ccls/src/import_pipeline.cc
Fangrui Song 86efddf032 README
2018-05-05 15:25:43 -07:00

578 lines
20 KiB
C++

#include "import_pipeline.h"
#include "cache_manager.h"
#include "config.h"
#include "diagnostics_engine.h"
#include "lsp.h"
#include "message_handler.h"
#include "platform.h"
#include "project.h"
#include "query_utils.h"
#include "queue_manager.h"
#include "timer.h"
#include <doctest/doctest.h>
#include <loguru.hpp>
#include <chrono>
namespace {
struct Out_Progress : public lsOutMessage<Out_Progress> {
struct Params {
int indexRequestCount = 0;
int loadPreviousIndexCount = 0;
int onIdMappedCount = 0;
int onIndexedCount = 0;
int activeThreads = 0;
};
std::string method = "$ccls/progress";
Params params;
};
MAKE_REFLECT_STRUCT(Out_Progress::Params,
indexRequestCount,
loadPreviousIndexCount,
onIdMappedCount,
onIndexedCount,
activeThreads);
MAKE_REFLECT_STRUCT(Out_Progress, jsonrpc, method, params);
long long GetCurrentTimeInMilliseconds() {
auto time_since_epoch = Timer::Clock::now().time_since_epoch();
long long elapsed_milliseconds =
std::chrono::duration_cast<std::chrono::milliseconds>(time_since_epoch)
.count();
return elapsed_milliseconds;
}
struct ActiveThread {
ActiveThread(ImportPipelineStatus* status)
: status_(status) {
if (g_config && g_config->progressReportFrequencyMs < 0)
return;
++status_->num_active_threads;
}
~ActiveThread() {
if (g_config && g_config->progressReportFrequencyMs < 0)
return;
--status_->num_active_threads;
EmitProgress();
}
// Send indexing progress to client if reporting is enabled.
void EmitProgress() {
auto* queue = QueueManager::instance();
Out_Progress out;
out.params.indexRequestCount = queue->index_request.Size();
out.params.onIdMappedCount = queue->on_id_mapped.Size();
out.params.onIndexedCount = queue->on_indexed.Size();
out.params.activeThreads = status_->num_active_threads;
// Ignore this progress update if the last update was too recent.
if (g_config && g_config->progressReportFrequencyMs != 0) {
// Make sure we output a status update if queue lengths are zero.
bool all_zero = out.params.indexRequestCount == 0 &&
out.params.loadPreviousIndexCount == 0 &&
out.params.onIdMappedCount == 0 &&
out.params.onIndexedCount == 0 &&
out.params.activeThreads == 0;
if (!all_zero &&
GetCurrentTimeInMilliseconds() < status_->next_progress_output)
return;
status_->next_progress_output =
GetCurrentTimeInMilliseconds() + g_config->progressReportFrequencyMs;
}
QueueManager::WriteStdout(kMethodType_Unknown, out);
}
ImportPipelineStatus* status_;
};
enum class ShouldParse { Yes, No, NoSuchFile };
// Checks if |path| needs to be reparsed. This will modify cached state
// such that calling this function twice with the same path may return true
// the first time but will return false the second.
//
// |from|: The file which generated the parse request for this file.
ShouldParse FileNeedsParse(
bool is_interactive,
TimestampManager* timestamp_manager,
const std::shared_ptr<ICacheManager>& cache_manager,
IndexFile* opt_previous_index,
const std::string& path,
const std::vector<std::string>& args,
const std::optional<std::string>& from) {
auto unwrap_opt = [](const std::optional<std::string>& opt) -> std::string {
if (opt)
return " (via " + *opt + ")";
return "";
};
std::optional<int64_t> modification_timestamp = LastWriteTime(path);
// Cannot find file.
if (!modification_timestamp)
return ShouldParse::NoSuchFile;
std::optional<int64_t> last_cached_modification =
timestamp_manager->GetLastCachedModificationTime(cache_manager.get(),
path);
// File has been changed.
if (!last_cached_modification ||
modification_timestamp != *last_cached_modification) {
LOG_S(INFO) << "Timestamp has changed for " << path << unwrap_opt(from);
return ShouldParse::Yes;
}
// Command-line arguments changed.
auto is_file = [](const std::string& arg) {
return EndsWithAny(arg, {".h", ".c", ".cc", ".cpp", ".hpp", ".m", ".mm"});
};
if (opt_previous_index) {
auto& prev_args = opt_previous_index->args;
bool same = prev_args.size() == args.size();
for (size_t i = 0; i < args.size() && same; ++i) {
same = prev_args[i] == args[i] ||
(is_file(prev_args[i]) && is_file(args[i]));
}
if (!same) {
LOG_S(INFO) << "Arguments have changed for " << path << unwrap_opt(from);
return ShouldParse::Yes;
}
}
// File has not changed, do not parse it.
return ShouldParse::No;
};
enum CacheLoadResult { Parse, DoNotParse };
CacheLoadResult TryLoadFromCache(
FileConsumerSharedState* file_consumer_shared,
TimestampManager* timestamp_manager,
const std::shared_ptr<ICacheManager>& cache_manager,
bool is_interactive,
const Project::Entry& entry,
const std::string& path_to_index) {
// Always run this block, even if we are interactive, so we can check
// dependencies and reset files in |file_consumer_shared|.
