28693295402

Committed 04 Jul 2026 03:17AM UTC coverage: 52.408% (+0.1%) from 52.278%

Build # 28693295402

Build Type

push

github

Committed by

hariharan-devarajan

Commit Message

feat: silence noisy warnings on aarch64

Coverage Stats

37318 of 92666 branches covered (40.27%)

Branch coverage included in aggregate %.

33462 of 42389 relevant lines covered (78.94%)

20557.64 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

46.71

/src/dftracer/utils/server/viz_api.cpp

#include <dftracer/utils/core/common/logging.h>
#include <dftracer/utils/core/common/to_chars.h>
#include <dftracer/utils/core/coro/channel.h>
#include <dftracer/utils/core/pipeline/executor.h>
#include <dftracer/utils/core/tasks/coro_scope.h>
#include <dftracer/utils/server/http_request.h>
#include <dftracer/utils/server/http_response.h>
#include <dftracer/utils/server/router.h>
#include <dftracer/utils/server/trace_index.h>
#include <dftracer/utils/server/viz_api.h>
#include <dftracer/utils/utilities/common/json/json_doc_guard.h>
#include <dftracer/utils/utilities/common/json/json_value.h>
#include <dftracer/utils/utilities/common/query/query.h>
#include <dftracer/utils/utilities/composites/dft/views/view_builder_utility.h>
#include <dftracer/utils/utilities/composites/dft/views/view_definition.h>
#include <dftracer/utils/utilities/composites/dft/views/view_reader_utility.h>
#include <dftracer/utils/utilities/fileio/lines/sources/async_streaming_gz_line_generator.h>
#include <simdjson.h>

#include <atomic>
#include <cstddef>
#include <cstdint>
#include <limits>
#include <memory>
#include <mutex>
#include <string>
#include <unordered_set>
#include <vector>

namespace dftracer::utils::server {

using namespace dftracer::utils::utilities::composites::dft;
using namespace dftracer::utils::utilities::composites::dft::views;

static const std::unordered_set<std::string> HASH_METADATA_NAMES = {"FH", "HH",
                                                                    "SH"};

/// Normalize the "ts" field in a Chrome Trace Event JSON string by
/// subtracting an offset.  Returns the modified JSON.  Falls back to
/// the original string on parse failure.
static std::string normalize_event_ts(const std::string& event_json,
                                      std::uint64_t offset) {
    thread_local simdjson::dom::parser tl_parser;
    auto result = tl_parser.parse(event_json);
    if (result.error()) return event_json;

    auto root = result.value_unsafe();
    if (!root.is_object()) return event_json;

    auto ts_result = root["ts"];
    if (ts_result.error()) return event_json;

    std::uint64_t old_ts = 0;
    if (ts_result.is_uint64()) {
        old_ts = ts_result.get_uint64().value_unsafe();
    } else if (ts_result.is_int64()) {
        auto val = ts_result.get_int64().value_unsafe();
        old_ts = val >= 0 ? static_cast<std::uint64_t>(val) : 0;
    } else {
        return event_json;
    }

    std::uint64_t new_ts = old_ts >= offset ? old_ts - offset : 0;

    // simdjson DOM is read-only, so we need to rebuild the JSON with the new ts
    // Find "ts": and replace the value
    std::string modified = event_json;
    auto pos = modified.find("\"ts\":");
    if (pos == std::string::npos) return event_json;

    pos += 5;  // Skip past "ts":
    while (pos < modified.size() && std::isspace(modified[pos])) ++pos;

    auto end_pos = pos;
    while (end_pos < modified.size() &&
           (std::isdigit(modified[end_pos]) || modified[end_pos] == '-')) {
        ++end_pos;
    }

    modified.replace(pos, end_pos - pos, std::to_string(new_ts));
    return modified;
}

/// Compute the minimum event duration threshold for a given summary level.
/// Level 1 = full detail, higher levels filter shorter events.
static double duration_threshold(double begin, double end, unsigned level,
                                 unsigned viewport_width = 1920) {
    if (level <= 1) return 0.0;
    double range = end - begin;
    return range /
           (static_cast<double>(viewport_width) * static_cast<double>(level));
}

static std::string extract_json_value(simdjson::dom::element val) {
    if (val.is_string()) {
        return std::string(val.get_string().value_unsafe());
    }
    if (val.is_int64()) {
        return std::to_string(val.get_int64().value_unsafe());
    }
    if (val.is_uint64()) {
        return std::to_string(val.get_uint64().value_unsafe());
    }
    return {};
}

static void append_lane_clause(std::string& dsl, const char* field,
                               const std::string& val) {
    if (!dsl.empty()) dsl += " and ";
    bool numeric =
        !val.empty() && std::all_of(val.begin(), val.end(),
                                    [](char c) { return std::isdigit(c); });
    if (numeric) {
        dsl += std::string(field) + " == " + val;
    } else {
        dsl += std::string(field) + " == \"" + val + "\"";
    }
}

static void apply_lanes(std::string& dsl, std::string_view lanes_str) {
    if (lanes_str.empty()) return;

    thread_local simdjson::dom::parser tl_parser;
    auto result = tl_parser.parse(lanes_str.data(), lanes_str.size());
    if (result.error()) return;

    auto root = result.value_unsafe();

    if (root.is_array()) {
        auto arr = root.get_array().value_unsafe();
        for (auto item : arr) {
            if (!item.is_object()) continue;
            auto obj = item.get_object().value_unsafe();

            auto field_result = obj["field"];
            if (field_result.error()) field_result = obj["fields"];
            auto value_result = obj["value"];
            if (field_result.error() || value_result.error()) continue;

