ccls/src/working_files.cc

#include "working_files.h"

#include "lex_utils.h"
#include "position.h"

#include <doctest/doctest.h>
#include <loguru.hpp>

#include <algorithm>
#include <climits>
#include <numeric>
#include <sstream>

namespace {

// When finding a best match of buffer line and index line, limit the max edit
// distance.
constexpr int kMaxDiff = 20;
// Don't align index line to buffer line if one of the lengths is larger than
// |kMaxColumnAlignSize|.
constexpr int kMaxColumnAlignSize = 200;

lsPosition GetPositionForOffset(const std::string& content, int offset) {
  if (offset >= content.size())
    offset = (int)content.size() - 1;

  lsPosition result;
  int i = 0;
  while (i < offset) {
    if (content[i] == '\n') {
      result.line += 1;
      result.character = 0;
    } else {
      result.character += 1;
    }
    ++i;
  }

  return result;
}

std::vector<std::string> ToLines(const std::string& content) {
  std::vector<std::string> result;
  std::istringstream lines(content);
  std::string line;
  while (getline(lines, line))
    result.push_back(line);
  return result;
}

// Computes the edit distance of strings [a,a+la) and [b,b+lb) with Eugene W.
// Myers' O(ND) diff algorithm.
// Costs: insertion=1, deletion=1, no substitution.
// If the distance is larger than threshold, returns threshould + 1.
int MyersDiff(const char* a, int la, const char* b, int lb, int threshold) {
  assert(threshold <= kMaxDiff);
  static int v_static[2 * kMaxColumnAlignSize + 2];
  const char *ea = a + la, *eb = b + lb;
  // Strip prefix
  for (; a < ea && b < eb && *a == *b; a++, b++) {
  }
  // Strip suffix
  for (; a < ea && b < eb && ea[-1] == eb[-1]; ea--, eb--) {
  }
  la = int(ea - a);
  lb = int(eb - b);
  // If the sum of lengths exceeds what we can handle, return a lower bound.
  if (la + lb > 2 * kMaxColumnAlignSize)
    return std::min(abs(la - lb), threshold + 1);

  int* v = v_static + lb;
  v[1] = 0;
  for (int di = 0; di <= threshold; di++) {
    int low = -di + 2 * std::max(0, di - lb),
        high = di - 2 * std::max(0, di - la);
    for (int i = low; i <= high; i += 2) {
      int x = i == -di || (i != di && v[i - 1] < v[i + 1]) ? v[i + 1]
                                                           : v[i - 1] + 1,
          y = x - i;
      while (x < la && y < lb && a[x] == b[y])
        x++, y++;
      v[i] = x;
      if (x == la && y == lb)
        return di;
    }
  }
  return threshold + 1;
}

int MyersDiff(const std::string& a, const std::string& b, int threshold) {
  return MyersDiff(a.data(), a.size(), b.data(), b.size(), threshold);
}

// Computes edit distance with O(N*M) Needleman-Wunsch algorithm
// and returns a distance vector where d[i] = cost of aligning a to b[0,i).
//
// Myers' diff algorithm is used to find best matching line while this one is
// used to align a single column because Myers' needs some twiddling to return
// distance vector.
std::vector<int> EditDistanceVector(std::string a, std::string b) {
  std::vector<int> d(b.size() + 1);
  std::iota(d.begin(), d.end(), 0);
  for (int i = 0; i < (int)a.size(); i++) {
    int ul = d[0];
    d[0] = i + 1;
    for (int j = 0; j < (int)b.size(); j++) {
      int t = d[j + 1];
      d[j + 1] = a[i] == b[j] ? ul : std::min(d[j], d[j + 1]) + 1;
      ul = t;
    }
  }
  return d;
}

// Find matching position of |a[column]| in |b|.
// This is actually a single step of Hirschberg's sequence alignment algorithm.
int AlignColumn(const std::string& a, int column, std::string b, bool is_end) {
  int head = 0, tail = 0;
  while (head < (int)a.size() && head < (int)b.size() && a[head] == b[head])
    head++;
  while (tail < (int)a.size() && tail < (int)b.size() &&
         a[a.size() - 1 - tail] == b[b.size() - 1 - tail])
    tail++;
  if (column < head)
    return column;
  if ((int)a.size() - tail < column)
    return column + b.size() - a.size();
  if (std::max(a.size(), b.size()) - head - tail >= kMaxColumnAlignSize)
    return std::min(column, (int)b.size());

  // b[head, b.size() - tail)
  b = b.substr(head, b.size() - tail - head);

  // left[i] = cost of aligning a[head, column) to b[head, head + i)
  std::vector<int> left = EditDistanceVector(a.substr(head, column - head), b);

