From 4676f759e8b514a71e63a039fe496c3ef1bd639e Mon Sep 17 00:00:00 2001
From: Niels Lohmann <mail@nlohmann.me>
Date: Thu, 4 Apr 2019 09:50:27 +0200
Subject: [PATCH] :arrow_up: updated fastcov

---
 test/CMakeLists.txt                       |   8 +-
 test/thirdparty/fastcov/fastcov.py        | 122 ++++++++----
 test/thirdparty/fastcov/fastcov_legacy.py | 218 ----------------------
 3 files changed, 92 insertions(+), 256 deletions(-)
 delete mode 100755 test/thirdparty/fastcov/fastcov_legacy.py

diff --git a/test/CMakeLists.txt b/test/CMakeLists.txt
index 5001be66..4d542917 100644
--- a/test/CMakeLists.txt
+++ b/test/CMakeLists.txt
@@ -35,9 +35,9 @@ if(JSON_Coverage)
     set(CMAKE_CXX_FLAGS "--coverage -g -O0 -fprofile-arcs -ftest-coverage")
 
     # from https://github.com/RWTH-HPC/CMake-codecov/blob/master/cmake/FindGcov.cmake
-    #get_filename_component(COMPILER_PATH "${CMAKE_CXX_COMPILER}" PATH)
-    #string(REGEX MATCH "^[0-9]+" GCC_VERSION "${CMAKE_CXX_COMPILER_VERSION}")
-    #find_program(GCOV_BIN NAMES gcov-${GCC_VERSION} gcov HINTS ${COMPILER_PATH})
+    get_filename_component(COMPILER_PATH "${CMAKE_CXX_COMPILER}" PATH)
+    string(REGEX MATCH "^[0-9]+" GCC_VERSION "${CMAKE_CXX_COMPILER_VERSION}")
+    find_program(GCOV_BIN NAMES gcov-${GCC_VERSION} gcov HINTS ${COMPILER_PATH})
 
     # collect all source files from the chosen include dir
     file(GLOB_RECURSE SOURCE_FILES ${NLOHMANN_JSON_INCLUDE_BUILD_DIR}*.hpp)
@@ -55,7 +55,7 @@ if(JSON_Coverage)
     # add target to collect coverage information and generate HTML file
     # (filter script from https://stackoverflow.com/a/43726240/266378)
     add_custom_target(lcov_html2
-        COMMAND ${CMAKE_SOURCE_DIR}/test/thirdparty/fastcov/fastcov.py --lcov -o json.info --gcov ${GCOV_BIN} --exclude-gcov /usr 9.0.1/ test/
+        COMMAND ${CMAKE_SOURCE_DIR}/test/thirdparty/fastcov/fastcov.py --branch-coverage --lcov -o json.info --gcov ${GCOV_BIN}
         COMMAND gsed -i 's%build_coverage/%%g' json.info
         COMMAND ${CMAKE_SOURCE_DIR}/test/thirdparty/imapdl/filterbr.py json.info > json.info.filtered.noexcept
         COMMAND genhtml --title "JSON for Modern C++" --legend --demangle-cpp --output-directory html --show-details --branch-coverage json.info.filtered.noexcept
diff --git a/test/thirdparty/fastcov/fastcov.py b/test/thirdparty/fastcov/fastcov.py
index ee7fbe78..dd4c9132 100755
--- a/test/thirdparty/fastcov/fastcov.py
+++ b/test/thirdparty/fastcov/fastcov.py
@@ -13,7 +13,7 @@
         $ cd build_dir
         $ ./fastcov.py --zerocounters
         $ <run unit tests>
-        $ ./fastcov.py --exclude-gcov /usr/include --lcov -o report.info
+        $ ./fastcov.py --exclude /usr/include test/ --lcov -o report.info
         $ genhtml -o code_coverage report.info
 """
 
@@ -64,24 +64,28 @@ def getGcdaFiles(cwd, gcda_files):
         gcda_files = glob.glob(os.path.join(cwd, "**/*.gcda"), recursive=True)
     return gcda_files
 