            if (!field_result.value_unsafe().is_string()) continue;
            const char* field =
                field_result.value_unsafe().get_c_str().value_unsafe();
            auto val = extract_json_value(value_result.value_unsafe());
            if (!val.empty()) append_lane_clause(dsl, field, val);
        }
    } else if (root.is_object()) {
        auto obj = root.get_object().value_unsafe();

        auto field_result = obj["field"];
        if (field_result.error()) field_result = obj["fields"];
        auto value_result = obj["value"];

        if (!field_result.error() && !value_result.error()) {
            if (field_result.value_unsafe().is_string()) {
                const char* field =
                    field_result.value_unsafe().get_c_str().value_unsafe();
                auto val = extract_json_value(value_result.value_unsafe());
                if (!val.empty()) append_lane_clause(dsl, field, val);
            }
        }
    }
}

static void apply_filters(std::string& dsl, std::string_view filters_str) {
    if (filters_str.empty()) return;

    thread_local simdjson::dom::parser tl_parser;
    auto result = tl_parser.parse(filters_str.data(), filters_str.size());
    if (result.error()) return;

    auto root = result.value_unsafe();
    if (!root.is_array()) return;

    auto arr = root.get_array().value_unsafe();
    for (auto item : arr) {
        if (!item.is_object()) continue;
        auto obj = item.get_object().value_unsafe();

        auto field_result = obj["field"];
        auto op_result = obj["op"];
        auto value_result = obj["value"];
        if (field_result.error() || op_result.error() || value_result.error())
            continue;

        if (!field_result.value_unsafe().is_string() ||
            !op_result.value_unsafe().is_string())
            continue;

        const char* field =
            field_result.value_unsafe().get_c_str().value_unsafe();
        const char* op = op_result.value_unsafe().get_c_str().value_unsafe();

        std::string val = extract_json_value(value_result.value_unsafe());
        if (val.empty()) continue;

        std::string op_str(op);
        std::string field_str(field);
        if (field_str == "begin") field_str = "ts";
        if (field_str == "end") field_str = "ts";
        if (field_str == "duration") field_str = "dur";

        std::string query_op;
        if (op_str == "=")
            query_op = "==";
        else if (op_str == ">=")
            query_op = ">=";
        else if (op_str == "<=")
            query_op = "<=";
        else if (op_str == ">")
            query_op = ">";
        else if (op_str == "<")
            query_op = "<";
        else
            continue;

        if (!dsl.empty()) dsl += " and ";
        bool numeric = !val.empty() && (std::isdigit(val[0]) || val[0] == '-');
        if (numeric || query_op != "==") {
            dsl += field_str + " " + query_op + " " + val;
        } else {
            dsl += field_str + " " + query_op + " \"" + val + "\"";
        }
    }
}

// --- GET /api/v1/viz/events ---
// Build the query view (time range + lane/filter/pid/tid/cat predicates) for a
// viz request from the parsed parameters.
static ViewDefinition build_viz_view(const QueryParams& params, double begin,
                                     double end, double min_dur) {
    ViewDefinition view;
    view.name = "viz_query";
    view.description = "Visualization query";

    // Reserve once and append in place: numbers go through to_chars into a
    // stack buffer (no per-number heap allocation, unlike std::to_string), and
    // literals/string_views are appended directly (no temporary
    // concatenations).
    std::string dsl;
    dsl.reserve(128);
    char numbuf[20];  // max digits of a uint64_t
    auto append_u64 = [&](std::uint64_t v) {
        dsl.append(numbuf, to_chars_u64(numbuf, numbuf + sizeof(numbuf), v));
    };

    dsl += "ts >= ";
    append_u64(static_cast<std::uint64_t>(begin));
    dsl += " and ts <= ";
    append_u64(static_cast<std::uint64_t>(end));
    if (min_dur > 0) {
        dsl += " and dur >= ";
        append_u64(static_cast<std::uint64_t>(min_dur));
    }

    apply_lanes(dsl, params.get("lanes"));
    apply_filters(dsl, params.get("filters"));

    auto pid = params.get("pid");
    if (!pid.empty()) {
        dsl += " and pid == ";
        dsl += pid;
    }

    auto tid = params.get("tid");
    if (!tid.empty()) {
        dsl += " and tid == ";
        dsl += tid;
    }

    auto cat = params.get("cat");
    if (!cat.empty()) {
        dsl += " and cat == \"";
        dsl += cat;
        dsl += '"';
    }

    view.with_query(dsl);
    return view;
}

// Select the files to scan: the explicit ?file= or all indexed files, then drop
// files whose cached time bounds don't overlap [begin, end]. Pure/synchronous.
static std::vector<const TraceIndex::FileInfo*> select_viz_target_files(
    TraceIndex& index, const QueryParams& params, double begin, double end) {
    auto target_files = collect_candidate_files(index, params);

    if (begin > 0 || end > 0) {
        std::vector<const TraceIndex::FileInfo*> filtered;
        filtered.reserve(target_files.size());
        for (auto* fi : target_files) {
            if (fi->min_timestamp_us == 0 && fi->max_timestamp_us == 0) {
                filtered.push_back(fi);
                continue;
            }
            double fi_min = static_cast<double>(fi->min_timestamp_us);
            double fi_max = static_cast<double>(fi->max_timestamp_us);
            if (fi_max < begin || fi_min > end) continue;
            filtered.push_back(fi);
        }
        target_files = std::move(filtered);
    }
    return target_files;
}

