|
| 1 | +From deeb435829d73524df851f6f4c2d4be552c99230 Mon Sep 17 00:00:00 2001 |
| 2 | +From: Dmitry Vedenko <dmitry@crsib.me> |
| 3 | +Date: Fri, 1 Oct 2021 16:21:22 +0300 |
| 4 | +Subject: [PATCH] Use a different approach to estimate the disk space usage |
| 5 | + |
| 6 | +New a approach is a bit less precise, but removes the requirement for the "private" SQLite3 table and allows Audacity to be built against system SQLite3. |
| 7 | +--- |
| 8 | + cmake-proxies/sqlite/CMakeLists.txt | 5 - |
| 9 | + src/DBConnection.h | 4 +- |
| 10 | + src/ProjectFileIO.cpp | 269 +++++----------------------- |
| 11 | + 3 files changed, 44 insertions(+), 234 deletions(-) |
| 12 | + |
| 13 | +diff --git a/cmake-proxies/sqlite/CMakeLists.txt b/cmake-proxies/sqlite/CMakeLists.txt |
| 14 | +index 63d70637c..d7b9b95ef 100644 |
| 15 | +--- a/cmake-proxies/sqlite/CMakeLists.txt |
| 16 | ++++ b/cmake-proxies/sqlite/CMakeLists.txt |
| 17 | +@@ -19,11 +19,6 @@ list( APPEND INCLUDES |
| 18 | + |
| 19 | + list( APPEND DEFINES |
| 20 | + PRIVATE |
| 21 | +- # |
| 22 | +- # We need the dbpage table for space calculations. |
| 23 | +- # |
| 24 | +- SQLITE_ENABLE_DBPAGE_VTAB=1 |
| 25 | +- |
| 26 | + # Can't be set after a WAL mode database is initialized, so change |
| 27 | + # the default here to ensure all project files get the same page |
| 28 | + # size. |
| 29 | +diff --git a/src/DBConnection.h b/src/DBConnection.h |
| 30 | +index 16a7fc9d4..07d3af95e 100644 |
| 31 | +--- a/src/DBConnection.h |
| 32 | ++++ b/src/DBConnection.h |
| 33 | +@@ -75,8 +75,8 @@ public: |
| 34 | + LoadSampleBlock, |
| 35 | + InsertSampleBlock, |
| 36 | + DeleteSampleBlock, |
| 37 | +- GetRootPage, |
| 38 | +- GetDBPage |
| 39 | ++ GetSampleBlockSize, |
| 40 | ++ GetAllSampleBlocksSize |
| 41 | + }; |
| 42 | + sqlite3_stmt *Prepare(enum StatementID id, const char *sql); |
| 43 | + |
| 44 | +diff --git a/src/ProjectFileIO.cpp b/src/ProjectFileIO.cpp |
| 45 | +index 3b3e2e1fd..c9bc45af4 100644 |
| 46 | +--- a/src/ProjectFileIO.cpp |
| 47 | ++++ b/src/ProjectFileIO.cpp |
| 48 | +@@ -35,6 +35,7 @@ Paul Licameli split from AudacityProject.cpp |
| 49 | + #include "widgets/ProgressDialog.h" |
| 50 | + #include "wxFileNameWrapper.h" |
| 51 | + #include "xml/XMLFileReader.h" |
| 52 | ++#include "MemoryX.h"` |
| 53 | + |
| 54 | + #undef NO_SHM |
| 55 | + #if !defined(__WXMSW__) |
| 56 | +@@ -2357,255 +2358,69 @@ int64_t ProjectFileIO::GetTotalUsage() |
| 57 | + } |
| 58 | + |
| 59 | + // |
| 60 | +-// Returns the amount of disk space used by the specified sample blockid or all |
| 61 | +-// of the sample blocks if the blockid is 0. It does this by using the raw SQLite |
| 62 | +-// pages available from the "sqlite_dbpage" virtual table to traverse the SQLite |
| 63 | +-// table b-tree described here: https://www.sqlite.org/fileformat.html |
| 64 | ++// Returns the estimation of disk space used by the specified sample blockid or all |
| 65 | ++// of the sample blocks if the blockid is 0. This does not include small overhead |
| 66 | ++// of the internal SQLite structures, only the size used by the data |
| 67 | + // |
| 68 | + int64_t ProjectFileIO::GetDiskUsage(DBConnection &conn, SampleBlockID blockid /* = 0 */) |
| 69 | + { |
| 70 | +- // Information we need to track our travels through the b-tree |
| 71 | +- typedef struct |
| 72 | +- { |
| 73 | +- int64_t pgno; |
| 74 | +- int currentCell; |
| 75 | +- int numCells; |
| 76 | +- unsigned char data[65536]; |
| 77 | +- } page; |
| 78 | +- std::vector<page> stack; |
| 79 | +- |
| 80 | +- int64_t total = 0; |
| 81 | +- int64_t found = 0; |
| 82 | +- int64_t right = 0; |
