storage/mozStorageSQLFunctions.cpp
author Kyle Huey <khuey@kylehuey.com>
Tue, 11 Aug 2015 06:10:46 -0700
changeset 257246 a13c1f26e351dd6251da641fe7a9eb53790fc2d0
parent 252629 91d6e262b662a0b4e47358665e222d3927337af9
child 270351 f623c844dca9a01dfa3f55791df285c4e16c3ef0
permissions -rw-r--r--
Bug 1179909: Refactor stable state handling. r=smaug This is motivated by three separate but related problems: 1. Our concept of recursion depth is broken for things that run from AfterProcessNextEvent observers (e.g. Promises). We decrement the recursionDepth counter before firing observers, so a Promise callback running at the lowest event loop depth has a recursion depth of 0 (whereas a regular nsIRunnable would be 1). This is a problem because it's impossible to distinguish a Promise running after a sync XHR's onreadystatechange handler from a top-level event (since the former runs with depth 2 - 1 = 1, and the latter runs with just 1). 2. The nsIThreadObserver mechanism that is used by a lot of code to run "after" the current event is a poor fit for anything that runs script. First, the order the observers fire in is the order they were added, not anything fixed by spec. Additionally, running script can cause the event loop to spin, which is a big source of pain here (bholley has some nasty bug caused by this). 3. We run Promises from different points in the code for workers and main thread. The latter runs from XPConnect's nsIThreadObserver callbacks, while the former runs from a hardcoded call to run Promises in the worker event loop. What workers do is particularly problematic because it means we can't get the right recursion depth no matter what we do to nsThread. The solve this, this patch does the following: 1. Consolidate some handling of microtasks and all handling of stable state from appshell and WorkerPrivate into CycleCollectedJSRuntime. 2. Make the recursionDepth counter only available to CycleCollectedJSRuntime (and its consumers) and remove it from the nsIThreadInternal and nsIThreadObserver APIs. 3. Adjust the recursionDepth counter so that microtasks run with the recursionDepth of the task they are associated with. 4. Introduce the concept of metastable state to replace appshell's RunBeforeNextEvent. Metastable state is reached after every microtask or task is completed. This provides the semantics that bent and I want for IndexedDB, where transactions autocommit at the end of a microtask and do not "spill" from one microtask into a subsequent microtask. This differs from appshell's RunBeforeNextEvent in two ways: a) It fires between microtasks, which was the motivation for starting this. b) It no longer ensures that we're at the same event loop depth in the native event queue. bent decided we don't care about this. 5. Reorder stable state to happen after microtasks such as Promises, per HTML. Right now we call the regular thread observers, including appshell, before the main thread observer (XPConnect), so stable state tasks happen before microtasks.

/* -*- Mode: C++; tab-width: 2; indent-tabs-mode: nil; c-basic-offset: 2 -*-
 * vim: sw=2 ts=2 et lcs=trail\:.,tab\:>~ :
 * This Source Code Form is subject to the terms of the Mozilla Public
 * License, v. 2.0. If a copy of the MPL was not distributed with this
 * file, You can obtain one at http://mozilla.org/MPL/2.0/. */

#include "mozilla/ArrayUtils.h"

#include "mozStorageSQLFunctions.h"
#include "nsUnicharUtils.h"
#include <algorithm>

namespace mozilla {
namespace storage {

////////////////////////////////////////////////////////////////////////////////
//// Local Helper Functions

namespace {

/**
 * Performs the LIKE comparison of a string against a pattern.  For more detail
 * see http://www.sqlite.org/lang_expr.html#like.
 *
 * @param aPatternItr
 *        An iterator at the start of the pattern to check for.
 * @param aPatternEnd
 *        An iterator at the end of the pattern to check for.
 * @param aStringItr
 *        An iterator at the start of the string to check for the pattern.
 * @param aStringEnd
 *        An iterator at the end of the string to check for the pattern.
 * @param aEscapeChar
 *        The character to use for escaping symbols in the pattern.
 * @return 1 if the pattern is found, 0 otherwise.
 */
int
likeCompare(nsAString::const_iterator aPatternItr,
            nsAString::const_iterator aPatternEnd,
            nsAString::const_iterator aStringItr,
            nsAString::const_iterator aStringEnd,
            char16_t aEscapeChar)
{
  const char16_t MATCH_ALL('%');
  const char16_t MATCH_ONE('_');