// Normalize event timestamps (when global_min > 0) and serialize the collected
// events plus metadata into the Chrome Trace Event Format body. `global_min` is
// the de-normalization base (already 0 unless normalization is active);
// `display_global_min` is the value reported in the metadata. Pure/synchronous.
static std::string build_viz_events_body(std::vector<std::string>& events,
                                         std::uint64_t global_min,
                                         double meta_begin, double meta_end,
                                         int limit, bool truncated,
                                         std::uint64_t display_global_min) {
    if (global_min > 0) {
        for (auto& event : events) {
            event = normalize_event_ts(event, global_min);
        }
    }

    // Pre-compute size to avoid repeated reallocations.
    std::size_t body_size = 256;
    for (const auto& ev : events) body_size += ev.size() + 1;
    std::string body;
    body.reserve(body_size);
    body += "{\"events\":[";
    for (std::size_t i = 0; i < events.size(); ++i) {
        if (i > 0) body += ',';
        body += events[i];  // Already JSON
    }
    body += "],\"metadata\":{\"begin\":";
    body += std::to_string(meta_begin);
    body += ",\"end\":";
    body += std::to_string(meta_end);
    body += ",\"count\":";
    body += std::to_string(events.size());
    body += ",\"limit\":";
    body += std::to_string(limit);
    body += ",\"truncated\":";
    body += truncated ? "true" : "false";
    body += ",\"ts_normalized\":";
    body += (global_min > 0) ? "true" : "false";
    body += ",\"global_min_timestamp_us\":";
    body += std::to_string(display_global_min);
    body += "}}";
    return body;
}

static coro::CoroTask<HttpResponse> handle_viz_events(
    const HttpRequest& /*req*/, const QueryParams& params, TraceIndex& index) {
    // Required: begin, end, summary
    if (!params.has("begin") || !params.has("end") || !params.has("summary")) {
        co_return HttpResponse::bad_request(
            "Missing required parameters: begin, end, summary");
    }

    double begin = params.get_double("begin", 0);
    double end = params.get_double("end", 0);
    int summary = params.get_int("summary", 1);
    if (summary < 1) summary = 1;

    // Timestamp normalization: default ON, opt-out with ?ts_normalize=0
    auto ts_norm_param = params.get("ts_normalize");
    bool normalize = ts_norm_param.empty() || ts_norm_param != "0";

    std::uint64_t global_min = 0;
    if (normalize) {
        global_min = index.global_min_timestamp_us();
        if (global_min == std::numeric_limits<std::uint64_t>::max()) {
            global_min = 0;  // No valid bounds, skip normalization
        }
    }

    // When normalization is active the user sends normalized
    // begin/end values (relative to global_min).  De-normalize them
    // so the predicate filters against absolute timestamps.
    double original_begin = begin;
    double original_end = end;
    if (normalize && global_min > 0) {
        begin += static_cast<double>(global_min);
        end += static_cast<double>(global_min);
    }

    double min_dur =
        duration_threshold(begin, end, static_cast<unsigned>(summary));

    ViewDefinition view = build_viz_view(params, begin, end, min_dur);

    // Optional limit: 0 (default) means no limit.
    int limit = params.get_int("limit", 0);
    if (limit < 0) limit = 0;

    std::vector<const TraceIndex::FileInfo*> target_files =
        select_viz_target_files(index, params, begin, end);

    std::vector<std::string> collected_events;

    bool truncated = false;

    if (target_files.size() <= 1) {
        for (auto* file_info : target_files) {
            if (limit > 0 &&
                static_cast<int>(collected_events.size()) >= limit) {
                truncated = true;
                break;
            }
            if (file_info->uncompressed_size == 0 &&
                file_info->num_checkpoints == 0)
                continue;

            ViewBuilderInput builder_input;
            builder_input.with_view(view)
                .with_file_path(file_info->path)
                .with_index_path(
                    file_info->has_bloom_data ? file_info->index_path : "")
                .with_uncompressed_size(file_info->uncompressed_size)
                .with_num_checkpoints(file_info->num_checkpoints)
                .with_bloom_cache(&index.bloom_cache())
                .with_time_range(begin, end);

            ViewBuilderUtility builder;
            auto build_output = co_await builder.process(builder_input);
            if (!build_output || !build_output->file_may_match) continue;

            for (const auto& candidate : build_output->candidates) {
                if (limit > 0 &&
                    static_cast<int>(collected_events.size()) >= limit) {
                    truncated = true;
                    break;
                }
                ViewReaderInput reader_input;
                reader_input.with_file_path(file_info->path)
                    .with_index_path(file_info->index_path)
                    .with_byte_range(candidate.start_byte, candidate.end_byte)
                    .with_checkpoint_idx(candidate.checkpoint_idx)
                    .with_view(view);

                ViewReaderUtility reader;
                auto gen = reader.process(reader_input);
                while (auto batch = co_await gen.next()) {
                    for (auto& event : batch->events) {
                        if (limit > 0 &&
                            static_cast<int>(collected_events.size()) >=
                                limit) {
                            truncated = true;
                            break;
                        }
                        collected_events.emplace_back(event);
                    }
                    if (truncated) break;
                }
            }
        }
    } else {
        std::size_t num_workers =
            std::min(index.max_concurrent(), target_files.size());
        auto* executor = Executor::current();

        auto file_chan = coro::make_channel<std::size_t>(num_workers * 2);
        auto collected_mutex = std::make_shared<std::mutex>();
        auto remaining = std::make_shared<std::atomic<int>>(
            limit > 0 ? limit : std::numeric_limits<int>::max());

        auto* target_files_ptr = &target_files;
        auto* collected_ptr = &collected_events;
        auto* view_ptr = &view;
        auto* bloom_cache_ptr = &index.bloom_cache();
        double t_begin = begin;
        double t_end = end;

