Handle UTF16 encodings in SQLite recover table.

SQLite databases use a consistent string encoding for the entire
database.  Detect the database's encoding and use it when setting text
results.

BUG=none

Review URL: https://chromiumcodereview.appspot.com/13940011

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@195999 0039d316-1c4b-4281-b951-d872f2087c98

2 files changed, 156 insertions, 1 deletions

diff --git a/third_party/sqlite/src/src/recover-alt.c b/third_party/sqlite/src/src/recover-alt.c
index 1a4ffe5..cc7a423 100644
--- a/third_party/sqlite/src/src/recover-alt.c
+++ b/third_party/sqlite/src/src/recover-alt.c
@@ -491,6 +491,56 @@ static int getRootPage(sqlite3 *db, const char *zDb, const char *zTable,
   return rc;
 }
 
+static int getEncoding(sqlite3 *db, const char *zDb, int* piEncoding){
+  char *zSql = sqlite3_mprintf("PRAGMA %s.encoding", zDb);
+  if( !zSql ){
+    return SQLITE_NOMEM;
+  }
+
+  sqlite3_stmt *pStmt = 0;
+  int rc = sqlite3_prepare_v2(db, zSql, -1, &pStmt, 0);
+  sqlite3_free(zSql);
+  if( rc!=SQLITE_OK ){
+    return rc;
+  }
+
+  /* Require a result. */
+  rc = sqlite3_step(pStmt);
+  if( rc==SQLITE_DONE ){
+    /* This case should not be possible. */
+    rc = SQLITE_CORRUPT;
+  }else if( rc==SQLITE_ROW ){
+    if( sqlite3_column_type(pStmt, 0)==SQLITE_TEXT ){
+      const char* z = sqlite3_column_text(pStmt, 0);
+      /* These strings match the literals in pragma.c. */
+      if( !strcmp(z, "UTF-16le") ){
+        *piEncoding = SQLITE_UTF16LE;
+      }else if( !strcmp(z, "UTF-16be") ){
+        *piEncoding = SQLITE_UTF16BE;
+      }else if( !strcmp(z, "UTF-8") ){
+        *piEncoding = SQLITE_UTF8;
+      }else{
+        /* This case should not be possible. */
+        *piEncoding = SQLITE_UTF8;
+      }
+    }else{
+      /* This case should not be possible. */
+      *piEncoding = SQLITE_UTF8;
+    }
+
+    /* Require only one result. */
+    rc = sqlite3_step(pStmt);
+    if( rc==SQLITE_DONE ){
+      rc = SQLITE_OK;
+    }else if( rc==SQLITE_ROW ){
+      /* This case should not be possible. */
+      rc = SQLITE_CORRUPT;
+    }
+  }
+  sqlite3_finalize(pStmt);
+  return rc;
+}
+
 /* Cursor for iterating interior nodes.  Interior page cells contain a
  * child page number and a rowid.  The child page contains items left
  * of the rowid (less than).  The rightmost page of the subtree is
@@ -1451,6 +1501,7 @@ typedef struct RecoverCursor RecoverCursor;
 struct RecoverCursor {
   sqlite3_vtab_cursor base;
   RecoverLeafCursor *pLeafCursor;
+  int iEncoding;
   int bEOF;
 };
 
@@ -1466,6 +1517,12 @@ static int recoverOpen(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCursor){
     return rc;
   }
 
+  int iEncoding = 0;
+  rc = getEncoding(pRecover->db, pRecover->zDb, &iEncoding);
+  if( rc!=SQLITE_OK ){
+    return rc;
+  }
+
   unsigned nPageSize;
   Pager *pPager;
   rc = GetPager(pRecover->db, pRecover->zDb, &pPager, &nPageSize);
@@ -1487,6 +1544,7 @@ static int recoverOpen(sqlite3_vtab *pVTab, sqlite3_vtab_cursor **ppCursor){
   memset(pCursor, 0, sizeof(*pCursor));
   pCursor->base.pVtab = pVTab;
   pCursor->pLeafCursor = pLeafCursor;
+  pCursor->iEncoding = iEncoding;
 