  bool lastWasEscape = false;
  while (aPatternItr != aPatternEnd) {
    /**
     * What we do in here is take a look at each character from the input
     * pattern, and do something with it.  There are 4 possibilities:
     * 1) character is an un-escaped match-all character
     * 2) character is an un-escaped match-one character
     * 3) character is an un-escaped escape character
     * 4) character is not any of the above
     */
    if (!lastWasEscape && *aPatternItr == MATCH_ALL) {
      // CASE 1
      /**
       * Now we need to skip any MATCH_ALL or MATCH_ONE characters that follow a
       * MATCH_ALL character.  For each MATCH_ONE character, skip one character
       * in the pattern string.
       */
      while (*aPatternItr == MATCH_ALL || *aPatternItr == MATCH_ONE) {
        if (*aPatternItr == MATCH_ONE) {
          // If we've hit the end of the string we are testing, no match
          if (aStringItr == aStringEnd)
            return 0;
          aStringItr++;
        }
        aPatternItr++;
      }

      // If we've hit the end of the pattern string, match
      if (aPatternItr == aPatternEnd)
        return 1;

      while (aStringItr != aStringEnd) {
        if (likeCompare(aPatternItr, aPatternEnd, aStringItr, aStringEnd,
                        aEscapeChar)) {
          // we've hit a match, so indicate this
          return 1;
        }
        aStringItr++;
      }

      // No match
      return 0;
    }
    else if (!lastWasEscape && *aPatternItr == MATCH_ONE) {
      // CASE 2
      if (aStringItr == aStringEnd) {
        // If we've hit the end of the string we are testing, no match
        return 0;
      }
      aStringItr++;
      lastWasEscape = false;
    }
    else if (!lastWasEscape && *aPatternItr == aEscapeChar) {
      // CASE 3
      lastWasEscape = true;
    }
    else {
      // CASE 4
      if (::ToUpperCase(*aStringItr) != ::ToUpperCase(*aPatternItr)) {
        // If we've hit a point where the strings don't match, there is no match
        return 0;
      }
      aStringItr++;
      lastWasEscape = false;
    }

    aPatternItr++;
  }

  return aStringItr == aStringEnd;
}

/**
 * This class manages a dynamic array.  It can represent an array of any 
 * reasonable size, but if the array is "N" elements or smaller, it will be
 * stored using fixed space inside the auto array itself.  If the auto array
 * is a local variable, this internal storage will be allocated cheaply on the
 * stack, similar to nsAutoString.  If a larger size is requested, the memory
 * will be dynamically allocated from the heap.  Since the destructor will
 * free any heap-allocated memory, client code doesn't need to care where the
 * memory came from.
 */
template <class T, size_t N> class AutoArray
{

public:

  explicit AutoArray(size_t size)
  : mBuffer(size <= N ? mAutoBuffer : new T[size])
  {
  }

  ~AutoArray()
  { 
    if (mBuffer != mAutoBuffer)
      delete[] mBuffer; 
  }

  /**
   * Return the pointer to the allocated array.
   * @note If the array allocation failed, get() will return nullptr!
   *
   * @return the pointer to the allocated array
   */
  T *get() 
  {
    return mBuffer; 
  }

private:
  T *mBuffer;           // Points to mAutoBuffer if we can use it, heap otherwise.
  T mAutoBuffer[N];     // The internal memory buffer that we use if we can.
};

/**
 * Compute the Levenshtein Edit Distance between two strings.
 * 
 * @param aStringS
 *        a string
 * @param aStringT
 *        another string
 * @param _result
 *        an outparam that will receive the edit distance between the arguments
 * @return a Sqlite result code, e.g. SQLITE_OK, SQLITE_NOMEM, etc.
 */
int
levenshteinDistance(const nsAString &aStringS,
                    const nsAString &aStringT,
                    int *_result)
{
    // Set the result to a non-sensical value in case we encounter an error.
    *_result = -1;

    const uint32_t sLen = aStringS.Length();
    const uint32_t tLen = aStringT.Length();

    if (sLen == 0) {
      *_result = tLen;
      return SQLITE_OK;
    }
    if (tLen == 0) {
      *_result = sLen;
      return SQLITE_OK;
    }

    // Notionally, Levenshtein Distance is computed in a matrix.  If we 
    // assume s = "span" and t = "spam", the matrix would look like this:
    //    s -->
    //  t          s   p   a   n
    //  |      0   1   2   3   4
    //  V  s   1   *   *   *   *
    //     p   2   *   *   *   *
    //     a   3   *   *   *   *
    //     m   4   *   *   *   *
    //
    // Note that the row width is sLen + 1 and the column height is tLen + 1,
    // where sLen is the length of the string "s" and tLen is the length of "t".
    // The first row and the first column are initialized as shown, and
    // the algorithm computes the remaining cells row-by-row, and
    // left-to-right within each row.  The computation only requires that
    // we be able to see the current row and the previous one.

