intermediate changes

ref:4c3887b325a2b9c0b2481e44130c092b3c8d545f

5 files changed, 120 insertions, 41 deletions

diff --git a/build/rules/flake8/migrations.yaml b/build/rules/flake8/migrations.yaml
index 206e5fd79c..6663023ff2 100644
--- a/build/rules/flake8/migrations.yaml
+++ b/build/rules/flake8/migrations.yaml
@@ -418,6 +418,7 @@ migrations:
     - market/mstat/ch-cache/lib/database
     - market/reductor/configure/lib
     - market/reductor/www
+    - market/report/lite
     - market/sre/library/python/maaslib
     - market/sre/services/balancer_api/lib
     - market/tools/report_stats/lib
@@ -3594,7 +3595,6 @@ migrations:
     - infra/walle/server/tests
     - market/forecaster/lite
     - market/idx/marketindexer/medium_tests/yatf
-    - market/report/lite
     - ofd/notifier/tests/test_unit
     - saas/rtyserver_test/tests
     - search/lingboost/saas/codecs/test_bundle_codecs/tests
diff --git a/contrib/python/six/.yandex_meta/yamaker.yaml b/contrib/python/six/.yandex_meta/yamaker.yaml
new file mode 100644
index 0000000000..2502bc85b9
--- /dev/null
+++ b/contrib/python/six/.yandex_meta/yamaker.yaml
@@ -0,0 +1,2 @@
+keep:
+- tests/test_six.py
diff --git a/contrib/python/six/patches/01-arcadia.patch b/contrib/python/six/patches/01-arcadia.patch
new file mode 100644
index 0000000000..57f85cb6d9
--- /dev/null
+++ b/contrib/python/six/patches/01-arcadia.patch
@@ -0,0 +1,11 @@
+--- contrib/python/six/six.py	(index)
++++ contrib/python/six/six.py	(working tree)
+@@ -274,7 +274,7 @@ _moved_attributes = [
+     MovedModule("collections_abc", "collections", "collections.abc" if sys.version_info >= (3, 3) else "collections"),
+     MovedModule("copyreg", "copy_reg"),
+     MovedModule("dbm_gnu", "gdbm", "dbm.gnu"),
+-    MovedModule("dbm_ndbm", "dbm", "dbm.ndbm"),
++    # MovedModule("dbm_ndbm", "dbm", "dbm.ndbm"),
+     MovedModule("_dummy_thread", "dummy_thread", "_dummy_thread" if sys.version_info < (3, 9) else "_thread"),
+     MovedModule("http_cookiejar", "cookielib", "http.cookiejar"),
+     MovedModule("http_cookies", "Cookie", "http.cookies"),
diff --git a/contrib/python/six/ya.make b/contrib/python/six/ya.make
index e0c7849214..096129b1e5 100644
--- a/contrib/python/six/ya.make
+++ b/contrib/python/six/ya.make
@@ -1,11 +1,15 @@
-OWNER(g:python-contrib)
+# Generated by devtools/yamaker (pypi).
 
 PY23_LIBRARY()
 
-LICENSE(MIT)
+OWNER(g:python-contrib)
 
 VERSION(1.16.0)
 
+LICENSE(MIT)
+
+NO_LINT()
+
 PY_SRCS(
     TOP_LEVEL
     six.py
@@ -17,8 +21,6 @@ RESOURCE_FILES(
     .dist-info/top_level.txt
 )
 
-NO_LINT()
-
 END()
 
 RECURSE_FOR_TESTS(
diff --git a/library/cpp/http/misc/httpreqdata.cpp b/library/cpp/http/misc/httpreqdata.cpp
index 38bdad3996..e8bf6df1a1 100644
--- a/library/cpp/http/misc/httpreqdata.cpp
+++ b/library/cpp/http/misc/httpreqdata.cpp
@@ -1,7 +1,15 @@
 #include "httpreqdata.h"
 
 #include <library/cpp/case_insensitive_string/case_insensitive_string.h>
+
 #include <util/stream/mem.h>
+#include <util/string/join.h>
+
+#include <array>
+
+#ifdef _sse4_2_
+#include <smmintrin.h>
+#endif
 
 TBaseServerRequestData::TBaseServerRequestData(SOCKET s)
     : Socket_(s)
@@ -98,25 +106,29 @@ TStringBuf TBaseServerRequestData::Environment(TStringBuf key) const {
 
 const TString& TBaseServerRequestData::GetCurPage() const {
     if (!CurPage_ && Host_) {
-        CurPage_ = "http://";
-        CurPage_ += Host_;
+        std::array<TStringBuf, 7> fragments;
+        auto fragmentIt = fragments.begin();
+        *fragmentIt++ = "http://"sv;
+        *fragmentIt++ = Host_;
         if (Port_) {
-            CurPage_ += ':';
-            CurPage_ += Port_;
+            *fragmentIt++ = ":"sv;
+            *fragmentIt++ = Port_;
         }
-        CurPage_ += Path_;
-        if (Query_) {
-            CurPage_ += '?';
-            CurPage_ += Query_;
+        *fragmentIt++ = Path_;
+        if (!Query_.empty()) {
+            *fragmentIt++ = "?"sv;
+            *fragmentIt++ = Query_;
         }
+
+        CurPage_ = JoinRange(""sv, fragments.begin(), fragmentIt);
     }
     return CurPage_;
 }
 
