smarter_csv 1.14.2 → 1.14.4

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 4b6c0ad6a61721dac33f4ef31cf34da3cdd221804aaa45ff8e49cf7b5894b539
-  data.tar.gz: 8c1307aa7a74fc4f434eec362519dcb4392774aa35f563fc2674d542f6b4ea62
+  metadata.gz: a84cfa57008a6f9f05ee82eeef9edb4dab5993874a38c456ce78d8da999280aa
+  data.tar.gz: 7e8569670615a6ff7fb63c152d8e849069bcde318edf885d95cef52b5513f52c
 SHA512:
-  metadata.gz: 8b7cef2ec65c990d3f6c8b05acaefa328952e70571c5868293531ac51c44fd13306dfad5e976f786eb61d206214bc7eae91cf8c2e00023a26c75d689769ed684
-  data.tar.gz: e204946071d76d264b0c8206b1d9170eb7b5c89bb25f22975322fbdc0c4d52869f4b7cf06245d5143298c6d3a89b8c2bc6aa6529fc32f0895e57e1d2ddef97d1
+  metadata.gz: 3d1baa73a0120824390f062e38d3acab8dfc7a09d48fbbdd647466605ccd543ca839e61cbefca9929d2dafce9a61b12e50b62d9e034aadd047a46e1886728980
+  data.tar.gz: 9756f2fdd15e619ba98011370a3410781b7771edd0656dd1bcb1226ebf899726ec00f43627dcc16756d986a962e9eb3e84c7e901ef5aecf12ede9f077f7a8423

data/CHANGELOG.md

@@ -1,8 +1,19 @@
 
 # SmarterCSV 1.x Change Log
 
+## 1.14.4 (2025-05-26)
+ * Bugfix: SmarterCSV::Reader fixing issue with header containing spaces ([PR 305](https://github.com/tilo/smarter_csv/pull/305) thanks to Felipe Cabezudo)
+
+## 1.14.3 (2025-05-04)
+ * Improved C-extension parsing logic:
+   - Added fast path for unquoted fields to avoid unnecessary quote checks.
+   - Aded inline whitespace stripping inside the C parser
+ * Performance
+   -  Significantly reduced per-line overhead in non-quoted, wide CSVs (e.g. fixed-width data exports).
+   - Benchmarks show ~10–40% speedup over v1.14.2 depending on structure and quoting.
+
 ## 1.14.2 (2025-04-10)
- * bugfix: SmarterCSV::Writer fixing corner case with `quote_headers: true`
+ * bugfix: SmarterCSV::Writer fixing corner case with `quote_headers: true` ([issue 301](https://github.com/tilo/smarter_csv/issues/301))
  * new option: `header_converter` allows to programatically modify the headers
 
 ## 1.14.1 (2025-04-09)

data/CONTRIBUTORS.md

@@ -59,3 +59,4 @@ A Big Thank you to everyone who filed issues, sent comments, and who contributed
  * [Randall B](https://github.com/randall-coding)
  * [Matthew Kennedy](https://github.com/MattKitmanLabs)
  * [Robert Reiz](https://github.com/reiz)
+ * [Felipe Cabezudo](https://github.com/felipekb)

data/ext/smarter_csv/extconf.rb

@@ -9,6 +9,8 @@ if RbConfig::MAKEFILE_CONFIG["CFLAGS"].include?("-g -O3")
   RbConfig::MAKEFILE_CONFIG["CFLAGS"] = fixed_CFLAGS
 end
 
-CONFIG["optflags"] = "-O3"
+# CONFIG["optflags"] = "-O3 -march=native -flto"
+CONFIG["optflags"] = "-O3 -march=native -flto -fomit-frame-pointer -DNDEBUG"
+CONFIG["debugflags"] = ""
 
 create_makefile('smarter_csv/smarter_csv')

data/ext/smarter_csv/smarter_csv.c

@@ -2,6 +2,7 @@
 #include "ruby/encoding.h"
 #include <stdio.h>
 #include <stdbool.h>
+#include <string.h>
 
 #ifndef bool
   #define bool int
@@ -12,8 +13,25 @@
 VALUE SmarterCSV = Qnil;
 VALUE eMalformedCSVError = Qnil;
 VALUE Parser = Qnil;
+VALUE Qempty_string = Qnil; // shared frozen empty string
+
+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];
+    }
+  }
+  VALUE out = rb_enc_str_new(buf, j, encoding);
+  xfree(buf);
+  return out;
+}
 
