ARROW-6825: [C++] Rework CSV reader IO around readahead iterator

Make the delimiting chunker a common facility used by CSV and JSON.

Closes #5727 from pitrou/ARROW-6825-delimited and squashes the following commits:

68a5a02 <Antoine Pitrou> Address review comments
394fecc <Antoine Pitrou> ARROW-6825:  Rework CSV reader IO around readahead iterator

Authored-by: Antoine Pitrou <antoine@python.org>
Signed-off-by: Benjamin Kietzman <bengilgit@gmail.com>

cpp/src/arrow/CMakeLists.txt

@@ -121,14 +121,14 @@ set(ARROW_SRCS
     io/hdfs_internal.cc
     io/interfaces.cc
     io/memory.cc
-    io/readahead.cc
     io/slow.cc
     testing/util.cc
     util/basic_decimal.cc
     util/bit_util.cc
     util/compression.cc
     util/cpu_info.cc
     util/decimal.cc
+    util/delimiting.cc
     util/formatting.cc
     util/int_util.cc
     util/io_util.cc

cpp/src/arrow/csv/chunker.cc

@@ -18,168 +18,248 @@
 #include "arrow/csv/chunker.h"
 
 #include <cstdint>
+#include <memory>
+#include <utility>
 
 #include "arrow/status.h"
 #include "arrow/util/logging.h"
+#include "arrow/util/stl.h"
+#include "arrow/util/string_view.h"
 
 namespace arrow {
 namespace csv {
 
 namespace {
 
-// Find the last newline character in the given data block.
-// nullptr is returned if not found (like memchr()).
-const char* FindNewlineReverse(const char* data, uint32_t size) {
-  if (size == 0) {
-    return nullptr;
-  }
-  const char* s = data + size - 1;
-  while (size > 0) {
-    if (*s == '\r' || *s == '\n') {
-      return s;
-    }
-    --s;
-    --size;
-  }
-  return nullptr;
-}
+// NOTE: csvmonkey (https://github.com/dw/csvmonkey) has optimization ideas
 
-}  // namespace
+template <bool quoting, bool escaping>
+class Lexer {
+ public:
+  enum State {
+    FIELD_START,
+    IN_FIELD,
+    AT_ESCAPE,
+    IN_QUOTED_FIELD,
+    AT_QUOTED_QUOTE,
+    AT_QUOTED_ESCAPE
+  };
 
-Chunker::Chunker(ParseOptions options) : options_(options) {}
+  explicit Lexer(const ParseOptions& options) : options_(options) {
+    DCHECK_EQ(quoting, options_.quoting);
+    DCHECK_EQ(escaping, options_.escaping);
+  }
 
-// NOTE: cvsmonkey (https://github.com/dw/csvmonkey) has optimization ideas
+  const char* ReadLine(const char* data, const char* data_end) {
+    // The parsing state machine
+    char c;
+    DCHECK_GT(data_end - data, 0);
+    if (ARROW_PREDICT_TRUE(state_ == FIELD_START)) {
+      goto FieldStart;
+    }
+    switch (state_) {
+      case FIELD_START:
+        goto FieldStart;
+      case IN_FIELD:
+        goto InField;
+      case AT_ESCAPE:
+        goto AtEscape;
+      case IN_QUOTED_FIELD:
+        goto InQuotedField;
+      case AT_QUOTED_QUOTE:
+        goto AtQuotedQuote;
+      case AT_QUOTED_ESCAPE:
+        goto AtQuotedEscape;
+    }
 
-template <bool quoting, bool escaping>
-inline const char* Chunker::ReadLine(const char* data, const char* data_end) {
-  DCHECK_EQ(quoting, options_.quoting);
-  DCHECK_EQ(escaping, options_.escaping);
+  FieldStart:
+    // At the start of a field
+    if (ARROW_PREDICT_FALSE(data == data_end)) {
+      state_ = FIELD_START;
+      goto AbortLine;
+    }
+    // Quoting is only recognized at start of field
+    if (quoting && *data == options_.quote_char) {
+      data++;
+      goto InQuotedField;
+    } else {
+      goto InField;
+    }
 
-  // The parsing state machine
-  char c;
+  InField:
+    // Inside a non-quoted part of a field
+    if (ARROW_PREDICT_FALSE(data == data_end)) {
+      state_ = IN_FIELD;
+      goto AbortLine;
+    }
+    c = *data++;
+    if (escaping && ARROW_PREDICT_FALSE(c == options_.escape_char)) {
+      if (ARROW_PREDICT_FALSE(data == data_end)) {
+        state_ = AT_ESCAPE;
+        goto AbortLine;
+      }
+      data++;
+      goto InField;
+    }
+    if (ARROW_PREDICT_FALSE(c == '\r')) {
+      if (ARROW_PREDICT_TRUE(data != data_end) && *data == '\n') {
+        data++;
+      }
+      goto LineEnd;
+    }
+    if (ARROW_PREDICT_FALSE(c == '\n')) {
+      goto LineEnd;
+    }
+    if (ARROW_PREDICT_FALSE(c == options_.delimiter)) {
+      goto FieldEnd;
+    }
+    goto InField;
 
-FieldStart:
-  // At the start of a field
-  // Quoting is only recognized at start of field
-  if (quoting && ARROW_PREDICT_TRUE(data != data_end) && *data == options_.quote_char) {
+  AtEscape:
+    // Coming here if last block ended on a non-quoted escape
     data++;
-    goto InQuotedField;
-  } else {
     goto InField;
-  }
 
-InField:
-  // Inside a non-quoted part of a field
-  if (ARROW_PREDICT_FALSE(data == data_end)) {
-    goto AbortLine;
-  }
-  c = *data++;
-  if (escaping && ARROW_PREDICT_FALSE(c == options_.escape_char)) {
+  InQuotedField:
+    // Inside a quoted part of a field
     if (ARROW_PREDICT_FALSE(data == data_end)) {
+      state_ = IN_QUOTED_FIELD;
       goto AbortLine;
     }
-    data++;
-    goto InField;
-  }
-  if (ARROW_PREDICT_FALSE(c == '\r')) {
-    if (ARROW_PREDICT_TRUE(data != data_end) && *data == '\n') {
+    c = *data++;
+    if (escaping && ARROW_PREDICT_FALSE(c == options_.escape_char)) {
+      if (ARROW_PREDICT_FALSE(data == data_end)) {
+        state_ = AT_QUOTED_ESCAPE;
+        goto AbortLine;
+      }
       data++;
+      goto InQuotedField;
     }
-    goto LineEnd;
-  }
-  if (ARROW_PREDICT_FALSE(c == '\n')) {
-    goto LineEnd;
-  }
-  if (ARROW_PREDICT_FALSE(c == options_.delimiter)) {
-    goto FieldEnd;
-  }
-  goto InField;
-
-InQuotedField:
-  // Inside a quoted part of a field
-  if (ARROW_PREDICT_FALSE(data == data_end)) {
-    goto AbortLine;
-  }
-  c = *data++;
-  if (escaping && ARROW_PREDICT_FALSE(c == options_.escape_char)) {
-    if (data == data_end) {
-      goto AbortLine;
+    if (ARROW_PREDICT_FALSE(c == options_.quote_char)) {
+      if (ARROW_PREDICT_FALSE(data == data_end)) {
+        state_ = AT_QUOTED_QUOTE;
+        goto AbortLine;
+      }
+      if (options_.double_quote && *data == options_.quote_char) {
+        // Double-quoting
+        data++;
+      } else {
+        // End of single-quoting
+        goto InField;
+      }
     }
+    goto InQuotedField;
+
+  AtQuotedEscape:
+    // Coming here if last block ended on a quoted escape
     data++;
     goto InQuotedField;
-  }
-  if (ARROW_PREDICT_FALSE(c == options_.quote_char)) {
-    if (options_.double_quote && data != data_end && *data == options_.quote_char) {
+
+  AtQuotedQuote:
+    // Coming here if last block ended on a quoted quote
+    if (options_.double_quote && *data == options_.quote_char) {
       // Double-quoting
       data++;
+      goto InQuotedField;
     } else {
       // End of single-quoting
       goto InField;
     }
-  }
-  goto InQuotedField;
 
-FieldEnd:
-  // At the end of a field
-  goto FieldStart;
+  FieldEnd:
+    // At the end of a field
+    goto FieldStart;
 
-LineEnd:
-  return data;
+  LineEnd:
+    return data;
 
-AbortLine:
-  // Truncated line at end of block
-  return nullptr;
-}
+  AbortLine:
+    // Truncated line
+    return nullptr;
+  }
 
+ protected:
+  const ParseOptions& options_;
+  State state_ = FIELD_START;
+};
+
+// A BoundaryFinder implementation that assumes CSV cells can contain raw newlines,
+// and uses actual CSV lexing to delimit them.
 template <bool quoting, bool escaping>
-Status Chunker::ProcessSpecialized(const char* start, uint32_t size, uint32_t* out_size) {
-  DCHECK_EQ(quoting, options_.quoting);
-  DCHECK_EQ(escaping, options_.escaping);
+class LexingBoundaryFinder : public BoundaryFinder {
+ public:
+  explicit LexingBoundaryFinder(ParseOptions options) : options_(std::move(options)) {}
+
+  Status FindFirst(util::string_view partial, util::string_view block,
+                   int64_t* out_pos) override {
+    Lexer<quoting, escaping> lexer(options_);
 
-  const char* data = start;
-  const char* data_end = start + size;
+    const char* line_end =
+        lexer.ReadLine(partial.data(), partial.data() + partial.size());
+    DCHECK_EQ(line_end, nullptr);  // Otherwise `partial` is a whole CSV line
+    line_end = lexer.ReadLine(block.data(), block.data() + block.size());
 
-  while (data < data_end) {
-    const char* line_end = ReadLine<quoting, escaping>(data, data_end);
     if (line_end == nullptr) {
-      // Cannot read any further
-      break;
+      // No complete CSV line
+      *out_pos = -1;
+    } else {
+      *out_pos = static_cast<int64_t>(line_end - block.data());
+      DCHECK_GT(*out_pos, 0);
     }
-    data = line_end;
+    return Status::OK();
   }
-  *out_size = static_cast<uint32_t>(data - start);
-  return Status::OK();
-}
 
-Status Chunker::Process(const char* start, uint32_t size, uint32_t* out_size) {
-  if (!options_.newlines_in_values) {
-    // In newlines are not accepted in CSV values, we can simply search for
-    // the last newline character.
-    // For common block sizes and CSV row sizes, this avoids reading
-    // most of the data block, making the chunker extremely fast compared
-    // to the rest of the CSV reading pipeline.
-    const char* nl = FindNewlineReverse(start, size);
-    if (nl == nullptr) {
-      *out_size = 0;
+  Status FindLast(util::string_view block, int64_t* out_pos) override {
+    Lexer<quoting, escaping> lexer(options_);
+
+    const char* data = block.data();
+    const char* const data_end = block.data() + block.size();
+
+    while (data < data_end) {
+      const char* line_end = lexer.ReadLine(data, data_end);
+      if (line_end == nullptr) {
+        // Cannot read any further
+        break;
+      }
+      DCHECK_GT(line_end, data);
+      data = line_end;
+    }
+    if (data == block.data()) {
+      // No complete CSV line
+      *out_pos = -1;
     } else {
-      *out_size = static_cast<uint32_t>(nl - start + 1);
+      *out_pos = static_cast<int64_t>(data - block.data());
+      DCHECK_GT(*out_pos, 0);
     }
     return Status::OK();
   }
 
-  if (options_.quoting) {
-    if (options_.escaping) {
-      return ProcessSpecialized<true, true>(start, size, out_size);
-    } else {
-      return ProcessSpecialized<true, false>(start, size, out_size);
-    }
+ protected:
+  ParseOptions options_;
+};
+
+}  // namespace
+
+std::unique_ptr<Chunker> MakeChunker(const ParseOptions& options) {
+  std::shared_ptr<BoundaryFinder> delimiter;
+  if (!options.newlines_in_values) {
+    delimiter = MakeNewlineBoundaryFinder();
   } else {
-    if (options_.escaping) {
-      return ProcessSpecialized<false, true>(start, size, out_size);
+    if (options.quoting) {
+      if (options.escaping) {
+        delimiter = std::make_shared<LexingBoundaryFinder<true, true>>(options);
+      } else {
+        delimiter = std::make_shared<LexingBoundaryFinder<true, false>>(options);
+      }
     } else {
-      return ProcessSpecialized<false, false>(start, size, out_size);
+      if (options.escaping) {
+        delimiter = std::make_shared<LexingBoundaryFinder<false, true>>(options);
+      } else {
+        delimiter = std::make_shared<LexingBoundaryFinder<false, false>>(options);
+      }
     }
   }
+  return internal::make_unique<Chunker>(std::move(delimiter));
 }
 
 }  // namespace csv