IndexFile* previous_index = cache_manager->TryLoad(path_to_index);
if (!previous_index)
return CacheLoadResult::Parse;
// If none of the dependencies have changed and the index is not
// interactive (ie, requested by a file save), skip parsing and just load
// from cache.
// Check timestamps and update |file_consumer_shared|.
ShouldParse path_state =
FileNeedsParse(is_interactive, timestamp_manager, cache_manager,
previous_index, path_to_index, entry.args, std::nullopt);
if (path_state == ShouldParse::Yes)
file_consumer_shared->Reset(path_to_index);
// Target file does not exist on disk, do not emit any indexes.
// TODO: Dependencies should be reassigned to other files. We can do this by
// updating the "primary_file" if it doesn't exist. Might not actually be a
// problem in practice.
if (path_state == ShouldParse::NoSuchFile)
return CacheLoadResult::DoNotParse;
bool needs_reparse = is_interactive || path_state == ShouldParse::Yes;
for (const std::string& dependency : previous_index->dependencies) {
assert(!dependency.empty());
if (FileNeedsParse(is_interactive, timestamp_manager, cache_manager,
previous_index, dependency, entry.args,
previous_index->path) == ShouldParse::Yes) {
needs_reparse = true;
// Do not break here, as we need to update |file_consumer_shared| for
// every dependency that needs to be reparsed.
file_consumer_shared->Reset(dependency);
}
}
// FIXME: should we still load from cache?
if (needs_reparse)
return CacheLoadResult::Parse;
// No timestamps changed - load directly from cache.
LOG_S(INFO) << "load index for " << path_to_index;
// TODO/FIXME: real perf
PerformanceImportFile perf;
std::vector<Index_OnIdMapped> result;
result.push_back(Index_OnIdMapped(
cache_manager, nullptr, cache_manager->TryTakeOrLoad(path_to_index), perf,
is_interactive, false /*write_to_disk*/));
for (const std::string& dependency : previous_index->dependencies) {
// Only load a dependency if it is not already loaded.
//
// This is important for perf in large projects where there are lots of
// dependencies shared between many files.
if (!file_consumer_shared->Mark(dependency))
continue;
LOG_S(INFO) << "emit index for " << dependency << " via "
<< previous_index->path;
// |dependency_index| may be null if there is no cache for it but
// another file has already started importing it.
if (std::unique_ptr<IndexFile> dependency_index =
cache_manager->TryTakeOrLoad(dependency)) {
result.push_back(
Index_OnIdMapped(cache_manager, nullptr, std::move(dependency_index),
perf, is_interactive, false /*write_to_disk*/));
}
}
QueueManager::instance()->on_id_mapped.EnqueueAll(std::move(result));
return CacheLoadResult::DoNotParse;
}
std::vector<FileContents> PreloadFileContents(
const std::shared_ptr<ICacheManager>& cache_manager,
const Project::Entry& entry,
const std::string& entry_contents,
const std::string& path_to_index) {
// Load file contents for all dependencies into memory. If the dependencies
// for the file changed we may not end up using all of the files we
// preloaded. If a new dependency was added the indexer will grab the file
// contents as soon as possible.
//
// We do this to minimize the race between indexing a file and capturing the
// file contents.
//
// TODO: We might be able to optimize perf by only copying for files in
// working_files. We can pass that same set of files to the indexer as
// well. We then default to a fast file-copy if not in working set.
// index->file_contents comes from cache, so we need to check if that cache is
// still valid. if so, we can use it, otherwise we need to load from disk.
auto get_latest_content = [](const std::string& path, int64_t cached_time,
const std::string& cached) -> std::string {
std::optional<int64_t> mod_time = LastWriteTime(path);
if (!mod_time)
return "";
if (*mod_time == cached_time)
return cached;
std::optional<std::string> fresh_content = ReadContent(path);
if (!fresh_content) {
LOG_S(ERROR) << "Failed to load content for " << path;
return "";
}
return *fresh_content;
};
std::vector<FileContents> file_contents;
file_contents.push_back(FileContents(entry.filename, entry_contents));
cache_manager->IterateLoadedCaches([&](IndexFile* index) {
if (index->path == entry.filename)
return;
file_contents.push_back(FileContents(
index->path,
get_latest_content(index->path, index->last_modification_time,
index->file_contents)));
});
return file_contents;
}
void ParseFile(DiagnosticsEngine* diag_engine,
WorkingFiles* working_files,
FileConsumerSharedState* file_consumer_shared,
TimestampManager* timestamp_manager,
IIndexer* indexer,
const Index_Request& request,
const Project::Entry& entry) {
// If the file is inferred, we may not actually be able to parse that file
// directly (ie, a header file, which are not listed in the project). If this
// file is inferred, then try to use the file which originally imported it.
