23935950925

Committed 03 Apr 2026 06:01AM UTC coverage: 51.257% (-0.2%) from 51.498%

Build # 23935950925

Build Type

Pull #63

github

Committed by

web-flow

Commit Message

Merge a8f78da48 into 773a62661

Pull Request Pull Request #63: feat(aggregator): support profile/system counter aggregation and custom metric Arrow output

Coverage Stats

23412 of 58927 branches covered (39.73%)

Branch coverage included in aggregate %.

344 of 571 new or added lines in 7 files covered. (60.25%)

9 existing lines in 4 files now uncovered.

20297 of 26348 relevant lines covered (77.03%)

13069.39 hits per line

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

47.52

/src/dftracer/utils/utilities/reader/internal/gzip_reader.cpp

#include <dftracer/utils/core/coro/task.h>
#include <dftracer/utils/core/utils/timer.h>
#include <dftracer/utils/utilities/indexer/internal/indexer.h>
#include <dftracer/utils/utilities/indexer/internal/indexer_factory.h>
#include <dftracer/utils/utilities/reader/internal/error.h>
#include <dftracer/utils/utilities/reader/internal/gzip_reader.h>
#include <dftracer/utils/utilities/reader/internal/stream_config.h>
#include <dftracer/utils/utilities/reader/internal/streams/gzip_byte_stream.h>
#include <dftracer/utils/utilities/reader/internal/streams/gzip_line_byte_stream.h>
#include <dftracer/utils/utilities/reader/internal/streams/line_stream.h>
#include <dftracer/utils/utilities/reader/internal/streams/multi_line_stream.h>
#include <dftracer/utils/utilities/reader/internal/string_line_processor.h>

#include <cstdio>
#include <cstring>
#include <limits>
#include <string_view>

static void validate_parameters(
    const char *buffer, std::size_t buffer_size, std::size_t start_bytes,
    std::size_t end_bytes,
    std::size_t max_bytes = std::numeric_limits<std::size_t>::max()) {
    if (!buffer || buffer_size == 0) {
        throw dftracer::utils::utilities::reader::internal::ReaderError(
            dftracer::utils::utilities::reader::internal::ReaderError::
                INVALID_ARGUMENT,
            "Invalid buffer parameters");
    }
    if (start_bytes >= end_bytes) {
        throw dftracer::utils::utilities::reader::internal::ReaderError(
            dftracer::utils::utilities::reader::internal::ReaderError::
                INVALID_ARGUMENT,
            "start_bytes must be less than end_bytes");
    }
    if (max_bytes != SIZE_MAX) {
        if (end_bytes > max_bytes) {
            throw dftracer::utils::utilities::reader::internal::ReaderError(
                dftracer::utils::utilities::reader::internal::ReaderError::
                    INVALID_ARGUMENT,
                "end_bytes exceeds maximum available bytes");
        }
        if (start_bytes > max_bytes) {
            throw dftracer::utils::utilities::reader::internal::ReaderError(
                dftracer::utils::utilities::reader::internal::ReaderError::
                    INVALID_ARGUMENT,
                "start_bytes exceeds maximum available bytes");
        }
    }
}

static void check_reader_state(bool is_open, const void *indexer) {
    if (!is_open || !indexer) {
        throw std::runtime_error("Reader is not open");
    }
}

static constexpr std::size_t DEFAULT_READER_BUFFER_SIZE = 1 * 1024 * 1024;

namespace dftracer::utils::utilities::reader::internal {

GzipReader::GzipReader(const std::string &gz_path_,
                       const std::string &idx_path_,
                       std::size_t index_ckpt_size)
    : gz_path(gz_path_),
      idx_path(idx_path_),
      is_open(false),
      default_buffer_size(DEFAULT_READER_BUFFER_SIZE),
      indexer(nullptr) {
    try {
        indexer = dftracer::utils::utilities::indexer::internal::
            IndexerFactory::create(gz_path, idx_path, index_ckpt_size, false);
        is_open = true;

        DFTRACER_UTILS_LOG_DEBUG(
            "Successfully created GZIP reader for gz: %s and index: %s",
            gz_path.c_str(), idx_path.c_str());
    } catch (const std::exception &e) {
        throw ReaderError(ReaderError::INITIALIZATION_ERROR,
                          "Failed to initialize reader with indexer: " +
                              std::string(e.what()));
    }
}

GzipReader::GzipReader(
    std::shared_ptr<dftracer::utils::utilities::indexer::internal::Indexer>
        indexer_)
    : default_buffer_size(DEFAULT_READER_BUFFER_SIZE),
      indexer(std::move(indexer_)) {
    if (!indexer) {
        throw ReaderError(ReaderError::INITIALIZATION_ERROR,
                          "Invalid indexer provided");
    }
    is_open = true;
    gz_path = indexer->get_archive_path();
    idx_path = indexer->get_idx_path();
}

GzipReader::~GzipReader() {
    DFTRACER_UTILS_LOG_DEBUG("Destroying GZIP reader for gz: %s and index: %s",
                             gz_path.c_str(), idx_path.c_str());
    reset();
    is_open = false;
}

GzipReader::GzipReader(GzipReader &&other) noexcept
    : gz_path(std::move(other.gz_path)),
      idx_path(std::move(other.idx_path)),
      is_open(other.is_open),
      default_buffer_size(other.default_buffer_size),
      indexer(std::move(other.indexer)) {
    other.is_open = false;
}

