X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=src%2FOCamlDriver.cpp;h=b93286cb1175b7603be98e044a09f4af2cb6bde6;hb=718a2eff89f4798ee47e055556f500dc950a82b7;hp=c1a4fdda93b726b40040dbb23b997328e1e6b900;hpb=a6baedc6c67cb1de0587a779f8bcddf276b0bf4c;p=SXSI%2Fxpathcomp.git diff --git a/src/OCamlDriver.cpp b/src/OCamlDriver.cpp index c1a4fdd..b93286c 100644 --- a/src/OCamlDriver.cpp +++ b/src/OCamlDriver.cpp @@ -40,9 +40,20 @@ extern "C" { #include #include +#include +#include } +extern "C" value caml_clz(value i) +{ + return Val_long( ((sizeof(unsigned long)*8) - __builtin_clzl(Long_val(i))) - 1); +} + +extern "C" value caml_leading_bit(value i) +{ + return Val_long( ( 1 << (sizeof(unsigned long)*8 - __builtin_clzl(Long_val(i)) - 1))); +} /** XMLTreeBuilder bindings * @@ -152,16 +163,13 @@ extern "C" value caml_xml_tree_save(value tree,value fd, value name){ extern "C" value caml_xml_tree_load(value fd, value name, value load_tc,value sf){ CAMLparam4(fd, name, load_tc, sf); - CAMLlocal2(result,tmp); + CAMLlocal1(result); XMLTree * tree; try { - tree = XMLTree::Load(Int_val(fd),Bool_val(load_tc),Int_val(sf), String_val(name)); + tree = XMLTree::Load(Int_val(fd), Bool_val(load_tc), Int_val(sf), String_val(name)); result = sxsi_alloc_custom(); - Obj_val(result) = tree; - tmp = sxsi_alloc_custom(); - Obj_val(tmp) = 3l; CAMLreturn(result); } catch (const std::exception& e){ CAMLRAISEMSG(e.what()); } @@ -863,28 +871,48 @@ extern "C" value caml_text_collection_lessthan(value tree,value str){ /** Full reporting into a bit vector */ +static std::vector sort_results(std::vector v) +{ + std::vector res; + std::sort(v.begin(), v.end()); + DocID prev = NULLT; + for(auto i = v.begin(); i != v.end(); ++i){ + while (prev == *i){ + ++i; + if (i == v.end()) return res; + }; + prev = *i; + res.push_back(prev); + }; + return res; +} #define BV_QUERY(pref, Pref) \ - extern "C" value caml_text_collection_## pref ##_bv(value tree, value str){ \ - CAMLparam2(tree, str); \ - CAMLlocal3(res, res_bv, res_array); \ - int j; \ - uchar * cstr = (uchar *) strdup(String_val(str)); \ - std::vector results = XMLTREE(tree)->Pref(cstr); \ - res_bv = caml_alloc_string((XMLTREE(tree)->Size() / 4) + 2); \ - unsigned long slen = caml_string_length(res_bv); \ - memset(&(Byte(res_bv,0)), 0, slen); \ - res_array = caml_alloc_shr(results.size(), 0); \ - for (unsigned int i = 0; i < results.size(); ++i) { \ - j = XMLTREE(tree)->ParentNode(results[i]); \ - Byte(res_bv, j >> 3) |= (1 << (j & 7)); \ - caml_initialize(&Field(res_array, i), Val_int(j)); \ - }; \ - free(cstr); \ - res = caml_alloc(2, 0); \ - Store_field(res, 0, res_bv); \ - Store_field(res, 1, res_array); \ - CAMLreturn(res); \ + extern "C" value caml_text_collection_## pref ##_bv(value tree, value str, value dobvv){ \ + CAMLparam3(tree, str, dobvv); \ + CAMLlocal3(res, res_bv, res_array); \ + int j; \ + uchar * cstr = (uchar *) strdup(String_val(str)); \ + std::vector uresults = XMLTREE(tree)->Pref(cstr); \ + std::vector results = sort_results(uresults); \ + bool dobv = Bool_val(dobvv); \ + res_bv = caml_alloc_string(dobv ? ((XMLTREE(tree)->Size() / 4) + 2) : 0); \ + unsigned long slen = caml_string_length(res_bv); \ + if (dobv) \ + memset(&(Byte(res_bv,0)), 