        CoroScope scope(executor);

        scope.spawn([ch = file_chan->producer(), target_files_ptr](
                        CoroScope&) mutable -> coro::CoroTask<void> {
            auto guard = ch.guard();
            for (std::size_t i = 0; i < target_files_ptr->size(); ++i) {
                if (!co_await ch.send(i)) co_return;
            }
            co_return;
        });

        for (std::size_t w = 0; w < num_workers; ++w) {
            scope.spawn([file_chan, target_files_ptr, collected_mutex,
                         collected_ptr, view_ptr, bloom_cache_ptr, remaining,
                         t_begin, t_end](CoroScope&) -> coro::CoroTask<void> {
                while (auto fi_opt = co_await file_chan->receive()) {
                    if (remaining->load(std::memory_order_relaxed) <= 0)
                        co_return;
                    auto* file_info = (*target_files_ptr)[*fi_opt];

                    if (file_info->uncompressed_size == 0 &&
                        file_info->num_checkpoints == 0)
                        continue;

                    ViewBuilderInput builder_input;
                    builder_input.with_view(*view_ptr)
                        .with_file_path(file_info->path)
                        .with_index_path(file_info->has_bloom_data
                                             ? file_info->index_path
                                             : "")
                        .with_uncompressed_size(file_info->uncompressed_size)
                        .with_num_checkpoints(file_info->num_checkpoints)
                        .with_bloom_cache(bloom_cache_ptr)
                        .with_time_range(t_begin, t_end);

                    ViewBuilderUtility builder;
                    auto build_output = co_await builder.process(builder_input);
                    if (!build_output || !build_output->file_may_match)
                        continue;

                    for (const auto& candidate : build_output->candidates) {
                        if (remaining->load(std::memory_order_relaxed) <= 0)
                            break;

                        ViewReaderInput reader_input;
                        reader_input.with_file_path(file_info->path)
                            .with_index_path(file_info->index_path)
                            .with_byte_range(candidate.start_byte,
                                             candidate.end_byte)
                            .with_checkpoint_idx(candidate.checkpoint_idx)
                            .with_view(*view_ptr);

                        ViewReaderUtility reader;
                        auto gen = reader.process(reader_input);
                        while (auto batch = co_await gen.next()) {
                            if (!batch->events.empty()) {
                                std::lock_guard<std::mutex> lock(
                                    *collected_mutex);
                                for (auto& event : batch->events) {
                                    collected_ptr->emplace_back(event);
                                }
                                remaining->fetch_sub(
                                    static_cast<int>(batch->events.size()));
                            }
                        }
                    }
                }
                co_return;
            });
        }

        co_await scope.join();

        if (limit > 0 && static_cast<int>(collected_events.size()) > limit) {
            collected_events.resize(static_cast<std::size_t>(limit));
            truncated = true;
        }
    }

    std::string body = build_viz_events_body(
        collected_events, global_min, original_begin, original_end, limit,
        truncated, index.global_min_timestamp_us());
    co_return HttpResponse::ok(body);
}

void register_viz_api(Router& router, TraceIndex& index) {
    auto* index_ptr = &index;

    router.get(
        "/api/v1/viz/events",
        [index_ptr](const HttpRequest& req,
                    const QueryParams& params) -> coro::CoroTask<HttpResponse> {
            co_return co_await handle_viz_events(req, params, *index_ptr);
        });
}

