +extern "C" CAMLprim value caml_xml_tree_tagged_foll_below(value tree, value id, value tag,value root){ + CAMLparam4(tree,id,tag,root); + CAMLreturn(Val_int (XMLTREE(tree)->TaggedFollBelow(TREENODEVAL(id),(TagType) Int_val(tag),TREENODEVAL(root)))); +} + + extern "C" CAMLprim value caml_xml_tree_next_sibling(value tree, value id){ CAMLparam2(tree,id); CAMLreturn(Val_int (XMLTREE(tree)->NextSibling(TREENODEVAL(id)))); @@ -190,9 +304,7 @@ extern "C" CAMLprim value caml_xml_tree_next_sibling(value tree, value id){ extern "C" CAMLprim value caml_xml_tree_prev_text(value tree, value id){ CAMLparam2(tree,id); - CAMLlocal1(res); CAMLreturn(Val_int((XMLTREE(tree)->PrevText(TREENODEVAL(id))))); - CAMLreturn(res); } extern "C" CAMLprim value caml_xml_tree_next_text(value tree, value id){ CAMLparam2(tree,id); @@ -211,21 +323,143 @@ extern "C" CAMLprim value caml_xml_tree_node_xml_id(value tree, value id){ CAMLparam2(tree,id); CAMLreturn(Val_int((XMLTREE(tree)->NodeXMLId(TREENODEVAL(id))))); } -extern "C" CAMLprim value caml_xml_tree_tag(value tree, value id){ - CAMLparam2(tree,id); - const char* tag; - tag =(const char*) XMLTREE(tree)->GetTagName(XMLTREE(tree)->Tag(TREENODEVAL(id))); - CAMLreturn (caml_copy_string(tag)); +extern "C" CAMLprim value caml_xml_tree_tag_name(value tree, value tagid){ + CAMLparam2(tree,tagid); + CAMLlocal1(str); + char* tag; + tag = (char*) XMLTREE(tree)->GetTagNameByRef((TagType) (Int_val(tagid))); + str = caml_copy_string((const char*) tag); + CAMLreturn (str); } + + extern "C" CAMLprim value caml_xml_tree_tag_id(value tree,value id){ - CAMLparam2(tree,id); - - int tag =XMLTREE(tree)->Tag(TREENODEVAL(id)); + CAMLparam2(tree,id); + CAMLreturn (Val_int(XMLTREE(tree)->Tag(TREENODEVAL(id)))); +} - CAMLreturn (Val_unit); +extern "C" CAMLprim value caml_xml_tree_subtree_tags(value tree,value id,value tag){ + CAMLparam3(tree,id,tag); + CAMLreturn (Val_int(XMLTREE(tree)->SubtreeTags(TREENODEVAL(id),Int_val(tag)))); } + + +extern "C" CAMLprim value caml_xml_tree_register_tag(value tree,value str){ + CAMLparam2(tree,str); + CAMLlocal1(id); + unsigned char* tag; + tag = (unsigned char*) (String_val(str)); + id = Val_int(XMLTREE(tree)->RegisterTag(tag)); + CAMLreturn (id); +} + extern "C" CAMLprim value caml_xml_tree_nullt(value unit){ CAMLparam1(unit); CAMLreturn (NULLT); } + +extern "C" CAMLprim value caml_xml_tree_save(value tree,value filename){ + CAMLparam2(tree,filename); + XMLTREE(tree)->Save((unsigned char *) String_val(filename)); + CAMLreturn (Val_unit); +} + +extern "C" CAMLprim value caml_xml_tree_load(value filename,value samplerate){ + CAMLparam2(filename,samplerate); + CAMLlocal1(doc); + XMLTree * tree; + tree = XMLTree::Load((unsigned char *) String_val(filename),Int_val(samplerate)); + caml_init_ops(); + doc = caml_alloc_custom(&ops,sizeof(XMLTree*),1,2); + memcpy(Data_custom_val(doc),&tree,sizeof(XMLTree*)); + CAMLreturn(doc); +} + +extern "C" { + static int caml_empty_vector[] = { 0 }; +} + +extern "C" CAMLprim value caml_int_vector_empty(value unit){ + CAMLparam1(unit); + CAMLreturn ((value) caml_empty_vector); +} + +extern "C" CAMLprim value caml_int_vector_length(value vec){ + CAMLparam1(vec); + CAMLreturn (Val_int( ((int*) caml_empty_vector)[0] )); +} +extern "C" CAMLprim value caml_int_vector_alloc(value len){ + CAMLparam1(len); + int * vec = (int *) malloc(sizeof(int)*(Int_val(len)+1)); + vec[0] = Int_val(len); + CAMLreturn ((value) vec); +} + +extern "C" CAMLprim value caml_int_vector_set(value vec, value i, value v){ + CAMLparam3(vec,i,v); + + ((int*) vec)[Int_val(i)+1] = Int_val(v); + CAMLreturn (Val_unit); +} + + +#define VECT(x) ((int*) (x)) +extern "C" CAMLprim value caml_xml_tree_select_below(value tree, value node, value ctags, value dtags){ + CAMLparam4(tree,node,ctags,dtags); + + CAMLreturn (Val_int ( + (XMLTREE(tree)->TaggedBelow(TREENODEVAL(node), + &(VECT(ctags)[1]), + VECT(ctags)[0], + &(VECT(dtags)[1]), + VECT(dtags)[0])))); +} + +extern "C" CAMLprim value caml_xml_tree_select_next(value tree, value node, value ctags, value ftags,value root){ + CAMLparam5(tree,node,ctags,ftags,root); + CAMLreturn (Val_int ( + (XMLTREE(tree)->TaggedNext(TREENODEVAL(node), + &(VECT(ctags)[1]), + VECT(ctags)[0], + &(VECT(ftags)[1]), + VECT(ftags)[0], + TREENODEVAL(root))))); +} + +extern "C" CAMLprim value caml_xml_tree_select_desc_only(value tree, value node,value dtags){ + CAMLparam3(tree,node,dtags); + + CAMLreturn (Val_int ( + (XMLTREE(tree)->TaggedDescOnly(TREENODEVAL(node), + &(VECT(dtags)[1]), + VECT(dtags)[0])))); +} + +extern "C" CAMLprim value caml_xml_tree_select_foll_only(value tree, value node, value ftags,value root){ + CAMLparam4(tree,node,ftags,root); + CAMLreturn (Val_int ( + (XMLTREE(tree)->TaggedFollOnly(TREENODEVAL(node), + &(VECT(ftags)[1]), + VECT(ftags)[0], + TREENODEVAL(root))))); +} + +extern "C" CAMLprim value caml_xml_tree_select_desc_or_foll_only(value tree, value node, value ftags,value root){ + CAMLparam4(tree,node,ftags,root); + CAMLreturn (Val_int ( + (XMLTREE(tree)->TaggedDescOrFollOnly(TREENODEVAL(node), + &(VECT(ftags)[1]), + VECT(ftags)[0], + TREENODEVAL(root))))); +} + +extern "C" CAMLprim value caml_xml_tree_doc_ids(value tree, value node){ + CAMLparam2(tree,node); + CAMLlocal1(tuple); + tuple = caml_alloc_tuple(2); + range r = (XMLTREE(tree)->DocIds(TREENODEVAL(node))); + caml_initialize(&Field(tuple,0),Val_int(r.min)); + caml_initialize(&Field(tuple,1),Val_int(r.max)); + CAMLreturn (tuple); +}