External build system generator release 44

Update tools: yexport

4 files changed, 528 insertions, 1 deletions

diff --git a/build/external_resources/yexport/resources.json b/build/external_resources/yexport/resources.json
index 2e83fa4247..571f01ee81 100644
--- a/build/external_resources/yexport/resources.json
+++ b/build/external_resources/yexport/resources.json
@@ -1,7 +1,7 @@
 {
     "by_platform": {
         "linux": {
-            "uri": "sbr:4771827313"
+            "uri": "sbr:4829977301"
         }
     }
 }
diff --git a/scripts/export_script_gen.py b/scripts/export_script_gen.py
new file mode 100644
index 0000000000..6bf68fef7c
--- /dev/null
+++ b/scripts/export_script_gen.py
@@ -0,0 +1,126 @@
+import argparse
+import collections
+import sys
+
+
+def parse_export_file(src):
+    for line in src:
+        line = line.strip()
+
+        if line and '#' not in line:
+            words = line.split()
+            if len(words) == 2 and words[0] == 'linux_version':
+                yield {'linux_version': words[1]}
+            elif len(words) == 2:
+                yield {'lang': words[0], 'sym': words[1]}
+            elif len(words) == 1:
+                yield {'lang': 'C', 'sym': words[0]}
+            else:
+                raise Exception('unsupported exports line: "{}"'.format(line))
+
+
+def to_c(sym):
+    symbols = collections.deque(sym.split('::'))
+    c_prefixes = [  # demangle prefixes for c++ symbols
+        '_ZN',      # namespace
+        '_ZTIN',    # typeinfo for
+        '_ZTSN',    # typeinfo name for
+        '_ZTTN',    # VTT for
+        '_ZTVN',    # vtable for
+        '_ZNK',     # const methods
+    ]
+    c_sym = ''
+    while symbols:
+        s = symbols.popleft()
+        if s == '*':
+            c_sym += '*'
+            break
+        if '*' in s and len(s) > 1:
+            raise Exception('Unsupported format, cannot guess length of symbol: ' + s)
+        c_sym += str(len(s)) + s
+    if symbols:
+        raise Exception('Unsupported format: ' + sym)
+    if c_sym[-1] != '*':
+        c_sym += 'E*'
+    return ['{prefix}{sym}'.format(prefix=prefix, sym=c_sym) for prefix in c_prefixes]
+
+
+def to_gnu(src, dest):
+    d = collections.defaultdict(list)
+    version = None
+    for item in parse_export_file(src):
+        if item.get('linux_version'):
+            if not version:
+                version = item.get('linux_version')
+            else:
+                raise Exception('More than one linux_version defined')
+        elif item['lang'] == 'C++':
+            d['C'].extend(to_c(item['sym']))
+        else:
+            d[item['lang']].append(item['sym'])
+
+    if version:
+        dest.write('{} {{\nglobal:\n'.format(version))
+    else:
+        dest.write('{\nglobal:\n')
+
+    for k, v in d.items():
+        dest.write('    extern "' + k + '" {\n')
+
+        for x in v:
+            dest.write('        ' + x + ';\n')
+
+        dest.write('    };\n')
+
+    dest.write('local: *;\n};\n')
+
+
+def to_msvc(src, dest):
+    dest.write('EXPORTS\n')
+    for item in parse_export_file(src):
+        if item.get('linux_version'):
+            continue
+        if item.get('lang') == 'C':
+            dest.write('    {}\n'.format(item.get('sym')))
+
+
+def to_darwin(src, dest):
+    pre = ''
+    for item in parse_export_file(src):
+        if item.get('linux_version'):
+            continue
+
+        if item['lang'] == 'C':
+            dest.write(pre + '-Wl,-exported_symbol,_' + item['sym'])
+        elif item['lang'] == 'C++':
+            for sym in to_c(item['sym']):
+                dest.write(pre + '-Wl,-exported_symbol,_' + sym)
+        else:
+            raise Exception('unsupported lang: ' + item['lang'])
+        if pre == '':
+            pre = ' '
+
+
+def main():
+    parser = argparse.ArgumentParser(description='Convert self-invented platform independent export file format to the format required by specific linker')
+    parser.add_argument('src', type=argparse.FileType('r', encoding='UTF-8'), help='platform independent export file path')