-def gcovWorker(cwd, gcov, files, chunk, exclude):
-    p = subprocess.Popen([gcov, "-it"] + chunk, cwd=cwd, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
+def gcovWorker(cwd, gcov, files, chunk, gcov_filter_options, branch_coverage):
+    gcov_args = "-it"
+    if branch_coverage:
+        gcov_args += "b"
+
+    p = subprocess.Popen([gcov, gcov_args] + chunk, cwd=cwd, stdout=subprocess.PIPE, stderr=subprocess.DEVNULL)
     for line in iter(p.stdout.readline, b''):
         intermediate_json = json.loads(line.decode(sys.stdout.encoding))
-        intermediate_json_files = processGcovs(intermediate_json["files"], exclude)
+        intermediate_json_files = processGcovs(intermediate_json["files"], gcov_filter_options)
         for f in intermediate_json_files:
             files.append(f) #thread safe, there might be a better way to do this though
         GCOVS_TOTAL.append(len(intermediate_json["files"]))
         GCOVS_SKIPPED.append(len(intermediate_json["files"])-len(intermediate_json_files))
     p.wait()
 
-def processGcdas(cwd, gcov, jobs, gcda_files, exclude):
+def processGcdas(cwd, gcov, jobs, gcda_files, gcov_filter_options, branch_coverage):
     chunk_size = max(MINIMUM_CHUNK_SIZE, int(len(gcda_files) / jobs) + 1)
 
     threads = []
     intermediate_json_files = []
     for chunk in chunks(gcda_files, chunk_size):
-        t = threading.Thread(target=gcovWorker, args=(cwd, gcov, intermediate_json_files, chunk, exclude))
+        t = threading.Thread(target=gcovWorker, args=(cwd, gcov, intermediate_json_files, chunk, gcov_filter_options, branch_coverage))
         threads.append(t)
         t.start()
 
@@ -91,39 +95,75 @@ def processGcdas(cwd, gcov, jobs, gcda_files, exclude):
 
     return intermediate_json_files
 
-def processGcov(gcov, files, exclude):
-    for ex in exclude:
+def processGcov(gcov, files, gcov_filter_options):
+    # If explicit sources were passed, check for match
+    source_file = os.path.abspath(gcov["file"])
+    if gcov_filter_options["sources"]:
+        if source_file in gcov_filter_options["sources"]:
+            files.append(gcov)
+        return
+
+    # Check include filter
+    if gcov_filter_options["include"]:
+        for ex in gcov_filter_options["include"]:
+            if ex in gcov["file"]:
+                files.append(gcov)
+                break
+        return
+
+    # Check exclude filter
+    for ex in gcov_filter_options["exclude"]:
         if ex in gcov["file"]:
             return
+
     files.append(gcov)
 
-def processGcovs(gcov_files, exclude):
+def processGcovs(gcov_files, gcov_filter_options):
     files = []
     for gcov in gcov_files:
-        processGcov(gcov, files, exclude)
+        processGcov(gcov, files, gcov_filter_options)
     return files
 