GzipReader &GzipReader::operator=(GzipReader &&other) noexcept {
    if (this != &other) {
        gz_path = std::move(other.gz_path);
        idx_path = std::move(other.idx_path);
        is_open = other.is_open;
        default_buffer_size = other.default_buffer_size;
        indexer = std::move(other.indexer);
        other.is_open = false;
    }
    return *this;
}

std::size_t GzipReader::get_max_bytes() const {
    check_reader_state(is_open, indexer.get());
    std::size_t max_bytes =
        static_cast<std::size_t>(indexer.get()->get_max_bytes());
    DFTRACER_UTILS_LOG_DEBUG("Maximum bytes available: %zu", max_bytes);
    return max_bytes;
}

std::size_t GzipReader::get_num_lines() const {
    check_reader_state(is_open, indexer.get());
    std::size_t num_lines = static_cast<std::size_t>(indexer->get_num_lines());
    DFTRACER_UTILS_LOG_DEBUG("Total lines available: %zu", num_lines);
    return num_lines;
}

const std::string &GzipReader::get_archive_path() const { return gz_path; }

const std::string &GzipReader::get_idx_path() const { return idx_path; }

void GzipReader::set_buffer_size(std::size_t size) {
    default_buffer_size = size;
}

void GzipReader::reset() {
    check_reader_state(is_open, indexer.get());
    stream_cache_.clear();
}

coro::CoroTask<std::size_t> GzipReader::read_async(std::size_t start_bytes,
                                                   std::size_t end_bytes,
                                                   char *buffer,
                                                   std::size_t buffer_size) {
    check_reader_state(is_open, indexer.get());
    validate_parameters(buffer, buffer_size, start_bytes, end_bytes,
                        indexer.get()->get_max_bytes());

    DFTRACER_UTILS_LOG_DEBUG(
        "GzipReader::read - request: start_bytes=%zu, end_bytes=%zu, "
        "buffer_size=%zu",
        start_bytes, end_bytes, buffer_size);

    // Check if we can reuse cached stream
    if (!stream_cache_.can_continue(StreamType::BYTES, gz_path, start_bytes,
                                    end_bytes)) {
        DFTRACER_UTILS_LOG_DEBUG("%s",
                                 "GzipReader::read - creating new byte stream");
        auto new_stream = stream(StreamConfig()
                                     .stream_type(StreamType::BYTES)
                                     .range_type(RangeType::BYTE_RANGE)
                                     .from(start_bytes)
                                     .to(end_bytes));
        stream_cache_.update(std::move(new_stream), StreamType::BYTES, gz_path,
                             start_bytes, end_bytes);
    } else {
        DFTRACER_UTILS_LOG_DEBUG(
            "%s", "GzipReader::read - reusing cached byte stream");
    }

    std::size_t result =
        co_await stream_cache_.get()->read_async(buffer, buffer_size);
    DFTRACER_UTILS_LOG_DEBUG("GzipReader::read - returned %zu bytes", result);

    // Update position for next potential read
    stream_cache_.update_position(start_bytes + result);

    co_return result;
}

coro::CoroTask<std::size_t> GzipReader::read_line_bytes_async(
    std::size_t start_bytes, std::size_t end_bytes, char *buffer,
    std::size_t buffer_size) {
    check_reader_state(is_open, indexer.get());

    if (end_bytes > indexer.get()->get_max_bytes()) {
        end_bytes = indexer.get()->get_max_bytes();
    }

    validate_parameters(buffer, buffer_size, start_bytes, end_bytes,
                        indexer.get()->get_max_bytes());

    // Check if we can reuse cached stream
    if (!stream_cache_.can_continue(StreamType::MULTI_LINES_BYTES, gz_path,
                                    start_bytes, end_bytes)) {
        auto new_stream = stream(StreamConfig()
                                     .stream_type(StreamType::MULTI_LINES_BYTES)
                                     .range_type(RangeType::BYTE_RANGE)
                                     .from(start_bytes)
                                     .to(end_bytes));
        stream_cache_.update(std::move(new_stream),
                             StreamType::MULTI_LINES_BYTES, gz_path,
                             start_bytes, end_bytes);
    }

    std::size_t result =
        co_await stream_cache_.get()->read_async(buffer, buffer_size);

    // Update position for next potential read
    stream_cache_.update_position(start_bytes + result);

    co_return result;
}

coro::CoroTask<std::string> GzipReader::read_lines_async(std::size_t start_line,
                                                         std::size_t end_line) {
    check_reader_state(is_open, indexer.get());

    if (start_line == 0 || end_line == 0) {
        throw std::runtime_error("Line numbers must be 1-based (start from 1)");
    }

    if (start_line > end_line) {
        throw std::runtime_error("Start line must be <= end line");
    }

    std::size_t total_lines = indexer.get()->get_num_lines();
    if (start_line > total_lines || end_line > total_lines) {
        throw std::runtime_error("Line numbers exceed total lines in file (" +
                                 std::to_string(total_lines) + ")");
    }

    // Check if we can reuse cached stream
    if (!stream_cache_.can_continue(StreamType::MULTI_LINES, gz_path,
                                    start_line, end_line)) {
        auto new_stream = stream(StreamConfig()
                                     .stream_type(StreamType::MULTI_LINES)
                                     .range_type(RangeType::LINE_RANGE)
                                     .from(start_line)
                                     .to(end_line));
        stream_cache_.update(std::move(new_stream), StreamType::MULTI_LINES,
                             gz_path, start_line, end_line);
    }

    std::string result;
    // Pre-allocate to avoid reallocations (like old StringLineProcessor)
    std::size_t estimated_lines = end_line - start_line + 1;
    result.reserve(estimated_lines * 100);  // Estimate ~100 bytes per line

    std::vector<char> buffer(default_buffer_size);

