Module: SmarterCSV

Defined in:
lib/smarter_csv.rb,
lib/smarter_csv/version.rb,
ext/smarter_csv/smarter_csv.c

Defined Under Namespace

Classes: DuplicateHeaders, HeaderSizeMismatch, IncorrectOption, KeyMappingError, MalformedCSVError, MissingHeaders, NoColSepDetected, SmarterCSVException

Constant Summary collapse

VERSION =
"1.7.1"

Class Method Summary collapse

Class Method Details

.has_acceleration?Boolean

Returns:

  • (Boolean)

186
187
188
# File 'lib/smarter_csv.rb', line 186

def has_acceleration?
  @has_acceleration ||= !!defined?(parse_csv_line_c)
end

.headersObject


194
195
196
# File 'lib/smarter_csv.rb', line 194

def headers
  @headers
end

.parse_csv_line_c(line, col_sep, quote_char, max_size) ⇒ Object

max_size: pass nil if no limit is specified


15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
# File 'ext/smarter_csv/smarter_csv.c', line 15

static VALUE rb_parse_csv_line(VALUE self, VALUE line, VALUE col_sep, VALUE quote_char, VALUE max_size) {
  if (RB_TYPE_P(line, T_NIL) == 1) {
    return rb_ary_new();

  } else if (RB_TYPE_P(line, T_STRING) == 1) {
    rb_encoding *encoding = rb_enc_get(line); /* get the encoding from the input line */
    char *startP = RSTRING_PTR(line); /* may not be null terminated */
    long line_len = RSTRING_LEN(line);
    char *endP = startP + line_len ; /* points behind the string */
    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_len = RSTRING_LEN(quote_char);
    long quote_count = 0;

    bool col_sep_found = true;

    VALUE elements = rb_ary_new();
    VALUE field;
    long i;

    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));
      }
      /* if col_sep was found and we have even quotes */
      if (col_sep_found && (quote_count % 2 == 0)) {
        /* if max_size != nil && lements.size >= header_size */
        if ((max_size != Qnil) && RARRAY_LEN(elements) >= NUM2INT(max_size)) {
          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);

          p += col_sep_len;
          startP = p;
        }
      } else {
        if (*p == *quoteP) {
          quote_count += 1;
        }
        p++;
      }
    } /* while */

    /* 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);
      rb_ary_push(elements, field);
    }

    return elements;
  }

  rb_raise(rb_eTypeError, "ERROR in SmarterCSV.parse_line: line has to be a string or nil");
}

.process(input, options = {}, &block) ⇒ Object

first parameter: filename or input object which responds to readline method


18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
# File 'lib/smarter_csv.rb', line 18

def SmarterCSV.process(input, options = {}, &block)
  options = default_options.merge(options)
  options[:invalid_byte_sequence] = '' if options[:invalid_byte_sequence].nil?
  puts "SmarterCSV OPTIONS: #{options.inspect}" if options[:verbose]

  headerA = []
  result = []
  @file_line_count = 0
  @csv_line_count = 0
  has_rails = !!defined?(Rails)
  begin
    fh = input.respond_to?(:readline) ? input : File.open(input, "r:#{options[:file_encoding]}")

    # auto-detect the row separator
    options[:row_sep] = guess_line_ending(fh, options) if options[:row_sep]&.to_sym == :auto
    # attempt to auto-detect column separator
    options[:col_sep] = guess_column_separator(fh, options) if options[:col_sep]&.to_sym == :auto

    if (options[:force_utf8] || options[:file_encoding] =~ /utf-8/i) && (fh.respond_to?(:external_encoding) && fh.external_encoding != Encoding.find('UTF-8') || fh.respond_to?(:encoding) && fh.encoding != Encoding.find('UTF-8'))
      puts 'WARNING: you are trying to process UTF-8 input, but did not open the input with "b:utf-8" option. See README file "NOTES about File Encodings".'
    end

    if options[:skip_lines].to_i > 0
      options[:skip_lines].to_i.times do
        readline_with_counts(fh, options)
      end
    end

    headerA, header_size = process_headers(fh, options)

    # in case we use chunking.. we'll need to set it up..
    if !options[:chunk_size].nil? && options[:chunk_size].to_i > 0
      use_chunks = true
      chunk_size = options[:chunk_size].to_i
      chunk_count = 0
      chunk = []
    else
      use_chunks = false
    end

    # now on to processing all the rest of the lines in the CSV file:
    until fh.eof? # we can't use fh.readlines() here, because this would read the whole file into memory at once, and eof => true
      line = readline_with_counts(fh, options)