}  // namespace dftracer::utils::server

1	#include <dftracer/utils/core/common/logging.h>
2	#include <dftracer/utils/core/common/to_chars.h>
3	#include <dftracer/utils/core/coro/channel.h>
4	#include <dftracer/utils/core/pipeline/executor.h>
5	#include <dftracer/utils/core/tasks/coro_scope.h>
6	#include <dftracer/utils/server/http_request.h>
7	#include <dftracer/utils/server/http_response.h>
8	#include <dftracer/utils/server/router.h>
9	#include <dftracer/utils/server/trace_index.h>
10	#include <dftracer/utils/server/viz_api.h>
11	#include <dftracer/utils/utilities/common/json/json_doc_guard.h>
12	#include <dftracer/utils/utilities/common/json/json_value.h>
13	#include <dftracer/utils/utilities/common/query/query.h>
14	#include <dftracer/utils/utilities/composites/dft/views/view_builder_utility.h>
15	#include <dftracer/utils/utilities/composites/dft/views/view_definition.h>
16	#include <dftracer/utils/utilities/composites/dft/views/view_reader_utility.h>
17	#include <dftracer/utils/utilities/fileio/lines/sources/async_streaming_gz_line_generator.h>
18	#include <simdjson.h>
19
20	#include <atomic>
21	#include <cstddef>
22	#include <cstdint>
23	#include <limits>
24	#include <memory>
25	#include <mutex>
26	#include <string>
27	#include <unordered_set>
28	#include <vector>
29
30	namespace dftracer::utils::server {
31
32	using namespace dftracer::utils::utilities::composites::dft;
33	using namespace dftracer::utils::utilities::composites::dft::views;
34
35	static const std::unordered_set<std::string> HASH_METADATA_NAMES = {"FH", "HH",	265!
36	"SH"};	265!
37
38	/// Normalize the "ts" field in a Chrome Trace Event JSON string by
39	/// subtracting an offset. Returns the modified JSON. Falls back to
40	/// the original string on parse failure.
41	static std::string normalize_event_ts(const std::string& event_json,	300✔
42	std::uint64_t offset) {
43	thread_local simdjson::dom::parser tl_parser;	300✔
44	auto result = tl_parser.parse(event_json);	300✔
45	if (result.error()) return event_json;	300!
46
47	auto root = result.value_unsafe();	300✔
48	if (!root.is_object()) return event_json;	300!
49
50	auto ts_result = root["ts"];	300✔
51	if (ts_result.error()) return event_json;	300!
52
53	std::uint64_t old_ts = 0;	300✔
54	if (ts_result.is_uint64()) {	300!
55	old_ts = ts_result.get_uint64().value_unsafe();	300✔
56	} else if (ts_result.is_int64()) {	150!
57	auto val = ts_result.get_int64().value_unsafe();	×
58	old_ts = val >= 0 ? static_cast<std::uint64_t>(val) : 0;	×
59	} else {
60	return event_json;	×
61	}
62
63	std::uint64_t new_ts = old_ts >= offset ? old_ts - offset : 0;	300!
64
65	// simdjson DOM is read-only, so we need to rebuild the JSON with the new ts
66	// Find "ts": and replace the value
67	std::string modified = event_json;	300!
68	auto pos = modified.find("\"ts\":");	300✔
69	if (pos == std::string::npos) return event_json;	300!
70
71	pos += 5; // Skip past "ts":	300✔
72	while (pos < modified.size() && std::isspace(modified[pos])) ++pos;	300!
73
74	auto end_pos = pos;	300✔
75	while (end_pos < modified.size() &&	5,100✔
76	(std::isdigit(modified[end_pos]) \|\| modified[end_pos] == '-')) {	3,300!
77	++end_pos;	3,000✔
78	}
79
80	modified.replace(pos, end_pos - pos, std::to_string(new_ts));	300!
81	return modified;	300✔
82	}	300✔
83
84	/// Compute the minimum event duration threshold for a given summary level.
85	/// Level 1 = full detail, higher levels filter shorter events.
86	static double duration_threshold(double begin, double end, unsigned level,	12✔
87	unsigned viewport_width = 1920) {
88	if (level <= 1) return 0.0;	12!
89	double range = end - begin;	×
90	return range /
91	(static_cast<double>(viewport_width) * static_cast<double>(level));	×
92	}	6✔
93
94	static std::string extract_json_value(simdjson::dom::element val) {	6✔
95	if (val.is_string()) {	6✔
96	return std::string(val.get_string().value_unsafe());	3!
97	}
98	if (val.is_int64()) {	4!
99	return std::to_string(val.get_int64().value_unsafe());	6!
100	}
101	if (val.is_uint64()) {	×
102	return std::to_string(val.get_uint64().value_unsafe());	×
103	}
104	return {};	×
105	}	3✔
106
107	static void append_lane_clause(std::string& dsl, const char* field,	2✔
108	const std::string& val) {
109	if (!dsl.empty()) dsl += " and ";	2✔
110	bool numeric =	1✔
111	!val.empty() && std::all_of(val.begin(), val.end(),	2!
112	[](char c) { return std::isdigit(c); });	2✔
113	if (numeric) {	2!
114	dsl += std::string(field) + " == " + val;	2!
115	} else {	1✔
116	dsl += std::string(field) + " == \"" + val + "\"";	×
117	}
118	}	2✔
119
120	static void apply_lanes(std::string& dsl, std::string_view lanes_str) {	12✔
121	if (lanes_str.empty()) return;	12✔
122
123	thread_local simdjson::dom::parser tl_parser;	2✔
124	auto result = tl_parser.parse(lanes_str.data(), lanes_str.size());	2✔
125	if (result.error()) return;	2!
126
127	auto root = result.value_unsafe();	2✔
128
129	if (root.is_array()) {	2!
130	auto arr = root.get_array().value_unsafe();	2✔
131	for (auto item : arr) {	4✔
132	if (!item.is_object()) continue;	2✔
133	auto obj = item.get_object().value_unsafe();	2✔
134