-static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quote_char, VALUE max_size) {
+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) {
   if (RB_TYPE_P(line, T_NIL) == 1) {
     return rb_ary_new();
   }
@@ -22,74 +40,180 @@ static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quot
     rb_raise(rb_eTypeError, "ERROR in SmarterCSV.parse_line: line has to be a string or nil");
   }
 
-  rb_encoding *encoding = rb_enc_get(line); /* get the encoding from the input line */
-  char *startP = RSTRING_PTR(line); /* may not be null terminated */
+  rb_encoding *encoding = rb_enc_get(line);
+  char *startP = RSTRING_PTR(line);
   long line_len = RSTRING_LEN(line);
-  char *endP = startP + line_len; /* points behind the string */
+  char *endP = startP + line_len;
   char *p = startP;
 
   char *col_sepP = RSTRING_PTR(col_sep);
   long col_sep_len = RSTRING_LEN(col_sep);
 
   char *quoteP = RSTRING_PTR(quote_char);
-  long quote_count = 0;
-
-  bool col_sep_found = true;
+  char quote_char_val = quoteP[0];
+  size_t quote_len = strlen(quoteP);
 
   VALUE elements = rb_ary_new();
   VALUE field;
-  long i;
 
-  /* Variables for escaped quote handling */
+  long element_count = 0;
+  int max_fields = -1;
+  if (max_size != Qnil) {
+    max_fields = NUM2INT(max_size);
+    if (max_fields < 0) {
+      return rb_ary_new();
+    }
+  }
+
+  bool has_quotes = RTEST(has_quotes_val);
+  bool strip_ws = RTEST(strip_ws_val);
+
+  // === FAST PATH: No quotes and single-character separator ===
+  if (__builtin_expect(!has_quotes && col_sep_len == 1, 1)) {
+    char sep = *col_sepP;
+    char *sep_pos = NULL;
+
+    while ((sep_pos = memchr(p, sep, endP - p))) {
+      if ((max_fields >= 0) && (element_count >= max_fields)) {
+        break;
+      }
+
+      long field_len = sep_pos - startP;
+      char *raw_field = startP;
+      char *trim_start = raw_field;
+      char *trim_end = raw_field + field_len - 1;
+
+      if (strip_ws) {
+        while (trim_start <= trim_end && (*trim_start == ' ' || *trim_start == '\t')) trim_start++;
+        while (trim_end >= trim_start && (*trim_end == ' ' || *trim_end == '\t')) trim_end--;
+      }
+
+      long trimmed_len = trim_end - trim_start + 1;
+
+      field = rb_enc_str_new(trim_start, trimmed_len, encoding);
+      rb_ary_push(elements, field);
+      element_count++;
+
+      p = sep_pos + 1;
+      startP = p;
+    }
+
+    if ((max_fields < 0) || (element_count < max_fields)) {
+      long field_len = endP - startP;
+      char *raw_field = startP;
+      char *trim_start = raw_field;
+      char *trim_end = raw_field + field_len - 1;
+
+      if (strip_ws) {
+        while (trim_start <= trim_end && (*trim_start == ' ' || *trim_start == '\t')) trim_start++;
+        while (trim_end >= trim_start && (*trim_end == ' ' || *trim_end == '\t')) trim_end--;
+      }
+
+      long trimmed_len = trim_end - trim_start + 1;
+
+      field = rb_enc_str_new(trim_start, trimmed_len, encoding);
+      rb_ary_push(elements, field);
+    }
+
+    return elements;
+  }
+
+  // === SLOW PATH: Quoted fields or multi-char separator ===
+  long i;
   long backslash_count = 0;
   bool in_quotes = false;
+  bool col_sep_found = true;
 
   while (p < endP) {
-    /* does the remaining string start with col_sep ? */
     col_sep_found = true;
-    for(i=0; (i < col_sep_len) && (p+i < endP); i++) {
-      col_sep_found = col_sep_found && (*(p+i) == *(col_sepP+i));
+    for (i = 0; (i < col_sep_len) && (p + i < endP); i++) {
+      if (*(p + i) != *(col_sepP + i)) {
+        col_sep_found = false;
+        break;
+      }
     }
-    /* if col_sep was found and we're not inside quotes */
+
     if (col_sep_found && !in_quotes) {
-      /* if max_size != nil && elements.size >= header_size */
-      if ((max_size != Qnil) && RARRAY_LEN(elements) >= NUM2INT(max_size)) {
+      if ((max_fields >= 0) && (element_count >= max_fields)) {
         break;
-      } else {
-        /* push that field with original encoding onto the results */
-        field = rb_enc_str_new(startP, p - startP, encoding);
-        rb_ary_push(elements, field);
+      }
+
+      long field_len = p - startP;
+      char *raw_field = startP;
 