| 83 | +- int rc; |
| 84 | ++ sqlite3_stmt* stmt = nullptr; |
| 85 | + |
| 86 | +- // Get the rootpage for the sampleblocks table. |
| 87 | +- sqlite3_stmt *stmt = |
| 88 | +- conn.Prepare(DBConnection::GetRootPage, |
| 89 | +- "SELECT rootpage FROM sqlite_master WHERE tbl_name = 'sampleblocks';"); |
| 90 | +- if (stmt == nullptr || sqlite3_step(stmt) != SQLITE_ROW) |
| 91 | ++ if (blockid == 0) |
| 92 | + { |
| 93 | +- return 0; |
| 94 | +- } |
| 95 | +- |
| 96 | +- // And store it in our first stack frame |
| 97 | +- stack.push_back({sqlite3_column_int64(stmt, 0)}); |
| 98 | ++ static const char* statement = |
| 99 | ++R"(SELECT |
| 100 | ++ sum(length(blockid) + length(sampleformat) + |
| 101 | ++ length(summin) + length(summax) + length(sumrms) + |
| 102 | ++ length(summary256) + length(summary64k) + |
| 103 | ++ length(samples)) |
| 104 | ++FROM sampleblocks;)"; |
| 105 | + |
| 106 | +- // All done with the statement |
| 107 | +- sqlite3_clear_bindings(stmt); |
| 108 | +- sqlite3_reset(stmt); |
| 109 | +- |
| 110 | +- // Prepare/retrieve statement to read raw database page |
| 111 | +- stmt = conn.Prepare(DBConnection::GetDBPage, |
| 112 | +- "SELECT data FROM sqlite_dbpage WHERE pgno = ?1;"); |
| 113 | +- if (stmt == nullptr) |
| 114 | +- { |
| 115 | +- return 0; |
| 116 | ++ stmt = conn.Prepare(DBConnection::GetAllSampleBlocksSize, statement); |
| 117 | + } |
| 118 | +- |
| 119 | +- // Traverse the b-tree until we've visited all of the leaf pages or until |
| 120 | +- // we find the one corresponding to the passed in sample blockid. Because we |
| 121 | +- // use an integer primary key for the sampleblocks table, the traversal will |
| 122 | +- // be in ascending blockid sequence. |
| 123 | +- do |
| 124 | ++ else |
| 125 | + { |
| 126 | +- // Acces the top stack frame |
| 127 | +- page &pg = stack.back(); |
| 128 | ++ static const char* statement = |
| 129 | ++R"(SELECT |
| 130 | ++ length(blockid) + length(sampleformat) + |
| 131 | ++ length(summin) + length(summax) + length(sumrms) + |
| 132 | ++ length(summary256) + length(summary64k) + |
| 133 | ++ length(samples) |
| 134 | ++FROM sampleblocks WHERE blockid = ?1;)"; |
| 135 | + |
| 136 | +- // Read the page from the sqlite_dbpage table if it hasn't yet been loaded |
| 137 | +- if (pg.numCells == 0) |
| 138 | +- { |
| 139 | +- // Bind the page number |
| 140 | +- sqlite3_bind_int64(stmt, 1, pg.pgno); |
| 141 | ++ stmt = conn.Prepare(DBConnection::GetSampleBlockSize, statement); |
| 142 | ++ } |
| 143 | + |
| 144 | +- // And retrieve the page |
| 145 | +- if (sqlite3_step(stmt) != SQLITE_ROW) |
| 146 | ++ auto cleanup = finally( |
| 147 | ++ [stmt]() { |
| 148 | ++ // Clear statement bindings and rewind statement |
| 149 | ++ if (stmt != nullptr) |
| 150 | + { |
| 151 | +- // REVIEW: Likely harmless failure - says size is zero on |
| 152 | +- // this error. |
| 153 | +- // LLL: Yea, but not much else we can do. |
| 154 | +- return 0; |
| 155 | ++ sqlite3_clear_bindings(stmt); |
| 156 | ++ sqlite3_reset(stmt); |
| 157 | + } |
| 158 | ++ }); |
| 159 | + |
| 160 | +- // Copy the page content to the stack frame |
| 161 | +- memcpy(&pg.data, |
| 162 | +- sqlite3_column_blob(stmt, 0), |
| 163 | +- sqlite3_column_bytes(stmt, 0)); |
| 164 | +- |
| 165 | +- // And retrieve the total number of cells within it |
| 166 | +- pg.numCells = get2(&pg.data[3]); |
| 167 | +- |
| 168 | +- // Reset statement for next usage |
| 169 | +- sqlite3_clear_bindings(stmt); |
| 170 | +- sqlite3_reset(stmt); |
| 171 | +- } |
| 172 | +- |
| 173 | +- //wxLogDebug("%*.*spgno %lld currentCell %d numCells %d", (stack.size() - 1) * 2, (stack.size() - 1) * 2, "", pg.pgno, pg.currentCell, pg.numCells); |