135	auto field_result = obj["field"];	2✔
136	if (field_result.error()) field_result = obj["fields"];	2✔
137	auto value_result = obj["value"];	2✔
138	if (field_result.error() \|\| value_result.error()) continue;	2!
139
140	if (!field_result.value_unsafe().is_string()) continue;	2✔
141	const char* field =	1✔
142	field_result.value_unsafe().get_c_str().value_unsafe();	2✔
143	auto val = extract_json_value(value_result.value_unsafe());	2!
144	if (!val.empty()) append_lane_clause(dsl, field, val);	2✔
145	}	2✔
146	} else if (root.is_object()) {	1!
147	auto obj = root.get_object().value_unsafe();	×
148
149	auto field_result = obj["field"];	×
150	if (field_result.error()) field_result = obj["fields"];	×
151	auto value_result = obj["value"];	×
152
153	if (!field_result.error() && !value_result.error()) {	×
154	if (field_result.value_unsafe().is_string()) {	×
155	const char* field =
156	field_result.value_unsafe().get_c_str().value_unsafe();	×
157	auto val = extract_json_value(value_result.value_unsafe());	×
158	if (!val.empty()) append_lane_clause(dsl, field, val);	×
159	}	×
160	}
161	}
162	}	6✔
163
164	static void apply_filters(std::string& dsl, std::string_view filters_str) {	12✔
165	if (filters_str.empty()) return;	12✔
166
167	thread_local simdjson::dom::parser tl_parser;	4✔
168	auto result = tl_parser.parse(filters_str.data(), filters_str.size());	4✔
169	if (result.error()) return;	4!
170
171	auto root = result.value_unsafe();	4✔
172	if (!root.is_array()) return;	4✔
173
174	auto arr = root.get_array().value_unsafe();	4✔
175	for (auto item : arr) {	8✔
176	if (!item.is_object()) continue;	4✔
177	auto obj = item.get_object().value_unsafe();	4✔
178
179	auto field_result = obj["field"];	4✔
180	auto op_result = obj["op"];	4✔
181	auto value_result = obj["value"];	4✔
182	if (field_result.error() \|\| op_result.error() \|\| value_result.error())	4!
183	continue;	×
184
185	if (!field_result.value_unsafe().is_string() \|\|	6!
186	!op_result.value_unsafe().is_string())	4!
187	continue;	×
188
189	const char* field =	2✔
190	field_result.value_unsafe().get_c_str().value_unsafe();	4✔
191	const char* op = op_result.value_unsafe().get_c_str().value_unsafe();	4✔
192
193	std::string val = extract_json_value(value_result.value_unsafe());	4!
194	if (val.empty()) continue;	4!
195
196	std::string op_str(op);	4✔
197	std::string field_str(field);	4✔
198	if (field_str == "begin") field_str = "ts";	4!
199	if (field_str == "end") field_str = "ts";	4!
200	if (field_str == "duration") field_str = "dur";	4!
201
202	std::string query_op;	4✔
203	if (op_str == "=")	4✔
204	query_op = "==";	2!
205	else if (op_str == ">=")	2!
206	query_op = ">=";	2!
207	else if (op_str == "<=")	×
208	query_op = "<=";	×
209	else if (op_str == ">")	×
210	query_op = ">";	×
211	else if (op_str == "<")	×
212	query_op = "<";	×
213	else
214	continue;	×
215
216	if (!dsl.empty()) dsl += " and ";	4!
217	bool numeric = !val.empty() && (std::isdigit(val[0]) \|\| val[0] == '-');	6!
218	if (numeric \|\| query_op != "==") {	4!
219	dsl += field_str + " " + query_op + " " + val;	4!
220	} else {	2✔
221	dsl += field_str + " " + query_op + " \"" + val + "\"";	×
222	}
223	}	4!
224	}	6✔
225
226	// --- GET /api/v1/viz/events ---
227	// Build the query view (time range + lane/filter/pid/tid/cat predicates) for a
228	// viz request from the parsed parameters.
229	static ViewDefinition build_viz_view(const QueryParams& params, double begin,	12✔
230	double end, double min_dur) {
231	ViewDefinition view;	12✔
232	view.name = "viz_query";	12!
233	view.description = "Visualization query";	12!
234
235	// Reserve once and append in place: numbers go through to_chars into a
236	// stack buffer (no per-number heap allocation, unlike std::to_string), and
237	// literals/string_views are appended directly (no temporary
238	// concatenations).
239	std::string dsl;	12✔
240	dsl.reserve(128);	12!
241	char numbuf[20]; // max digits of a uint64_t
242	auto append_u64 = [&](std::uint64_t v) {	30✔
243	dsl.append(numbuf, to_chars_u64(numbuf, numbuf + sizeof(numbuf), v));	24✔
244	};	30✔
245
246	dsl += "ts >= ";	12!
247	append_u64(static_cast<std::uint64_t>(begin));	12!
248	dsl += " and ts <= ";	12!
249	append_u64(static_cast<std::uint64_t>(end));	12!
250	if (min_dur > 0) {	12!
251	dsl += " and dur >= ";	×
252	append_u64(static_cast<std::uint64_t>(min_dur));	×
253	}
254
255	apply_lanes(dsl, params.get("lanes"));	12!
256	apply_filters(dsl, params.get("filters"));	12!
257
258	auto pid = params.get("pid");	12!
259	if (!pid.empty()) {	12!
260	dsl += " and pid == ";	×
261	dsl += pid;	×
262	}
263
264	auto tid = params.get("tid");	12!
265	if (!tid.empty()) {	12!
266	dsl += " and tid == ";	×
267	dsl += tid;	×
268	}
269
270	auto cat = params.get("cat");	12!
271	if (!cat.empty()) {	12!
272	dsl += " and cat == \"";	×
273	dsl += cat;	×
274	dsl += '"';	×
275	}
276
277	view.with_query(dsl);	12!
278	return view;	18✔
279	}	12!
280
281	// Select the files to scan: the explicit ?file= or all indexed files, then drop
282	// files whose cached time bounds don't overlap [begin, end]. Pure/synchronous.
283	static std::vector<const TraceIndex::FileInfo*> select_viz_target_files(	12✔