  // right[i] = cost of aligning a[column, a.size() - tail) to b[head + i,
  // b.size() - tail)
  std::string a_rev = a.substr(column, a.size() - tail - column);
  std::reverse(a_rev.begin(), a_rev.end());
  std::reverse(b.begin(), b.end());
  std::vector<int> right = EditDistanceVector(a_rev, b);
  std::reverse(right.begin(), right.end());

  int best = 0, best_cost = INT_MAX;
  for (size_t i = 0; i < left.size(); i++) {
    int cost = left[i] + right[i];
    if (is_end ? cost < best_cost : cost <= best_cost) {
      best_cost = cost;
      best = i;
    }
  }
  return head + best;
}

// Find matching buffer line of index_lines[line].
// By symmetry, this can also be used to find matching index line of a buffer
// line.
std::optional<int> FindMatchingLine(const std::vector<std::string>& index_lines,
                               const std::vector<int>& index_to_buffer,
                               int line,
                               int* column,
                               const std::vector<std::string>& buffer_lines,
                               bool is_end) {
  // If this is a confident mapping, returns.
  if (index_to_buffer[line] >= 0) {
    int ret = index_to_buffer[line];
    if (column)
      *column =
          AlignColumn(index_lines[line], *column, buffer_lines[ret], is_end);
    return ret;
  }

  // Find the nearest two confident lines above and below.
  int up = line, down = line;
  while (--up >= 0 && index_to_buffer[up] < 0) {
  }
  while (++down < int(index_to_buffer.size()) && index_to_buffer[down] < 0) {
  }
  up = up < 0 ? 0 : index_to_buffer[up];
  down = down >= int(index_to_buffer.size()) ? int(buffer_lines.size()) - 1
                                             : index_to_buffer[down];
  if (up > down)
    return std::nullopt;

  // Search for lines [up,down] and use Myers's diff algorithm to find the best
  // match (least edit distance).
  int best = up, best_dist = kMaxDiff + 1;
  const std::string& needle = index_lines[line];
  for (int i = up; i <= down; i++) {
    int dist = MyersDiff(needle, buffer_lines[i], kMaxDiff);
    if (dist < best_dist) {
      best_dist = dist;
      best = i;
    }
  }
  if (column)
    *column =
        AlignColumn(index_lines[line], *column, buffer_lines[best], is_end);
  return best;
}

}  // namespace

std::vector<CXUnsavedFile> WorkingFiles::Snapshot::AsUnsavedFiles() const {
  std::vector<CXUnsavedFile> result;
  result.reserve(files.size());
  for (auto& file : files) {
    CXUnsavedFile unsaved;
    unsaved.Filename = file.filename.c_str();
    unsaved.Contents = file.content.c_str();
    unsaved.Length = (unsigned long)file.content.size();

    result.push_back(unsaved);
  }
  return result;
}

WorkingFile::WorkingFile(const std::string& filename,
                         const std::string& buffer_content)
    : filename(filename), buffer_content(buffer_content) {
  OnBufferContentUpdated();

  // SetIndexContent gets called when the file is opened.
}

void WorkingFile::SetIndexContent(const std::string& index_content) {
  index_lines = ToLines(index_content);

  index_to_buffer.clear();
  buffer_to_index.clear();
}

void WorkingFile::OnBufferContentUpdated() {
  buffer_lines = ToLines(buffer_content);

  index_to_buffer.clear();
  buffer_to_index.clear();
}

// Variant of Paul Heckel's diff algorithm to compute |index_to_buffer| and
// |buffer_to_index|.
// The core idea is that if a line is unique in both index and buffer,
// we are confident that the line appeared in index maps to the one appeared in
// buffer. And then using them as start points to extend upwards and downwards
// to align other identical lines (but not unique).
void WorkingFile::ComputeLineMapping() {
  std::unordered_map<uint64_t, int> hash_to_unique;
  std::vector<uint64_t> index_hashes(index_lines.size());
  std::vector<uint64_t> buffer_hashes(buffer_lines.size());
  index_to_buffer.resize(index_lines.size());
  buffer_to_index.resize(buffer_lines.size());
  hash_to_unique.reserve(
      std::max(index_to_buffer.size(), buffer_to_index.size()));

  // For index line i, set index_to_buffer[i] to -1 if line i is duplicated.
  int i = 0;
  for (auto& line : index_lines) {
    uint64_t h = HashUsr(line);
    auto it = hash_to_unique.find(h);
    if (it == hash_to_unique.end()) {
      hash_to_unique[h] = i;
      index_to_buffer[i] = i;
    } else {
      if (it->second >= 0)
        index_to_buffer[it->second] = -1;
      index_to_buffer[i] = it->second = -1;
    }
    index_hashes[i++] = h;
  }