    // Allocate memory for two rows.  Use AutoArray's to manage the memory
    // so we don't have to explicitly free it, and so we can avoid the expense
    // of memory allocations for relatively small strings.
    AutoArray<int, nsAutoString::kDefaultStorageSize> row1(sLen + 1);
    AutoArray<int, nsAutoString::kDefaultStorageSize> row2(sLen + 1);

    // Declare the raw pointers that will actually be used to access the memory.
    int *prevRow = row1.get();
    NS_ENSURE_TRUE(prevRow, SQLITE_NOMEM);
    int *currRow = row2.get();
    NS_ENSURE_TRUE(currRow, SQLITE_NOMEM);

    // Initialize the first row.
    for (uint32_t i = 0; i <= sLen; i++)
        prevRow[i] = i;

    const char16_t *s = aStringS.BeginReading();
    const char16_t *t = aStringT.BeginReading();

    // Compute the empty cells in the "matrix" row-by-row, starting with
    // the second row.
    for (uint32_t ti = 1; ti <= tLen; ti++) {

        // Initialize the first cell in this row.
        currRow[0] = ti;

        // Get the character from "t" that corresponds to this row.
        const char16_t tch = t[ti - 1];

        // Compute the remaining cells in this row, left-to-right,
        // starting at the second column (and first character of "s").
        for (uint32_t si = 1; si <= sLen; si++) {
            
            // Get the character from "s" that corresponds to this column,
            // compare it to the t-character, and compute the "cost".
            const char16_t sch = s[si - 1];
            int cost = (sch == tch) ? 0 : 1;

            // ............ We want to calculate the value of cell "d" from
            // ...ab....... the previously calculated (or initialized) cells
            // ...cd....... "a", "b", and "c", where d = min(a', b', c').
            // ............ 
            int aPrime = prevRow[si - 1] + cost;
            int bPrime = prevRow[si] + 1;
            int cPrime = currRow[si - 1] + 1;
            currRow[si] = std::min(aPrime, std::min(bPrime, cPrime));
        }

        // Advance to the next row.  The current row becomes the previous
        // row and we recycle the old previous row as the new current row.
        // We don't need to re-initialize the new current row since we will
        // rewrite all of its cells anyway.
        int *oldPrevRow = prevRow;
        prevRow = currRow;
        currRow = oldPrevRow;
    }

    // The final result is the value of the last cell in the last row.
    // Note that that's now in the "previous" row, since we just swapped them.
    *_result = prevRow[sLen];
    return SQLITE_OK;
}

// This struct is used only by registerFunctions below, but ISO C++98 forbids
// instantiating a template dependent on a locally-defined type.  Boo-urns!
struct Functions {
  const char *zName;
  int nArg;
  int enc;
  void *pContext;
  void (*xFunc)(::sqlite3_context*, int, sqlite3_value**);
};

} // namespace

////////////////////////////////////////////////////////////////////////////////
//// Exposed Functions

int
registerFunctions(sqlite3 *aDB)
{
  Functions functions[] = {
    {"lower",               
      1, 
      SQLITE_UTF16, 
      0,        
      caseFunction},
    {"lower",               
      1, 
      SQLITE_UTF8,  
      0,        
      caseFunction},
    {"upper",               
      1, 
      SQLITE_UTF16, 
      (void*)1, 
      caseFunction},
    {"upper",               
      1, 
      SQLITE_UTF8,  
      (void*)1, 
      caseFunction},

    {"like",                
      2, 
      SQLITE_UTF16, 
      0,        
      likeFunction},
    {"like",                
      2, 
      SQLITE_UTF8,  
      0,        
      likeFunction},
    {"like",                
      3, 
      SQLITE_UTF16, 
      0,        
      likeFunction},
    {"like",                
      3, 
      SQLITE_UTF8,  
      0,        
      likeFunction},

