text_property
oriental_language_rep::advance (tree t, int& pos) {
string s= t->label;
if (pos == N(s)) return &tp_normal_rep;
if (s[pos] == ' ') {
pos++;
return &tp_space_rep;
}
if (pos < N(s) && !test (s, pos, "<#")) {
while (pos < N(s) && s[pos] != ' ' && !test (s, pos, "<#"))
tm_char_forwards (s, pos);
return &tp_cjk_no_break_rep;
}
int start= pos;
tm_char_forwards (s, pos);
string c= s (start, pos);
int next= pos;
tm_char_forwards (s, next);
string x= s (pos, next);
if (punct->contains (c)) {
if (punct->contains (x) || pos == N(s))
return &tp_cjk_no_break_period_rep;
else return &tp_cjk_period_rep;
}
else {
if (punct->contains (x) || pos == N(s))
return &tp_cjk_no_break_rep;
else return &tp_cjk_normal_rep;
}
}
void
edit_env_rep::local_update (hashmap<string,tree>& old_patch,
hashmap<string,tree>& change)
{
old_patch->pre_patch (back, env);
old_patch->post_patch (change, env);
change= invert (back, env);
}
language
ad_hoc_language (language base, tree hyphs) {
static hashmap<tree,int> abbrevs;
if (!abbrevs->contains (hyphs))
abbrevs (hyphs)= N (abbrevs);
string name= base->res_name * "-" * as_string (abbrevs [hyphs]);
if (language::instances -> contains (name)) return language (name);
return tm_new<ad_hoc_language_rep> (name, base, hyphs);
}
void
unregister_pointer (string id, observer which) {
// cout << "Unregister: " << id << " -> " << which << "\n";
// cout << "Unregister: " << id << " -> " << obtain_tree (which) << "\n";
list<observer>& l1= id_resolve (id);
l1= remove (l1, which);
if (is_nil (l1)) id_resolve->reset (id);
list<string>& l2= pointer_resolve (which);
l2= remove (l2, id);
if (is_nil (l2)) pointer_resolve->reset (which);
}
void
window_delete (int win) {
static hashmap<int,bool> busy (false);
if (busy->contains (win)) return;
busy (win)= true;
ASSERT (window_table->contains (win), "window does not exist");
widget pww= window_table [win];
window_table->reset (win);
send_destroy (pww);
destroy_window_widget (pww);
busy (win)= false;
}
std::vector<int> search(std::string s){
if (s.length() == 0) {
return indexes;
}else{
char first = s.at(0);
if (children.find(first) != children.end()) {
std::string remain(s.substr(1));
return children[first]->search(remain);
}
}
std::vector<int> tmp;
return tmp;
}
void insertString(std::string s, int index){
indexes.push_back(index);
if (s.size()>0) {
value = s.at(0);
std::shared_ptr<SuffixTreeNode> child;
if (children.find(value) != children.end()) {
child = children[value];
}else{
child.reset(new SuffixTreeNode());
children[value] = child;
}
std::string remain(s.substr(1));
child->insertString(remain, index);
}
}
static void
parse_other_op_index (hashmap<string,string>& t, string s, int& pos) {
int i;
i=13 ;
if( i > N(s) ) i=N(s) ;
for (i=13; i >= 1; i--) {
string r=s(pos,pos+i);
if (t->contains(r) && (t(r)=="op index")) {
pos=pos+i; return; }
if (t->contains(r) && (t(r)=="operator")) {
return; }
if (t->contains(r) && (t(r)=="op assign")) {
return; }
}
}
Future<http::Response> MetricsProcess::__snapshot(
const http::Request& request,
const hashmap<string, Future<double> >& metrics,
const hashmap<string, Option<Statistics<double> > >& statistics)
{
JSON::Object object;
foreachpair (const string& key, const Future<double>& value, metrics) {
// Value.
if (value.isReady()) {
object.values[key] = value.get();
}
// Statistics.