| 174 | +- |
| 175 | +- // Process an interior table b-tree page |
| 176 | +- if (pg.data[0] == 0x05) |
| 177 | +- { |
| 178 | +- // Process the next cell if we haven't examined all of them yet |
| 179 | +- if (pg.currentCell < pg.numCells) |
| 180 | +- { |
| 181 | +- // Remember the right-most leaf page number. |
| 182 | +- right = get4(&pg.data[8]); |
| 183 | +- |
| 184 | +- // Iterate over the cells. |
| 185 | +- // |
| 186 | +- // If we're not looking for a specific blockid, then we always push the |
| 187 | +- // target page onto the stack and leave the loop after a single iteration. |
| 188 | +- // |
| 189 | +- // Otherwise, we match the blockid against the highest integer key contained |
| 190 | +- // within the cell and if the blockid falls within the cell, we stack the |
| 191 | +- // page and stop the iteration. |
| 192 | +- // |
| 193 | +- // In theory, we could do a binary search for a specific blockid here, but |
| 194 | +- // because our sample blocks are always large, we will get very few cells |
| 195 | +- // per page...usually 6 or less. |
| 196 | +- // |
| 197 | +- // In both cases, the stacked page can be either an internal or leaf page. |
| 198 | +- bool stacked = false; |
| 199 | +- while (pg.currentCell < pg.numCells) |
| 200 | +- { |
| 201 | +- // Get the offset to this cell using the offset in the cell pointer |
| 202 | +- // array. |
| 203 | +- // |
| 204 | +- // The cell pointer array starts immediately after the page header |
| 205 | +- // at offset 12 and the retrieved offset is from the beginning of |
| 206 | +- // the page. |
| 207 | +- int celloff = get2(&pg.data[12 + (pg.currentCell * 2)]); |
| 208 | +- |
| 209 | +- // Bump to the next cell for the next iteration. |
| 210 | +- pg.currentCell++; |
| 211 | +- |
| 212 | +- // Get the page number this cell describes |
| 213 | +- int pagenum = get4(&pg.data[celloff]); |
| 214 | +- |
| 215 | +- // And the highest integer key, which starts at offset 4 within the cell. |
| 216 | +- int64_t intkey = 0; |
| 217 | +- get_varint(&pg.data[celloff + 4], &intkey); |
| 218 | +- |
| 219 | +- //wxLogDebug("%*.*sinternal - right %lld celloff %d pagenum %d intkey %lld", (stack.size() - 1) * 2, (stack.size() - 1) * 2, " ", right, celloff, pagenum, intkey); |
| 220 | +- |
| 221 | +- // Stack the described page if we're not looking for a specific blockid |
| 222 | +- // or if this page contains the given blockid. |
| 223 | +- if (!blockid || blockid <= intkey) |
| 224 | +- { |
| 225 | +- stack.push_back({pagenum, 0, 0}); |
| 226 | +- stacked = true; |
| 227 | +- break; |
| 228 | +- } |
| 229 | +- } |
| 230 | +- |
| 231 | +- // If we pushed a new page onto the stack, we need to jump back up |
| 232 | +- // to read the page |
| 233 | +- if (stacked) |
| 234 | +- { |
| 235 | +- continue; |
| 236 | +- } |
| 237 | +- } |
| 238 | ++ if (blockid != 0) |
| 239 | ++ { |
| 240 | ++ int rc = sqlite3_bind_int64(stmt, 1, blockid); |
| 241 | + |
| 242 | +- // We've exhausted all the cells with this page, so we stack the right-most |
| 243 | +- // leaf page. Ensure we only process it once. |
| 244 | +- if (right) |
| 245 | +- { |
| 246 | +- stack.push_back({right, 0, 0}); |
| 247 | +- right = 0; |
| 248 | +- continue; |
| 249 | +- } |
| 250 | +- } |
| 251 | +- // Process a leaf table b-tree page |
| 252 | +- else if (pg.data[0] == 0x0d) |
| 253 | ++ if (rc != SQLITE_OK) |
| 254 | + { |
| 255 | +- // Iterate over the cells |
| 256 | +- // |
| 257 | +- // If we're not looking for a specific blockid, then just accumulate the |
| 258 | +- // payload sizes. We will be reading every leaf page in the sampleblocks |