    while (!stream_cache_.get()->done()) {
        std::size_t bytes_read = co_await stream_cache_.get()->read_async(
            buffer.data(), buffer.size());
        if (bytes_read == 0) break;

        result.append(buffer.data(), bytes_read);
    }

    co_return result;
}

coro::CoroTask<void> GzipReader::read_lines_with_processor_async(
    std::size_t start_line, std::size_t end_line, LineProcessor &processor) {
    check_reader_state(is_open, indexer.get());

    if (start_line == 0 || end_line == 0) {
        throw std::runtime_error("Line numbers must be 1-based (start from 1)");
    }

    if (start_line > end_line) {
        throw std::runtime_error("Start line must be <= end line");
    }

    std::size_t total_lines = indexer.get()->get_num_lines();
    if (start_line > total_lines || end_line > total_lines) {
        throw std::runtime_error("Line numbers exceed total lines in file (" +
                                 std::to_string(total_lines) + ")");
    }

    processor.begin(start_line, end_line);

    // Create a LineStream that returns one line at a time
    auto line_stream = stream(StreamConfig()
                                  .stream_type(StreamType::LINE)
                                  .range_type(RangeType::LINE_RANGE)
                                  .from(start_line)
                                  .to(end_line));

    std::vector<char> buffer(default_buffer_size);

    while (!line_stream->done()) {
        std::size_t bytes_read =
            co_await line_stream->read_async(buffer.data(), buffer.size());
        if (bytes_read == 0) break;

        // LineStream returns one complete line with \n
        // Processor expects line without \n
        std::size_t line_length = bytes_read;
        if (line_length > 0 && buffer[line_length - 1] == '\n') {
            line_length--;
        }

        if (!co_await processor.process(buffer.data(), line_length)) {
            processor.end();
            co_return;
        }
    }

    processor.end();
}

coro::CoroTask<void> GzipReader::read_line_bytes_with_processor_async(
    std::size_t start_bytes, std::size_t end_bytes, LineProcessor &processor) {
    check_reader_state(is_open, indexer.get());

    if (end_bytes > indexer.get()->get_max_bytes()) {
        end_bytes = indexer.get()->get_max_bytes();
    }

    if (start_bytes >= end_bytes) {
        co_return;
    }

    processor.begin(start_bytes, end_bytes);

    auto lines_stream = stream(StreamConfig()
                                   .stream_type(StreamType::LINE_BYTES)
                                   .range_type(RangeType::BYTE_RANGE)
                                   .from(start_bytes)
                                   .to(end_bytes));

    std::vector<char> buffer(default_buffer_size);

    while (!lines_stream->done()) {
        std::size_t bytes_read =
            co_await lines_stream->read_async(buffer.data(), buffer.size());
        if (bytes_read == 0) break;
        co_await processor.process(buffer.data(), bytes_read);
    }

    processor.end();
}

bool GzipReader::is_valid() const { return is_open && indexer.get(); }

std::string GzipReader::get_format_name() const { return "GZIP"; }

std::unique_ptr<ReaderStream> GzipReader::stream(const StreamConfig &config) {
    check_reader_state(is_open, indexer.get());

    // Extract config parameters
    StreamType stream_type = config.stream_type();
    RangeType range_type = config.range_type();
    std::size_t start = config.start();
    std::size_t end = config.end();
    std::size_t buffer_size = config.buffer_size();

    // Convert line range to byte range if needed
    std::size_t start_bytes = start;
    std::size_t end_bytes = end;
    std::size_t actual_start_line =
        1;  // Track what line number start_bytes corresponds to

    if (range_type == RangeType::LINE_RANGE) {
        // Convert line numbers to byte offsets using checkpoints
        if (start == 0 || end == 0) {
            throw ReaderError(ReaderError::INVALID_ARGUMENT,
                              "Line numbers must be 1-based (start from 1)");
        }
        if (start > end) {
            throw ReaderError(ReaderError::INVALID_ARGUMENT,
                              "Start line must be <= end line");
        }

        std::size_t total_lines = indexer->get_num_lines();
        if (start > total_lines || end > total_lines) {
            throw ReaderError(ReaderError::INVALID_ARGUMENT,
                              "Line numbers exceed total lines in file (" +
                                  std::to_string(total_lines) + ")");
        }

        // Get checkpoints for the line range
        std::vector<
            dftracer::utils::utilities::indexer::internal::IndexerCheckpoint>
            checkpoints = indexer->get_checkpoints_for_line_range(start, end);

        DFTRACER_UTILS_LOG_DEBUG("Line range %zu-%zu: found %zu checkpoints",
                                 start, end, checkpoints.size());

        if (checkpoints.empty()) {
            // No checkpoints, read from beginning
            start_bytes = 0;
            end_bytes = indexer->get_max_bytes();
            actual_start_line = 1;
            DFTRACER_UTILS_LOG_DEBUG(
                "No checkpoints found, using full file: start_bytes=%zu, "
                "end_bytes=%zu, max_bytes=%zu",
                start_bytes, end_bytes, indexer->get_max_bytes());
        } else {
            // Use checkpoint to determine byte range.
            //
            // Checkpoint uc_offset values fall at deflate block boundaries
            // which may land in the middle of a text line.  When we start
            // decompressing from such a mid-line position the first "line"
            // seen by MultiLineStream is a partial fragment.  If
            // actual_start_line == start_line the fragment is emitted as
            // the requested first line, producing wrong content.
            //
            // To avoid this we choose a checkpoint whose last_line_num is
            // strictly less than (start - 1), guaranteeing
            // actual_start_line < start.  MultiLineStream then filters
            // out the (potentially partial) early lines before reaching
            // the requested range.
            auto all_checkpoints = indexer->get_checkpoints();
            bool found_start = false;