-def dumpToLcovInfo(cwd, intermediate, output):
+def dumpBranchCoverageToLcovInfo(f, source):
+    branch_miss  = 0
+    branch_total = 0
+    for line in source["lines"]:
+        if not line["branches"]:
+            continue
+        branch_total += len(line["branches"])
+        for i, branch in enumerate(line["branches"]):
+            #Branch (<line number>, <block number>, <branch number>, <taken>)
+            f.write("BRDA:%d,%d,%d,%d\n" % (line["line_number"], int(i/2), i, branch["count"]))
+            branch_miss += int(branch["count"] == 0)
+    f.write("BRF:%d\n" % branch_total)                 #Branches Found
+    f.write("BRH:%d\n" % (branch_total - branch_miss)) #Branches Hit
+
+def dumpToLcovInfo(cwd, intermediate, output, branch_coverage):
     with open(output, "w") as f:
-        for file in intermediate:
+        for source in intermediate:
             #Convert to absolute path so it plays nice with genhtml
-            sf = file["file"]
-            if not os.path.isabs(file["file"]):
-                sf = os.path.abspath(os.path.join(cwd, file["file"]))
-            f.write("SF:%s\n" % sf)
+            sf = source["file"]
+            if not os.path.isabs(source["file"]):
+                sf = os.path.abspath(os.path.join(cwd, source["file"]))
+            f.write("SF:%s\n" % sf) #Source File
+
             fn_miss = 0
-            for function in file["functions"]:
-                f.write("FN:%s,%s\n" % (function["start_line"], function["name"]))
-                f.write("FNDA:%s,%s\n" % (function["execution_count"], function["name"]))
-                fn_miss += int(not function["execution_count"] == 0)
-            f.write("FNF:%s\n" % len(file["functions"]))
-            f.write("FNH:%s\n" % (len(file["functions"]) - fn_miss))
+            for function in source["functions"]:
+                f.write("FN:%d,%s\n" % (function["start_line"], function["name"]))          #Function Start Line
+                f.write("FNDA:%d,%s\n" % (function["execution_count"], function["name"]))   #Function Hits
+                fn_miss += int(function["execution_count"] == 0)
+            f.write("FNF:%d\n" % len(source["functions"]))                #Functions Found
+            f.write("FNH:%d\n" % (len(source["functions"]) - fn_miss))    #Functions Hit
+
+            if branch_coverage:
+                dumpBranchCoverageToLcovInfo(f, source)
+
             line_miss = 0
-            for line in file["lines"]:
-                f.write("DA:%s,%s\n" % (line["line_number"], line["count"]))
-                line_miss += int(not line["count"] == 0)
-            f.write("LF:%s\n" % len(file["lines"]))
-            f.write("LH:%s\n" % (len(file["lines"]) - line_miss))
+            for line in source["lines"]:
+                f.write("DA:%d,%d\n" % (line["line_number"], line["count"])) #Line
+                line_miss += int(line["count"] == 0)
+            f.write("LF:%d\n" % len(source["lines"]))                 #Lines Found
+            f.write("LH:%d\n" % (len(source["lines"]) - line_miss))   #Lines Hit
             f.write("end_of_record\n")
 
 def dumpToGcovJson(intermediate, output):
@@ -134,6 +174,13 @@ def log(line):
     if not args.quiet:
         print(line)
 
+def getGcovFilterOptions(args):
+    return {
+        "sources": set([os.path.abspath(s) for s in args.sources]), #Make paths absolute
+        "include": args.includepost,
+        "exclude": args.excludepost,
+    }
+
 def main(args):
     # Need at least gcov 9.0.0 because that's when gcov JSON and stdout streaming was introduced
     current_gcov_version = getGcovVersion(args.gcov)
@@ -154,7 +201,8 @@ def main(args):
         log("%d .gcda files removed" % len(gcda_files))
         return
 
-    intermediate_json_files = processGcdas(args.cdirectory, args.gcov, args.jobs, gcda_files, args.excludepost)
+    gcov_filter_options = getGcovFilterOptions(args)
+    intermediate_json_files = processGcdas(args.cdirectory, args.gcov, args.jobs, gcda_files, gcov_filter_options, args.branchcoverage)
 
     gcov_total = sum(GCOVS_TOTAL)
     gcov_skipped = sum(GCOVS_SKIPPED)
@@ -162,7 +210,7 @@ def main(args):
     log("%d .gcov files processed by fastcov (%d skipped)" % (gcov_total - gcov_skipped, gcov_skipped))
 
     if args.lcov:
-        dumpToLcovInfo(args.cdirectory, intermediate_json_files, args.output)
+        dumpToLcovInfo(args.cdirectory, intermediate_json_files, args.output, args.branchcoverage)
         log("Created lcov info file '%s'" % args.output)
     else:
         dumpToGcovJson(intermediate_json_files, args.output)
@@ -172,18 +220,24 @@ if __name__ == '__main__':
     parser = argparse.ArgumentParser(description='A parallel gcov wrapper for fast coverage report generation')
     parser.add_argument('-z', '--zerocounters', dest='zerocounters', action="store_true", help='Recursively delete all gcda files')
 