Option<Statistics<double> > statistics_ = statistics.get(key).get();
if (statistics_.isSome()) {
object.values[key + "/count"] = statistics_.get().count;
object.values[key + "/min"] = statistics_.get().min;
object.values[key + "/max"] = statistics_.get().max;
object.values[key + "/p50"] = statistics_.get().p50;
object.values[key + "/p90"] = statistics_.get().p90;
object.values[key + "/p95"] = statistics_.get().p95;
object.values[key + "/p99"] = statistics_.get().p99;
object.values[key + "/p999"] = statistics_.get().p999;
object.values[key + "/p9999"] = statistics_.get().p9999;
}
}
return http::OK(object, request.query.get("jsonp"));
}
Future<hashmap<string, double>> MetricsProcess::__snapshot(
const Option<Duration>& timeout,
const hashmap<string, Future<double>>& metrics,
const hashmap<string, Option<Statistics<double>>>& statistics)
{
hashmap<string, double> snapshot;
foreachpair (const string& key, const Future<double>& value, metrics) {
// TODO(dhamon): Maybe add the failure message for this metric to the
// response if value.isFailed().
if (value.isPending()) {
CHECK_SOME(timeout);
VLOG(1) << "Exceeded timeout of " << timeout.get()
<< " when attempting to get metric '" << key << "'";
} else if (value.isReady()) {
snapshot[key] = value.get();
}
Option<Statistics<double>> statistics_ = statistics.get(key).get();
if (statistics_.isSome()) {
snapshot[key + "/count"] = static_cast<double>(statistics_.get().count);
snapshot[key + "/min"] = statistics_.get().min;
snapshot[key + "/max"] = statistics_.get().max;
snapshot[key + "/p50"] = statistics_.get().p50;
snapshot[key + "/p90"] = statistics_.get().p90;
snapshot[key + "/p95"] = statistics_.get().p95;
snapshot[key + "/p99"] = statistics_.get().p99;
snapshot[key + "/p999"] = statistics_.get().p999;
snapshot[key + "/p9999"] = statistics_.get().p9999;
}
}
return snapshot;
}
tree
texmacs_invarianted (tree t, tree p, int c, string src,
hashmap<tree,tree> corr,
hashmap<tree,tree> pred,
hashmap<tree,tree> succ) {
if (corr->contains (t)) {
tree oids= corr[t], ids (TUPLE);
for (int i=0; i<N(oids); i++) {
int b, e;
if (get_range (oids[i], b, e, src)) ids << oids[i];
}
if (N(ids) >= 1) {
tree id= texmacs_best_match (ids, p, c, corr, pred, succ);
if (id != tree (UNINIT)) return compound ("ilx", id);
}
}
if (is_atomic (t)) return t;
else {
int i, n= N(t);
tree r (t, n);
for (i=0; i<n; i++)
r[i]= texmacs_invarianted (t[i], t, i, src, corr, pred, succ);
return r;
}
}
foreach (const Future<TaskStatus>& taskStatus, taskStatuses) {
AWAIT_READY(taskStatus);
Option<TaskState> taskState = taskStates.get(taskStatus->task_id());
ASSERT_SOME(taskState);
switch (taskState.get()) {
case TASK_STAGING: {
ASSERT_EQ(TASK_STARTING, taskStatus->state())
<< taskStatus->DebugString();
taskStates[taskStatus->task_id()] = TASK_STARTING;
break;
}
case TASK_STARTING: {
ASSERT_EQ(TASK_RUNNING, taskStatus->state())
<< taskStatus->DebugString();
taskStates[taskStatus->task_id()] = TASK_RUNNING;
break;
}
default: {
FAIL() << "Unexpected task update: " << taskStatus->DebugString();
break;
}
}
}
std::string stringify(const hashmap<K, V>& map)
{
std::ostringstream out;
out << "{ ";
typename hashmap<K, V>::const_iterator iterator = map.begin();
while (iterator != map.end()) {
out << stringify(iterator->first);
out << ": ";
out << stringify(iterator->second);
if (++iterator != map.end()) {
out << ", ";
}
}
out << " }";
return out.str();
}
static bool
is_special (string s) {
if (N (special_table) == 0) {
special_table ("*")= "";
special_table ("<noplus>")= "";
special_table ("<nocomma>")= "";
special_table ("<nospace>")= "";
special_table ("<nobracket>")= "";
special_table ("<nosymbol>")= "";
special_table ("-")= "<minus>";
special_table ("|")= "<mid>";
special_table ("'")= "<#2B9>";
special_table ("`")= "<backprime>";
}
if (starts (s, "<big-."))