    {"levenshteinDistance", 
      2, 
      SQLITE_UTF16, 
      0,        
      levenshteinDistanceFunction},
    {"levenshteinDistance", 
      2, 
      SQLITE_UTF8,  
      0,        
      levenshteinDistanceFunction},
  };

  int rv = SQLITE_OK;
  for (size_t i = 0; SQLITE_OK == rv && i < ArrayLength(functions); ++i) {
    struct Functions *p = &functions[i];
    rv = ::sqlite3_create_function(aDB, p->zName, p->nArg, p->enc, p->pContext,
                                   p->xFunc, nullptr, nullptr);
  }

  return rv;
}

////////////////////////////////////////////////////////////////////////////////
//// SQL Functions

void
caseFunction(sqlite3_context *aCtx,
             int aArgc,
             sqlite3_value **aArgv)
{
  NS_ASSERTION(1 == aArgc, "Invalid number of arguments!");

  nsAutoString data(static_cast<const char16_t *>(::sqlite3_value_text16(aArgv[0])));
  bool toUpper = ::sqlite3_user_data(aCtx) ? true : false;

  if (toUpper)
    ::ToUpperCase(data);
  else
    ::ToLowerCase(data);

  // Set the result.
  ::sqlite3_result_text16(aCtx, data.get(), -1, SQLITE_TRANSIENT);
}

/**
 * This implements the like() SQL function.  This is used by the LIKE operator.
 * The SQL statement 'A LIKE B' is implemented as 'like(B, A)', and if there is
 * an escape character, say E, it is implemented as 'like(B, A, E)'.
 */
void
likeFunction(sqlite3_context *aCtx,
             int aArgc,
             sqlite3_value **aArgv)
{
  NS_ASSERTION(2 == aArgc || 3 == aArgc, "Invalid number of arguments!");

  if (::sqlite3_value_bytes(aArgv[0]) > SQLITE_MAX_LIKE_PATTERN_LENGTH) {
    ::sqlite3_result_error(aCtx, "LIKE or GLOB pattern too complex",
                           SQLITE_TOOBIG);
    return;
  }

  if (!::sqlite3_value_text16(aArgv[0]) || !::sqlite3_value_text16(aArgv[1]))
    return;

  nsDependentString A(static_cast<const char16_t *>(::sqlite3_value_text16(aArgv[1])));
  nsDependentString B(static_cast<const char16_t *>(::sqlite3_value_text16(aArgv[0])));
  NS_ASSERTION(!B.IsEmpty(), "LIKE string must not be null!");

  char16_t E = 0;
  if (3 == aArgc)
    E = static_cast<const char16_t *>(::sqlite3_value_text16(aArgv[2]))[0];

  nsAString::const_iterator itrString, endString;
  A.BeginReading(itrString);
  A.EndReading(endString);
  nsAString::const_iterator itrPattern, endPattern;
  B.BeginReading(itrPattern);
  B.EndReading(endPattern);
  ::sqlite3_result_int(aCtx, likeCompare(itrPattern, endPattern, itrString,
                                         endString, E));
}

void levenshteinDistanceFunction(sqlite3_context *aCtx,
                                 int aArgc,
                                 sqlite3_value **aArgv)
{
  NS_ASSERTION(2 == aArgc, "Invalid number of arguments!");

  // If either argument is a SQL NULL, then return SQL NULL.
  if (::sqlite3_value_type(aArgv[0]) == SQLITE_NULL ||
      ::sqlite3_value_type(aArgv[1]) == SQLITE_NULL) {
    ::sqlite3_result_null(aCtx);
    return;
  }

  int aLen = ::sqlite3_value_bytes16(aArgv[0]) / sizeof(char16_t);
  const char16_t *a = static_cast<const char16_t *>(::sqlite3_value_text16(aArgv[0]));

  int bLen = ::sqlite3_value_bytes16(aArgv[1]) / sizeof(char16_t);
  const char16_t *b = static_cast<const char16_t *>(::sqlite3_value_text16(aArgv[1]));

  // Compute the Levenshtein Distance, and return the result (or error).
  int distance = -1;
  const nsDependentString A(a, aLen);
  const nsDependentString B(b, bLen);
  int status = levenshteinDistance(A, B, &distance);
  if (status == SQLITE_OK) {
    ::sqlite3_result_int(aCtx, distance);    
  }
  else if (status == SQLITE_NOMEM) {
    ::sqlite3_result_error_nomem(aCtx);
  }
  else {
    ::sqlite3_result_error(aCtx, "User function returned error code", -1);
  }
}

} // namespace storage
} // namespace mozilla