DataStore.cpp 13.6 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
#include <vector>
#include <functional>
#include <iostream>
#include <yaml-cpp/yaml.h>
#include <sdskv-client.h>
#include <bake-client.h>
#include <ch-placement.h>
#include "hepnos/Exception.hpp"
#include "hepnos/DataStore.hpp"
#include "hepnos/DataSet.hpp"
11
#include "private/DataStoreImpl.hpp"
12 13 14

namespace hepnos {

15 16 17 18
////////////////////////////////////////////////////////////////////////////////////////////
// DataStore implementation
////////////////////////////////////////////////////////////////////////////////////////////

19
DataStore::DataStore(const std::string& configFile) 
20
: m_impl(std::make_unique<DataStore::Impl>(this)) {
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41
    m_impl->init(configFile);
}

DataStore::DataStore(DataStore&& other)
: m_impl(std::move(other.m_impl)) {}

DataStore& DataStore::operator=(DataStore&& other) {
    if(&other == this) return *this;
    if(m_impl) {
        m_impl->cleanup();
    }
    m_impl = std::move(other.m_impl);
}
    
DataStore::~DataStore() {
    if(m_impl) {
        m_impl->cleanup();
    }
}

DataStore::iterator DataStore::find(const std::string& datasetName) {
42
    int ret;
43 44 45
    if(datasetName.find('/') != std::string::npos
    || datasetName.find('%') != std::string::npos) {
        throw Exception("Invalid character ('/' or '%') in dataset name");
46
    }
47
    std::vector<char> data;
48
    bool b = m_impl->load(1, "", datasetName, data);
49 50
    if(!b) {
        return m_impl->m_end;
51
    }
52
    return iterator(DataSet(this, 1, datasetName));
53 54
}

Matthieu Dorier's avatar
Matthieu Dorier committed
55 56 57 58 59
DataSet DataStore::operator[](const std::string& datasetName) const {
    auto it = find(datasetName);
    return std::move(*it);
}

60
DataStore::const_iterator DataStore::find(const std::string& datasetName) const {
61 62
    DataStore::iterator it = const_cast<DataStore*>(this)->find(datasetName);
    return it;
63 64 65
}

DataStore::iterator DataStore::begin() {
66
    DataSet ds(this, 1, "", "");
67
    ds = ds.next();
68
    if(ds.valid()) return iterator(std::move(ds));
69
    else return end();
70 71
}

72 73 74 75 76
DataStore::const_iterator DataStore::begin() const {
    iterator it = const_cast<DataStore*>(this)->begin();
    return const_iterator(std::move(it));
}

77 78 79 80
DataStore::iterator DataStore::end() {
    return m_impl->m_end;
}

81 82 83 84
DataStore::const_iterator DataStore::end() const {
    return m_impl->m_end;
}

85
DataStore::const_iterator DataStore::cbegin() const {
86
    return const_cast<DataStore*>(this)->begin();
87 88 89
}

DataStore::const_iterator DataStore::cend() const {
90
    return const_cast<DataStore*>(this)->end();
91 92 93
}

DataStore::iterator DataStore::lower_bound(const std::string& lb) {
94 95 96 97 98 99 100 101 102
    std::string lb2 = lb;
    size_t s = lb2.size();
    lb2[s-1] -= 1; // sdskv_list_keys's start_key is exclusive
    iterator it = find(lb2);
    if(it != end()) {
        // we found something before the specified lower bound
        ++it;
        return it;
    }
103
    DataSet ds(this, 1, "", lb2);
104 105
    ds = ds.next();
    if(!ds.valid()) return end();
106
    else return iterator(std::move(ds));
107 108 109
}

DataStore::const_iterator DataStore::lower_bound(const std::string& lb) const {
110 111
    iterator it = const_cast<DataStore*>(this)->lower_bound(lb);
    return it;
112 113 114
}

DataStore::iterator DataStore::upper_bound(const std::string& ub) {
115
    DataSet ds(this, 1, "", ub);
116 117
    ds = ds.next();
    if(!ds.valid()) return end();
118
    else return iterator(std::move(ds));
119 120 121
}