special_table (s)= "";
if (starts (s, "<big-") && (ends (s, "-1>") || ends (s, "-2>"))) {
string ss= s (0, N(s)-3) * ">";
//cout << "Search " << ss << "\n";
if (unicode_provides (ss))
special_table (s)= ss;
ss= "<big" * s (5, N(s)-3) * ">";
//cout << "Search " << ss << "\n";
if (unicode_provides (ss))
special_table (s)= ss;
ss= "<" * s (5, N(s)-3) * ">";
if (ends (ss, "lim>")) ss= ss (0, N(ss)-4) * ">";
//cout << "Search " << ss << "\n";
if (unicode_provides (ss))
special_table (s)= ss;
}
return special_table->contains (s);
}
// Mapping from module ID to the actual module name.
Try<string> getModuleName(ModuleID id)
{
if (!moduleNames.contains(id)) {
return Error("Module '" + stringify(id) + "' not found");
}
return moduleNames[id];
}
bool
is_up_to_date (url dir) {
string name_dir= concretize (dir);
if (cache_valid->contains (name_dir)) return cache_valid [name_dir];
int l= last_modified (dir, false);
if (is_cached ("validate_cache.scm", name_dir)) {
int r= as_int (cache_get ("validate_cache.scm", name_dir) -> label);
if (l == r) {
cache_valid (name_dir)= true;
return true;
}
//cout << name_dir << " no longer up to date " << r << " -> " << l << "\n";
}
//else cout << name_dir << " not up to date " << l << "\n";
cache_set ("validate_cache.scm", name_dir, as_string (l));
cache_valid (name_dir)= false;
// FIXME: we should explicitly remove all data concerning files in 'dir'
// from the various caches. Indeed, at a next run of TeXmacs, the directory
// will be regarded as up to date, but the other caches may still contain
// outdated data. Careful: invalidating the cache lines should not
// give rise to a performance penaly (e.g. go through all entries of
// 'cache_data', or reloading unchanged files many times).
// See also 'declare_out_of_date'.
return false;
}
string
uni_upcase_char (string s) {
if (N(s) == 1) {
unsigned char c= s[0];
if ((c >= 'a' && c <= 'z') ||
(c >= ((unsigned char) 0xA0) && (c <= ((unsigned char) 0xBF))) ||
(c >= ((unsigned char) 0xE0)))
return string ((char) (c - 0x20));
return s;
}
else if (starts (s, "<#") && ends (s, ">")) {
int code= from_hexadecimal (s (2, N(s) - 1));
if (code >= 0x3AC && code <= 0x3CE) {
if (code >= 0x3B1 && code <= 0x3CB) code -= 0x20;
else if (code >= 0x3AC && code <= 0x3AC) code -= 0x26;
else if (code >= 0x3AD && code <= 0x3AF) code -= 0x25;
else if (code >= 0x3CC && code <= 0x3CC) code -= 0x40;
else if (code >= 0x3CD && code <= 0x3CE) code -= 0x3f;
}
else if (code >= 0x450 && code <= 0x45F) code -= 0x50;
else if (code >= 0x430 && code <= 0x44F) code -= 0x20;
else if (code >= 0x460 && code <= 0x4FF) {
if ((code & 1) == 1) code -= 1;
}
return "<#" * as_hexadecimal (code) * ">";
}
else {
init_case_tables ();
if (upcase_tab->contains (s)) return upcase_tab[s];
return s;
}
}
string
uni_locase_char (string s) {
if (N(s) == 1) {
unsigned char c= s[0];
if ((c >= 'A' && c <= 'Z') ||
(c >= ((unsigned char) 0x80) && (c <= ((unsigned char) 0x9F))) ||
(c >= ((unsigned char) 0xC0) && (c <= ((unsigned char) 0xDF))))