+    parser.add_argument('dest', type=argparse.FileType('w', encoding='UTF-8'), help='destination export file for required linker')
+    parser.add_argument('--format', help='destination file type format: gnu, msvc or darwin')
+
+    args = parser.parse_args()
+    if args.format == 'gnu':
+        to_gnu(args.src, args.dest)
+    elif args.format == 'msvc':
+        to_msvc(args.src, args.dest)
+    elif args.format == 'darwin':
+        to_darwin(args.src, args.dest)
+    else:
+        print('Unknown destination file format: {}'.format(args.format), file=sys.stderr)
+        sys.exit(1)
+
+    args.src.close()
+    args.dest.close()
+
+
+if __name__ == '__main__':
+    main()
diff --git a/scripts/generate_vcs_info.py b/scripts/generate_vcs_info.py
new file mode 100644
index 0000000000..770d2ec802
--- /dev/null
+++ b/scripts/generate_vcs_info.py
@@ -0,0 +1,321 @@
+# coding: utf-8
+import json
+import locale
+import re
+import os
+import subprocess
+import sys
+import time
+import six as six_
+
+
+INDENT = " " * 4
+
+
+def _get_vcs_dictionary(vcs_type, *arg):
+    if vcs_type == 'git':
+        return _GitVersion.parse(*arg)
+    else:
+        raise Exception("Unknown VCS type {}".format(str(vcs_type)))
+
+
+def _get_user_locale():
+    try:
+        if six_.PY3:
+            return [locale.getencoding()]
+        else:
+            return [locale.getdefaultlocale()[1]]
+    except Exception:
+        return []
+
+
+class _GitVersion():
+    @classmethod
+    def parse(cls, commit_hash, author_info, summary_info, body_info, tag_info, branch_info, depth=None):
+        r""" Parses output of
+            git rev-parse HEAD
+            git log -1 --format='format:%an <%ae>'
+            git log -1 --format='format:%s'
+            git log -1 --grep='^git-svn-id: ' --format='format:%b' or
+            git log -1 --grep='^Revision: r?\d*' --format='format:%b
+            git describe --exact-match --tags HEAD
+            git describe --exact-match --all HEAD
+            and depth as computed by _get_git_depth
+            '"""
+
+        info = {}
+        info['hash'] = commit_hash
+        info['commit_author'] = _SystemInfo._to_text(author_info)
+        info['summary'] = _SystemInfo._to_text(summary_info)
+
+        if 'svn_commit_revision' not in info:
+            url = re.search("git?-svn?-id: (.*)@(\\d*).*", body_info)
+            if url:
+                info['svn_url'] = url.group(1)
+                info['svn_commit_revision'] = int(url.group(2))
+
+        if 'svn_commit_revision' not in info:
+            rev = re.search('Revision: r?(\\d*).*', body_info)
+            if rev:
+                info['svn_commit_revision'] = int(rev.group(1))
+
+        info['tag'] = tag_info
+        info['branch'] = branch_info
+        info['scm_text'] = cls._format_scm_data(info)
+        info['vcs'] = 'git'
+
+        if depth:
+            info['patch_number'] = int(depth)
+        return info
+
+    @staticmethod
+    def _format_scm_data(info):
+        scm_data = "Git info:\n"
+        scm_data += INDENT + "Commit: " + info['hash'] + "\n"
+        scm_data += INDENT + "Branch: " + info['branch'] + "\n"
+        scm_data += INDENT + "Author: " + info['commit_author'] + "\n"
+        scm_data += INDENT + "Summary: " + info['summary'] + "\n"
+        if 'svn_commit_revision' in info or 'svn_url' in info:
+            scm_data += INDENT + "git-svn info:\n"
+        if 'svn_url' in info:
+            scm_data += INDENT + "URL: " + info['svn_url'] + "\n"
+        if 'svn_commit_revision' in info:
+            scm_data += INDENT + "Last Changed Rev: " + str(info['svn_commit_revision']) + "\n"
+        return scm_data
+
+    @staticmethod
+    def external_data(arc_root):
+        env = os.environ.copy()
+        env['TZ'] = ''
+
+        hash_args = ['rev-parse', 'HEAD']
+        author_args = ['log', '-1', '--format=format:%an <%ae>']
+        summary_args = ['log', '-1', '--format=format:%s']