-bool TBaseServerRequestData::Parse(TStringBuf origReqBuf) {
-    ParseBuf_.reserve(origReqBuf.size() + 1);
-    ParseBuf_.assign(origReqBuf.begin(), origReqBuf.end());
-    ParseBuf_.push_back('\0');
+bool TBaseServerRequestData::Parse(TStringBuf origReq) {
+    ParseBuf_.reserve(origReq.size() + 16);
+    ParseBuf_.assign(origReq.begin(), origReq.end());
+    ParseBuf_.insert(ParseBuf_.end(), 16, ' ');
     char* req = ParseBuf_.data();
 
     while (*req == ' ' || *req == '\t')
@@ -126,33 +138,85 @@ bool TBaseServerRequestData::Parse(TStringBuf origReqBuf) {
     while (req[1] == '/') // remove redundant slashes
         req++;
 
-    // detect url end (can contain some garbage after whitespace, e.g. 'HTTP 1.1')
-    char* urlEnd = req;
-    while (*urlEnd && *urlEnd != ' ' && *urlEnd != '\t')
-        urlEnd++;
-    if (*urlEnd)
-        *urlEnd = 0;
-
-    // cut fragment if exists
-    char* fragment = strchr(req, '#');
-    if (fragment)
-        *fragment = 0; // ignore fragment
-    else
-        fragment = urlEnd;
-    char* path = req;
-
-    // calculate Search length without additional strlen-ing
-    char* query = strchr(path, '?');
-    if (query) {
-        *query++ = 0;
-        ptrdiff_t delta = fragment - query;
-        // indeed, second case is a parse error
-        Query_ = {query, static_cast<size_t>(delta >= 0 ? delta : (urlEnd - query))};
+    char* pathBegin = req;
+    char* queryBegin = nullptr;
+
+#ifdef _sse4_2_
+    const __m128i simdSpace = _mm_set1_epi8(' ');
+    const __m128i simdTab = _mm_set1_epi8('\t');
+    const __m128i simdHash = _mm_set1_epi8('#');
+    const __m128i simdQuestion = _mm_set1_epi8('?');
+
+    auto isEnd = [=](__m128i x) {
+        const auto v = _mm_or_si128(
+                _mm_or_si128(
+                    _mm_cmpeq_epi8(x, simdSpace), _mm_cmpeq_epi8(x, simdTab)),
+                _mm_cmpeq_epi8(x, simdHash));
+        return !_mm_testz_si128(v, v);
+    };
+
+    // No need for the range check because we have padding of spaces at the end.
+    for (;; req += 16) {
+        const auto x = _mm_loadu_si128(reinterpret_cast<const __m128i *>(req));
+        const auto isQuestionSimd = _mm_cmpeq_epi8(x, simdQuestion);
+        const auto isQuestion = !_mm_testz_si128(isQuestionSimd, isQuestionSimd);
+        if (isEnd(x)) {
+            if (isQuestion) {
+                // The prospective query end and a question sign are both in the
+                // current block. Need to find out which comes first.
+                for (;*req != ' ' && *req != '\t' && *req != '#'; ++req) {
+                    if (*req == '?') {
+                        queryBegin = req + 1;
+                        break;
+                    }
+                }
+            }
+            break;
+        }
+        if (isQuestion) {
+            // Find the exact query beginning
+            for (queryBegin = req; *queryBegin != '?'; ++queryBegin) {
+            }
+            ++queryBegin;
+
+            break;
+        }
+    }
+
+    // If we bailed out because we found query string begin. Now look for the the end of the query
+    if (queryBegin) {
+        for (;; req += 16) {
+            const auto x = _mm_loadu_si128(reinterpret_cast<const __m128i *>(req));
+            if (isEnd(x)) {
+                break;
+            }
+        }
+    }
+#else
+    for (;*req != ' ' && *req != '\t' && *req != '#'; ++req) {
+        if (*req == '?') {
+            queryBegin = req + 1;
+            break;
+        }
+    }
+#endif
+
+    while (*req != ' ' && *req != '\t' && *req != '#') {
+        ++req;
+    }
+
+    char* pathEnd = queryBegin ? queryBegin - 1 : req;
+    // Make sure Path_ and Query_ are actually zero-reminated.
+    *pathEnd = '\0';
+    *req = '\0';
+    Path_ = TStringBuf{pathBegin, pathEnd};
+    if (queryBegin) {
+        Query_ = TStringBuf{queryBegin, req};
+        OrigQuery_ = Query_;
     } else {
         Query_ = {};
+        OrigQuery_ = {};
     }
-    Path_ = path;
-    OrigQuery_ = Query_;
 
     return true;
 }