return string ((char) (c + 0x20));
return s;
}
else if (starts (s, "<#") && ends (s, ">")) {
int code= from_hexadecimal (s (2, N(s) - 1));
if (code >= 0x386 && code <= 0x3AB) {
if (code >= 0x391 && code <= 0x3AB) code += 0x20;
else if (code >= 0x386 && code <= 0x386) code += 0x26;
else if (code >= 0x388 && code <= 0x38A) code += 0x25;
else if (code >= 0x38C && code <= 0x38C) code += 0x40;
else if (code >= 0x38E && code <= 0x38F) code += 0x3f;
}
else if (code >= 0x400 && code <= 0x40F) code += 0x50;
else if (code >= 0x410 && code <= 0x42F) code += 0x20;
else if (code >= 0x460 && code <= 0x4FF) {
if ((code & 1) == 0) code += 1;
}
return "<#" * as_hexadecimal (code) * ">";
}
else {
init_case_tables ();
if (locase_tab->contains (s)) return locase_tab[s];
return s;
}
}
// Returns P_i(perm)
inline unsigned short getPval(perm_t perm, int i, const hashmap &Phashmap) {
// if (Phashmap.getpayload(perm) == NULL) { // !!
// cout<<"perm read failed: "<<endl;
// displayperm(perm);
// }
// assert(Phashmap.getpayload(perm) != NULL); // !!
return ((unsigned short*)Phashmap.getpayload(perm))[i];
}
static void
get_subtree_paths (tree t, path p, hashmap<tree,path>& h) {
if (h->contains (t)) h (t)= path (-1);
else h (t)= p;
if (is_compound (t))
for (int i=0; i<N(t); i++)
get_subtree_paths (t[i], p * i, h);
}
void
web_cache_invalidate (url name) {
for (int i=0; i<MAX_CACHED; i++)
if (web_cache[i] == name->t) {
web_cache[i]= tree ("");
web_cache_resolve->reset (name->t);
}
}
void
persistent_reset (url dir, string key) {
string v= local_prefix (dir) * key;
persistent_init_key (dir, key);
persistent_remove (dir, key, persistent_file [v], persistent_hash [v]);
persistent_has (v)= false;
persistent_cache->reset (v);
}
static url
set_cache (url name, url tmp) {
web_cache_resolve->reset (web_cache [web_nr]);
web_cache [web_nr]= name->t;
web_cache_resolve (name->t)= tmp->t;
web_nr= (web_nr+1) % MAX_CACHED;
return tmp;
}
void hashmap_iterator_remove(id self) {
hashmap_iterator itr = cast(hashmap_iterator, self);
hashmap hm = cast(hashmap, itr->host);
assert (itr->aux != NULL);
hashmap_entry curr = cast(hashmap_entry, itr->aux);
hm->remove(hm, curr->key);
itr->aux = NULL;
}
static void
parse_constant (hashmap<string,string>& t, string s, int& pos) {
int i=pos;
if (pos>=N(s)) return;
while ((i<N(s)) && belongs_to_identifier (s[i])) i++;
string r= s (pos, i);
if (t->contains (r) && t(r)=="#2060c0") { pos=i; return; }
}
array<int>
get_hyphens (string s,
hashmap<string,string> patterns,
hashmap<string,string> hyphenations, bool utf8) {
ASSERT (N(s) != 0, "hyphenation of empty string");
if (utf8) s= cork_to_utf8 (s);
if (hyphenations->contains (s)) {
string h= hyphenations [s];
array<int> penalty (str_length (s, utf8)-1);
int i=0, j=0;
while (h[j] == '-') j++;
i++; goto_next_char (h, j, utf8);
while (i < N(penalty)+1) {
penalty[i-1]= HYPH_INVALID;
while (j < N(h) && h[j] == '-') {
penalty[i-1]= HYPH_STD;
j++;
}
i++;
goto_next_char (h, j, utf8);