            // Walk checkpoints from the end to find the latest one whose
            // line range ends before (start - 1).
            for (auto it = all_checkpoints.rbegin();
                 it != all_checkpoints.rend(); ++it) {
                if (it->last_line_num < start - 1) {
                    start_bytes = it->uc_offset;
                    actual_start_line = it->last_line_num + 1;
                    found_start = true;
                    break;
                }
            }

            if (!found_start) {
                // No suitable checkpoint found -- start from beginning
                start_bytes = 0;
                actual_start_line = 1;
            }

            const auto &last_checkpoint = checkpoints.back();
            end_bytes = last_checkpoint.uc_offset + last_checkpoint.uc_size;

            DFTRACER_UTILS_LOG_DEBUG(
                "Using checkpoints: matched_first_idx=%zu "
                "(first_line=%zu, last_line=%zu), "
                "end_checkpoint_idx=%zu (first_line=%zu, last_line=%zu), "
                "byte_range=%zu-%zu, actual_start_line=%zu",
                checkpoints[0].checkpoint_idx, checkpoints[0].first_line_num,
                checkpoints[0].last_line_num, last_checkpoint.checkpoint_idx,
                last_checkpoint.first_line_num, last_checkpoint.last_line_num,
                start_bytes, end_bytes, actual_start_line);
        }
    }

    // Create appropriate stream type
    switch (stream_type) {
        case StreamType::BYTES: {
            auto byte_stream = std::make_unique<GzipByteStream>(buffer_size);
            byte_stream->initialize(gz_path, start_bytes, end_bytes, *indexer);
            return byte_stream;
        }
        case StreamType::LINE_BYTES: {
            // Single line-aligned bytes at a time
            auto line_byte_stream =
                std::make_unique<GzipLineByteStream>(buffer_size);
            line_byte_stream->initialize(gz_path, start_bytes, end_bytes,
                                         *indexer);

            // Wrap with LineStream to return one line-aligned chunk at a time
            if (range_type == RangeType::LINE_RANGE) {
                return std::make_unique<LineStream>(
                    std::move(line_byte_stream), start, end, actual_start_line);
            } else {
                return std::make_unique<LineStream>(
                    std::move(line_byte_stream));
            }
        }
        case StreamType::MULTI_LINES_BYTES: {
            // Multiple line-aligned bytes per read
            auto line_byte_stream =
                std::make_unique<GzipLineByteStream>(buffer_size);
            line_byte_stream->initialize(gz_path, start_bytes, end_bytes,
                                         *indexer);
            return line_byte_stream;
        }
        case StreamType::LINE: {
            // Single parsed line per read
            auto line_byte_stream =
                std::make_unique<GzipLineByteStream>(buffer_size);
            line_byte_stream->initialize(gz_path, start_bytes, end_bytes,
                                         *indexer);

            if (range_type == RangeType::LINE_RANGE) {
                return std::make_unique<LineStream>(
                    std::move(line_byte_stream), start, end, actual_start_line);
            } else {
                return std::make_unique<LineStream>(
                    std::move(line_byte_stream));
            }
        }
        case StreamType::MULTI_LINES: {
            // Multiple parsed lines per read
            auto line_byte_stream =
                std::make_unique<GzipLineByteStream>(buffer_size);
            line_byte_stream->initialize(gz_path, start_bytes, end_bytes,
                                         *indexer);

            if (range_type == RangeType::LINE_RANGE) {
                return std::make_unique<MultiLineStream>(
                    std::move(line_byte_stream), start, end, actual_start_line);
            } else {
                return std::make_unique<MultiLineStream>(
                    std::move(line_byte_stream));
            }
        }
        default:
            throw ReaderError(ReaderError::INVALID_ARGUMENT,
                              "Invalid stream type");
    }
}