-    parser.add_argument('-f', '--gcda-files', dest='gcda_files', nargs="+", default=[], help='Specify exactly which gcda files should be processed instead of recursivly searching the search directory.')
-    parser.add_argument('-E', '--exclude-gcda', dest='excludepre', nargs="+", default=[], help='.gcda filter - Exclude gcda files from being processed via simple find matching (not regex)')
-    parser.add_argument('-e', '--exclude-gcov', dest='excludepost', nargs="+", default=[], help='.gcov filter - Exclude gcov files from being processed via simple find matching (not regex)')
+    # Enable Branch Coverage
+    parser.add_argument('-b', '--branch-coverage', dest='branchcoverage', action="store_true", help='Include branch counts in the coverage report')
 
-    parser.add_argument('-g', '--gcov', dest='gcov', default='gcov', help='which gcov binary to use')
+    # Filtering Options
+    parser.add_argument('-s', '--source-files', dest='sources',     nargs="+", default=[], help='Filter: Specify exactly which source files should be included in the final report. Paths must be either absolute or relative to current directory.')
+    parser.add_argument('-e', '--exclude',      dest='excludepost', nargs="+", default=[], help='Filter: Exclude source files from final report if they contain one of the provided substrings (i.e. /usr/include test/, etc.)')
+    parser.add_argument('-i', '--include',      dest='includepost', nargs="+", default=[], help='Filter: Only include source files in final report that contain one of the provided substrings (i.e. src/ etc.)')
+    parser.add_argument('-f', '--gcda-files',   dest='gcda_files',  nargs="+", default=[], help='Filter: Specify exactly which gcda files should be processed instead of recursively searching the search directory.')
+    parser.add_argument('-E', '--exclude-gcda', dest='excludepre',  nargs="+", default=[], help='Filter: Exclude gcda files from being processed via simple find matching (not regex)')
+
+    parser.add_argument('-g', '--gcov', dest='gcov', default='gcov', help='Which gcov binary to use')
 
     parser.add_argument('-d', '--search-directory', dest='directory', default=".", help='Base directory to recursively search for gcda files (default: .)')
     parser.add_argument('-c', '--compiler-directory', dest='cdirectory', default=".", help='Base directory compiler was invoked from (default: .)')
     parser.add_argument('-j', '--jobs', dest='jobs', type=int, default=multiprocessing.cpu_count(), help='Number of parallel gcov to spawn (default: %d).' % multiprocessing.cpu_count())
 