-        p += col_sep_len;
-        startP = p;
-        backslash_count = 0; // Reset backslash count at the start of a new field
+      bool quoted = (field_len >= 2 && raw_field[0] == quote_char_val && raw_field[field_len - 1] == quote_char_val);
+      if (quoted) {
+        raw_field++;
+        field_len -= 2;
+      }
+
+      char *trim_start = raw_field;
+      char *trim_end = raw_field + field_len - 1;
+
+      if (strip_ws) {
+        while (trim_start <= trim_end && (*trim_start == ' ' || *trim_start == '\t')) trim_start++;
+        while (trim_end >= trim_start && (*trim_end == ' ' || *trim_end == '\t')) trim_end--;
+      }
+
+      long trimmed_len = trim_end - trim_start + 1;
+
+      if (quoted || memchr(trim_start, quote_char_val, trimmed_len)) {
+        field = unescape_quotes(trim_start, trimmed_len, quote_char_val, encoding);
+      } else {
+        field = rb_enc_str_new(trim_start, trimmed_len, encoding);
       }
+
+      rb_ary_push(elements, field);
+      element_count++;
+
+      p += col_sep_len;
+      startP = p;
+      backslash_count = 0;
     } else {
       if (*p == '\\') {
         backslash_count++;
       } else {
-        if (*p == *quoteP) {
+        if (*p == quote_char_val) {
           if (backslash_count % 2 == 0) {
-            /* Even number of backslashes means quote is not escaped */
             in_quotes = !in_quotes;
           }
-          /* Else, quote is escaped; do nothing */
         }
-        backslash_count = 0; // Reset after any character other than backslash
+        backslash_count = 0;
       }
       p++;
     }
-  } /* while */
+  }
 
-  /* Check for unclosed quotes at the end of the line */
   if (in_quotes) {
     rb_raise(eMalformedCSVError, "Unclosed quoted field detected in line: %s", StringValueCStr(line));
   }
 
-  /* check if the last part of the line needs to be processed */
-  if ((max_size == Qnil) || RARRAY_LEN(elements) < NUM2INT(max_size)) {
-    /* copy the remaining line as a field with original encoding onto the results */
-    field = rb_enc_str_new(startP, endP - startP, encoding);
+  if ((max_fields < 0) || (element_count < max_fields)) {
+    long field_len = endP - startP;
+    char *raw_field = startP;
+
+    bool quoted = (field_len >= 2 && raw_field[0] == quote_char_val && raw_field[field_len - 1] == quote_char_val);
+    if (quoted) {
+      raw_field++;
+      field_len -= 2;
+    }
+
+    char *trim_start = raw_field;
+    char *trim_end = raw_field + field_len - 1;
+
+    if (strip_ws) {
+      while (trim_start <= trim_end && (*trim_start == ' ' || *trim_start == '\t')) trim_start++;
+      while (trim_end >= trim_start && (*trim_end == ' ' || *trim_end == '\t')) trim_end--;
+    }
+
+    long trimmed_len = trim_end - trim_start + 1;
+
+    if (quoted || memchr(trim_start, quote_char_val, trimmed_len)) {
+      field = unescape_quotes(trim_start, trimmed_len, quote_char_val, encoding);
+    } else {
+      field = rb_enc_str_new(trim_start, trimmed_len, encoding);
+    }
+
     rb_ary_push(elements, field);
   }
 