}  // namespace dftracer::utils::utilities::reader::internal

1	#include <dftracer/utils/core/coro/task.h>
2	#include <dftracer/utils/core/utils/timer.h>
3	#include <dftracer/utils/utilities/indexer/internal/indexer.h>
4	#include <dftracer/utils/utilities/indexer/internal/indexer_factory.h>
5	#include <dftracer/utils/utilities/reader/internal/error.h>
6	#include <dftracer/utils/utilities/reader/internal/gzip_reader.h>
7	#include <dftracer/utils/utilities/reader/internal/stream_config.h>
8	#include <dftracer/utils/utilities/reader/internal/streams/gzip_byte_stream.h>
9	#include <dftracer/utils/utilities/reader/internal/streams/gzip_line_byte_stream.h>
10	#include <dftracer/utils/utilities/reader/internal/streams/line_stream.h>
11	#include <dftracer/utils/utilities/reader/internal/streams/multi_line_stream.h>
12	#include <dftracer/utils/utilities/reader/internal/string_line_processor.h>
13
14	#include <cstdio>
15	#include <cstring>
16	#include <limits>
17	#include <string_view>
18
19	static void validate_parameters(	154,656✔
20	const char *buffer, std::size_t buffer_size, std::size_t start_bytes,
21	std::size_t end_bytes,
22	std::size_t max_bytes = std::numeric_limits<std::size_t>::max()) {
23	if (!buffer \|\| buffer_size == 0) {	154,656!
24	throw dftracer::utils::utilities::reader::internal::ReaderError(	12!
25	dftracer::utils::utilities::reader::internal::ReaderError::
26	INVALID_ARGUMENT,
27	"Invalid buffer parameters");	×
28	}
29	if (start_bytes >= end_bytes) {	154,656✔
30	throw dftracer::utils::utilities::reader::internal::ReaderError(	20!
31	dftracer::utils::utilities::reader::internal::ReaderError::
32	INVALID_ARGUMENT,
33	"start_bytes must be less than end_bytes");	30!
34	}
35	if (max_bytes != SIZE_MAX) {	154,636✔
36	if (end_bytes > max_bytes) {	154,636✔
37	throw dftracer::utils::utilities::reader::internal::ReaderError(	4!
38	dftracer::utils::utilities::reader::internal::ReaderError::
39	INVALID_ARGUMENT,
40	"end_bytes exceeds maximum available bytes");	6!
41	}
42	if (start_bytes > max_bytes) {	154,632✔
43	throw dftracer::utils::utilities::reader::internal::ReaderError(	×
44	dftracer::utils::utilities::reader::internal::ReaderError::
45	INVALID_ARGUMENT,
46	"start_bytes exceeds maximum available bytes");	×
47	}
48	}	77,315✔
49	}	154,644✔
50
51	static void check_reader_state(bool is_open, const void *indexer) {	160,183✔
52	if (!is_open \|\| !indexer) {	160,183!
UNCOV 53	throw std::runtime_error("Reader is not open");	×
54	}
55	}	160,183✔
56
57	static constexpr std::size_t DEFAULT_READER_BUFFER_SIZE = 1 * 1024 * 1024;
58
59	namespace dftracer::utils::utilities::reader::internal {
60
61	GzipReader::GzipReader(const std::string &gz_path_,	1,843!
62	const std::string &idx_path_,
63	std::size_t index_ckpt_size)	459✔
64	: gz_path(gz_path_),	922!
65	idx_path(idx_path_),	922!
66	is_open(false),	923✔
67	default_buffer_size(DEFAULT_READER_BUFFER_SIZE),	923✔
68	indexer(nullptr) {	2,305!
69	try {
70	indexer = dftracer::utils::utilities::indexer::internal::	461✔
71	IndexerFactory::create(gz_path, idx_path, index_ckpt_size, false);	921✔
72	is_open = true;	910✔
73
74	DFTRACER_UTILS_LOG_DEBUG(
75	"Successfully created GZIP reader for gz: %s and index: %s",
76	gz_path.c_str(), idx_path.c_str());
77	} catch (const std::exception &e) {	467!
78	throw ReaderError(ReaderError::INITIALIZATION_ERROR,	18!
79	"Failed to initialize reader with indexer: " +	18!
80	std::string(e.what()));	18!
81	}	12!
82	}	1,420✔
83
84	GzipReader::GzipReader(	660!
85	std::shared_ptr<dftracer::utils::utilities::indexer::internal::Indexer>
86	indexer_)	132✔
87	: default_buffer_size(DEFAULT_READER_BUFFER_SIZE),	264✔
88	indexer(std::move(indexer_)) {	528✔
89	if (!indexer) {	264✔
90	throw ReaderError(ReaderError::INITIALIZATION_ERROR,	×
91	"Invalid indexer provided");	×
92	}
93	is_open = true;	264✔
94	gz_path = indexer->get_archive_path();	264!
95	idx_path = indexer->get_idx_path();	264!
96	}	396✔
97
98	GzipReader::~GzipReader() {	1,764✔
99	DFTRACER_UTILS_LOG_DEBUG("Destroying GZIP reader for gz: %s and index: %s",
100	gz_path.c_str(), idx_path.c_str());
101	reset();	1,176!
102	is_open = false;	1,176✔
103	}	1,764✔
104
105	GzipReader::GzipReader(GzipReader &&other) noexcept	×
106	: gz_path(std::move(other.gz_path)),	×
107	idx_path(std::move(other.idx_path)),	×
108	is_open(other.is_open),	×
109	default_buffer_size(other.default_buffer_size),	×
110	indexer(std::move(other.indexer)) {	×
111	other.is_open = false;	×
112	}	×
113
114	GzipReader &GzipReader::operator=(GzipReader &&other) noexcept {	×
115	if (this != &other) {	×
116	gz_path = std::move(other.gz_path);	×
117	idx_path = std::move(other.idx_path);	×
118	is_open = other.is_open;	×
119	default_buffer_size = other.default_buffer_size;	×
120	indexer = std::move(other.indexer);	×
121	other.is_open = false;	×
122	}
123	return *this;	×
124	}
125
126	std::size_t GzipReader::get_max_bytes() const {	376✔
127	check_reader_state(is_open, indexer.get());	376✔
128	std::size_t max_bytes =	188✔