     parser.add_argument('-o', '--output', dest='output', default="coverage.json", help='Name of output file (default: coverage.json)')
-    parser.add_argument('-i', '--lcov', dest='lcov', action="store_true", help='Output in lcov info format instead of gcov json')
+    parser.add_argument('-l', '--lcov', dest='lcov', action="store_true", help='Output in lcov info format instead of gcov json')
     parser.add_argument('-q', '--quiet', dest='quiet', action="store_true", help='Suppress output to stdout')
     args = parser.parse_args()
     main(args)
\ No newline at end of file
diff --git a/test/thirdparty/fastcov/fastcov_legacy.py b/test/thirdparty/fastcov/fastcov_legacy.py
deleted file mode 100755
index 3e0151ab..00000000
--- a/test/thirdparty/fastcov/fastcov_legacy.py
+++ /dev/null
@@ -1,218 +0,0 @@
-#!/usr/bin/env python3
-"""
-    Author: Bryan Gillespie
-
-    Legacy version... supports versions 7.1.0 <= GCC < 9.0.0
-
-    A massively parallel gcov wrapper for generating intermediate coverage formats fast
-
-    The goal of fastcov is to generate code coverage intermediate formats as fast as possible
-    (ideally < 1 second), even for large projects with hundreds of gcda objects. The intermediate
-    formats may then be consumed by a report generator such as lcov's genhtml, or a dedicated front
-    end such as coveralls.
-
-    Sample Usage:
-        $ cd build_dir
-        $ ./fastcov.py --exclude-gcov /usr/include --lcov -o report.info
-        $ genhtml -o code_coverage report.info
-"""
-
-import re
-import os
-import glob
-import json
-import argparse
-import subprocess
-import multiprocessing
-from random import shuffle
-
-MINIMUM_GCOV = (7,1,0)
-MINIMUM_CHUNK_SIZE = 10
-
-def chunks(l, n):
-    """Yield successive n-sized chunks from l."""
-    for i in range(0, len(l), n):
-        yield l[i:i + n]
-
-def getGcovVersion(gcov):
-    p = subprocess.Popen([gcov, "-v"], stdout=subprocess.PIPE)
-    output = p.communicate()[0].decode('UTF-8')
-    p.wait()
-    version_str = re.search(r'\s([\d.]+)\s', output.split("\n")[0]).group(1)
-    version = tuple(map(int, version_str.split(".")))
-    return version
-
-def removeFiles(files):
-    for file in files:
-        os.remove(file)
-
-def getFilteredGcdaFiles(gcda_files, exclude):
-    def excludeGcda(gcda):
-        for ex in exclude:
-            if ex in gcda:
-                return False
-        return True
-    return list(filter(excludeGcda, gcda_files))
-
-def getGcdaFiles(cwd, gcda_files, exclude):
-    if not gcda_files:
-        gcda_files = glob.glob(os.path.join(cwd, "**/*.gcda"), recursive=True)
-    if exclude:
-        return getFilteredGcdaFiles(gcda_files, exclude)
-    return gcda_files
-
-def getGcovFiles(cwd):
-    return glob.glob(os.path.join(cwd, "*.gcov"))
-
-def filterGcovFiles(gcov):
-    with open(gcov) as f:
-        path = f.readline()[5:]
-        for ex in args.exclude:
-            if ex in path:
-                return False
-        return True
-
-def processGcdasPre9(cwd, gcov, jobs, gcda_files):
-    chunk_size = min(MINIMUM_CHUNK_SIZE, int(len(gcda_files) / jobs) + 1)
-
-    processes = []
-    # shuffle(gcda_files) # improves performance by preventing any one gcov from bottlenecking on a list of sequential, expensive gcdas (?)
-    for chunk in chunks(gcda_files, chunk_size):
-        processes.append(subprocess.Popen([gcov, "-i"] + chunk, cwd=cwd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL))
-
-    for p in processes:
-        p.wait()
-
-def processGcdasPre9Accurate(cwd, gcov, gcda_files, exclude):
-    intermediate_json_files = []
-    for gcda in gcda_files:
-        subprocess.Popen([gcov, "-i", gcda], cwd=cwd, stdout=subprocess.DEVNULL, stderr=subprocess.DEVNULL).wait()
-        gcov_files = getGcovFiles(cwd)
-        intermediate_json_files += processGcovs(gcov_files, exclude)
-        removeFiles(gcov_files)
-    return intermediate_json_files
-
-def processGcovLine(file, line):
-    line_type, data = line.split(":", 1)
-    if line_type == "lcount":
-        num, count = data.split(",")
-        hit = (count != 0)
-        file["lines_hit"] += int(hit)
-        file["lines"].append({
-            "branches": [],
-            "line_number": num,
-            "count": count,
-            "unexecuted_block": not hit
-        })
-    elif line_type == "function":
-        num, count, name = data.split(",")
-        hit = (count != 0)
-        file["functions_hit"] += int(hit)
-        file["functions"].append({
-            "name": name,
-            "execution_count": count,
-            "start_line": num,
-            "end_line": None,
-            "blocks": None,
-            "blocks_executed": None,
-            "demangled_name": None
-        })
-
-def processGcov(files, gcov, exclude):
-    with open(gcov) as f:
-        path = f.readline()[5:].rstrip()
-        for ex in exclude:
-            if ex in path:
-                return False
-        file = {
-            "file": path,
-            "functions": [],
-            "functions_hit": 0,
-            "lines": [],
-            "lines_hit": 0
-        }
-        for line in f:
-            processGcovLine(file, line.rstrip())
-    files.append(file)
-    return True
-
-def processGcovs(gcov_files, exclude):
-    files = []
-    filtered = 0
-    for gcov in gcov_files:
-        filtered += int(not processGcov(files, gcov, exclude))
-    print("Skipped %d .gcov files" % filtered)
-    return files
-
-def dumpToLcovInfo(intermediate, output):
-    with open(output, "w") as f:
-        for file in intermediate:
-            f.write("SF:%s\n" % file["file"])
-            for function in file["functions"]:
-                f.write("FN:%s,%s\n" % (function["start_line"], function["name"]))
-                f.write("FNDA:%s,%s\n" % (function["execution_count"], function["name"]))
-            f.write("FNF:%s\n" % len(file["functions"]))
-            f.write("FNH:%s\n" % file["functions_hit"])
-            for line in file["lines"]:
-                f.write("DA:%s,%s\n" % (line["line_number"], line["count"]))
-            f.write("LF:%s\n" % len(file["lines"]))
-            f.write("LH:%s\n" % file["lines_hit"])
-            f.write("end_of_record\n")
-
-def dumpToGcovJson(intermediate, output):
-    with open(output, "w") as f:
-        json.dump(intermediate, f)
-
-def main(args):
-    # Need at least gcov 7.1.0 because of bug not allowing -i in conjunction with multiple files
-    # See: https://github.com/gcc-mirror/gcc/commit/41da7513d5aaaff3a5651b40edeccc1e32ea785a
-    current_gcov_version = getGcovVersion(args.gcov)
-    if current_gcov_version < MINIMUM_GCOV:
-        print("Minimum gcov version {} required, found {}".format(".".join(map(str, MINIMUM_GCOV)), ".".join(map(str, current_gcov_version))))
-        exit(1)
-
-    gcda_files = getGcdaFiles(args.directory, args.gcda_files, args.excludepre)
-    print("Found %d .gcda files" % len(gcda_files))
-
-    # We "zero" the "counters" by simply deleting all gcda files
-    if args.zerocounters:
-        removeFiles(gcda_files)
-        print("Removed %d .gcda files" % len(gcda_files))
-        return
-
-    # If we are less than gcov 9.0.0, convert .gcov files to GCOV 9 JSON format
-    processGcdasPre9(args.cdirectory, args.gcov, args.jobs, gcda_files)
-    gcov_files = getGcovFiles(args.cdirectory)
-
-    print("Found %d .gcov files" % len(gcov_files))
-
-    intermediate_json_files = processGcovs(gcov_files, args.excludepost)
-    removeFiles(gcov_files)
-
-    intermediate_json_files += processGcdasPre9Accurate(args.cdirectory, args.gcov, args.gcda_files_accurate, args.excludepost)
-
-    if args.lcov:
-        dumpToLcovInfo(intermediate_json_files, args.output)
-    else:
-        dumpToGcovJson(intermediate_json_files, args.output)
-
-if __name__ == '__main__':
-    parser = argparse.ArgumentParser(description='A parallel gcov wrapper for fast coverage report generation')
-    parser.add_argument('-z', '--zerocounters', dest='zerocounters', action="store_true", help='Recursively delete all gcda files')
-
-    parser.add_argument('-f', '--gcda-files', dest='gcda_files', nargs="+", default=[], help='Specify exactly which gcda files should be processed instead of recursivly searching the search directory.')
-    parser.add_argument('-F', '--gcda-files-accurate', dest='gcda_files_accurate', nargs="+", default=[], help='(< gcov 9.0.0) Get accurate header coverage information for just these. These files cannot be processed in parallel')
-    parser.add_argument('-E', '--exclude-gcda', dest='excludepre', nargs="+", default=[], help='.gcda filter - Exclude gcda files from being processed via simple find matching (not regex)')
-    parser.add_argument('-e', '--exclude-gcov', dest='excludepost', nargs="+", default=[], help='.gcov filter - Exclude gcov files from being processed via simple find matching (not regex)')
-
-    parser.add_argument('-g', '--gcov', dest='gcov', default='gcov', help='which gcov binary to use')
-
-    parser.add_argument('-d', '--search-directory', dest='directory', default=".", help='Base directory to recursively search for gcda files (default: .)')
-    parser.add_argument('-c', '--compiler-directory', dest='cdirectory', default=".", help='Base directory compiler was invoked from (default: .)')
-    parser.add_argument('-j', '--jobs', dest='jobs', type=int, default=multiprocessing.cpu_count(), help='Number of parallel gcov to spawn (default: %d).' % multiprocessing.cpu_count())
-
-
-    parser.add_argument('-o', '--output', dest='output', default="coverage.json", help='Name of output file (default: coverage.json)')
-    parser.add_argument('-i', '--lcov', dest='lcov', action="store_true", help='Output in lcov info format instead of gcov json')
-    args = parser.parse_args()
-    main(args)
\ No newline at end of file