Make it compile

simdjson · jkeiser · Aug 2, 2023 · Aug 2, 2023 · Aug 2, 2023 · Aug 2, 2023
commit 76dd137c87bf489cb536a6dea4e29868eec21e34
diff --git a/include/simdjson/generic/lookup_table.h b/include/simdjson/generic/lookup_table.h
@@ -218,7 +218,7 @@ struct low_nibble_lookup {
   simdjson_inline simd8<uint8_t> operator[](const simd8<uint8_t>& keys) const noexcept { return lookup(keys); }
   /** Look up the value corresponding the lower 4 bits of each input byte, and return it. */
   simdjson_inline simd8<uint8_t> lookup(const simd8<uint8_t>& keys) const noexcept {
-    return keys.lookup_low_nibble_ascii(keys);
+    return keys.lookup_low_nibble_ascii(table);
   }
   /**
    * Look up the value in the table. Behavior is system-dependent for indexes greater than 16.
@@ -227,15 +227,15 @@ struct low_nibble_lookup {
    * - On platforms like Intel, index bits 4-6 will be ignored, but if the high bit is set, it
    *   will not match anything in the table. greater than 16 will be ignored, *except* if the high bit is 1,
    */
-  simdjson_inline simd8<uint8_t> lookup_unsafe(const simd8<uint8_t>& shifted_keys) const noexcept {
-    return shifted_keys.lookup_16(table);
+  simdjson_inline simd8<uint8_t> lookup_unsafe(const simd8<uint8_t>& keys) const noexcept {
+    return keys.lookup_16(table);
   }
 
   /** Look up the value corresponding the lower 4 bits of each input byte, and return it. */
   simdjson_inline simd8x64<uint8_t> operator[](const simd8x64<uint8_t>& keys) const noexcept { return lookup(keys); }
   /** Look up the value corresponding the lower 4 bits of each input byte, and return it. */
   simdjson_inline simd8x64<uint8_t> lookup(const simd8x64<uint8_t>& keys) const noexcept {
-    return keys.lookup_low_nibble_ascii(keys);
+    return keys.lookup_low_nibble_ascii(table);
   }
   /**
    * Look up the value in the table. Behavior is system-dependent for indexes greater than 16.

diff --git a/include/simdjson/westmere/begin.h b/include/simdjson/westmere/begin.h
@@ -6,7 +6,6 @@
 SIMDJSON_TARGET_REGION("sse4.2,pclmul,popcnt")
 #endif
 
-#include "simdjson/westmere/bitmask.h"
 #include "simdjson/westmere/bitmask.h"
 #include "simdjson/westmere/numberparsing_defs.h"
 #include "simdjson/westmere/simd.h"

diff --git a/include/simdjson/westmere/bitmask.h b/include/simdjson/westmere/bitmask.h
@@ -10,6 +10,11 @@ namespace simdjson {
 namespace westmere {
 namespace bitmask {
 
+simdjson_constinit uint64_t ALL  = 0xFFFFFFFFFFFFFFFF;
+simdjson_constinit uint64_t NONE = 0xFFFFFFFFFFFFFFFF;
+simdjson_constinit uint64_t EVEN = 0x5555555555555555;
+simdjson_constinit uint64_t ODD  = 0xAAAAAAAAAAAAAAAA;
+
 // We sometimes call trailing_zero on inputs that are zero,
 // but the algorithms do not end up using the returned value.
 // Sadly, sanitizers are not smart enough to figure it out.

diff --git a/src/generic/amalgamated.h b/src/generic/amalgamated.h
@@ -4,4 +4,3 @@
 
 #include <generic/base.h>
 #include <generic/dom_parser_implementation.h>
-#include <generic/json_character_block.h>
diff --git a/src/generic/base.h b/src/generic/base.h
@@ -10,8 +10,6 @@ namespace simdjson {
 namespace SIMDJSON_IMPLEMENTATION {
 namespace {
 
-struct json_character_block;
-
 } // unnamed namespace
 } // namespace SIMDJSON_IMPLEMENTATION
 } // namespace simdjson

diff --git a/src/generic/json_character_block.h b/src/generic/json_character_block.h
diff --git a/src/generic/stage1/base.h b/src/generic/stage1/base.h
@@ -13,10 +13,8 @@ namespace stage1 {
 class bit_indexer;
 template<size_t STEP_SIZE>
 struct buf_block_reader;
-struct json_block;
 class json_minifier;
 class json_scanner;
-struct json_string_block;
 class json_string_scanner;
 class json_structural_indexer;
 

diff --git a/src/generic/stage1/json_minifier.h b/src/generic/stage1/json_minifier.h
@@ -28,15 +28,14 @@ class json_minifier {
   {}
   template<size_t STEP_SIZE>
   simdjson_inline void step(const uint8_t *block_buf, buf_block_reader<STEP_SIZE> &reader) noexcept;
-  simdjson_inline void next(const simd::simd8x64<uint8_t>& in, const json_block& block);
+  simdjson_inline void next(const simd::simd8x64<uint8_t>& in, uint64_t whitespace);
   simdjson_inline error_code finish(uint8_t *dst_start, size_t &dst_len);
   json_scanner scanner{};
   uint8_t *dst;
 };
 
-simdjson_inline void json_minifier::next(const simd::simd8x64<uint8_t>& in, const json_block& block) {
-  uint64_t mask = block.whitespace();
-  dst += in.compress(mask, dst);
+simdjson_inline void json_minifier::next(const simd::simd8x64<uint8_t>& in, uint64_t ws) {
+  dst += in.compress(ws, dst);
 }
 
 simdjson_inline error_code json_minifier::finish(uint8_t *dst_start, size_t &dst_len) {
@@ -50,18 +49,18 @@ template<>
 simdjson_inline void json_minifier::step<128>(const uint8_t *block_buf, buf_block_reader<128> &reader) noexcept {
   simd::simd8x64<uint8_t> in_1(block_buf);
   simd::simd8x64<uint8_t> in_2(block_buf+64);
-  json_block block_1 = scanner.next(in_1);
-  json_block block_2 = scanner.next(in_2);
-  this->next(in_1, block_1);
-  this->next(in_2, block_2);
+  uint64_t ws_1 = scanner.next_whitespace(in_1);
+  uint64_t ws_2 = scanner.next_whitespace(in_2);
+  this->next(in_1, ws_1);
+  this->next(in_2, ws_2);
   reader.advance();
 }
 
 template<>
 simdjson_inline void json_minifier::step<64>(const uint8_t *block_buf, buf_block_reader<64> &reader) noexcept {
   simd::simd8x64<uint8_t> in_1(block_buf);
-  json_block block_1 = scanner.next(in_1);
-  this->next(block_buf, block_1);
+  uint64_t ws_1 = scanner.next(in_1);
+  this->next(block_buf, ws_1);
   reader.advance();
 }
 

diff --git a/src/generic/stage1/json_scanner.h b/src/generic/stage1/json_scanner.h
@@ -4,7 +4,6 @@
 #ifndef SIMDJSON_CONDITIONAL_INCLUDE
 #define SIMDJSON_SRC_GENERIC_STAGE1_JSON_SCANNER_H
 #include <generic/stage1/base.h>
-#include <generic/json_character_block.h>
 #include <generic/stage1/json_string_scanner.h>
 #include <simdjson/generic/lookup_table.h>
 #endif // SIMDJSON_CONDITIONAL_INCLUDE
@@ -31,6 +30,7 @@ class json_scanner {
 public:
   json_scanner() = default;
   simdjson_inline uint64_t next(const simd::simd8x64<uint8_t>& in) noexcept;
+  simdjson_inline uint64_t next_whitespace(const simd::simd8x64<uint8_t>& in) noexcept;
 
   // Returns either UNCLOSED_STRING or SUCCESS
   simdjson_inline error_code finish() const noexcept;
@@ -64,14 +64,14 @@ class json_scanner {
     { '\\', BACKSLASH },
   };
   static simdjson_constinit low_nibble_lookup WHITESPACE_MATCH = {
-    { ' '-1, 0xFF },
-    { '\t'-1, 0xFF },
-    { '\r'-1, 0xFF },
-    { '\n'-1, 0xFF },
+    { ' ', ' ' },
+    { '\t', '\t' },
+    { '\r', '\r' },
+    { '\n', '\n' },
   };
 
   simdjson_inline uint64_t next_separated_values(uint64_t sep_open, uint64_t scalar_close) noexcept;
-  simdjson_inline void check_errors(uint64_t scalar, uint64_t sep, uint64_t open, uint64_t raw_quote, uint64_t separated_values, uint64_t in_string) noexcept;
+  simdjson_inline void check_errors(uint64_t scalar, uint64_t ctrl, uint64_t sep, uint64_t open, uint64_t raw_quote, uint64_t separated_values, uint64_t in_string) noexcept;
 
   // Whether the last character of the previous iteration is part of a scalar token
   // (anything except whitespace or a structural character/'operator').
@@ -100,7 +100,7 @@ simdjson_inline uint64_t json_scanner::next(
   uint64_t backslash = in.eq('\\'); // 3+LN (LN+simd:N)
   uint64_t raw_quote = in.eq('"');  // 3+LN (LN+simd:N)
   uint64_t in_string = string_scanner.next(backslash, raw_quote, separated_values); // 10+LN (+6) or (14+LN or 18+LN (+8+simd:3))
-  // critical path = 10+LN (+6+2LN+simd:2N) or (14+LN or 18+LN (+8+2LN+simd:2N+3))
+  // critical path = 10+LN (+6+2LN+simd:2N) or (14+LN or 18+LN) (+8+2LN+simd:2N+3)
 
   uint64_t lead_value = scalar_close & separated_values;    // 8+LN (+1)
   // uint64_t op_without_comma = colon | open | close;         // 8+LN (+1) (ternary)
@@ -111,14 +111,16 @@ simdjson_inline uint64_t json_scanner::next(
   // critical path = 11+LN or (15+LN or 19+LN) (+3)
 
   uint64_t scalar = scalar_close & ~close;                  // 8+LN (+1)
-  check_errors(scalar, sep, open, raw_quote, separated_values, in_string); // 14+LN (+9)
-  // critical path = 14+LN (+10)
+  uint64_t ws = in.eq(WHITESPACE_MATCH.lookup(in));         // 6+LN (+LN+simd:2N)
+  uint64_t ctrl = ws_ctrl & ~ws;                            // 7+LN (+1)
+  check_errors(scalar, ctrl, sep, open, raw_quote, separated_values, in_string); // 14+LN (+9)
+  // critical path = 14+LN (+11+LN+simd:2N)
 
   return structurals;
-  // structurals: critical path = 11+LN (+24+7LN+simd:8N) or (15+LN or 19+LN) (+26+7LN+simd:8N+3)
-  // = icelake:  11 (24+simd:8 ) or (15 or 19) (26+simd:11)
-  // = haswell:  12 (31+simd:16) or (16 or 21) (33+simd:19)
-  // = westmere: 13 (38+simd:32) or (17 or 22) (40+simd:35)
+  // structurals: critical path = 11+LN (+25+8LN+simd:10N) or (15+LN or 19+LN) (+27+8LN+simd:10N+3)
+  // = icelake:  11 (24+simd:10) or (15 or 19) (27+simd:13)
+  // = haswell:  12 (31+simd:20) or (16 or 21) (35+simd:23)
+  // = westmere: 13 (38+simd:40) or (17 or 22) (43+simd:43)
 }
 
 simdjson_inline uint64_t json_scanner::next_separated_values(
@@ -138,6 +140,7 @@ simdjson_inline uint64_t json_scanner::next_separated_values(
 
 simdjson_inline void json_scanner::check_errors(
   uint64_t scalar,           // 8+LN
+  uint64_t ctrl,             // 7+LN
   uint64_t sep,              // 4+LN
   uint64_t open,             // 6+LN
   uint64_t raw_quote,        // 3+LN
@@ -164,7 +167,7 @@ simdjson_inline void json_scanner::check_errors(
   uint64_t raw_separator_error = missing_separator_error | extra_separator_error | missing_separator_before_open_error;
   // flip lead quote off and trail quote on: lead quote errors
   uint64_t separator_error = raw_separator_error & (in_string ^ raw_quote); // 13+LN (+1) (ternary)
-  this->error |= separator_error;                                           // 14+LN (+1)
+  this->error |= separator_error | ctrl;                                    // 14+LN (+1) (ternary)
   // critical path = 14+LN (+3)
 
   // NOT validated:
@@ -178,10 +181,37 @@ simdjson_inline void json_scanner::check_errors(
   // critical path = 14+LN (+9)
 }
 
+simdjson_inline uint64_t json_scanner::next_whitespace(
+  const simd::simd8x64<uint8_t>& in
+) noexcept {
+  simd8x64<uint8_t> curlified = in | ('{' - '[');   // 3    (+simd:N)
+  uint64_t open   = curlified.eq('{');              // 6+LN (+LN+simd:N)
+  uint64_t close  = curlified.eq('}');              // 6+LN (+LN+simd:N)
+  uint64_t comma = in.eq(',');                      // 3+LN (+LN+simd:N)
+  uint64_t colon = in.eq(':');                      // 3+LN (+LN+simd:N)
+  uint64_t sep   = comma | colon;                   // 4+LN (+1)
+  uint64_t ws = in.eq(WHITESPACE_MATCH.lookup(in)); // 6+LN (+LN+simd:2N)
+  uint64_t sep_open = sep | open;                   // 7+LN (+1)
+  uint64_t scalar_close = ~sep & ~open & ~ws;       // 7+LN (+1) (ternary)
+  // total 7+LN (+3+5LN+simd:7N)
+
+  uint64_t separated_values = next_separated_values(sep_open, scalar_close); // 8+LN (+2)
+
+  uint64_t backslash = in.eq('\\'); // 3+LN (LN+simd:N)
+  uint64_t raw_quote = in.eq('"');  // 3+LN (LN+simd:N)
+  uint64_t in_string = string_scanner.next(backslash, raw_quote, separated_values); // 10+LN (+6) or (14+LN or 18+LN (+8+simd:3))
+  // total 10+LN (+6+2LN+simd:2N) or (14+LN or 18+LN) (+8+2LN+simd:2N+3)
+
+  return ws & ~in_string;
+  // critical path = 10+LN (+11+7LN+simd:9N) or (14+LN or 18+LN) (+13+7LN+simd:9N+3)
+}
+
+
 simdjson_inline error_code json_scanner::finish() const noexcept {
   if (this->error | this->string_scanner.finish()) {
     return TAPE_ERROR;
   }
+  return SUCCESS;
 }
 
 } // namespace stage1

diff --git a/src/generic/stage1/json_structural_indexer.h b/src/generic/stage1/json_structural_indexer.h
@@ -140,14 +140,13 @@ class json_structural_indexer {
   simdjson_inline json_structural_indexer(uint32_t *structural_indexes);
   template<size_t STEP_SIZE>
   simdjson_inline void step(const uint8_t *block, buf_block_reader<STEP_SIZE> &reader) noexcept;
-  simdjson_inline void next(const simd::simd8x64<uint8_t>& in, const json_block& block, size_t idx);
+  simdjson_inline void next(const simd::simd8x64<uint8_t>& in, uint64_t structurals, size_t idx);
   simdjson_inline error_code finish(dom_parser_implementation &parser, size_t idx, size_t len, stage1_mode partial);
 
   json_scanner scanner{};
   utf8_checker checker{};
   bit_indexer indexer;
   uint64_t prev_structurals = 0;
-  uint64_t unescaped_chars_error = 0;
 };
 
 simdjson_inline json_structural_indexer::json_structural_indexer(uint32_t *structural_indexes) : indexer{structural_indexes} {}
@@ -221,29 +220,28 @@ template<>
 simdjson_inline void json_structural_indexer::step<128>(const uint8_t *block, buf_block_reader<128> &reader) noexcept {
   simd::simd8x64<uint8_t> in_1(block);
   simd::simd8x64<uint8_t> in_2(block+64);
-  json_block block_1 = scanner.next(in_1);
-  json_block block_2 = scanner.next(in_2);
-  this->next(in_1, block_1, reader.block_index());
-  this->next(in_2, block_2, reader.block_index()+64);
+  uint64_t structurals_1 = scanner.next(in_1);
+  uint64_t structurals_2 = scanner.next(in_2);
+  this->next(in_1, structurals_1, reader.block_index());
+  this->next(in_2, structurals_2, reader.block_index()+64);
   reader.advance();
 }
 
 template<>
 simdjson_inline void json_structural_indexer::step<64>(const uint8_t *block, buf_block_reader<64> &reader) noexcept {
   simd::simd8x64<uint8_t> in_1(block);
-  json_block block_1 = scanner.next(in_1);
-  this->next(in_1, block_1, reader.block_index());
+  uint64_t structurals_1 = scanner.next(in_1);
+  this->next(in_1, structurals_1, reader.block_index());
   reader.advance();
 }
 
-simdjson_inline void json_structural_indexer::next(const simd::simd8x64<uint8_t>& in, const json_block& block, size_t idx) {
+simdjson_inline void json_structural_indexer::next(const simd::simd8x64<uint8_t>& in,uint64_t structurals, size_t idx) {
   uint64_t unescaped = in.lteq(0x1F);
 #if SIMDJSON_UTF8VALIDATION
   checker.check_next_input(in);
 #endif
   indexer.write(uint32_t(idx-64), prev_structurals); // Output *last* iteration's structurals to the parser
-  prev_structurals = block.structural_start();
-  unescaped_chars_error |= block.non_quote_inside_string(unescaped);
+  prev_structurals = structurals;
 }
 
 simdjson_inline error_code json_structural_indexer::finish(dom_parser_implementation &parser, size_t idx, size_t len, stage1_mode partial) {
@@ -258,9 +256,6 @@ simdjson_inline error_code json_structural_indexer::finish(dom_parser_implementa
   const bool have_unclosed_string = (error == UNCLOSED_STRING);
   if (simdjson_unlikely(should_we_exit)) { return error; }
 
-  if (unescaped_chars_error) {
-    return UNESCAPED_CHARS;
-  }
   parser.n_structural_indexes = uint32_t(indexer.tail - parser.structural_indexes.get());
   /***
    * The On Demand API requires special padding.

diff --git a/src/haswell.cpp b/src/haswell.cpp
@@ -38,53 +38,6 @@ namespace {
 
 using namespace simd;
 
-enum class ws_op_t {
-  COMMA     = 1 << 0,
-  COLON     = 1 << 1,
-  OPEN      = 1 << 2,
-  CLOSE     = 1 << 3,
-  QUOTE     = 1 << 4,
-  BACKSLASH = 1 << 5,
-  SPACE     = 1 << 6,
-  TAB_CR_LF = 1 << 7,
-  OP = COMMA | COLON | OPEN | CLOSE,
-  SEPARATOR = COMMA | COLON,
-  WHITESPACE = SPACE | TAB_CR_LF,
-};
-
-struct json_character_block {
-  simd8x64<uint8_t> classified;
-
-  static simdjson_constinit byte_classifier WS_MATCH {
-    { ',',  ws_op_t::COMMA },
-    { ':',  ws_op_t::COLON },
-    { '[',  ws_op_t::OPEN },
-    { '{',  ws_op_t::OPEN },
-    { ']',  ws_op_t::CLOSE },
-    { '}',  ws_op_t::CLOSE },
-    { '\"', ws_op_t::QUOTE },
-    { ' ',  ws_op_t::SPACE },
-    { '\t', ws_op_t::TAB_CR_LF },
-    { '\r', ws_op_t::TAB_CR_LF },
-    { '\n', ws_op_t::TAB_CR_LF },
-    { '\\', ws_op_t::BACKSLASH },
-  };
-
-  // These are all mutually exclusive, so we can use > to
-  simdjson_inline uint64_t separators_or_open() const noexcept { return ; }
-  simdjson_inline uint64_t separators_or_open() const noexcept { return classified.any_bits_set(ws_op_t::OP); }
-  simdjson_inline uint64_t operators() const noexcept { return classified.any_bits_set(ws_op_t::OP); }
-  simdjson_inline uint64_t whitespace() const noexcept { return classified.any_bits_set(ws_op_t::WHITESPACE); }
-};
-
-// This identifies structural characters (comma, colon, braces, brackets),
-// and ASCII white-space ('\r','\n','\t',' ').
-simdjson_inline json_character_block json_character_block::classify(const simd::simd8x64<uint8_t>& in) {
-  // Turn [ and ] into { and }
-  uint64_t sep_and_open = in.
-  return { in.eq(WS_MATCH[in]), curlified.eq(OP_MATCH[in]) };
-}
-
 simdjson_inline bool is_ascii(const simd8x64<uint8_t>& in) {
   return in.reduce_or().is_ascii();
 }
Original file line number	Diff line number	Diff line change
Expand Up		@@ -4,4 +4,3 @@

		#include <generic/base.h>
		#include <generic/dom_parser_implementation.h>
		#include <generic/json_character_block.h>