129	static_cast<std::size_t>(indexer.get()->get_max_bytes());	376✔
130	DFTRACER_UTILS_LOG_DEBUG("Maximum bytes available: %zu", max_bytes);
131	return max_bytes;	376✔
132	}
133
134	std::size_t GzipReader::get_num_lines() const {	106✔
135	check_reader_state(is_open, indexer.get());	106✔
136	std::size_t num_lines = static_cast<std::size_t>(indexer->get_num_lines());	106✔
137	DFTRACER_UTILS_LOG_DEBUG("Total lines available: %zu", num_lines);
138	return num_lines;	106✔
139	}
140
141	const std::string &GzipReader::get_archive_path() const { return gz_path; }	14✔
142
143	const std::string &GzipReader::get_idx_path() const { return idx_path; }	8✔
144
145	void GzipReader::set_buffer_size(std::size_t size) {	×
146	default_buffer_size = size;	×
147	}	×
148
149	void GzipReader::reset() {	1,184✔
150	check_reader_state(is_open, indexer.get());	1,184✔
151	stream_cache_.clear();	1,184✔
152	}	1,184✔
153
154	coro::CoroTask<std::size_t> GzipReader::read_async(std::size_t start_bytes,	74,892!
155	std::size_t end_bytes,
156	char *buffer,
157	std::size_t buffer_size) {	12,482!
158	check_reader_state(is_open, indexer.get());	12,482!
159	validate_parameters(buffer, buffer_size, start_bytes, end_bytes,	12,482✔
160	indexer.get()->get_max_bytes());	12,482!
161
162	DFTRACER_UTILS_LOG_DEBUG(
163	"GzipReader::read - request: start_bytes=%zu, end_bytes=%zu, "
164	"buffer_size=%zu",
165	start_bytes, end_bytes, buffer_size);
166
167	// Check if we can reuse cached stream
168	if (!stream_cache_.can_continue(StreamType::BYTES, gz_path, start_bytes,	24,940!
169	end_bytes)) {	12,470✔
170	DFTRACER_UTILS_LOG_DEBUG("%s",
171	"GzipReader::read - creating new byte stream");
172	auto new_stream = stream(StreamConfig()	322!
173	.stream_type(StreamType::BYTES)	161!
174	.range_type(RangeType::BYTE_RANGE)	161!
175	.from(start_bytes)	161!
176	.to(end_bytes));	161!
177	stream_cache_.update(std::move(new_stream), StreamType::BYTES, gz_path,	322!
178	start_bytes, end_bytes);	161✔
179	} else {	161✔
180	DFTRACER_UTILS_LOG_DEBUG(
181	"%s", "GzipReader::read - reusing cached byte stream");
182	}
183
184	std::size_t result =	24,940✔
185	co_await stream_cache_.get()->read_async(buffer, buffer_size);	37,422!
186	DFTRACER_UTILS_LOG_DEBUG("GzipReader::read - returned %zu bytes", result);
187
188	// Update position for next potential read
189	stream_cache_.update_position(start_bytes + result);	12,470!
190
191	co_return result;	12,470!
192	}	62,386!
193
194	coro::CoroTask<std::size_t> GzipReader::read_line_bytes_async(	518,762!
195	std::size_t start_bytes, std::size_t end_bytes, char *buffer,
196	std::size_t buffer_size) {	64,845!
197	check_reader_state(is_open, indexer.get());	194,535✔
198
199	if (end_bytes > indexer.get()->get_max_bytes()) {	64,845!
200	end_bytes = indexer.get()->get_max_bytes();	×
201	}
202
203	validate_parameters(buffer, buffer_size, start_bytes, end_bytes,	64,845!
204	indexer.get()->get_max_bytes());	64,845!
205
206	// Check if we can reuse cached stream
207	if (!stream_cache_.can_continue(StreamType::MULTI_LINES_BYTES, gz_path,	129,690✔
208	start_bytes, end_bytes)) {	64,845✔
209	auto new_stream = stream(StreamConfig()	2,032!
210	.stream_type(StreamType::MULTI_LINES_BYTES)	1,016!
211	.range_type(RangeType::BYTE_RANGE)	1,016!
212	.from(start_bytes)	1,016!
213	.to(end_bytes));	1,016!
214	stream_cache_.update(std::move(new_stream),	2,032!
215	StreamType::MULTI_LINES_BYTES, gz_path,	1,016✔
216	start_bytes, end_bytes);	1,016✔
217	}	1,016✔
218
219	std::size_t result =	259,380✔
220	co_await stream_cache_.get()->read_async(buffer, buffer_size);	324,225!
221
222	// Update position for next potential read
223	stream_cache_.update_position(start_bytes + result);	64,845!
224
225	co_return result;	64,845!
226	}	842,989!
227
228	coro::CoroTask<std::string> GzipReader::read_lines_async(std::size_t start_line,	2,006!
229	std::size_t end_line) {	163!
230	check_reader_state(is_open, indexer.get());	163!
231
232	if (start_line == 0 \|\| end_line == 0) {	163✔
233	throw std::runtime_error("Line numbers must be 1-based (start from 1)");	6!
234	}
235
236	if (start_line > end_line) {	157✔
237	throw std::runtime_error("Start line must be <= end line");	3!
238	}
239
240	std::size_t total_lines = indexer.get()->get_num_lines();	154!
241	if (start_line > total_lines \|\| end_line > total_lines) {	154✔
242	throw std::runtime_error("Line numbers exceed total lines in file (" +	4!
243	std::to_string(total_lines) + ")");	4!
244	}
245
246	// Check if we can reuse cached stream
247	if (!stream_cache_.can_continue(StreamType::MULTI_LINES, gz_path,	304!
248	start_line, end_line)) {	152✔
249	auto new_stream = stream(StreamConfig()	304!
250	.stream_type(StreamType::MULTI_LINES)	152!
251	.range_type(RangeType::LINE_RANGE)	152!
252	.from(start_line)	152!
253	.to(end_line));	152!
254	stream_cache_.update(std::move(new_stream), StreamType::MULTI_LINES,	304!
255	gz_path, start_line, end_line);	152✔
256	}	152✔
257
258	std::string result;	152✔
259	// Pre-allocate to avoid reallocations (like old StringLineProcessor)