std::string path_to_index = entry.filename;
if (entry.is_inferred) {
IndexFile* entry_cache = request.cache_manager->TryLoad(entry.filename);
if (entry_cache)
path_to_index = entry_cache->import_file;
}
// Try to load the file from cache.
if (TryLoadFromCache(file_consumer_shared, timestamp_manager,
request.cache_manager, request.is_interactive, entry,
path_to_index) == CacheLoadResult::DoNotParse)
return;
LOG_S(INFO) << "parse " << path_to_index;
std::vector<FileContents> file_contents = PreloadFileContents(
request.cache_manager, entry, request.contents, path_to_index);
std::vector<Index_OnIdMapped> result;
PerformanceImportFile perf;
auto indexes = indexer->Index(file_consumer_shared, path_to_index, entry.args,
file_contents, &perf);
if (indexes.empty()) {
if (g_config->index.enabled && request.id.Valid()) {
Out_Error out;
out.id = request.id;
out.error.code = lsErrorCodes::InternalError;
out.error.message = "Failed to index " + path_to_index;
QueueManager::WriteStdout(kMethodType_Unknown, out);
}
return;
}
for (std::unique_ptr<IndexFile>& new_index : indexes) {
Timer time;
// Only emit diagnostics for non-interactive sessions, which makes it easier
// to identify indexing problems. For interactive sessions, diagnostics are
// handled by code completion.
if (!request.is_interactive)
diag_engine->Publish(working_files, new_index->path,
new_index->diagnostics_);
// When main thread does IdMap request it will request the previous index if
// needed.
LOG_S(INFO) << "emit index for " << new_index->path;
result.push_back(
Index_OnIdMapped(request.cache_manager,
request.cache_manager->TryTakeOrLoad(path_to_index),
std::move(new_index), perf, request.is_interactive,
true /*write_to_disk*/));
}
QueueManager::instance()->on_id_mapped.EnqueueAll(std::move(result),
request.is_interactive);
}
bool IndexMain_DoParse(
DiagnosticsEngine* diag_engine,
WorkingFiles* working_files,
FileConsumerSharedState* file_consumer_shared,
TimestampManager* timestamp_manager,
IIndexer* indexer) {
auto* queue = QueueManager::instance();
std::optional<Index_Request> request = queue->index_request.TryPopFront();
if (!request)
return false;
Project::Entry entry;
entry.filename = request->path;
entry.args = request->args;
ParseFile(diag_engine, working_files, file_consumer_shared, timestamp_manager,
indexer, request.value(), entry);
return true;
}
bool IndexMain_DoCreateIndexUpdate(TimestampManager* timestamp_manager) {
auto* queue = QueueManager::instance();
bool did_work = false;
for (int i = 100; i--; ) {
std::optional<Index_OnIdMapped> response = queue->on_id_mapped.TryPopFront();
if (!response)
return did_work;
did_work = true;
Timer time;
// Write current index to disk if requested.
std::string path = response->current->path;
if (response->write_to_disk) {
LOG_S(INFO) << "store index for " << path;
time.Reset();
response->cache_manager->WriteToCache(*response->current);
response->perf.index_save_to_disk = time.ElapsedMicrosecondsAndReset();
timestamp_manager->UpdateCachedModificationTime(
path, response->current->last_modification_time);
}
// Build delta update.
IndexUpdate update = IndexUpdate::CreateDelta(response->previous.get(),
response->current.get());
response->perf.index_make_delta = time.ElapsedMicrosecondsAndReset();
LOG_S(INFO) << "built index for " << path
<< " (is_delta=" << !!response->previous << ")";
Index_OnIndexed reply(std::move(update), response->perf);
queue->on_indexed.PushBack(std::move(reply), response->is_interactive);
}
return did_work;
}
} // namespace
std::optional<int64_t> TimestampManager::GetLastCachedModificationTime(
ICacheManager* cache_manager,
const std::string& path) {
{
std::lock_guard<std::mutex> guard(mutex_);
auto it = timestamps_.find(path);
if (it != timestamps_.end())
return it->second;
}
IndexFile* file = cache_manager->TryLoad(path);
if (!file)
return std::nullopt;
UpdateCachedModificationTime(path, file->last_modification_time);
return file->last_modification_time;
}
void TimestampManager::UpdateCachedModificationTime(const std::string& path,
int64_t timestamp) {
std::lock_guard<std::mutex> guard(mutex_);
timestamps_[path] = timestamp;
}
ImportPipelineStatus::ImportPipelineStatus()
: num_active_threads(0), next_progress_output(0) {}
// Index a file using an already-parsed translation unit from code completion.