   *ppCursor = (sqlite3_vtab_cursor*)pCursor;
   return SQLITE_OK;
@@ -1645,7 +1703,13 @@ static int recoverColumn(sqlite3_vtab_cursor *cur, sqlite3_context *ctx, int i){
       if( SerialTypeIsBlob(iColType) ){
         sqlite3_result_blob(ctx, pColData, l, pFn);
       }else{
-        sqlite3_result_text(ctx, (const char*)pColData, l, pFn);
+        if( pCursor->iEncoding==SQLITE_UTF16LE ){
+          sqlite3_result_text16le(ctx, (const void*)pColData, l, pFn);
+        }else if( pCursor->iEncoding==SQLITE_UTF16BE ){
+          sqlite3_result_text16be(ctx, (const void*)pColData, l, pFn);
+        }else{
+          sqlite3_result_text(ctx, (const char*)pColData, l, pFn);
+        }
       }
     } break;
   }
diff --git a/third_party/sqlite/src/test/recover1.test b/third_party/sqlite/src/test/recover1.test
index c3541b7..1d90f09 100644
--- a/third_party/sqlite/src/test/recover1.test
+++ b/third_party/sqlite/src/test/recover1.test
@@ -335,4 +335,95 @@ do_test recover-types-7.0 {
   execsql {SELECT rowid, value FROM integers_recover}
 } {1 0 2 1 3 2 4 -2 5 127 6 -128 7 12345 8 -12345 9 32767 10 -32768 11 1234567 12 -1234567 13 8388607 14 -8388608 15 1234567890 16 -1234567890 17 2147483647 18 -2147483648 19 123456789012345 20 -123456789012345 21 140737488355327 22 -140737488355328 23 9223372036854775807 24 -9223372036854775808}
 
+# If UTF16 support is disabled, ignore the rest of the tests.
+#
+ifcapable {!utf16} {
+  finish_test
+  return
+}
+
+# Baseline UTF-8.
+file delete -force test.db
+sqlite3 db test.db;
+db eval {
+  PRAGMA encoding = 'UTF-8';
+}
+
+do_test recover-encoding-1.0 {
+  execsql {
+    DROP TABLE IF EXISTS e;
+    CREATE TABLE e (v TEXT);
+    INSERT INTO e VALUES('Mjollnir');
+    INSERT INTO e VALUES('Mjölnir');
+    INSERT INTO e VALUES('Mjǫlnir');
+    INSERT INTO e VALUES('Mjölner');
+    INSERT INTO e VALUES('Mjølner');
+    INSERT INTO e VALUES('ハンマー');
+    PRAGMA encoding;
+
+    DROP TABLE IF EXISTS e_recover;
+    CREATE VIRTUAL TABLE temp.e_recover USING recover(
+      e,
+      v TEXT
+    );
+    SELECT rowid, v FROM e_recover ORDER BY rowid;
+  }
+} {UTF-8 1 Mjollnir 2 Mjölnir 3 Mjǫlnir 4 Mjölner 5 Mjølner 6 ハンマー}
+
+# Reset the database to UTF-16LE.
+file delete -force test.db
+sqlite3 db test.db;
+db eval {
+  PRAGMA encoding = 'UTF-16LE';
+}
+
+do_test recover-encoding-2.0 {
+  execsql {
+    DROP TABLE IF EXISTS e;
+    CREATE TABLE e (v TEXT);
+    INSERT INTO e VALUES('Mjollnir');
+    INSERT INTO e VALUES('Mjölnir');
+    INSERT INTO e VALUES('Mjǫlnir');
+    INSERT INTO e VALUES('Mjölner');
+    INSERT INTO e VALUES('Mjølner');
+    INSERT INTO e VALUES('ハンマー');
+    PRAGMA encoding;
+
+    DROP TABLE IF EXISTS e_recover;
+    CREATE VIRTUAL TABLE temp.e_recover USING recover(
+      e,
+      v TEXT
+    );
+    SELECT rowid, v FROM e_recover ORDER BY rowid;
+  }
+} {UTF-16le 1 Mjollnir 2 Mjölnir 3 Mjǫlnir 4 Mjölner 5 Mjølner 6 ハンマー}
+
+# Reset the database to UTF-16BE.
+file delete -force test.db
+sqlite3 db test.db;
+db eval {
+  PRAGMA encoding = 'UTF-16BE';
+}
+
+do_test recover-encoding-3.0 {
+  execsql {
+    DROP TABLE IF EXISTS e;
+    CREATE TABLE e (v TEXT);
+    INSERT INTO e VALUES('Mjollnir');
+    INSERT INTO e VALUES('Mjölnir');
+    INSERT INTO e VALUES('Mjǫlnir');
+    INSERT INTO e VALUES('Mjölner');
+    INSERT INTO e VALUES('Mjølner');
+    INSERT INTO e VALUES('ハンマー');
+    PRAGMA encoding;
+
+    DROP TABLE IF EXISTS e_recover;
+    CREATE VIRTUAL TABLE temp.e_recover USING recover(
+      e,
+      v TEXT
+    );
+    SELECT rowid, v FROM e_recover ORDER BY rowid;
+  }
+} {UTF-16be 1 Mjollnir 2 Mjölnir 3 Mjǫlnir 4 Mjölner 5 Mjølner 6 ハンマー}
+
 finish_test