260	std::size_t estimated_lines = end_line - start_line + 1;	152✔
261	result.reserve(estimated_lines * 100); // Estimate ~100 bytes per line	152!
262
263	std::vector<char> buffer(default_buffer_size);	152!
264
265	while (!stream_cache_.get()->done()) {	561!
266	std::size_t bytes_read = co_await stream_cache_.get()->read_async(	1,799!
267	buffer.data(), buffer.size());	409✔
268	if (bytes_read == 0) break;	409!
269
270	result.append(buffer.data(), bytes_read);	409!
271	}	409✔
272
273	co_return result;	1,180!
274	}	3,459!
275
276	coro::CoroTask<void> GzipReader::read_lines_with_processor_async(	×
277	std::size_t start_line, std::size_t end_line, LineProcessor &processor) {	×
278	check_reader_state(is_open, indexer.get());	×
279
280	if (start_line == 0 \|\| end_line == 0) {	×
281	throw std::runtime_error("Line numbers must be 1-based (start from 1)");	×
282	}
283
284	if (start_line > end_line) {	×
285	throw std::runtime_error("Start line must be <= end line");	×
286	}
287
288	std::size_t total_lines = indexer.get()->get_num_lines();	×
289	if (start_line > total_lines \|\| end_line > total_lines) {	×
290	throw std::runtime_error("Line numbers exceed total lines in file (" +	×
291	std::to_string(total_lines) + ")");	×
292	}
293
294	processor.begin(start_line, end_line);	×
295
296	// Create a LineStream that returns one line at a time
297	auto line_stream = stream(StreamConfig()	×
298	.stream_type(StreamType::LINE)	×
299	.range_type(RangeType::LINE_RANGE)	×
300	.from(start_line)	×
301	.to(end_line));	×
302
303	std::vector<char> buffer(default_buffer_size);	×
304
305	while (!line_stream->done()) {	×
306	std::size_t bytes_read =
307	co_await line_stream->read_async(buffer.data(), buffer.size());	×
308	if (bytes_read == 0) break;	×
309
310	// LineStream returns one complete line with \n
311	// Processor expects line without \n
312	std::size_t line_length = bytes_read;
313	if (line_length > 0 && buffer[line_length - 1] == '\n') {	×
314	line_length--;
315	}
316
317	if (!co_await processor.process(buffer.data(), line_length)) {	×
318	processor.end();	×
319	co_return;
320	}
321	}	×
322
323	processor.end();	×
324	}	×
325
326	coro::CoroTask<void> GzipReader::read_line_bytes_with_processor_async(	×
327	std::size_t start_bytes, std::size_t end_bytes, LineProcessor &processor) {	×
328	check_reader_state(is_open, indexer.get());	×
329
330	if (end_bytes > indexer.get()->get_max_bytes()) {	×
331	end_bytes = indexer.get()->get_max_bytes();	×
332	}
333
334	if (start_bytes >= end_bytes) {	×
335	co_return;
336	}
337
338	processor.begin(start_bytes, end_bytes);	×
339
340	auto lines_stream = stream(StreamConfig()	×
341	.stream_type(StreamType::LINE_BYTES)	×
342	.range_type(RangeType::BYTE_RANGE)	×
343	.from(start_bytes)	×
344	.to(end_bytes));	×
345
346	std::vector<char> buffer(default_buffer_size);	×
347
348	while (!lines_stream->done()) {	×
349	std::size_t bytes_read =
350	co_await lines_stream->read_async(buffer.data(), buffer.size());	×
351	if (bytes_read == 0) break;	×
352	co_await processor.process(buffer.data(), bytes_read);	×
353	}	×
354
355	processor.end();	×
356	}	×
357
358	bool GzipReader::is_valid() const { return is_open && indexer.get(); }	74!
359
360	std::string GzipReader::get_format_name() const { return "GZIP"; }	2!
361
362	std::unique_ptr<ReaderStream> GzipReader::stream(const StreamConfig &config) {	3,538✔
363	check_reader_state(is_open, indexer.get());	3,538!
364
365	// Extract config parameters
366	StreamType stream_type = config.stream_type();	3,538✔
367	RangeType range_type = config.range_type();	3,538✔
368	std::size_t start = config.start();	3,538✔
369	std::size_t end = config.end();	3,538✔
370	std::size_t buffer_size = config.buffer_size();	3,537✔
371
372	// Convert line range to byte range if needed
373	std::size_t start_bytes = start;	3,538✔
374	std::size_t end_bytes = end;	3,538✔
375	std::size_t actual_start_line =	3,538✔
376	1; // Track what line number start_bytes corresponds to
377
378	if (range_type == RangeType::LINE_RANGE) {	3,538✔
379	// Convert line numbers to byte offsets using checkpoints
380	if (start == 0 \|\| end == 0) {	452!
381	throw ReaderError(ReaderError::INVALID_ARGUMENT,	×
382	"Line numbers must be 1-based (start from 1)");	×
383	}
384	if (start > end) {	452✔
385	throw ReaderError(ReaderError::INVALID_ARGUMENT,	×
386	"Start line must be <= end line");	×
387	}
388
389	std::size_t total_lines = indexer->get_num_lines();	452!
390	if (start > total_lines \|\| end > total_lines) {	452!
391	throw ReaderError(ReaderError::INVALID_ARGUMENT,	×
392	"Line numbers exceed total lines in file (" +	×
393	std::to_string(total_lines) + ")");	×
394	}
395
396	// Get checkpoints for the line range
397	std::vector<
398	dftracer::utils::utilities::indexer::internal::IndexerCheckpoint>
399	checkpoints = indexer->get_checkpoints_for_line_range(start, end);	452!
400
401	DFTRACER_UTILS_LOG_DEBUG("Line range %zu-%zu: found %zu checkpoints",
402	start, end, checkpoints.size());
403
404	if (checkpoints.empty()) {	452✔