+        svn_args = ['log', '-1', '--grep=^git-svn-id: ',  '--format=format:%b']
+        svn_args_alt = ['log', '-1', '--grep=^Revision: r\\?\\d*',  '--format=format:%b']
+        tag_args = ['describe', '--exact-match', '--tags', 'HEAD']
+        branch_args = ['describe', '--exact-match', '--all', 'HEAD']
+
+        # using local 'Popen' wrapper
+        commit = _SystemInfo._system_command_call(['git'] + hash_args, env=env, cwd=arc_root).rstrip()
+        author = _SystemInfo._system_command_call(['git'] + author_args, env=env, cwd=arc_root)
+        commit = _SystemInfo._system_command_call(['git'] + hash_args, env=env, cwd=arc_root).rstrip()
+        author = _SystemInfo._system_command_call(['git'] + author_args, env=env, cwd=arc_root)
+        summary = _SystemInfo._system_command_call(['git'] + summary_args, env=env, cwd=arc_root)
+        svn_id = _SystemInfo._system_command_call(['git'] + svn_args, env=env, cwd=arc_root)
+        if not svn_id:
+            svn_id = _SystemInfo._system_command_call(['git'] + svn_args_alt, env=env, cwd=arc_root)
+
+        try:
+            tag_info = _SystemInfo._system_command_call(['git'] + tag_args, env=env, cwd=arc_root).splitlines()
+        except Exception:
+            tag_info = [''.encode('utf-8')]
+
+        try:
+            branch_info = _SystemInfo._system_command_call(['git'] + branch_args, env=env, cwd=arc_root).splitlines()
+        except Exception:
+            branch_info = [''.encode('utf-8')]
+
+        depth = six_.text_type(_GitVersion._get_git_depth(env, arc_root)).encode('utf-8')
+
+        # logger.debug('Git info commit:{}, author:{}, summary:{}, svn_id:{}'.format(commit, author, summary, svn_id))
+        return [commit, author, summary, svn_id, tag_info[0], branch_info[0], depth]
+
+    # YT's patch number.
+    @staticmethod
+    def _get_git_depth(env, arc_root):
+        graph = {}
+        full_history_args = ["log", "--full-history", "--format=%H %P", "HEAD"]
+        history = _SystemInfo._system_command_call(['git'] + full_history_args, env=env, cwd=arc_root).decode('utf-8')
+
+        head = None
+        for line in history.splitlines():
+            values = line.split()
+            if values:
+                if head is None:
+                    head = values[0]
+                graph[values[0]] = values[1:]
+
+        assert head
+        cache = {}
+        stack = [(head, None, False)]
+        while stack:
+            commit, child, calculated = stack.pop()
+            if commit in cache:
+                calculated = True
+            if calculated:
+                if child is not None:
+                    cache[child] = max(cache.get(child, 0), cache[commit] + 1)
+            else:
+                stack.append((commit, child, True))
+                parents = graph[commit]
+                if not parents:
+                    cache[commit] = 0
+                else:
+                    for parent in parents:
+                        stack.append((parent, commit, False))
+        return cache[head]
+
+
+class _SystemInfo:
+    LOCALE_LIST = _get_user_locale() + [sys.getfilesystemencoding(), 'utf-8']
+
+    @classmethod
+    def get_locale(cls):
+        import codecs
+        for i in cls.LOCALE_LIST:
+            if not i:
+                continue
+            try:
+                codecs.lookup(i)
+                return i
+            except LookupError:
+                continue
+
+    @staticmethod
+    def _to_text(s):
+        if isinstance(s, six_.binary_type):
+            return s.decode(_SystemInfo.get_locale(), errors='replace')
+        return s
+
+    @staticmethod
+    def get_user():
+        sys_user = os.environ.get("USER")
+        if not sys_user:
+            sys_user = os.environ.get("USERNAME")
+        if not sys_user:
+            sys_user = os.environ.get("LOGNAME")
+        if not sys_user:
+            sys_user = "Unknown user"
+        return sys_user
+
+    @staticmethod
+    def get_date(stamp=None):
+        # Format compatible with SVN-xml format.
