Merge branch 'smalton/date-lib-enscapsulation' into 'master'

Encapsulate `date` dependency See merge request machine-learning/dorado!673
nanoporetech · Nov 2, 2023 · 52cbabf · 52cbabf
2 parents ac2a90a + 5857d94
commit 52cbabf
Show file tree

Hide file tree

Showing 5 changed files with 100 additions and 84 deletions.
diff --git a/CMakeLists.txt b/CMakeLists.txt
@@ -295,7 +295,6 @@ target_link_libraries(dorado_lib
         htslib
         vbz_hdf_plugin
         edlib
-        date::date
         dorado_utils
     PRIVATE
         ${KOI_LIBRARIES}    
@@ -465,7 +464,6 @@ if(NOT CMAKE_SYSTEM_NAME STREQUAL "iOS")
             ${POD5_LIBRARIES}
             ${HDF5_C_LIBRARIES}
             ${CMAKE_DL_LIBS}
-            date::date
     )
 
     if(WIN32)

diff --git a/dorado/cli/summary.cpp b/dorado/cli/summary.cpp
@@ -2,10 +2,9 @@
 #include "read_pipeline/HtsReader.h"
 #include "utils/bam_utils.h"
 #include "utils/log_utils.h"
+#include "utils/time_utils.h"
 
 #include <argparse.hpp>
-#include <date/date.h>
-#include <date/tz.h>
 #include <spdlog/spdlog.h>
 
 #include <cctype>
@@ -16,34 +15,6 @@ namespace dorado {
 
 volatile sig_atomic_t interrupt = 0;
 
-// todo: move to time_utils after !273
-double time_difference_seconds(const std::string &timestamp1, const std::string &timestamp2) {
-    using namespace date;
-    using namespace std::chrono;
-    try {
-        std::istringstream ss1(timestamp1);
-        std::istringstream ss2(timestamp2);
-        sys_time<microseconds> time1, time2;
-        ss1 >> parse("%FT%T%Ez", time1);
-        ss2 >> parse("%FT%T%Ez", time2);
-        // If parsing with timezone offset failed, try parsing with 'Z' format
-        if (ss1.fail()) {
-            ss1.clear();
-            ss1.str(timestamp1);
-            ss1 >> parse("%FT%TZ", time1);
-        }
-        if (ss2.fail()) {
-            ss2.clear();
-            ss2.str(timestamp2);
-            ss2 >> parse("%FT%TZ", time2);
-        }
-        duration<double> diff = time1 - time2;
-        return diff.count();
-    } catch (const std::exception &e) {
-        throw std::runtime_error("Failed to parse timestamps");
-    }
-}
-
 int summary(int argc, char *argv[]) {
     utils::InitLogging();
 
@@ -154,7 +125,7 @@ int summary(int argc, char *argv[]) {
         float sample_rate = num_samples / duration;
         float template_duration = (num_samples - trim_samples) / sample_rate;
         auto exp_start_dt = read_group_exp_start_time.at(rg_value);
-        auto start_time = time_difference_seconds(start_time_dt, exp_start_dt);
+        auto start_time = utils::time_difference_seconds(start_time_dt, exp_start_dt);
         auto template_start_time = start_time + (duration - template_duration);
 
         std::cout << filename << separator << read_id << separator << run_id << separator << channel

diff --git a/dorado/utils/CMakeLists.txt b/dorado/utils/CMakeLists.txt
@@ -33,6 +33,7 @@ add_library(dorado_utils
     stats.h
     tensor_utils.cpp
     tensor_utils.h
+    time_utils.cpp
     time_utils.h
     torch_utils.h
     trim.cpp
@@ -85,11 +86,11 @@ target_link_libraries(dorado_utils
     PUBLIC
         ${TORCH_LIBRARIES}
         edlib
-        date::date
         spdlog::spdlog
     PRIVATE
         minimap2
         OpenSSL::SSL
+        date::date
         htslib
 )
 