405	// No checkpoints, read from beginning
406	start_bytes = 0;	180✔
407	end_bytes = indexer->get_max_bytes();	180!
408	actual_start_line = 1;	180✔
409	DFTRACER_UTILS_LOG_DEBUG(
410	"No checkpoints found, using full file: start_bytes=%zu, "
411	"end_bytes=%zu, max_bytes=%zu",
412	start_bytes, end_bytes, indexer->get_max_bytes());
413	} else {	90✔
414	// Use checkpoint to determine byte range.
415	//
416	// Checkpoint uc_offset values fall at deflate block boundaries
417	// which may land in the middle of a text line. When we start
418	// decompressing from such a mid-line position the first "line"
419	// seen by MultiLineStream is a partial fragment. If
420	// actual_start_line == start_line the fragment is emitted as
421	// the requested first line, producing wrong content.
422	//
423	// To avoid this we choose a checkpoint whose last_line_num is
424	// strictly less than (start - 1), guaranteeing
425	// actual_start_line < start. MultiLineStream then filters
426	// out the (potentially partial) early lines before reaching
427	// the requested range.
428	auto all_checkpoints = indexer->get_checkpoints();	272!
429	bool found_start = false;	272✔
430
431	// Walk checkpoints from the end to find the latest one whose
432	// line range ends before (start - 1).
433	for (auto it = all_checkpoints.rbegin();	13,668✔
434	it != all_checkpoints.rend(); ++it) {	19,703!
435	if (it->last_line_num < start - 1) {	12,940!
436	start_bytes = it->uc_offset;	139!
437	actual_start_line = it->last_line_num + 1;	139!
438	found_start = true;	139✔
439	break;	139✔
440	}
441	}	6,630✔
442
443	if (!found_start) {	272✔
444	// No suitable checkpoint found -- start from beginning
445	start_bytes = 0;	133✔
446	actual_start_line = 1;	133✔
447	}	67✔
448
449	const auto &last_checkpoint = checkpoints.back();	272✔
450	end_bytes = last_checkpoint.uc_offset + last_checkpoint.uc_size;	272✔
451
452	DFTRACER_UTILS_LOG_DEBUG(
453	"Using checkpoints: matched_first_idx=%zu "
454	"(first_line=%zu, last_line=%zu), "
455	"end_checkpoint_idx=%zu (first_line=%zu, last_line=%zu), "
456	"byte_range=%zu-%zu, actual_start_line=%zu",
457	checkpoints[0].checkpoint_idx, checkpoints[0].first_line_num,
458	checkpoints[0].last_line_num, last_checkpoint.checkpoint_idx,
459	last_checkpoint.first_line_num, last_checkpoint.last_line_num,
460	start_bytes, end_bytes, actual_start_line);
461	}	272✔
462	}	452✔
463
464	// Create appropriate stream type
465	switch (stream_type) {	3,537!
466	case StreamType::BYTES: {	193✔
467	auto byte_stream = std::make_unique<GzipByteStream>(buffer_size);	386!
468	byte_stream->initialize(gz_path, start_bytes, end_bytes, *indexer);	386!
469	return byte_stream;	386✔
470	}	386✔
471	case StreamType::LINE_BYTES: {	155✔
472	// Single line-aligned bytes at a time
473	auto line_byte_stream =
474	std::make_unique<GzipLineByteStream>(buffer_size);	310!
475	line_byte_stream->initialize(gz_path, start_bytes, end_bytes,	619!
476	*indexer);	310✔
477
478	// Wrap with LineStream to return one line-aligned chunk at a time
479	if (range_type == RangeType::LINE_RANGE) {	310✔
480	return std::make_unique<LineStream>(	6!
481	std::move(line_byte_stream), start, end, actual_start_line);	4✔
482	} else {
483	return std::make_unique<LineStream>(	459!
484	std::move(line_byte_stream));	306✔
485	}
486	}	310✔
487	case StreamType::MULTI_LINES_BYTES: {	1,128✔
488	// Multiple line-aligned bytes per read
489	auto line_byte_stream =
490	std::make_unique<GzipLineByteStream>(buffer_size);	2,257!
491	line_byte_stream->initialize(gz_path, start_bytes, end_bytes,	4,515!
492	*indexer);	2,258✔
493	return line_byte_stream;	2,258✔
494	}	2,257✔
495	case StreamType::LINE: {	54✔
496	// Single parsed line per read
497	auto line_byte_stream =
498	std::make_unique<GzipLineByteStream>(buffer_size);	108!
499	line_byte_stream->initialize(gz_path, start_bytes, end_bytes,	216!
500	*indexer);	108✔
501
502	if (range_type == RangeType::LINE_RANGE) {	108✔
503	return std::make_unique<LineStream>(	159!
504	std::move(line_byte_stream), start, end, actual_start_line);	106✔
505	} else {
506	return std::make_unique<LineStream>(	3!
507	std::move(line_byte_stream));	2✔
508	}
509	}	108✔
510	case StreamType::MULTI_LINES: {	238✔
511	// Multiple parsed lines per read
512	auto line_byte_stream =
513	std::make_unique<GzipLineByteStream>(buffer_size);	476!
514	line_byte_stream->initialize(gz_path, start_bytes, end_bytes,	952!
515	*indexer);	476✔
516
517	if (range_type == RangeType::LINE_RANGE) {	476✔
518	return std::make_unique<MultiLineStream>(	507!
519	std::move(line_byte_stream), start, end, actual_start_line);	337✔
520	} else {
521	return std::make_unique<MultiLineStream>(	207!
522	std::move(line_byte_stream));	138✔
523	}
524	}	476✔
525	default:
526	throw ReaderError(ReaderError::INVALID_ARGUMENT,	×
527	"Invalid stream type");	×
528	}
529	}	1,769✔
530
531	} // namespace dftracer::utils::utilities::reader::internal

llnl / dftracer-utils / 23935950925

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