// Since most of the time for indexing a file comes from parsing, we can do
// real-time indexing.
// TODO: add option to disable this.
void IndexWithTuFromCodeCompletion(
FileConsumerSharedState* file_consumer_shared,
ClangTranslationUnit* tu,
const std::vector<CXUnsavedFile>& file_contents,
const std::string& path,
const std::vector<std::string>& args) {
file_consumer_shared->Reset(path);
PerformanceImportFile perf;
ClangIndex index;
auto indexes = ParseWithTu(file_consumer_shared, &perf, tu, &index, path,
args, file_contents);
if (indexes.empty())
return;
std::vector<Index_OnIdMapped> result;
for (std::unique_ptr<IndexFile>& new_index : indexes) {
Timer time;
std::shared_ptr<ICacheManager> cache_manager;
assert(false && "FIXME cache_manager");
// When main thread does IdMap request it will request the previous index if
// needed.
LOG_S(INFO) << "Emitting index for " << new_index->path;
result.push_back(Index_OnIdMapped(
cache_manager, cache_manager->TryTakeOrLoad(path), std::move(new_index),
perf, true /*is_interactive*/, true /*write_to_disk*/));
}
LOG_IF_S(WARNING, result.size() > 1)
<< "Code completion index update generated more than one index";
QueueManager::instance()->on_id_mapped.EnqueueAll(std::move(result));
}
void Indexer_Main(DiagnosticsEngine* diag_engine,
FileConsumerSharedState* file_consumer_shared,
TimestampManager* timestamp_manager,
ImportPipelineStatus* status,
Project* project,
WorkingFiles* working_files,
MultiQueueWaiter* waiter) {
auto* queue = QueueManager::instance();
// Build one index per-indexer, as building the index acquires a global lock.
auto indexer = std::make_unique<ClangIndexer>();
while (true) {
bool did_work = false;
{
ActiveThread active_thread(status);
// TODO: process all off IndexMain_DoIndex before calling
// IndexMain_DoCreateIndexUpdate for better icache behavior. We need to
// have some threads spinning on both though otherwise memory usage will
// get bad.
// We need to make sure to run both IndexMain_DoParse and
// IndexMain_DoCreateIndexUpdate so we don't starve querydb from doing any
// work. Running both also lets the user query the partially constructed
// index.
did_work = IndexMain_DoParse(diag_engine, working_files,
file_consumer_shared, timestamp_manager,
indexer.get()) ||
did_work;
did_work = IndexMain_DoCreateIndexUpdate(timestamp_manager) || did_work;
}
// We didn't do any work, so wait for a notification.
if (!did_work) {
waiter->Wait(&queue->on_indexed, &queue->index_request,
&queue->on_id_mapped);
}
}
}
namespace {
void QueryDb_OnIndexed(QueueManager* queue,
QueryDatabase* db,
ImportPipelineStatus* status,
SemanticHighlightSymbolCache* semantic_cache,
WorkingFiles* working_files,
Index_OnIndexed* response) {
Timer time;
db->ApplyIndexUpdate(&response->update);
// Update indexed content, inactive lines, and semantic highlighting.
if (response->update.files_def_update) {
auto& update = *response->update.files_def_update;
time.ResetAndPrint("apply index for " + update.value.path);
WorkingFile* working_file =
working_files->GetFileByFilename(update.value.path);
if (working_file) {
// Update indexed content.
working_file->SetIndexContent(update.file_content);
// Inactive lines.
EmitInactiveLines(working_file, update.value.inactive_regions);
// Semantic highlighting.
int file_id =
db->name2file_id[LowerPathIfInsensitive(working_file->filename)];
QueryFile* file = &db->files[file_id];
EmitSemanticHighlighting(db, semantic_cache, working_file, file);
}
}
}
} // namespace
bool QueryDb_ImportMain(QueryDatabase* db,
ImportPipelineStatus* status,
SemanticHighlightSymbolCache* semantic_cache,
WorkingFiles* working_files) {
auto* queue = QueueManager::instance();
ActiveThread active_thread(status);
bool did_work = false;
for (int i = 80; i--; ) {
std::optional<Index_OnIndexed> response = queue->on_indexed.TryPopFront();
if (!response)
break;
did_work = true;
QueryDb_OnIndexed(queue, db, status, semantic_cache, working_files,
&*response);
}
return did_work;
}