| 259 | +- // table. |
| 260 | +- // |
| 261 | +- // Otherwise we break out when we find the matching blockid. In this case, |
| 262 | +- // we only ever look at 1 leaf page. |
| 263 | +- bool stop = false; |
| 264 | +- for (int i = 0; i < pg.numCells; i++) |
| 265 | +- { |
| 266 | +- // Get the offset to this cell using the offset in the cell pointer |
| 267 | +- // array. |
| 268 | +- // |
| 269 | +- // The cell pointer array starts immediately after the page header |
| 270 | +- // at offset 8 and the retrieved offset is from the beginning of |
| 271 | +- // the page. |
| 272 | +- int celloff = get2(&pg.data[8 + (i * 2)]); |
| 273 | +- |
| 274 | +- // Get the total payload size in bytes of the described row. |
| 275 | +- int64_t payload = 0; |
| 276 | +- int digits = get_varint(&pg.data[celloff], &payload); |
| 277 | +- |
| 278 | +- // Get the integer key for this row. |
| 279 | +- int64_t intkey = 0; |
| 280 | +- get_varint(&pg.data[celloff + digits], &intkey); |
| 281 | +- |
| 282 | +- //wxLogDebug("%*.*sleaf - celloff %4d intkey %lld payload %lld", (stack.size() - 1) * 2, (stack.size() - 1) * 2, " ", celloff, intkey, payload); |
| 283 | +- |
| 284 | +- // Add this payload size to the total if we're not looking for a specific |
| 285 | +- // blockid |
| 286 | +- if (!blockid) |
| 287 | +- { |
| 288 | +- total += payload; |
| 289 | +- } |
| 290 | +- // Otherwise, return the payload size for a matching row |
| 291 | +- else if (blockid == intkey) |
| 292 | +- { |
| 293 | +- return payload; |
| 294 | +- } |
| 295 | +- } |
| 296 | ++ conn.ThrowException(false); |
| 297 | + } |
| 298 | ++ } |
| 299 | + |
| 300 | +- // Done with the current branch, so pop back up to the previous one (if any) |
| 301 | +- stack.pop_back(); |
| 302 | +- } while (!stack.empty()); |
| 303 | +- |
| 304 | +- // Return the total used for all sample blocks |
| 305 | +- return total; |
| 306 | +-} |
| 307 | +- |
| 308 | +-// Retrieves a 2-byte big-endian integer from the page data |
| 309 | +-unsigned int ProjectFileIO::get2(const unsigned char *ptr) |
| 310 | +-{ |
| 311 | +- return (ptr[0] << 8) | ptr[1]; |
| 312 | +-} |
| 313 | +- |
| 314 | +-// Retrieves a 4-byte big-endian integer from the page data |
| 315 | +-unsigned int ProjectFileIO::get4(const unsigned char *ptr) |
| 316 | +-{ |
| 317 | +- return ((unsigned int) ptr[0] << 24) | |
| 318 | +- ((unsigned int) ptr[1] << 16) | |
| 319 | +- ((unsigned int) ptr[2] << 8) | |
| 320 | +- ((unsigned int) ptr[3]); |
| 321 | +-} |
| 322 | +- |
| 323 | +-// Retrieves a variable length integer from the page data. Returns the |
| 324 | +-// number of digits used to encode the integer and the stores the |
| 325 | +-// value at the given location. |
| 326 | +-int ProjectFileIO::get_varint(const unsigned char *ptr, int64_t *out) |
| 327 | +-{ |
| 328 | +- int64_t val = 0; |
| 329 | +- int i; |
| 330 | ++ int rc = sqlite3_step(stmt); |
| 331 | + |
| 332 | +- for (i = 0; i < 8; ++i) |
| 333 | ++ if (rc != SQLITE_ROW) |
| 334 | + { |
| 335 | +- val = (val << 7) + (ptr[i] & 0x7f); |
| 336 | +- if ((ptr[i] & 0x80) == 0) |
| 337 | +- { |
| 338 | +- *out = val; |
| 339 | +- return i + 1; |
| 340 | +- } |
| 341 | ++ conn.ThrowException(false); |
| 342 | + } |
| 343 | + |
| 344 | +- val = (val << 8) + (ptr[i] & 0xff); |
| 345 | +- *out = val; |
| 346 | ++ const int64_t size = sqlite3_column_int64(stmt, 0); |
| 347 | + |
| 348 | +- return 9; |
| 349 | ++ return size; |
| 350 | + } |
| 351 | + |
| 352 | + InvisibleTemporaryProject::InvisibleTemporaryProject() |
| 353 | +-- |
| 354 | +2.33.1 |
| 355 | + |
0 commit comments