str_to_date function push down (#1961)

dbms/src/Common/MyTime.h

@@ -3,6 +3,7 @@
 #include <Core/Field.h>
 #include <common/DateLUTImpl.h>
 
+struct StringRef;
 namespace DB
 {
 
@@ -133,6 +134,24 @@ struct MyDateTimeFormatter
     }
 };
 
+struct MyDateTimeParser
+{
+    explicit MyDateTimeParser(String format_);
+
+    std::optional<UInt64> parseAsPackedUInt(const StringRef & str_view) const;
+
+    struct Context;
+
+private:
+    const String format;
+
+    // Parsing method. Parse from ctx.view[ctx.pos].
+    // If success, update `datetime`, `ctx` and return true.
+    // If fail, return false.
+    using ParserCallback = std::function<bool(MyDateTimeParser::Context & ctx, MyTimeBase & datetime)>;
+    std::vector<ParserCallback> parsers;
+};
+
 Field parseMyDateTime(const String & str, int8_t fsp = 6);
 
 void convertTimeZone(UInt64 from_time, UInt64 & to_time, const DateLUTImpl & time_zone_from, const DateLUTImpl & time_zone_to);

dbms/src/Common/StringUtils/StringRefUtils.h

@@ -0,0 +1,30 @@
+#pragma once
+
+#include <Common/StringUtils/StringUtils.h>
+#include <common/StringRef.h>
+
+inline bool startsWith(const StringRef & view, const StringRef & prefix)
+{
+    return detail::startsWith(view.data, view.size, prefix.data, prefix.size);
+}
+
+// case insensitive version of startsWith
+inline bool startsWithCI(const StringRef & view, const StringRef & prefix)
+{
+    return detail::startsWithCI(view.data, view.size, prefix.data, prefix.size);
+}
+
+inline bool endsWith(const StringRef & view, const char * prefix)
+{
+    return detail::endsWith(view.data, view.size, prefix, strlen(prefix)); //
+}
+
+// case insensitive version of endsWith
+inline bool endsWithCI(const StringRef & view, const char * prefix)
+{
+    return detail::endsWithCI(view.data, view.size, prefix, strlen(prefix));
+}
+
+// n - number of characters to remove from the start of the view,
+//     The behavior is undefined if `n > view.size`
+inline StringRef removePrefix(const StringRef & view, size_t n) { return StringRef{view.data + n, view.size - n}; }

dbms/src/Common/StringUtils/StringUtils.cpp

@@ -1,16 +1,44 @@
 #include "StringUtils.h"
 
+#include <cctype>
+
 namespace detail
 {
 
-bool startsWith(const std::string & s, const char * prefix, size_t prefix_size)
+bool startsWith(const char * s, size_t size, const char * prefix, size_t prefix_size)
+{
+    return size >= prefix_size && 0 == memcmp(s, prefix, prefix_size);
+}
+
+bool endsWith(const char * s, size_t size, const char * suffix, size_t suffix_size)
 {
-    return s.size() >= prefix_size && 0 == memcmp(s.data(), prefix, prefix_size);
+    return size >= suffix_size && 0 == memcmp(s + size - suffix_size, suffix, suffix_size);
 }
 
-bool endsWith(const std::string & s, const char * suffix, size_t suffix_size)
+bool startsWithCI(const char * s, size_t size, const char * prefix, size_t prefix_size)
 {
-    return s.size() >= suffix_size && 0 == memcmp(s.data() + s.size() - suffix_size, suffix, suffix_size);
+    if (size < prefix_size)
+        return false;
+    // case insensitive compare
+    for (size_t i = 0; i < prefix_size; ++i)
+    {
+        if (std::tolower(s[i]) != std::tolower(prefix[i]))
+            return false;
+    }
+    return true;
 }
 
+bool endsWithCI(const char * s, size_t size, const char * suffix, size_t suffix_size)
+{
+    if (size < suffix_size)
+        return false;
+    // case insensitive compare
+    for (size_t i = 0; i < suffix_size; ++i)
+    {
+        if (std::tolower(s[i]) != std::tolower(suffix[i]))
+            return false;
+    }
+    return true;
 }
+
+} // namespace detail

dbms/src/Common/StringUtils/StringUtils.h

@@ -1,38 +1,43 @@
 #pragma once
 
 #include <string>
+#include <string_view>
 #include <cstring>
 #include <cstddef>
 
 
 namespace detail
 {
-    bool startsWith(const std::string & s, const char * prefix, size_t prefix_size);
-    bool endsWith(const std::string & s, const char * suffix, size_t suffix_size);
-}
+bool startsWith(const char * s, size_t size, const char * prefix, size_t prefix_size);
+bool endsWith(const char * s, size_t size, const char * suffix, size_t suffix_size);
+
+// case insensitive version
+bool startsWithCI(const char * s, size_t size, const char * prefix, size_t prefix_size);
+bool endsWithCI(const char * s, size_t size, const char * suffix, size_t suffix_size);
+} // namespace detail
 
 
 inline bool startsWith(const std::string & s, const std::string & prefix)
 {
-    return detail::startsWith(s, prefix.data(), prefix.size());
+    return detail::startsWith(s.data(), s.size(), prefix.data(), prefix.size());
 }
 
 inline bool endsWith(const std::string & s, const std::string & suffix)
 {
-    return detail::endsWith(s, suffix.data(), suffix.size());
+    return detail::endsWith(s.data(), s.size(), suffix.data(), suffix.size());
 }
 
 
 /// With GCC, strlen is evaluated compile time if we pass it a constant
 /// string that is known at compile time.
 inline bool startsWith(const std::string & s, const char * prefix)
 {
-    return detail::startsWith(s, prefix, strlen(prefix));
+    return detail::startsWith(s.data(), s.size(), prefix, strlen(prefix));
 }
 
 inline bool endsWith(const std::string & s, const char * suffix)
 {
-    return detail::endsWith(s, suffix, strlen(suffix));
+    return detail::endsWith(s.data(), s.size(), suffix, strlen(suffix)); //
 }
 
 /// Given an integer, return the adequate suffix for