DataStore::const_iterator DataStore::upper_bound(const std::string& ub) const {
122 123
    iterator it = const_cast<DataStore*>(this)->upper_bound(ub);
    return it;
124 125 126
}

DataSet DataStore::createDataSet(const std::string& name) {
127 128 129
    if(name.find('/') != std::string::npos
    || name.find('%') != std::string::npos) {
        throw Exception("Invalid character ('/' or '%') in dataset name");
130
    }
131
    m_impl->store(1, "", name, std::vector<char>());
132
    return DataSet(this, 1, "", name);
133
}
134
#if 0
135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154
bool DataStore::load(uint8_t level, const std::string& containerName,
        const std::string& objectName, std::vector<char>& data) const {
    int ret;
    // build full name
    std::stringstream ss;
    if(!containerName.empty())
        ss << containerName << "/";
    ss << objectName;
    // hash the name to get the provider id
    long unsigned provider_idx = 0;
    if(level != 0) {
        uint64_t h = std::hash<std::string>()(containerName);
        ch_placement_find_closest(m_impl->m_chi_sdskv, h, 1, &provider_idx);
    } else {
        // use the complete name for final objects (level 255)
        uint64_t h = std::hash<std::string>()(ss.str());
        ch_placement_find_closest(m_impl->m_chi_sdskv, h, 1, &provider_idx);
    }
    // make corresponding datastore entry
    DataStoreEntryPtr entry = make_datastore_entry(level, ss.str());
155 156
    auto ph = m_impl->m_sdskv_ph[provider_idx];
    auto db_id = m_impl->m_sdskv_db[provider_idx];
157 158 159 160 161 162 163 164 165
    // find the size of the value, as a way to check if the key exists
    hg_size_t vsize;
    std::cerr << "[LOG] load (level=" << (int)level
        << ", container=\"" << containerName << "\", object=\""
        << objectName << "\")" << std::endl;
    ret = sdskv_length(ph, db_id, entry->raw(), entry->length(), &vsize);
    if(ret == SDSKV_ERR_UNKNOWN_KEY) {
        return false;
    }
166
    if(ret != SDSKV_SUCCESS) {
167
        throw Exception("Error occured when calling sdskv_length");
168
    }
169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206
    // read the value
    data.resize(vsize);
    ret = sdskv_get(ph, db_id, entry->raw(), entry->length(), data.data(), &vsize);
    if(ret != SDSKV_SUCCESS) {
        throw Exception("Error occured when calling sdskv_get");
    }
    return true;
}

bool DataStore::store(uint8_t level, const std::string& containerName,
        const std::string& objectName, const std::vector<char>& data) {
    // build full name
    std::stringstream ss;
    if(!containerName.empty())
        ss << containerName << "/";
    ss << objectName;
    // hash the name to get the provider id
    long unsigned provider_idx = 0;
    if(level != 0) {
        uint64_t h = std::hash<std::string>()(containerName);
        ch_placement_find_closest(m_impl->m_chi_sdskv, h, 1, &provider_idx);
    } else {
        // use the complete name for final objects (level 0)
        uint64_t h = std::hash<std::string>()(ss.str());
        ch_placement_find_closest(m_impl->m_chi_sdskv, h, 1, &provider_idx);
    }
    // make corresponding datastore entry
    DataStoreEntryPtr entry = make_datastore_entry(level, ss.str());
    auto ph = m_impl->m_sdskv_ph[provider_idx];
    auto db_id = m_impl->m_sdskv_db[provider_idx];
    std::cerr << "[LOG] store (level=" << (int)level
        << ", container=\"" << containerName << "\", object=\""
        << objectName << "\")" << std::endl;
    int ret = sdskv_put(ph, db_id, entry->raw(), entry->length(), data.data(), data.size());
    if(ret != SDSKV_SUCCESS) {
        throw Exception("Could not put key/value pair in SDSKV (sdskv_put error)");
    }
    return true;
207 208
}