  // For buffer line i, set buffer_to_index[i] to -1 if line i is duplicated.
  i = 0;
  hash_to_unique.clear();
  for (auto& line : buffer_lines) {
    uint64_t h = HashUsr(line);
    auto it = hash_to_unique.find(h);
    if (it == hash_to_unique.end()) {
      hash_to_unique[h] = i;
      buffer_to_index[i] = i;
    } else {
      if (it->second >= 0)
        buffer_to_index[it->second] = -1;
      buffer_to_index[i] = it->second = -1;
    }
    buffer_hashes[i++] = h;
  }

  // If index line i is the identical to buffer line j, and they are both
  // unique, align them by pointing from_index[i] to j.
  i = 0;
  for (auto h : index_hashes) {
    if (index_to_buffer[i] >= 0) {
      auto it = hash_to_unique.find(h);
      if (it != hash_to_unique.end() && it->second >= 0 &&
          buffer_to_index[it->second] >= 0)
        index_to_buffer[i] = it->second;
      else
        index_to_buffer[i] = -1;
    }
    i++;
  }

  // Starting at unique lines, extend upwards and downwards.
  for (i = 0; i < (int)index_hashes.size() - 1; i++) {
    int j = index_to_buffer[i];
    if (0 <= j && j + 1 < buffer_hashes.size() &&
        index_hashes[i + 1] == buffer_hashes[j + 1])
      index_to_buffer[i + 1] = j + 1;
  }
  for (i = (int)index_hashes.size(); --i > 0;) {
    int j = index_to_buffer[i];
    if (0 < j && index_hashes[i - 1] == buffer_hashes[j - 1])
      index_to_buffer[i - 1] = j - 1;
  }

  // |buffer_to_index| is a inverse mapping of |index_to_buffer|.
  std::fill(buffer_to_index.begin(), buffer_to_index.end(), -1);
  for (i = 0; i < (int)index_hashes.size(); i++)
    if (index_to_buffer[i] >= 0)
      buffer_to_index[index_to_buffer[i]] = i;
}

std::optional<int> WorkingFile::GetBufferPosFromIndexPos(int line,
                                                    int* column,
                                                    bool is_end) {
  // The implementation is simple but works pretty well for most cases. We
  // lookup the line contents in the indexed file contents, and try to find the
  // most similar line in the current buffer file.
  //
  // Previously, this was implemented by tracking edits and by running myers
  // diff algorithm. They were complex implementations that did not work as
  // well.

  // Note: |index_line| and |buffer_line| are 1-based.

  // TODO: reenable this assert once we are using the real indexed file.
  // assert(index_line >= 1 && index_line <= index_lines.size());
  if (line < 0 || line >= (int)index_lines.size()) {
    loguru::Text stack = loguru::stacktrace();
    LOG_S(WARNING) << "Bad index_line (got " << line << ", expected [0, "
                   << index_lines.size() << ")) in " << filename
                   << stack.c_str();
    return std::nullopt;
  }

  if (index_to_buffer.empty())
    ComputeLineMapping();
  return FindMatchingLine(index_lines, index_to_buffer, line, column,
                          buffer_lines, is_end);
}

std::optional<int> WorkingFile::GetIndexPosFromBufferPos(int line,
                                                    int* column,
                                                    bool is_end) {
  // See GetBufferLineFromIndexLine for additional comments.
  if (line < 0 || line >= (int)buffer_lines.size())
    return std::nullopt;

  if (buffer_to_index.empty())
    ComputeLineMapping();
  return FindMatchingLine(buffer_lines, buffer_to_index, line, column,
                          index_lines, is_end);
}

std::string WorkingFile::FindClosestCallNameInBuffer(
    lsPosition position,
    int* active_parameter,
    lsPosition* completion_position) const {
  *active_parameter = 0;

  int offset = GetOffsetForPosition(position, buffer_content);

  // If vscode auto-inserts closing ')' we will begin on ')' token in foo()
  // which will make the below algorithm think it's a nested call.
  if (offset > 0 && buffer_content[offset] == ')')
    --offset;

  // Scan back out of call context.
  int balance = 0;
  while (offset > 0) {
    char c = buffer_content[offset];
    if (c == ')')
      ++balance;
    else if (c == '(')
      --balance;

    if (balance == 0 && c == ',')
      *active_parameter += 1;

    --offset;

    if (balance == -1)
      break;
  }

  if (offset < 0)
    return "";