+        return time.strftime("%Y-%m-%dT%H:%M:%S.000000Z", time.gmtime(stamp))
+
+    @staticmethod
+    def get_timestamp():
+        # Unix timestamp.
+        return int(time.time())
+
+    @staticmethod
+    def get_other_data(src_dir, data_file='local.ymake'):
+        other_data = "Other info:\n"
+        other_data += INDENT + "Build by: " + _SystemInfo.get_user() + "\n"
+        other_data += INDENT + "Top src dir: {}\n".format(src_dir)
+
+        # logger.debug("Other data: %s", other_data)
+
+        return other_data
+
+    @staticmethod
+    def _get_host_info(fake_build_info=False):
+        if fake_build_info:
+            host_info = '*sys localhost 1.0.0 #dummy information '
+        elif not on_win():
+            host_info = ' '.join(os.uname())
+        else:
+            host_info = _SystemInfo._system_command_call("VER")  # XXX: check shell from cygwin to call VER this way!
+        return INDENT + INDENT + host_info.strip() + "\n" if host_info else ""
+
+    @staticmethod
+    def _system_command_call(command, **kwargs):
+        if isinstance(command, list):
+            command = subprocess.list2cmdline(command)
+        try:
+            process = subprocess.Popen(command, stdout=subprocess.PIPE, stderr=subprocess.PIPE, shell=True, **kwargs)
+            stdout, stderr = process.communicate()
+            if process.returncode != 0:
+                # logger.debug('{}\nRunning {} failed with exit code {}\n'.format(stderr, command, process.returncode))
+                raise get_svn_exception()(stdout=stdout, stderr=stderr, rc=process.returncode, cmd=[command])
+            return stdout
+        except OSError as e:
+            msg = e.strerror
+            errcodes = 'error {}'.format(e.errno)
+            if on_win() and isinstance(e, WindowsError):
+                errcodes += ', win-error {}'.format(e.winerror)
+                try:
+                    import ctypes
+                    msg = six_.text_type(ctypes.FormatError(e.winerror), _SystemInfo.get_locale()).encode('utf-8')
+                except ImportError:
+                    pass
+            # logger.debug('System command call {} failed [{}]: {}\n'.format(command, errcodes, msg))
+            return None
+
+
+def _get_raw_data(vcs_type, vcs_root):
+    lines = []
+    if vcs_type == 'git':
+        lines = _GitVersion.external_data(vcs_root)
+
+    return [l.decode('utf-8') for l in lines]
+
+
+def _get_json(vcs_root):
+    try:
+        vcs_type = "git"
+        info = _get_vcs_dictionary(vcs_type, *_get_raw_data(vcs_type, vcs_root))
+        return info, vcs_root
+    except Exception:
+        return None, ""
+
+
+def _dump_json(
+    arc_root, info,
+    other_data=None,
+    build_user=None,
+    build_date=None,
+    build_timestamp=0,
+    custom_version='',
+):
+    j = {}
+    j['PROGRAM_VERSION'] = info['scm_text'] + "\n" + _SystemInfo._to_text(other_data)
+    j['CUSTOM_VERSION'] = str(_SystemInfo._to_text(custom_version))
+    j['SCM_DATA'] = info['scm_text']
+    j['ARCADIA_SOURCE_PATH'] = _SystemInfo._to_text(arc_root)
+    j['ARCADIA_SOURCE_URL'] = info.get('url', info.get('svn_url', ''))
+    j['ARCADIA_SOURCE_REVISION'] = info.get('revision', -1)
+    j['ARCADIA_SOURCE_HG_HASH'] = info.get('hash', '')
+    j['ARCADIA_SOURCE_LAST_CHANGE'] = info.get('commit_revision', info.get('svn_commit_revision', -1))
+    j['ARCADIA_SOURCE_LAST_AUTHOR'] = info.get('commit_author', '')
+    j['ARCADIA_PATCH_NUMBER'] = info.get('patch_number', 0)
+    j['BUILD_USER'] = _SystemInfo._to_text(build_user)
+    j['VCS'] = info.get('vcs', '')
+    j['BRANCH'] = info.get('branch', '')
+    j['ARCADIA_TAG'] = info.get('tag', '')