0, slen); \ + res_array = caml_alloc_shr(results.size(), 0); \ + for (unsigned int i = 0; i < results.size(); ++i) { \ + j = XMLTREE(tree)->ParentNode(results[i]); \ + if (dobv) { \ + Byte(res_bv, j >> 3) |= (1 << (j & 7)); \ + }; \ + caml_initialize(&Field(res_array, i), Val_int(j)); \ + }; \ + free(cstr); \ + res = caml_alloc(2, 0); \ + Store_field(res, 0, res_bv); \ + Store_field(res, 1, res_array); \ + CAMLreturn(res); \ } \ @@ -893,3 +921,113 @@ BV_QUERY(suffix, Suffix) BV_QUERY(equals, Equals) BV_QUERY(contains, Contains) BV_QUERY(lessthan, LessThan) + + +////////////////////// BP + +extern "C" value caml_bitmap_create(value size) +{ + CAMLparam1(size); + size_t bits = Long_val(size); + size_t words = bits / (8*sizeof(unsigned int)); + unsigned int *buffer = (unsigned int*) calloc(words+1, sizeof(unsigned int)); + if (buffer == NULL) + CAMLRAISEMSG("BP: cannot allocate memory"); + CAMLreturn( (value) buffer); +} + +extern "C" value caml_bitmap_resize(value bitmap, value nsize) +{ + CAMLparam2(bitmap, nsize); + size_t bits = Long_val(nsize); + size_t bytes = (bits / (8 * sizeof(unsigned int)) + 1 ) * sizeof(unsigned int); + unsigned int * buffer = (unsigned int*) realloc((void *) bitmap, bytes); + if (buffer == NULL) + CAMLRAISEMSG("BP: cannot reallocate memory"); + CAMLreturn((value) buffer); +} + +extern "C" value caml_bitmap_setbit(value bitmap, value i, value b) +{ + CAMLparam3(bitmap, i, b); + unsigned int j = Int_val(i); + unsigned int x = Bool_val(b); + bp_setbit ((unsigned int*) bitmap, j, x); + CAMLreturn(Val_unit); +} + +extern "C" void caml_bp_delete(value b) +{ + CAMLparam1(b); + bp * B = Obj_val(b); + bp_delete(B); + CAMLreturn0; +} + +extern "C" value caml_bp_construct(value bitmap, value npar) +{ + CAMLparam2(bitmap, npar); + CAMLlocal1(res); + bp * b = bp_construct(Int_val(npar), (unsigned int *) bitmap, OPT_DEGREE); + res = sxsi_alloc_custom(caml_bp_delete); + Obj_val(res) = b; + CAMLreturn(res); +} + +extern "C" value caml_bp_first_child(value b, value idx) +{ + CAMLparam2(b, idx); + CAMLreturn (Val_int( bp_first_child(Obj_val(b), Int_val(idx)))); +} + + +extern "C" value caml_bp_next_sibling(value b, value idx) +{ + CAMLparam2(b, idx); + CAMLreturn (Val_int(bp_next_sibling(Obj_val(b), Int_val(idx)))); +} + +extern "C" value caml_bp_preorder_rank(value b, value idx) +{ + CAMLparam2(b, idx); + CAMLreturn (Val_int(bp_preorder_rank(Obj_val(b), Int_val(idx)) - 1)); +} + + +extern "C" value caml_bp_load(value file) +{ + CAMLparam1(file); + CAMLlocal1(result); + bp *B; + int f1 = Int_val(file); + int f2 = dup(f1); + FILE * fd = fdopen(f2, "r"); + if (fd == NULL) + CAMLRAISEMSG("Error opening bp file"); + B = loadTree(fd); + fclose(fd); + result = sxsi_alloc_custom(caml_bp_delete); + Obj_val(result) = B; + CAMLreturn(result); +} + +extern "C" value caml_bp_save(value b, value file) +{ + CAMLparam2(b, file); + bp *B = Obj_val(b); + int f1 = Int_val(file); + int f2 = dup(f1); + FILE * fd = fdopen(f2, "a"); + fflush(stderr); + if (fd == NULL) + CAMLRAISEMSG("Error saving bp file"); + saveTree(B, fd); + fclose(fd); + CAMLreturn(Val_unit); +} + +extern "C" value caml_bp_alloc_stats(value unit) +{ + CAMLparam1(unit); + CAMLreturn (Val_long(bp_get_alloc_stats())); +}