@@ -97,10 +221,10 @@ static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quot
 }
 
 void Init_smarter_csv(void) {
-  // these modules and the error class are already defined in Ruby code, make them accessible:
   SmarterCSV = rb_const_get(rb_cObject, rb_intern("SmarterCSV"));
   Parser = rb_const_get(SmarterCSV, rb_intern("Parser"));
   eMalformedCSVError = rb_const_get(SmarterCSV, rb_intern("MalformedCSV"));
-
-  rb_define_module_function(Parser, "parse_csv_line_c", rb_parse_csv_line, 4);
+  Qempty_string = rb_str_new_literal("");
+  rb_gc_register_address(&Qempty_string);
+  rb_define_module_function(Parser, "parse_csv_line_c", rb_parse_csv_line, 6);
 }

data/lib/smarter_csv/auto_detection.rb

@@ -32,7 +32,7 @@ module SmarterCSV
 
       if candidates.values.max == 0
         # if the header only contains
-        return ',' if line.chomp(options[:row_sep]) =~ /^\w+$/
+        return ',' if line.chomp(options[:row_sep]) =~ /^[\w\s]+$/
 
         raise SmarterCSV::NoColSepDetected
       end

data/lib/smarter_csv/parser.rb

@@ -2,6 +2,8 @@
 
 module SmarterCSV
   module Parser
+    EMPTY_STRING = ''.freeze
+
     protected
 
     ###
@@ -11,17 +13,16 @@ module SmarterCSV
     ###
     def parse(line, options, header_size = nil)
       # puts "SmarterCSV.parse OPTIONS: #{options[:acceleration]}" if options[:verbose]
+      has_quotes = line.include?(options[:quote_char])
 
       if options[:acceleration] && has_acceleration
         # :nocov:
-        has_quotes = line =~ /#{options[:quote_char]}/
-        elements = parse_csv_line_c(line, options[:col_sep], options[:quote_char], header_size)
-        elements.map!{|x| cleanup_quotes(x, options[:quote_char])} if has_quotes
+        elements = parse_csv_line_c(line, options[:col_sep], options[:quote_char], header_size, has_quotes, options[:strip_whitespace])
         [elements, elements.size]
         # :nocov:
       else
         # puts "WARNING: SmarterCSV is using un-accelerated parsing of lines. Check options[:acceleration]"
-        parse_csv_line_ruby(line, options, header_size)
+        parse_csv_line_ruby(line, options, header_size, has_quotes)
       end
     end
 
@@ -46,7 +47,7 @@ module SmarterCSV
     #
     # Our convention is that empty fields are returned as empty strings, not as nil.
 
-    def parse_csv_line_ruby(line, options, header_size = nil)
+    def parse_csv_line_ruby(line, options, header_size = nil, has_quotes = false)
       return [[], 0] if line.nil?
 
       line_size = line.size
@@ -98,11 +99,13 @@ module SmarterCSV
         elements << cleanup_quotes(line[start..-1], quote)
       end
 
+      elements.map!(&:strip) if options[:strip_whitespace]
       [elements, elements.size]
     end
 
     def cleanup_quotes(field, quote)
-      return field if field.nil?
+      return nil if field.nil?
+      return EMPTY_STRING if field.empty?
 
       # Remove surrounding quotes if present
       if field.start_with?(quote) && field.end_with?(quote)
@@ -110,9 +113,13 @@ module SmarterCSV
       end
 
       # Replace double quotes with a single quote
-      field.gsub!((quote * 2).to_s, quote)
+      field.gsub!(doubled_quote(quote), quote)
 
       field
     end
+
+    def doubled_quote(quote)
+      @doubled_quote ||= (quote * 2).to_s.freeze
+    end
   end
 end

data/lib/smarter_csv/reader.rb

@@ -128,6 +128,7 @@ module SmarterCSV
           line.chomp!(options[:row_sep])
 
           # --- SPLIT LINE & DATA TRANSFORMATIONS ------------------------------------------------------------
+          # we are now stripping whitespace inside the parse() methods
           dataA, data_size = parse(line, options) # we parse the extra columns
 
           if options[:strict]
@@ -141,8 +142,6 @@ module SmarterCSV
             end
           end
 
-          dataA.map!{|x| x.strip} if options[:strip_whitespace]
-
           # if all values are blank, then ignore this line
           next if options[:remove_empty_hashes] && (dataA.empty? || blank?(dataA))
 

data/lib/smarter_csv/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module SmarterCSV
-  VERSION = "1.14.2"
+  VERSION = "1.14.4"
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: smarter_csv
 version: !ruby/object:Gem::Version
-  version: 1.14.2
+  version: 1.14.4
 platform: ruby
 authors:
 - Tilo Sloboda
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2025-04-10 00:00:00.000000000 Z
+date: 2025-05-29 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: awesome_print