284	TraceIndex& index, const QueryParams& params, double begin, double end) {
285	auto target_files = collect_candidate_files(index, params);	12✔
286
287	if (begin > 0 \|\| end > 0) {	12!
288	std::vector<const TraceIndex::FileInfo*> filtered;	12✔
289	filtered.reserve(target_files.size());	12!
290	for (auto* fi : target_files) {	24✔
291	if (fi->min_timestamp_us == 0 && fi->max_timestamp_us == 0) {	12!
292	filtered.push_back(fi);	×
293	continue;	×
294	}
295	double fi_min = static_cast<double>(fi->min_timestamp_us);	12✔
296	double fi_max = static_cast<double>(fi->max_timestamp_us);	12✔
297	if (fi_max < begin \|\| fi_min > end) continue;	12!
298	filtered.push_back(fi);	10!
299	}
300	target_files = std::move(filtered);	12✔
301	}	12✔
302	return target_files;	12✔
303	}	6!
304
305	// Normalize event timestamps (when global_min > 0) and serialize the collected
306	// events plus metadata into the Chrome Trace Event Format body. `global_min` is
307	// the de-normalization base (already 0 unless normalization is active);
308	// `display_global_min` is the value reported in the metadata. Pure/synchronous.
309	static std::string build_viz_events_body(std::vector<std::string>& events,	12✔
310	std::uint64_t global_min,
311	double meta_begin, double meta_end,
312	int limit, bool truncated,
313	std::uint64_t display_global_min) {
314	if (global_min > 0) {	12✔
315	for (auto& event : events) {	310✔
316	event = normalize_event_ts(event, global_min);	300!
317	}
318	}	5✔
319
320	// Pre-compute size to avoid repeated reallocations.
321	std::size_t body_size = 256;	12✔
322	for (const auto& ev : events) body_size += ev.size() + 1;	312✔
323	std::string body;	12✔
324	body.reserve(body_size);	12!
325	body += "{\"events\":[";	12!
326	for (std::size_t i = 0; i < events.size(); ++i) {	312✔
327	if (i > 0) body += ',';	300!
328	body += events[i]; // Already JSON	300!
329	}	150✔
330	body += "],\"metadata\":{\"begin\":";	12!
331	body += std::to_string(meta_begin);	12!
332	body += ",\"end\":";	12!
333	body += std::to_string(meta_end);	12!
334	body += ",\"count\":";	12!
335	body += std::to_string(events.size());	12!
336	body += ",\"limit\":";	12!
337	body += std::to_string(limit);	12!
338	body += ",\"truncated\":";	12!
339	body += truncated ? "true" : "false";	12!
340	body += ",\"ts_normalized\":";	12!
341	body += (global_min > 0) ? "true" : "false";	12!
342	body += ",\"global_min_timestamp_us\":";	12!
343	body += std::to_string(display_global_min);	12!
344	body += "}}";	12!
345	return body;	12✔
346	}	6!
347
348	static coro::CoroTask<HttpResponse> handle_viz_events(	66!
349	const HttpRequest& /req/, const QueryParams& params, TraceIndex& index) {	7!
350	// Required: begin, end, summary
351	if (!params.has("begin") \|\| !params.has("end") \|\| !params.has("summary")) {	7!
352	co_return HttpResponse::bad_request(	8!
353	"Missing required parameters: begin, end, summary");	1!
354	}
355
356	double begin = params.get_double("begin", 0);	6!
357	double end = params.get_double("end", 0);	6!
358	int summary = params.get_int("summary", 1);	6!
359	if (summary < 1) summary = 1;	6!
360
361	// Timestamp normalization: default ON, opt-out with ?ts_normalize=0
362	auto ts_norm_param = params.get("ts_normalize");	6!
363	bool normalize = ts_norm_param.empty() \|\| ts_norm_param != "0";	6!
364
365	std::uint64_t global_min = 0;	6✔
366	if (normalize) {	6✔
367	global_min = index.global_min_timestamp_us();	5!
368	if (global_min == std::numeric_limits<std::uint64_t>::max()) {	5!
369	global_min = 0; // No valid bounds, skip normalization
370	}
371	}	5✔
372
373	// When normalization is active the user sends normalized
374	// begin/end values (relative to global_min). De-normalize them
375	// so the predicate filters against absolute timestamps.
376	double original_begin = begin;	6✔
377	double original_end = end;	6✔
378	if (normalize && global_min > 0) {	6!
379	begin += static_cast<double>(global_min);	5✔
380	end += static_cast<double>(global_min);	5✔
381	}	5✔
382
383	double min_dur =	6✔
384	duration_threshold(begin, end, static_cast<unsigned>(summary));	6!
385
386	ViewDefinition view = build_viz_view(params, begin, end, min_dur);	6!
387
388	// Optional limit: 0 (default) means no limit.
389	int limit = params.get_int("limit", 0);	6!
390	if (limit < 0) limit = 0;	6!
391
392	std::vector<const TraceIndex::FileInfo*> target_files =	6✔
393	select_viz_target_files(index, params, begin, end);	6!
394
395	std::vector<std::string> collected_events;	6✔
396
397	bool truncated = false;	6✔
398
399	if (target_files.size() <= 1) {	6!
400	for (auto* file_info : target_files) {	23✔
401	if (limit > 0 &&	5!
402	static_cast<int>(collected_events.size()) >= limit) {
403	truncated = true;
404	break;
405	}
406	if (file_info->uncompressed_size == 0 &&	5!
407	file_info->num_checkpoints == 0)
408	continue;
409
410	ViewBuilderInput builder_input;	5✔
411	builder_input.with_view(view)	10!
412	.with_file_path(file_info->path)	5!
413	.with_index_path(	5!
414	file_info->has_bloom_data ? file_info->index_path : "")	5!
415	.with_uncompressed_size(file_info->uncompressed_size)	5!