+    j['DIRTY'] = info.get('dirty', '')
+
+    if 'url' in info or 'svn_url' in info:
+       j['SVN_REVISION'] = info.get('svn_commit_revision', info.get('revision', -1))
+       j['SVN_ARCROOT'] = info.get('url', info.get('svn_url', ''))
+       j['SVN_TIME'] = info.get('commit_date', info.get('svn_commit_date', ''))
+
+    j['BUILD_DATE'] = build_date
+    j['BUILD_TIMESTAMP'] = build_timestamp
+
+    return json.dumps(j,  sort_keys=True, indent=4, separators=(',', ': '))
+
+
+def get_version_info(arc_root, custom_version=""):
+    info, vcs_root = _get_json(arc_root)
+    if info is None:
+        return ""
+
+    return _dump_json(
+        vcs_root,
+        info,
+        other_data=_SystemInfo.get_other_data(
+            src_dir=vcs_root,
+        ),
+        build_user=_SystemInfo.get_user(),
+        build_date=_SystemInfo.get_date(None),
+        build_timestamp=_SystemInfo.get_timestamp(),
+        custom_version=custom_version,
+    )
+
+
+if __name__ == '__main__':
+    with open(sys.argv[1], 'w') as f:
+        f.write(get_version_info(sys.argv[2]))
+
diff --git a/scripts/split_unittest.py b/scripts/split_unittest.py
new file mode 100644
index 0000000000..8874b8b915
--- /dev/null
+++ b/scripts/split_unittest.py
@@ -0,0 +1,80 @@
+import argparse
+import tempfile
+import shlex
+import subprocess
+
+
+def parse_args():
+    parser = argparse.ArgumentParser()
+    parser.add_argument("--split-factor", type=int, default=0)
+    parser.add_argument("--shard", type=int, default=0)
+    parser.add_argument("--fork-mode", type=str, default="SEQUENTIAL")
+    parser.add_argument("command", nargs=argparse.REMAINDER)
+    return parser.parse_args()
+
+
+def get_sequential_chunk(tests, modulo, modulo_index):
+    chunk_size = len(tests) // modulo
+    not_used = len(tests) % modulo
+    shift = chunk_size + (modulo_index < not_used)
+    start = chunk_size * modulo_index + min(modulo_index, not_used)
+    end = start + shift
+    return [] if end > len(tests) else tests[start:end]
+
+
+def get_shuffled_chunk(tests, modulo, modulo_index):
+    result_tests = []
+    for i, test in enumerate(tests):
+        if i % modulo == modulo_index:
+            result_tests.append(test)
+    return result_tests
+
+
+def list_tests(binary):
+    with tempfile.NamedTemporaryFile() as tmpfile:
+        cmd = [binary, "--list-verbose", "--list-path", tmpfile.name]
+        subprocess.check_call(cmd)
+
+        with open(tmpfile.name) as afile:
+            lines = afile.read().strip().split("\n")
+            lines = [x.strip() for x in lines]
+            return [x for x in lines if x]
+
+
+def get_shard_tests(args):
+    test_names = list_tests(args.command[0])
+    test_names = sorted(test_names)
+
+    if args.fork_mode == "MODULO":
+        return get_shuffled_chunk(test_names, args.split_factor, args.shard)
+    elif args.fork_mode == "SEQUENTIAL":
+        return get_sequential_chunk(test_names, args.split_factor, args.shard)
+    else:
+        raise ValueError("detected unknown partition mode: {}".format(args.fork_mode))
+
+
+def get_shard_cmd_args(args):
+    return ["+{}".format(x) for x in get_shard_tests(args)]
+
+
+def main():
+    args = parse_args()
+
+    if args.split_factor:
+        shard_cmd = get_shard_cmd_args(args)
+        if shard_cmd:
+            cmd = args.command + shard_cmd
+        else:
+            print("No tests for {} shard".format(args.shard))
+            return 0
+    else:
+        cmd = args.command
+
+    rc = subprocess.call(cmd)
+    if rc:
+        print("Some tests failed. To reproduce run: {}".format(shlex.join(cmd)))
+    return rc
+
+
+if __name__ == "__main__":
+    exit(main())