  // Scan back entire identifier.
  int start_offset = offset;
  while (offset > 0) {
    char c = buffer_content[offset - 1];
    if (isalnum(c) == false && c != '_')
      break;
    --offset;
  }

  if (completion_position)
    *completion_position = GetPositionForOffset(buffer_content, offset);

  return buffer_content.substr(offset, start_offset - offset + 1);
}

lsPosition WorkingFile::FindStableCompletionSource(
    lsPosition position,
    bool* is_global_completion,
    std::string* existing_completion) const {
  *is_global_completion = true;

  int start_offset = GetOffsetForPosition(position, buffer_content);
  int offset = start_offset;

  while (offset > 0) {
    char c = buffer_content[offset - 1];
    if (!isalnum(c) && c != '_') {
      // Global completion is everything except for dot (.), arrow (->), and
      // double colon (::)
      if (c == '.')
        *is_global_completion = false;
      if (offset > 2) {
        char pc = buffer_content[offset - 2];
        if (pc == ':' && c == ':')
          *is_global_completion = false;
        else if (pc == '-' && c == '>')
          *is_global_completion = false;
      }

      break;
    }
    --offset;
  }

  *existing_completion = buffer_content.substr(offset, start_offset - offset);
  return GetPositionForOffset(buffer_content, offset);
}

WorkingFile* WorkingFiles::GetFileByFilename(const std::string& filename) {
  std::lock_guard<std::mutex> lock(files_mutex);
  return GetFileByFilenameNoLock(filename);
}

WorkingFile* WorkingFiles::GetFileByFilenameNoLock(
    const std::string& filename) {
  for (auto& file : files) {
    if (file->filename == filename)
      return file.get();
  }
  return nullptr;
}

void WorkingFiles::DoAction(const std::function<void()>& action) {
  std::lock_guard<std::mutex> lock(files_mutex);
  action();
}

void WorkingFiles::DoActionOnFile(
    const std::string& filename,
    const std::function<void(WorkingFile* file)>& action) {
  std::lock_guard<std::mutex> lock(files_mutex);
  WorkingFile* file = GetFileByFilenameNoLock(filename);
  action(file);
}

WorkingFile* WorkingFiles::OnOpen(const lsTextDocumentItem& open) {
  std::lock_guard<std::mutex> lock(files_mutex);

  std::string filename = open.uri.GetPath();
  std::string content = open.text;

  // The file may already be open.
  if (WorkingFile* file = GetFileByFilenameNoLock(filename)) {
    file->version = open.version;
    file->buffer_content = content;
    file->OnBufferContentUpdated();
    return file;
  }

  files.push_back(std::make_unique<WorkingFile>(filename, content));
  return files[files.size() - 1].get();
}

void WorkingFiles::OnChange(const lsTextDocumentDidChangeParams& change) {
  std::lock_guard<std::mutex> lock(files_mutex);

  std::string filename = change.textDocument.uri.GetPath();
  WorkingFile* file = GetFileByFilenameNoLock(filename);
  if (!file) {
    LOG_S(WARNING) << "Could not change " << filename
                   << " because it was not open";
    return;
  }

  // version: number | null
  if (std::holds_alternative<int>(change.textDocument.version))
    file->version = std::get<int>(change.textDocument.version);

  for (const lsTextDocumentContentChangeEvent& diff : change.contentChanges) {
    // Per the spec replace everything if the rangeLength and range are not set.
    // See https://github.com/Microsoft/language-server-protocol/issues/9.
    if (!diff.range) {
      file->buffer_content = diff.text;
      file->OnBufferContentUpdated();
    } else {
      int start_offset =
          GetOffsetForPosition(diff.range->start, file->buffer_content);
      // Ignore TextDocumentContentChangeEvent.rangeLength which causes trouble
      // when UTF-16 surrogate pairs are used.
      int end_offset =
          GetOffsetForPosition(diff.range->end, file->buffer_content);
      file->buffer_content.replace(file->buffer_content.begin() + start_offset,
                                   file->buffer_content.begin() + end_offset,
                                   diff.text);
      file->OnBufferContentUpdated();
    }
  }
}

void WorkingFiles::OnClose(const lsTextDocumentIdentifier& close) {
  std::lock_guard<std::mutex> lock(files_mutex);

  std::string filename = close.uri.GetPath();

  for (int i = 0; i < files.size(); ++i) {
    if (files[i]->filename == filename) {
      files.erase(files.begin() + i);
      return;
    }
  }

  LOG_S(WARNING) << "Could not close " << filename
                 << " because it was not open";
}

WorkingFiles::Snapshot WorkingFiles::AsSnapshot(
    const std::vector<std::string>& filter_paths) {
  std::lock_guard<std::mutex> lock(files_mutex);

  Snapshot result;
  result.files.reserve(files.size());
  for (const auto& file : files) {
    if (filter_paths.empty() || FindAnyPartial(file->filename, filter_paths))
      result.files.push_back({file->filename, file->buffer_content});
  }
  return result;
}