      # replace invalid byte sequence in UTF-8 with question mark to avoid errors
      line = line.force_encoding('utf-8').encode('utf-8', invalid: :replace, undef: :replace, replace: options[:invalid_byte_sequence]) if options[:force_utf8] || options[:file_encoding] !~ /utf-8/i

      print "processing file line %10d, csv line %10d\r" % [@file_line_count, @csv_line_count] if options[:verbose]

      next if options[:comment_regexp] && line =~ options[:comment_regexp] # ignore all comment lines if there are any

      # cater for the quoted csv data containing the row separator carriage return character
      # in which case the row data will be split across multiple lines (see the sample content in spec/fixtures/carriage_returns_rn.csv)
      # by detecting the existence of an uneven number of quote characters

      multiline = line.count(options[:quote_char]).odd? # should handle quote_char nil
      while line.count(options[:quote_char]).odd? # should handle quote_char nil
        next_line = fh.readline(options[:row_sep])
        next_line = next_line.force_encoding('utf-8').encode('utf-8', invalid: :replace, undef: :replace, replace: options[:invalid_byte_sequence]) if options[:force_utf8] || options[:file_encoding] !~ /utf-8/i
        line += next_line
        @file_line_count += 1
      end
      print "\nline contains uneven number of quote chars so including content through file line %d\n" % @file_line_count if options[:verbose] && multiline

      line.chomp!(options[:row_sep])

      dataA, _data_size = parse(line, options, header_size)

      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))

      hash = Hash.zip(headerA, dataA) # from Facets of Ruby library

      # make sure we delete any key/value pairs from the hash, which the user wanted to delete:
      # Note: Ruby < 1.9 doesn't allow empty symbol literals!
      hash.delete(nil)
      hash.delete('')
      eval('hash.delete(:"")') if RUBY_VERSION.to_f > 1.8

      if options[:remove_empty_values] == true
        hash.delete_if{|_k, v| has_rails ? v.blank? : blank?(v)}
      end

      hash.delete_if{|_k, v| !v.nil? && v =~ /^(\d+|\d+\.\d+)$/ && v.to_f == 0} if options[:remove_zero_values] # values are typically Strings!
      hash.delete_if{|_k, v| v =~ options[:remove_values_matching]} if options[:remove_values_matching]

      if options[:convert_values_to_numeric]
        hash.each do |k, v|
          # deal with the :only / :except options to :convert_values_to_numeric
          next if only_or_except_limit_execution(options, :convert_values_to_numeric, k)

          # convert if it's a numeric value:
          case v
          when /^[+-]?\d+\.\d+$/
            hash[k] = v.to_f
          when /^[+-]?\d+$/
            hash[k] = v.to_i
          end
        end
      end

      if options[:value_converters]
        hash.each do |k, v|
          converter = options[:value_converters][k]
          next unless converter

          hash[k] = converter.convert(v)
        end
      end

      next if options[:remove_empty_hashes] && hash.empty?

      if use_chunks
        chunk << hash # append temp result to chunk

        if chunk.size >= chunk_size || fh.eof? # if chunk if full, or EOF reached
          # do something with the chunk
          if block_given?
            yield chunk # do something with the hashes in the chunk in the block
          else
            result << chunk # not sure yet, why anybody would want to do this without a block
          end
          chunk_count += 1
          chunk = [] # initialize for next chunk of data
        else

          # the last chunk may contain partial data, which also needs to be returned (BUG / ISSUE-18)

        end

        # while a chunk is being filled up we don't need to do anything else here

      else # no chunk handling
        if block_given?
          yield [hash] # do something with the hash in the block (better to use chunking here)
        else
          result << hash
        end
      end
    end

    # print new line to retain last processing line message
    print "\n" if options[:verbose]

    # last chunk:
    if !chunk.nil? && chunk.size > 0
      # do something with the chunk
      if block_given?
        yield chunk # do something with the hashes in the chunk in the block
      else
        result << chunk # not sure yet, why anybody would want to do this without a block
      end
      chunk_count += 1
      chunk = [] # initialize for next chunk of data
    end
  ensure
    fh.close if fh.respond_to?(:close)
  end
  if block_given?
    return chunk_count # when we do processing through a block we only care how many chunks we processed
  else
    return result # returns either an Array of Hashes, or an Array of Arrays of Hashes (if in chunked mode)
  end
end

.raw_headerObject


190
191
192
# File 'lib/smarter_csv.rb', line 190

def raw_header
  @raw_header
end