416	.with_num_checkpoints(file_info->num_checkpoints)	5!
417	.with_bloom_cache(&index.bloom_cache())	5!
418	.with_time_range(begin, end);	5!
419
420	ViewBuilderUtility builder;	5!
421	auto build_output = co_await builder.process(builder_input);	10!
422	if (!build_output \|\| !build_output->file_may_match) continue;	5!
423
424	for (const auto& candidate : build_output->candidates) {	18!
425	if (limit > 0 &&	3!
426	static_cast<int>(collected_events.size()) >= limit) {
427	truncated = true;
428	break;
429	}
430	ViewReaderInput reader_input;	3✔
431	reader_input.with_file_path(file_info->path)	3!
432	.with_index_path(file_info->index_path)	3!
433	.with_byte_range(candidate.start_byte, candidate.end_byte)	3!
434	.with_checkpoint_idx(candidate.checkpoint_idx)	3!
435	.with_view(view);	3!
436
437	ViewReaderUtility reader;	3!
438	auto gen = reader.process(reader_input);	3!
439	while (auto batch = co_await gen.next()) {	24!
440	for (auto& event : batch->events) {	153✔
441	if (limit > 0 &&	150!
442	static_cast<int>(collected_events.size()) >=
443	limit) {
444	truncated = true;
445	break;
446	}
447	collected_events.emplace_back(event);	150!
448	}	150!
449	if (truncated) break;	3!
450	}	6✔
451	}	15✔
452	}	17✔
453	} else {	6✔
454	std::size_t num_workers =
455	std::min(index.max_concurrent(), target_files.size());	×
456	auto* executor = Executor::current();
457
458	auto file_chan = coro::make_channel<std::size_t>(num_workers * 2);	×
459	auto collected_mutex = std::make_shared<std::mutex>();	×
460	auto remaining = std::make_shared<std::atomic<int>>(	×
461	limit > 0 ? limit : std::numeric_limits<int>::max());	×
462
463	auto* target_files_ptr = &target_files;
464	auto* collected_ptr = &collected_events;
465	auto* view_ptr = &view;
466	auto* bloom_cache_ptr = &index.bloom_cache();	×
467	double t_begin = begin;
468	double t_end = end;
469
470	CoroScope scope(executor);	×
471
472	scope.spawn([ch = file_chan->producer(), target_files_ptr](	×
473	CoroScope&) mutable -> coro::CoroTask<void> {	×
474	auto guard = ch.guard();	×
475	for (std::size_t i = 0; i < target_files_ptr->size(); ++i) {	×
476	if (!co_await ch.send(i)) co_return;	×
477	}
478	co_return;
479	});	×
480
481	for (std::size_t w = 0; w < num_workers; ++w) {	×
482	scope.spawn([file_chan, target_files_ptr, collected_mutex,	×
483	collected_ptr, view_ptr, bloom_cache_ptr, remaining,
484	t_begin, t_end](CoroScope&) -> coro::CoroTask<void> {	×
485	while (auto fi_opt = co_await file_chan->receive()) {	×
486	if (remaining->load(std::memory_order_relaxed) <= 0)	×
487	co_return;
488	auto* file_info = (target_files_ptr)[fi_opt];
489
490	if (file_info->uncompressed_size == 0 &&	×
491	file_info->num_checkpoints == 0)
492	continue;
493
494	ViewBuilderInput builder_input;
495	builder_input.with_view(*view_ptr)	×
496	.with_file_path(file_info->path)	×
497	.with_index_path(file_info->has_bloom_data	×
498	? file_info->index_path	×
499	: "")	×
500	.with_uncompressed_size(file_info->uncompressed_size)	×
501	.with_num_checkpoints(file_info->num_checkpoints)	×
502	.with_bloom_cache(bloom_cache_ptr)	×
503	.with_time_range(t_begin, t_end);	×
504
505	ViewBuilderUtility builder;	×
506	auto build_output = co_await builder.process(builder_input);	×
507	if (!build_output \|\| !build_output->file_may_match)	×
508	continue;
509
510	for (const auto& candidate : build_output->candidates) {	×
511	if (remaining->load(std::memory_order_relaxed) <= 0)	×
512	break;
513
514	ViewReaderInput reader_input;
515	reader_input.with_file_path(file_info->path)	×
516	.with_index_path(file_info->index_path)	×
517	.with_byte_range(candidate.start_byte,	×
518	candidate.end_byte)
519	.with_checkpoint_idx(candidate.checkpoint_idx)	×
520	.with_view(*view_ptr);	×
521
522	ViewReaderUtility reader;	×
523	auto gen = reader.process(reader_input);	×
524	while (auto batch = co_await gen.next()) {	×
525	if (!batch->events.empty()) {	×
526	std::lock_guard<std::mutex> lock(	×
527	*collected_mutex);
528	for (auto& event : batch->events) {	×
529	collected_ptr->emplace_back(event);	×
530	}
531	remaining->fetch_sub(
532	static_cast<int>(batch->events.size()));
533	}
534	}	×
535	}	×
536	}	×
537	co_return;
538	});	×
539	}
540
541	co_await scope.join();	×
542
543	if (limit > 0 && static_cast<int>(collected_events.size()) > limit) {	×
544	collected_events.resize(static_cast<std::size_t>(limit));	×
545	truncated = true;
546	}
547	}	×
548
549	std::string body = build_viz_events_body(	6!
550	collected_events, global_min, original_begin, original_end, limit,	6✔
551	truncated, index.global_min_timestamp_us());	6!
552	co_return HttpResponse::ok(body);	6!
553	}	66!
554
555	void register_viz_api(Router& router, TraceIndex& index) {	4✔
556	auto* index_ptr = &index;	4✔
557
558	router.get(	4!
559	"/api/v1/viz/events",	2✔
560	[index_ptr](const HttpRequest& req,	58!
561	const QueryParams& params) -> coro::CoroTask<HttpResponse> {	7!
562	co_return co_await handle_viz_events(req, params, *index_ptr);	35!
563	});	28!
564	}	4✔
565
566	} // namespace dftracer::utils::server

llnl / dftracer-utils / 28693295402

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous