RubyGems - smarter_csv - Versions diffs - 1.16.4 → 1.17.0 - Mend

smarter_csv 1.16.4 → 1.17.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.

Files changed (46) hide show

checksums.yaml +4 -4
data/.rubocop.yml +10 -1
data/CHANGELOG.md +54 -0
data/Gemfile +10 -5
data/README.md +98 -14
data/TO_DO.md +109 -0
data/docs/_introduction.md +1 -0
data/docs/bad_row_quarantine.md +2 -1
data/docs/basic_read_api.md +6 -1
data/docs/basic_write_api.md +30 -0
data/docs/batch_processing.md +25 -0
data/docs/column_selection.md +1 -0
data/docs/data_transformations.md +1 -0
data/docs/examples.md +126 -0
data/docs/header_transformations.md +23 -0
data/docs/header_validations.md +1 -0
data/docs/history.md +1 -0
data/docs/instrumentation.md +2 -1
data/docs/migrating_from_csv.md +1 -0
data/docs/options.md +20 -18
data/docs/parsing_strategy.md +1 -0
data/docs/real_world_csv.md +51 -1
data/docs/releases/1.16.0/performance_notes.md +15 -15
data/docs/releases/1.17.0/benchmarks.md +121 -0
data/docs/releases/1.17.0/changes.md +161 -0
data/docs/releases/1.17.0/performance_notes.md +126 -0
data/docs/row_col_sep.md +21 -1
data/docs/ruby_csv_pitfalls.md +1 -0
data/docs/value_converters.md +24 -0
data/docs/warnings.md +141 -0
data/ext/smarter_csv/smarter_csv.c +98 -32
data/images/SmarterCSV_1.17.0_vs_RubyCSV_3.3.5_speedup.svg +106 -0
data/images/SmarterCSV_1.17.0_vs_previous_C-speedup.svg +181 -0
data/images/SmarterCSV_1.17.0_vs_previous_Rb-speedup.svg +179 -0
data/lib/smarter_csv/auto_detection.rb +215 -30
data/lib/smarter_csv/file_io.rb +2 -2
data/lib/smarter_csv/hash_transformations.rb +29 -13
data/lib/smarter_csv/parser.rb +42 -33
data/lib/smarter_csv/peekable_io.rb +453 -0
data/lib/smarter_csv/reader.rb +119 -23
data/lib/smarter_csv/reader_options.rb +61 -1
data/lib/smarter_csv/version.rb +1 -1
data/lib/smarter_csv.rb +40 -12
metadata +12 -5
data/TO_DO_v2.md +0 -14
data/ext/smarter_csv/Makefile +0 -270

data/ext/smarter_csv/smarter_csv.c CHANGED Viewed

@@ -134,24 +134,51 @@ static const rb_data_type_t parse_context_type = {
 };
 static VALUE unescape_quotes(char *str, long len, char quote_char, rb_encoding *encoding) {
-  char *buf = ALLOC_N(char, len);
-  long j = 0;
-  for (long i = 0; i < len; i++) {
-    if (str[i] == quote_char && i + 1 < len && str[i + 1] == quote_char) {
-      buf[j++] = quote_char;
-      i++; // skip second quote
-    } else {
-      buf[j++] = str[i];
+  // Fast path: scan for any doubled quote pair. If none present, the field has
+  // nothing to unescape — emit it directly via rb_enc_str_new and skip the
+  // temp buffer + byte-by-byte copy. memchr is SIMD-optimized; the scan cost
+  // is far less than the malloc/free pair this avoids.
+  char *p = str;
+  char *end = str + len;
+  while ((p = memchr(p, quote_char, end - p))) {
+    if (p + 1 < end && *(p + 1) == quote_char) goto needs_unescape;
+    p++;
+  }
+  return rb_enc_str_new(str, len, encoding);
+needs_unescape:
+  // Slow path: at least one doubled quote pair was found. Allocate a temp
+  // buffer and walk byte-by-byte, collapsing "" → ".
+  {
+    char *buf = ALLOC_N(char, len);
+    long j = 0;
+    for (long i = 0; i < len; i++) {
+      if (str[i] == quote_char && i + 1 < len && str[i + 1] == quote_char) {
+        buf[j++] = quote_char;
+        i++; // skip second quote
+      } else {
+        buf[j++] = str[i];
+      }
     }
+    VALUE out = rb_enc_str_new(buf, j, encoding);
+    xfree(buf);
+    return out;
   }
-  VALUE out = rb_enc_str_new(buf, j, encoding);
-  xfree(buf);
-  return out;
+}
+/* Helper: build the 2-element [elements, data_size] tuple returned by rb_parse_csv_line.
+ * Aligns this function's return shape with parse_csv_line_ruby and rb_parse_line_to_hash_ctx:
+ * data_size = -1 signals "unclosed quoted field — needs more data". */
+static inline VALUE make_parse_result(VALUE elements, long data_size) {
+  VALUE result = rb_ary_new_capa(2);
+  rb_ary_push(result, elements);
+  rb_ary_push(result, LONG2FIX(data_size));
+  return result;
 }
 static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quote_char, VALUE max_size, VALUE has_quotes_val, VALUE strip_ws_val, VALUE allow_escaped_quotes_val, VALUE quote_boundary_standard_val, VALUE row_sep_val) {
   if (RB_TYPE_P(line, T_NIL) == 1) {
-    return rb_ary_new();
+    return make_parse_result(rb_ary_new(), 0);
   }
   if (RB_TYPE_P(line, T_STRING) != 1) {
@@ -178,7 +205,7 @@ static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quot
   if (max_size != Qnil) {
     max_fields = NUM2INT(max_size);
     if (max_fields < 0) {
-      return rb_ary_new();
+      return make_parse_result(rb_ary_new(), 0);
     }
   }
@@ -237,7 +264,7 @@ static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quot
       rb_ary_push(elements, field);
     }
-    return elements;
+    return make_parse_result(elements, RARRAY_LEN(elements));
   }
   // === SLOW PATH: Quoted fields or multi-char separator ===
@@ -350,7 +377,13 @@ static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quot
   }
   if (in_quotes) {
-    rb_raise(eMalformedCSVError, "Unclosed quoted field detected in line: %s", StringValueCStr(line));
+    /* Unclosed quoted field at EOL: signal "needs more data" rather than raising.
+     * Aligns with parse_csv_line_ruby and rb_parse_line_to_hash_ctx, which both
+     * return data_size = -1 on this condition. The Reader's stitch loop consumes
+     * the signal: append the next physical line and re-parse, or raise MalformedCSV
+     * at EOF if the field never closes. The parser does not decide "ultimately
+     * malformed"; the caller does. */
+    return make_parse_result(rb_ary_new(), -1);
   }
   if ((max_fields < 0) || (element_count < max_fields)) {
@@ -384,7 +417,7 @@ static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quot
     rb_ary_push(elements, field);
   }
-  return elements;
+  return make_parse_result(elements, RARRAY_LEN(elements));
 }
 // Efficiently combine two arrays into a hash (replaces headers.zip(values).to_h)
@@ -484,6 +517,37 @@ static inline VALUE try_numeric_conversion(char *trim_start, long trimmed_len) {
   return Qundef;  // not numeric
 }
+/*
+ * leading_whitespace_len - byte length (1, 2, or 3) of the whitespace character at the start of
+ * `s` (with `len` bytes available), or 0 if `s` does not start with whitespace.
+ *
+ * "Whitespace" here matches Ruby's [[:space:]] / Rails' String#blank? — the Unicode White_Space
+ * set — so the C blank check stays consistent with the Ruby fallback path (hash_transformations).
+ * ASCII bytes are handled with a single comparison; the multibyte arms are only reached when a
+ * byte >= 0x80 appears, so all-ASCII fields pay nothing extra.
+ */
+static inline int leading_whitespace_len(const char *s, long len) {
+  if (len < 1) return 0;
+  unsigned char b0 = (unsigned char)s[0];
+  if (b0 == 0x20 || (b0 >= 0x09 && b0 <= 0x0D)) return 1;  // space (most common) then \t \n \v \f \r
+  if (b0 < 0x80) return 0;                                 // any other ASCII byte: not whitespace
+  if (len < 2) return 0;
+  unsigned char b1 = (unsigned char)s[1];
+  if (b0 == 0xC2 && (b1 == 0x85 || b1 == 0xA0)) return 2;  // U+0085 NEL, U+00A0 NBSP
+  if (len < 3) return 0;
+  unsigned char b2 = (unsigned char)s[2];
+  if (b0 == 0xE1 && b1 == 0x9A && b2 == 0x80) return 3;    // U+1680 OGHAM SPACE MARK
+  if (b0 == 0xE2) {
+    // U+2000..U+200A (E2 80 80..8A) — note: 0x8B is U+200B ZERO WIDTH SPACE, NOT whitespace.
+    // U+2028 LINE SEP (A8), U+2029 PARA SEP (A9), U+202F NARROW NBSP (AF), U+205F MMSP (E2 81 9F).
+    if (b1 == 0x80 && ((b2 >= 0x80 && b2 <= 0x8A) || b2 == 0xA8 || b2 == 0xA9 || b2 == 0xAF)) return 3;
+    if (b1 == 0x81 && b2 == 0x9F) return 3;
+    return 0;
+  }
+  if (b0 == 0xE3 && b1 == 0x80 && b2 == 0x80) return 3;    // U+3000 IDEOGRAPHIC SPACE
+  return 0;
+}
 /*
  * ================================================================================
  * Transformation options struct - passed to insert_field_into_hash to avoid
@@ -541,17 +605,16 @@ static inline __attribute__((always_inline)) bool insert_field_into_hash(
   VALUE key = get_key_for_index(element_count, opts->headers, opts->headers_len, opts->prefix_str);
   // 1. Empty/blank field handling
-  // Check if field is blank: either zero-length, or all whitespace characters.
-  // This matches Ruby's blank? behavior (BLANK_RE = /\A\s*\z/) which considers
-  // spaces, tabs, \r, \n, \v, \f as whitespace.
+  // A field is blank if it is zero-length or consists entirely of whitespace characters.
+  // "Whitespace" matches Ruby's BLANK_RE = /\A[[:space:]]*\z/ (and Rails' String#blank?) — the
+  // Unicode White_Space set — so this stays consistent with the Ruby fallback path.
   if (opts->remove_empty_values) {
     bool is_blank = true;
-    for (long i = 0; i < trimmed_len; i++) {
-      char c = trim_start[i];
-      if (c != ' ' && c != '\t' && c != '\r' && c != '\n' && c != '\v' && c != '\f') {
-        is_blank = false;
-        break;
-      }
+    long i = 0;
+    while (i < trimmed_len) {
+      int w = leading_whitespace_len(trim_start + i, trimmed_len - i);
+      if (w == 0) { is_blank = false; break; }
+      i += w;
     }
     if (is_blank) return false;  // skip blank value
   }
@@ -562,22 +625,25 @@ static inline __attribute__((always_inline)) bool insert_field_into_hash(
     return false;  // not a non-blank value
   }
-  // 2. String-based zero check — matches /\A0+(?:\.0+)?\z/
-  // Works independently of numeric conversion: "0", "00", "0.0", "00.00" etc.
+  // 2. String-based zero check — matches /\A[+-]?0+(?:\.0+)?\z/
+  // Works independently of numeric conversion: "0", "00", "0.0", "00.00", "+0", "-0.00" etc.
   // Outer quotes are stripped before this call, so the check applies equally
   // to quoted ("0") and unquoted (0) fields.
   if (opts->remove_zero_values) {
-    long i = 0;
-    // Must start with at least one '0'
-    if (trimmed_len > 0 && trim_start[0] == '0') {
+    char c0 = trim_start[0];  // trimmed_len > 0 guaranteed (zero-length handled above)
+    // Index i skips an optional leading sign; bail right away if the first byte can't begin a zero.
+    long i = (c0 == '0') ? 0
+           : (c0 == '+' || c0 == '-') ? 1
+           : trimmed_len;
+    if (i < trimmed_len && trim_start[i] == '0') {
       while (i < trimmed_len && trim_start[i] == '0') i++;
-      if (i == trimmed_len) return false;  // all zeros, e.g. "0", "00"
+      if (i == trimmed_len) return false;  // all zeros, e.g. "0", "00", "+0", "-00"
       if (trim_start[i] == '.') {
         i++;
         long dot_pos = i;
         while (i < trimmed_len && trim_start[i] == '0') i++;
         // Valid if we consumed everything AND had at least one zero after dot
-        if (i == trimmed_len && i > dot_pos) return false;  // e.g. "0.0", "00.00"
+        if (i == trimmed_len && i > dot_pos) return false;  // e.g. "0.0", "00.00", "+0.0", "-0.00"
       }
     }
   }

data/images/SmarterCSV_1.17.0_vs_RubyCSV_3.3.5_speedup.svg ADDED Viewed

@@ -0,0 +1,106 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="764" height="632"
+     font-family="ui-monospace, 'Cascadia Code', 'Courier New', monospace" font-size="12">
+<rect width="764" height="632" fill="#ffffff"/>
+<text x="382" y="20" text-anchor="middle" font-size="14" font-weight="bold" fill="#212121">SmarterCSV 1.17.0 vs Ruby CSV.read 3.3.5</text>
+<text x="382" y="36" text-anchor="middle" font-size="10" fill="#9e9e9e">Speedup = CSV.read time ÷ SmarterCSV time  (higher = SmarterCSV is faster) · Ruby 3.4.7 · best of 40</text>
+<text x="490" y="620" text-anchor="middle" font-size="11" fill="#616161">Speedup (CSV.read ÷ SmarterCSV 1.17.0 C)</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">0×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">1×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">2×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">3×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">4×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">5×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">6×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">7×</text>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="240" y="606" text-anchor="middle" font-size="11" fill="#757575">8×</text>
+<line x1="740" y1="62" x2="740" y2="594" stroke="#e0e0e0" stroke-width="1"/>
+<text x="740" y="606" text-anchor="middle" font-size="11" fill="#757575">9×</text>
+<line x1="296" y1="62" x2="296" y2="594" stroke="#9e9e9e" stroke-width="1.5" stroke-dasharray="4,3"/>
+<line x1="240" y1="594" x2="740" y2="594" stroke="#bdbdbd" stroke-width="1"/>
+<line x1="240" y1="62" x2="240" y2="594" stroke="#bdbdbd" stroke-width="1"/>
+<rect x="0" y="62" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="80" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_C.csv</text>
+<rect x="240" y="67" width="437" height="18" fill="#1565C0" rx="2"/>
+<text x="673" y="80" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">7.87×</text>
+<rect x="0" y="90" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="108" text-anchor="end" font-size="11" fill="#424242">uszips.csv</text>
+<rect x="240" y="95" width="435" height="18" fill="#1565C0" rx="2"/>
+<text x="671" y="108" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">7.83×</text>
+<rect x="0" y="118" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="136" text-anchor="end" font-size="11" fill="#424242">long_fields_40k.csv</text>
+<rect x="240" y="123" width="380" height="18" fill="#1565C0" rx="2"/>
+<text x="616" y="136" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">6.83×</text>
+<rect x="0" y="146" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="164" text-anchor="end" font-size="11" fill="#424242">worldcities.csv</text>
+<rect x="240" y="151" width="378" height="18" fill="#1565C0" rx="2"/>
+<text x="614" y="164" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">6.80×</text>
+<rect x="0" y="174" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="192" text-anchor="end" font-size="11" fill="#424242">uscities.csv</text>
+<rect x="240" y="179" width="371" height="18" fill="#1565C0" rx="2"/>
+<text x="607" y="192" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">6.68×</text>
+<rect x="0" y="202" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="220" text-anchor="end" font-size="11" fill="#424242">embedded_separators_60k.csv</text>
+<rect x="240" y="207" width="307" height="18" fill="#1565C0" rx="2"/>
+<text x="543" y="220" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">5.53×</text>
+<rect x="0" y="230" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="248" text-anchor="end" font-size="11" fill="#424242">utf8_multibyte_60k.csv</text>
+<rect x="240" y="235" width="216" height="18" fill="#1565C0" rx="2"/>
+<text x="452" y="248" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">3.89×</text>
+<rect x="0" y="258" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="276" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_NC.csv</text>
+<rect x="240" y="263" width="212" height="18" fill="#1565C0" rx="2"/>
+<text x="448" y="276" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">3.81×</text>
+<rect x="0" y="286" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="304" text-anchor="end" font-size="11" fill="#424242">heavy_quoting_60k.csv</text>
+<rect x="240" y="291" width="195" height="18" fill="#1565C0" rx="2"/>
+<text x="431" y="304" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">3.51×</text>
+<rect x="0" y="314" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="332" text-anchor="end" font-size="11" fill="#424242">many_empty_fields_60k.csv</text>
+<rect x="240" y="319" width="194" height="18" fill="#1565C0" rx="2"/>
+<text x="430" y="332" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">3.49×</text>
+<rect x="0" y="342" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="360" text-anchor="end" font-size="11" fill="#424242">sample_100k.csv</text>
+<rect x="240" y="347" width="191" height="18" fill="#1565C0" rx="2"/>
+<text x="427" y="360" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">3.44×</text>
+<rect x="0" y="370" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="388" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_NB.csv</text>
+<rect x="240" y="375" width="175" height="18" fill="#1565C0" rx="2"/>
+<text x="411" y="388" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">3.14×</text>
+<rect x="0" y="398" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="416" text-anchor="end" font-size="11" fill="#424242">whitespace_heavy_60k.csv</text>
+<rect x="240" y="403" width="166" height="18" fill="#1565C0" rx="2"/>
+<text x="402" y="416" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">2.98×</text>
+<rect x="0" y="426" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="444" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_B.csv</text>
+<rect x="240" y="431" width="164" height="18" fill="#1565C0" rx="2"/>
+<text x="400" y="444" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">2.95×</text>
+<rect x="0" y="454" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="472" text-anchor="end" font-size="11" fill="#424242">embedded_newlines_60k.csv</text>
+<rect x="240" y="459" width="162" height="18" fill="#1565C0" rx="2"/>
+<text x="398" y="472" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">2.91×</text>
+<rect x="0" y="482" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="500" text-anchor="end" font-size="11" fill="#424242">sensor_data_50krows_50cols.csv</text>
+<rect x="240" y="487" width="141" height="18" fill="#1565C0" rx="2"/>
+<text x="377" y="500" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">2.54×</text>
+<rect x="0" y="510" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="528" text-anchor="end" font-size="11" fill="#424242">wide_500_cols_20k.csv</text>
+<rect x="240" y="515" width="94" height="18" fill="#1565C0" rx="2"/>
+<text x="330" y="528" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">1.69×</text>
+<rect x="0" y="538" width="764" height="28" fill="#ffffff"/>
+<text x="232" y="556" text-anchor="end" font-size="11" fill="#424242">tab_separated_60k.tsv</text>
+<rect x="240" y="543" width="87" height="18" fill="#1565C0" rx="2"/>
+<text x="323" y="556" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">1.56×</text>
+<rect x="0" y="566" width="764" height="28" fill="#f5f5f5"/>
+<text x="232" y="584" text-anchor="end" font-size="11" fill="#424242">multi_char_separator_60k.csv</text>
+<rect x="240" y="571" width="75" height="18" fill="#1565C0" rx="2"/>
+<text x="311" y="584" text-anchor="end" font-size="10" fill="#ffffff" font-weight="bold">1.35×</text>
+</svg>

data/images/SmarterCSV_1.17.0_vs_previous_C-speedup.svg ADDED Viewed

@@ -0,0 +1,181 @@
+<svg xmlns="http://www.w3.org/2000/svg" width="820" height="668"
+     font-family="ui-monospace, 'Cascadia Code', 'Courier New', monospace" font-size="12">
+<rect width="820" height="668" fill="#ffffff"/>
+<text x="410" y="18" text-anchor="middle" font-size="13" font-weight="bold" fill="#212121">SmarterCSV improvements 1.15.2, 1.16.4, 1.17.0 vs 1.14.4 — C accelerated</text>
+<text x="410" y="32" text-anchor="middle" font-size="10" fill="#9e9e9e">Speedup ratio = baseline version time ÷ newer version time  (higher = newer version is faster)</text>
+<text x="410" y="48" text-anchor="middle" font-size="11" fill="#616161">Ruby 3.4.7  [log scale, best of 40]</text>
+<line x1="220" y1="68" x2="220" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="220" y="64" text-anchor="middle" font-size="11" fill="#757575">1×</text>
+<line x1="307" y1="68" x2="307" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="307" y="64" text-anchor="middle" font-size="11" fill="#757575">2×</text>
+<line x1="423" y1="68" x2="423" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="423" y="64" text-anchor="middle" font-size="11" fill="#757575">5×</text>
+<line x1="510" y1="68" x2="510" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="510" y="64" text-anchor="middle" font-size="11" fill="#757575">10×</text>
+<line x1="597" y1="68" x2="597" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="597" y="64" text-anchor="middle" font-size="11" fill="#757575">20×</text>
+<line x1="713" y1="68" x2="713" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="713" y="64" text-anchor="middle" font-size="11" fill="#757575">50×</text>
+<line x1="800" y1="68" x2="800" y2="580" stroke="#e0e0e0" stroke-width="1"/>
+<text x="800" y="64" text-anchor="middle" font-size="11" fill="#757575">100×</text>
+<line x1="220" y1="68" x2="220" y2="580" stroke="#9e9e9e" stroke-width="1.5"/>
+<line x1="220" y1="68" x2="800" y2="68" stroke="#bdbdbd" stroke-width="1"/>
+<rect x="0" y="86" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="103" text-anchor="end" font-size="11" fill="#424242">long_fields_40k</text>
+<circle cx="645" cy="99" r="5" fill="#1565C0"/>
+<text x="616" y="103" font-size="10" fill="#1565C0">29×</text>
+<circle cx="744" cy="99" r="5" fill="#BF360C"/>
+<text x="715" y="103" font-size="10" fill="#BF360C">64×</text>
+<circle cx="775" cy="99" r="5" fill="#2E7D32"/>
+<text x="783" y="103" font-size="10" fill="#2E7D32">82×</text>
+<rect x="0" y="112" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="129" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_C</text>
+<circle cx="674" cy="125" r="5" fill="#1565C0"/>
+<circle cx="691" cy="125" r="5" fill="#2E7D32"/>
+<circle cx="693" cy="125" r="5" fill="#BF360C"/>
+<text x="645" y="129" font-size="10" fill="#1565C0">37×</text>
+<text x="701" y="129" font-size="10" fill="#2E7D32">42×</text>
+<text x="728" y="129" font-size="10" fill="#BF360C">43×</text>
+<rect x="0" y="138" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="155" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_NC</text>
+<circle cx="568" cy="151" r="5" fill="#1565C0"/>
+<circle cx="592" cy="151" r="5" fill="#2E7D32"/>
+<circle cx="596" cy="151" r="5" fill="#BF360C"/>
+<text x="539" y="155" font-size="10" fill="#1565C0">16×</text>
+<text x="604" y="155" font-size="10" fill="#2E7D32">19×</text>
+<text x="631" y="155" font-size="10" fill="#BF360C">20×</text>
+<rect x="0" y="164" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="181" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_NB</text>
+<circle cx="566" cy="177" r="5" fill="#1565C0"/>
+<circle cx="589" cy="177" r="5" fill="#2E7D32"/>
+<circle cx="591" cy="177" r="5" fill="#BF360C"/>
+<text x="537" y="181" font-size="10" fill="#1565C0">16×</text>
+<text x="599" y="181" font-size="10" fill="#2E7D32">19×</text>
+<text x="626" y="181" font-size="10" fill="#BF360C">19×</text>
+<rect x="0" y="190" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="207" text-anchor="end" font-size="11" fill="#424242">PEOPLE_IMPORT_B</text>
+<circle cx="565" cy="203" r="5" fill="#1565C0"/>
+<circle cx="587" cy="203" r="5" fill="#2E7D32"/>
+<circle cx="588" cy="203" r="5" fill="#BF360C"/>
+<text x="536" y="207" font-size="10" fill="#1565C0">15×</text>
+<text x="596" y="207" font-size="10" fill="#2E7D32">18×</text>
+<text x="623" y="207" font-size="10" fill="#BF360C">19×</text>
+<rect x="0" y="216" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="233" text-anchor="end" font-size="11" fill="#424242">whitespace_heavy_60k</text>
+<circle cx="551" cy="229" r="5" fill="#1565C0"/>
+<circle cx="574" cy="229" r="5" fill="#2E7D32"/>
+<circle cx="576" cy="229" r="5" fill="#BF360C"/>
+<text x="522" y="233" font-size="10" fill="#1565C0">14×</text>
+<text x="584" y="233" font-size="10" fill="#2E7D32">17×</text>
+<text x="611" y="233" font-size="10" fill="#BF360C">17×</text>
+<rect x="0" y="242" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="259" text-anchor="end" font-size="11" fill="#424242">sensor_data_50krows_50cols</text>
+<circle cx="560" cy="255" r="5" fill="#1565C0"/>
+<circle cx="571" cy="255" r="5" fill="#BF360C"/>
+<circle cx="571" cy="255" r="5" fill="#2E7D32"/>
+<text x="531" y="259" font-size="10" fill="#1565C0">15×</text>
+<text x="579" y="259" font-size="10" fill="#BF360C">16×</text>
+<text x="606" y="259" font-size="10" fill="#2E7D32">16×</text>
+<rect x="0" y="268" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="285" text-anchor="end" font-size="11" fill="#424242">tab_separated_60k</text>
+<circle cx="540" cy="281" r="5" fill="#1565C0"/>
+<circle cx="563" cy="281" r="5" fill="#2E7D32"/>
+<circle cx="565" cy="281" r="5" fill="#BF360C"/>
+<text x="511" y="285" font-size="10" fill="#1565C0">13×</text>
+<text x="573" y="285" font-size="10" fill="#2E7D32">15×</text>
+<text x="600" y="285" font-size="10" fill="#BF360C">15×</text>
+<rect x="0" y="294" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="311" text-anchor="end" font-size="11" fill="#424242">multi_char_separator_60k</text>
+<circle cx="538" cy="307" r="5" fill="#1565C0"/>
+<circle cx="561" cy="307" r="5" fill="#2E7D32"/>
+<circle cx="563" cy="307" r="5" fill="#BF360C"/>
+<text x="509" y="311" font-size="10" fill="#1565C0">13×</text>
+<text x="571" y="311" font-size="10" fill="#2E7D32">15×</text>
+<text x="598" y="311" font-size="10" fill="#BF360C">15×</text>
+<rect x="0" y="320" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="337" text-anchor="end" font-size="11" fill="#424242">heavy_quoting_60k</text>
+<circle cx="503" cy="333" r="5" fill="#1565C0"/>
+<text x="467" y="337" font-size="10" fill="#1565C0">9.4×</text>
+<circle cx="538" cy="333" r="5" fill="#BF360C"/>
+<circle cx="560" cy="333" r="5" fill="#2E7D32"/>
+<text x="568" y="337" font-size="10" fill="#BF360C">13×</text>
+<text x="595" y="337" font-size="10" fill="#2E7D32">15×</text>
+<rect x="0" y="346" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="363" text-anchor="end" font-size="11" fill="#424242">uszips</text>
+<circle cx="508" cy="359" r="5" fill="#1565C0"/>
+<text x="472" y="363" font-size="10" fill="#1565C0">9.8×</text>
+<circle cx="542" cy="359" r="5" fill="#BF360C"/>
+<circle cx="555" cy="359" r="5" fill="#2E7D32"/>
+<text x="563" y="363" font-size="10" fill="#BF360C">13×</text>
+<text x="590" y="363" font-size="10" fill="#2E7D32">14×</text>
+<rect x="0" y="372" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="389" text-anchor="end" font-size="11" fill="#424242">uscities</text>
+<circle cx="495" cy="385" r="5" fill="#1565C0"/>
+<text x="459" y="389" font-size="10" fill="#1565C0">8.9×</text>
+<circle cx="531" cy="385" r="5" fill="#BF360C"/>
+<circle cx="548" cy="385" r="5" fill="#2E7D32"/>
+<text x="556" y="389" font-size="10" fill="#BF360C">12×</text>
+<text x="583" y="389" font-size="10" fill="#2E7D32">13×</text>
+<rect x="0" y="398" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="415" text-anchor="end" font-size="11" fill="#424242">embedded_separators_60k</text>
+<circle cx="482" cy="411" r="5" fill="#1565C0"/>
+<text x="446" y="415" font-size="10" fill="#1565C0">8.0×</text>
+<circle cx="530" cy="411" r="5" fill="#BF360C"/>
+<circle cx="543" cy="411" r="5" fill="#2E7D32"/>
+<text x="551" y="415" font-size="10" fill="#BF360C">12×</text>
+<text x="578" y="415" font-size="10" fill="#2E7D32">13×</text>
+<rect x="0" y="424" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="441" text-anchor="end" font-size="11" fill="#424242">many_empty_fields_60k</text>
+<circle cx="514" cy="437" r="5" fill="#1565C0"/>
+<circle cx="540" cy="437" r="5" fill="#2E7D32"/>
+<circle cx="541" cy="437" r="5" fill="#BF360C"/>
+<text x="485" y="441" font-size="10" fill="#1565C0">10×</text>
+<text x="549" y="441" font-size="10" fill="#2E7D32">13×</text>
+<text x="576" y="441" font-size="10" fill="#BF360C">13×</text>
+<rect x="0" y="450" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="467" text-anchor="end" font-size="11" fill="#424242">worldcities</text>
+<circle cx="484" cy="463" r="5" fill="#1565C0"/>
+<text x="448" y="467" font-size="10" fill="#1565C0">8.1×</text>
+<circle cx="523" cy="463" r="5" fill="#BF360C"/>
+<circle cx="536" cy="463" r="5" fill="#2E7D32"/>
+<text x="544" y="467" font-size="10" fill="#BF360C">11×</text>
+<text x="571" y="467" font-size="10" fill="#2E7D32">12×</text>
+<rect x="0" y="476" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="493" text-anchor="end" font-size="11" fill="#424242">wide_500_cols_20k</text>
+<circle cx="525" cy="489" r="5" fill="#1565C0"/>
+<circle cx="527" cy="489" r="5" fill="#BF360C"/>
+<circle cx="527" cy="489" r="5" fill="#2E7D32"/>
+<text x="496" y="493" font-size="10" fill="#1565C0">11×</text>
+<text x="535" y="493" font-size="10" fill="#BF360C">11×</text>
+<text x="562" y="493" font-size="10" fill="#2E7D32">11×</text>
+<rect x="0" y="502" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="519" text-anchor="end" font-size="11" fill="#424242">utf8_multibyte_60k</text>
+<circle cx="494" cy="515" r="5" fill="#1565C0"/>
+<text x="458" y="519" font-size="10" fill="#1565C0">8.8×</text>
+<circle cx="525" cy="515" r="5" fill="#BF360C"/>
+<circle cx="525" cy="515" r="5" fill="#2E7D32"/>
+<text x="533" y="519" font-size="10" fill="#BF360C">11×</text>
+<text x="560" y="519" font-size="10" fill="#2E7D32">11×</text>
+<rect x="0" y="528" width="820" height="26" fill="#ffffff"/>
+<text x="212" y="545" text-anchor="end" font-size="11" fill="#424242">embedded_newlines_60k</text>
+<circle cx="479" cy="541" r="5" fill="#1565C0"/>
+<circle cx="495" cy="541" r="5" fill="#BF360C"/>
+<circle cx="503" cy="541" r="5" fill="#2E7D32"/>
+<text x="443" y="545" font-size="10" fill="#1565C0">7.8×</text>
+<text x="511" y="545" font-size="10" fill="#BF360C">8.9×</text>
+<text x="545" y="545" font-size="10" fill="#2E7D32">9.5×</text>
+<rect x="0" y="554" width="820" height="26" fill="#f5f5f5"/>
+<text x="212" y="571" text-anchor="end" font-size="11" fill="#424242">sample_100k</text>
+<circle cx="470" cy="567" r="5" fill="#1565C0"/>
+<circle cx="500" cy="567" r="5" fill="#2E7D32"/>
+<circle cx="502" cy="567" r="5" fill="#BF360C"/>
+<text x="434" y="571" font-size="10" fill="#1565C0">7.3×</text>
+<text x="510" y="571" font-size="10" fill="#2E7D32">9.2×</text>
+<text x="544" y="571" font-size="10" fill="#BF360C">9.4×</text>
+<circle cx="228" cy="594" r="5" fill="#1565C0"/>
+<text x="240" y="598" font-size="11" fill="#1565C0">C accelerated (v1.15.2)</text>
+<circle cx="228" cy="614" r="5" fill="#BF360C"/>
+<text x="240" y="618" font-size="11" fill="#BF360C">C accelerated (v1.16.4)</text>
+<circle cx="228" cy="634" r="5" fill="#2E7D32"/>
+<text x="240" y="638" font-size="11" fill="#2E7D32">C accelerated (v1.17.0)</text>
+</svg>