209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273
size_t DataStore::nextKeys(uint8_t level, const std::string& containerName,
        const std::string& lower,
        std::vector<std::string>& keys, size_t maxKeys) const {
    int ret;
    if(level == 0) return 0; // cannot iterate at object level
    // build full name from lower bound key
    std::stringstream ss;
    if(!containerName.empty())
        ss << containerName << "/";
    ss << lower;
    // hash the name to get the provider id
    long unsigned provider_idx = 0;
    uint64_t h = std::hash<std::string>()(containerName);
    ch_placement_find_closest(m_impl->m_chi_sdskv, h, 1, &provider_idx);
    // make an entry for the lower bound
    DataStoreEntryPtr lb_entry = make_datastore_entry(level, ss.str());
    // create data structures to receive keys
    std::vector<DataStoreEntryPtr> keys_ent;
    std::vector<void*>             keys_ptr(maxKeys);
    std::vector<hg_size_t>         keys_len(maxKeys);
    for(unsigned i=0; i < maxKeys; i++) {
        keys_ent.push_back(make_datastore_entry(level, 1024));
        keys_ptr[i] = keys_ent[i]->raw();
        keys_len[i] = sizeof(DataStoreEntry) + 1024;
    }
    // get provider and database
    auto ph = m_impl->m_sdskv_ph[provider_idx];
    auto db_id = m_impl->m_sdskv_db[provider_idx];
    // issue an sdskv_list_keys
    hg_size_t max_keys = maxKeys;
    std::cerr << "[LOG] list keys (level=" << (int)level
        << ", container=\"" << containerName << "\", greaterthan=\""
        << lower << "\")" << std::endl;
    ret = sdskv_list_keys(ph, db_id, lb_entry->raw(), lb_entry->length(),
                          keys_ptr.data(), keys_len.data(), &max_keys);
    if(ret != HG_SUCCESS) {
        throw Exception("Error occured when calling sdskv_list_keys");
    }
    unsigned i = max_keys - 1;
    if(max_keys == 0) return 0;
    // remove keys that don't have the same level or the same prefix
    std::string prefix = containerName + "/";
    keys.resize(0);
    for(unsigned i = 0; i < max_keys; i++) {
        if(keys_ent[i]->m_level != level) {
            max_keys = i;
            break;
        }
        if(!containerName.empty()) {
            size_t lenpre = prefix.size();
            size_t lenstr = strlen(keys_ent[i]->m_fullname);
            if(lenstr < lenpre) {
                max_keys = i;
                break;
            }
            if(strncmp(prefix.c_str(), keys_ent[i]->m_fullname, lenpre) != 0) {
                max_keys = i;
                break;
            }
        }
        keys.push_back(keys_ent[i]->m_fullname);
    }
    // set the resulting keys
    return max_keys;
}
274
#endif
275 276 277 278
////////////////////////////////////////////////////////////////////////////////////////////
// DataStore::const_iterator::Impl implementation
////////////////////////////////////////////////////////////////////////////////////////////

279 280 281 282
class DataStore::const_iterator::Impl {
    public:
        DataSet    m_current_dataset;

283 284
        Impl()
        : m_current_dataset()
285
        {}
286

287 288 289 290 291 292
        Impl(const DataSet& dataset)
        : m_current_dataset(dataset)
        {}

        Impl(DataSet&& dataset)
        : m_current_dataset(std::move(dataset))
293
        {}
294 295

        Impl(const Impl& other)
296
        : m_current_dataset(other.m_current_dataset) 
297 298 299
        {}

        bool operator==(const Impl& other) const {
300
            return m_current_dataset == other.m_current_dataset;
301
        }
302 303
};

304 305 306
////////////////////////////////////////////////////////////////////////////////////////////
// DataStore::const_iterator::Impl implementation
////////////////////////////////////////////////////////////////////////////////////////////
307

308 309
DataStore::const_iterator::const_iterator()
: m_impl(std::make_unique<Impl>()) {}
310

311 312 313 314 315
DataStore::const_iterator::const_iterator(const DataSet& dataset)
: m_impl(std::make_unique<Impl>(dataset)) {}