diff --git a/dorado/utils/time_utils.cpp b/dorado/utils/time_utils.cpp
@@ -0,0 +1,91 @@
+#include "time_utils.h"
+
+#include <date/date.h>
+#include <date/tz.h>
+
+#include <chrono>
+#include <sstream>
+
+namespace dorado::utils {
+
+std::string get_string_timestamp_from_unix_time(time_t time_stamp_ms) {
+    auto tp = std::chrono::system_clock::from_time_t(time_stamp_ms / 1000);
+    tp += std::chrono::milliseconds(time_stamp_ms % 1000);
+    auto dp = date::floor<date::days>(tp);
+    auto time = date::make_time(std::chrono::duration_cast<std::chrono::milliseconds>(tp - dp));
+    auto ymd = date::year_month_day{dp};
+
+    std::ostringstream date_time_ss;
+    date_time_ss << ymd << "T" << time << "+00:00";
+    return date_time_ss.str();
+}
+
+// Expects the time to be encoded like "2017-09-12T09:50:12.456+00:00" or "2017-09-12T09:50:12Z".
+// Time stamp can be specified up to microseconds
+time_t get_unix_time_from_string_timestamp(const std::string & time_stamp) {
+    std::istringstream ss(time_stamp);
+    date::sys_time<std::chrono::microseconds> time_us;
+    ss >> date::parse("%FT%T%Ez", time_us);
+    // If parsing with timezone offset failed, try parsing with 'Z' format
+    if (ss.fail()) {
+        ss.clear();
+        ss.str(time_stamp);
+        ss >> date::parse("%FT%TZ", time_us);
+    }
+
+    auto epoch = time_us.time_since_epoch();
+    auto value = std::chrono::duration_cast<std::chrono::milliseconds>(epoch);
+    return value.count();
+}
+
+std::string adjust_time_ms(const std::string & time_stamp, uint64_t offset_ms) {
+    return get_string_timestamp_from_unix_time(get_unix_time_from_string_timestamp(time_stamp) +
+                                               offset_ms);
+}
+
+std::string adjust_time(const std::string & time_stamp, uint32_t offset) {
+    // Expects the time to be encoded like "2017-09-12T9:50:12Z".
+    // Adds the offset (in seconds) to the timeStamp.
+
+    date::sys_time<std::chrono::seconds> time_s;
+    std::istringstream ss(time_stamp);
+    ss >> date::parse("%FT%TZ", time_s);
+    time_s += std::chrono::seconds(offset);
+
+    auto dp = date::floor<date::days>(time_s);
+    auto time = date::make_time(time_s - dp);
+    auto ymd = date::year_month_day{dp};
+
+    std::ostringstream date_time_ss;
+    date_time_ss << ymd << "T" << time << "Z";
+    return date_time_ss.str();
+}
+
+double time_difference_seconds(const std::string & timestamp1, const std::string & timestamp2) {
+    using namespace date;
+    using namespace std::chrono;
+    try {
+        std::istringstream ss1(timestamp1);
+        std::istringstream ss2(timestamp2);
+        sys_time<microseconds> time1, time2;
+        ss1 >> parse("%FT%T%Ez", time1);
+        ss2 >> parse("%FT%T%Ez", time2);
+        // If parsing with timezone offset failed, try parsing with 'Z' format
+        if (ss1.fail()) {
+            ss1.clear();
+            ss1.str(timestamp1);
+            ss1 >> parse("%FT%TZ", time1);
+        }
+        if (ss2.fail()) {
+            ss2.clear();
+            ss2.str(timestamp2);
+            ss2 >> parse("%FT%TZ", time2);
+        }
+        duration<double> diff = time1 - time2;
+        return diff.count();
+    } catch (const std::exception & e) {
+        throw std::runtime_error("Failed to parse timestamps");
+    }
+}
+
+}  // namespace dorado::utils
diff --git a/dorado/utils/time_utils.h b/dorado/utils/time_utils.h
@@ -1,65 +1,20 @@
 #pragma once
-#include <date/date.h>
-#include <date/tz.h>
 
-#include <chrono>
 #include <ctime>
-#include <sstream>
 #include <string>
 
 namespace dorado::utils {
 
-inline std::string get_string_timestamp_from_unix_time(time_t time_stamp_ms) {
-    auto tp = std::chrono::system_clock::from_time_t(time_stamp_ms / 1000);
-    tp += std::chrono::milliseconds(time_stamp_ms % 1000);
-    auto dp = date::floor<date::days>(tp);
-    auto time = date::make_time(std::chrono::duration_cast<std::chrono::milliseconds>(tp - dp));
-    auto ymd = date::year_month_day{dp};
-
-    std::ostringstream date_time_ss;
-    date_time_ss << ymd << "T" << time << "+00:00";
-    return date_time_ss.str();
-}
+std::string get_string_timestamp_from_unix_time(time_t time_stamp_ms);
 
 // Expects the time to be encoded like "2017-09-12T09:50:12.456+00:00" or "2017-09-12T09:50:12Z".
 // Time stamp can be specified up to microseconds
-inline time_t get_unix_time_from_string_timestamp(const std::string& time_stamp) {
-    std::istringstream ss(time_stamp);
-    date::sys_time<std::chrono::microseconds> time_us;
-    ss >> date::parse("%FT%T%Ez", time_us);
-    // If parsing with timezone offset failed, try parsing with 'Z' format
-    if (ss.fail()) {
-        ss.clear();
-        ss.str(time_stamp);
-        ss >> date::parse("%FT%TZ", time_us);
-    }
-
-    auto epoch = time_us.time_since_epoch();
-    auto value = std::chrono::duration_cast<std::chrono::milliseconds>(epoch);
-    return value.count();
-}
-
-inline std::string adjust_time_ms(const std::string& time_stamp, uint64_t offset_ms) {
-    return get_string_timestamp_from_unix_time(get_unix_time_from_string_timestamp(time_stamp) +
-                                               offset_ms);
-}
-
-inline std::string adjust_time(const std::string& time_stamp, uint32_t offset) {
-    // Expects the time to be encoded like "2017-09-12T9:50:12Z".
-    // Adds the offset (in seconds) to the timeStamp.
+time_t get_unix_time_from_string_timestamp(const std::string& time_stamp);
 
-    date::sys_time<std::chrono::seconds> time_s;
-    std::istringstream ss(time_stamp);
-    ss >> date::parse("%FT%TZ", time_s);
-    time_s += std::chrono::seconds(offset);
+std::string adjust_time_ms(const std::string& time_stamp, uint64_t offset_ms);
 
-    auto dp = date::floor<date::days>(time_s);
-    auto time = date::make_time(time_s - dp);
-    auto ymd = date::year_month_day{dp};
+std::string adjust_time(const std::string& time_stamp, uint32_t offset);
 
-    std::ostringstream date_time_ss;
-    date_time_ss << ymd << "T" << time << "Z";
-    return date_time_ss.str();
-}
+double time_difference_seconds(const std::string& timestamp1, const std::string& timestamp2);
 
 }  // namespace dorado::utils