}
//cout << s << " --> " << penalty << "\n";
return penalty;
}
else {
s= "." * lower_case (s) * ".";
// cout << s << "\n";
int i, j, k, l, m, len;
array<int> T (str_length (s, utf8)+1);
for (i=0; i<N(T); i++) T[i]=0;
for (len=1; len < MAX_SEARCH; len++)
for (i=0, l=0; i<N(s) - len; goto_next_char (s, i, utf8), l++) {
string r= patterns [s (i, i+len)];
if (!(r == "?")) {
// cout << " " << s (i, i+len) << " => " << r << "\n";
for (j=0, k=0; j<=len; j++, k++) {
if ((k<N(r)) && (r[k]>='0') && (r[k]<='9')) {
m=((int) r[k])-((int) '0');
k++;
}
else m=0;
if (m>T[l+j]) T[l+j]=m;
}
}
}
array<int> penalty (N(T)-4);
for (i=2; i < N(T)-4; i++)
penalty [i-2]= (((T[i]&1)==1)? HYPH_STD: HYPH_INVALID);
if (N(penalty)>0) penalty[0] = penalty[N(penalty)-1] = HYPH_INVALID;
if (N(penalty)>1) penalty[1] = penalty[N(penalty)-2] = HYPH_INVALID;
if (N(penalty)>2) penalty[N(penalty)-3] = HYPH_INVALID;
// cout << s << " --> " << penalty << "\n";
return penalty;
}
}
string
persistent_get (url dir, string key) {
string v= local_prefix (dir) * key;
if (!persistent_has->contains (v)) {
persistent_init_key (dir, key);
persistent_retrieve (dir, key, persistent_file [v], persistent_hash [v]);
}
return persistent_cache [v];
}
void
edit_env_rep::local_end (hashmap<string,tree>& prev_back) {
int i=0, n=back->n;
for (; i<n; i++) {
list<hashentry<string,tree> > l=back->a[i];
for (; !is_nil(l); l=l->next)
prev_back->write_back (l->item.key, back);
}
back= prev_back;
}
void
ad_hoc_language_rep::hyphenate (
string s, int after, string& left, string& right)
{
if (hyphens->contains (s)) {
array<int> penalty= get_hyphens (s);
std_hyphenate (s, after, left, right, penalty[after]);
}
else base->hyphenate (s, after, left, right);
}
// SPIMI-1: write current postings to intermediate files
//
// {word} => {ndoc => [doc][frq]}
// pst.trm.x pst.doc.x
//
// where x is the id of current intermediate file
//
void IndexWriter::flushPSTBlk(hashmap<string, vector<pair<int, int> > >&pst, int turn) {
ofstream ftrm, fdoc;
string prefix = path+"/"+POSTINGS_FILE;
ftrm.open((prefix+".trm."+itoa(turn)).c_str(), ios::binary);
fdoc.open((prefix+".doc."+itoa(turn)).c_str(), ios::binary);
hashmap<string, vector<pair<int, int> > >::iterator it;
vector<pair<int, int> >::iterator jt;
vector<string> list;
for (it = pst.begin(); it != pst.end(); ++it) {
list.push_back(it->first);
}
sort(list.begin(), list.end());
for (unsigned i = 0; i < list.size(); ++i) {
it = pst.find(list[i]);
string &term = list[i];
int ndoc = it->second.size(), cnt = 0, accum = 0;
for (jt = it->second.begin(); jt != it->second.end(); ++jt) {
int did = jt->first;
int frq = jt->second;
didbuf[cnt] = did;
frqbuf[cnt] = frq;
accum += frq;
assert(accum > 0 || !(cout << accum << endl));
cnt++;
}
fwrite(fdoc, &ndoc, sizeof(ndoc));
fwrite(fdoc, didbuf, sizeof(didbuf[0])*ndoc);
fwrite(fdoc, frqbuf, sizeof(frqbuf[0])*ndoc);
TermAttr attr;
attr.str = term;
attr.df = ndoc;
attr.cf = accum;
attr.flush(ftrm);
}
ftrm.close();
fdoc.close();
}