DataStore::const_iterator::const_iterator(DataSet&& dataset)
: m_impl(std::make_unique<Impl>(std::move(dataset))) {}
316

317
DataStore::const_iterator::~const_iterator() {}
318 319

DataStore::const_iterator::const_iterator(const DataStore::const_iterator& other) 
320
: m_impl(std::make_unique<Impl>(*other.m_impl)) {}
321 322

DataStore::const_iterator::const_iterator(DataStore::const_iterator&& other) 
323
: m_impl(std::move(other.m_impl)) {}
324

325 326 327 328
DataStore::const_iterator& DataStore::const_iterator::operator=(const DataStore::const_iterator& other) {
    if(&other == this) return *this;
    m_impl = std::make_unique<Impl>(*other.m_impl);
    return *this;
329 330
}

331 332 333 334
DataStore::const_iterator& DataStore::const_iterator::operator=(DataStore::const_iterator&& other) {
    if(&other == this) return *this;
    m_impl = std::move(other.m_impl);
    return *this;
335 336 337
}

DataStore::const_iterator::self_type DataStore::const_iterator::operator++() {
338 339 340 341 342
    if(!m_impl) {
        throw Exception("Trying to increment an invalid iterator");
    }
    m_impl->m_current_dataset = m_impl->m_current_dataset.next();
    return *this;
343 344 345
}

DataStore::const_iterator::self_type DataStore::const_iterator::operator++(int) {
346 347 348
    const_iterator copy = *this;
    ++(*this);
    return copy;
349 350 351
}

const DataStore::const_iterator::reference DataStore::const_iterator::operator*() {
352 353 354 355
    if(!m_impl) {
        throw Exception("Trying to dereference an invalid iterator");
    }
    return m_impl->m_current_dataset;
356 357 358
}

const DataStore::const_iterator::pointer DataStore::const_iterator::operator->() {
359 360
    if(!m_impl) return nullptr;
    return &(m_impl->m_current_dataset);
361 362 363
}

bool DataStore::const_iterator::operator==(const self_type& rhs) const {
364 365 366 367
    if(!m_impl && !rhs.m_impl)  return true;
    if(m_impl  && !rhs.m_impl)  return false;
    if(!m_impl && rhs.m_impl)   return false;
    return *m_impl == *(rhs.m_impl);
368 369 370
}

bool DataStore::const_iterator::operator!=(const self_type& rhs) const {
371
    return !(*this == rhs);
372 373
}

374 375 376 377
////////////////////////////////////////////////////////////////////////////////////////////
// DataStore::iterator implementation
////////////////////////////////////////////////////////////////////////////////////////////

378 379 380 381 382
DataStore::iterator::iterator(const DataSet& current)
: const_iterator(current) {}

DataStore::iterator::iterator(DataSet&& current)
: const_iterator(std::move(current)) {}
383

384 385
DataStore::iterator::iterator()
: const_iterator() {}
386

387
DataStore::iterator::~iterator() {}
388 389

DataStore::iterator::iterator(const DataStore::iterator& other)
390
: const_iterator(other) {}
391 392

DataStore::iterator::iterator(DataStore::iterator&& other) 
393
: const_iterator(std::move(other)) {}
394 395

DataStore::iterator& DataStore::iterator::operator=(const DataStore::iterator& other) {
396 397 398
    if(this == &other) return *this;
    m_impl = std::make_unique<Impl>(*other.m_impl);
    return *this;
399 400 401
}

DataStore::iterator& DataStore::iterator::operator=(DataStore::iterator&& other) {
402 403 404
    if(this == &other) return *this;
    m_impl = std::move(other.m_impl);
    return *this;
405 406 407
}

DataStore::iterator::reference DataStore::iterator::operator*() {
408
    return const_cast<reference>(const_iterator::operator*());
409 410 411
}

DataStore::iterator::pointer DataStore::iterator::operator->() {
412
    return const_cast<pointer>(const_iterator::operator->());
413 414 415 416
}

}