bamfcsv 0.1.0 → 0.1.1

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    bamfcsv (0.0.2)
+    bamfcsv (0.1.0)
 
 GEM
   remote: http://rubygems.org/

data/ext/bamfcsv/bamfcsv_ext.c

@@ -1,19 +1,7 @@
 #include <stdlib.h>
 #include "bamfcsv_ext.h"
 
-struct s_Row *alloc_row() {
-
-  struct s_Row *new_row = malloc(sizeof(struct s_Row));
-
-  new_row -> first_cell = 0;
-  new_row -> next_row = 0;
-  new_row -> cell_count = 0;
-
-  return new_row;
-
-}
-
-struct s_Cell *alloc_cell() {
+struct s_Cell *alloc_cell(struct s_Row *row, struct s_Cell *prev_cell) {
 
   struct s_Cell *new_cell = malloc(sizeof(struct s_Cell));
 
@@ -21,11 +9,26 @@ struct s_Cell *alloc_cell() {
   new_cell -> len = 0;
   new_cell -> next_cell = 0;
   new_cell -> has_quotes = 0;
+  row->cell_count++;
+  if (prev_cell) prev_cell->next_cell = new_cell;
 
   return new_cell;
 
 }
 
+struct s_Row *alloc_row(struct s_Row *prev_row) {
+
+  struct s_Row *new_row = malloc(sizeof(struct s_Row));
+
+  new_row -> next_row = 0;
+  new_row -> cell_count = 0;
+  new_row -> first_cell = alloc_cell(new_row, 0);
+  if (prev_row) prev_row->next_row = new_row;
+
+  return new_row;
+
+}
+
 void free_cell(struct s_Cell *cell) {
 
   if (cell != 0) {
@@ -66,21 +69,20 @@ VALUE build_matrix_from_pointer_tree(struct s_Row *first_row, int num_rows) {
     cur_cell = cur_row->first_cell;
     row = rb_ary_new2(cur_row->cell_count);
     rb_ary_store(matrix,i,row);
-    for (j = 0; j < cur_row->cell_count; j++) {
-      if (*(cur_cell->start) == '"' 
-          && *((cur_cell->start)+(cur_cell->len-1)) == '"')
-        new_string = rb_str_new(cur_cell->start+1, cur_cell->len-2);
-      else {
-        if (cur_cell->len)
-          new_string = rb_str_new(cur_cell->start, cur_cell->len);
-        else
-          new_string = Qnil; /* Empty, unquoted cells are nil, for default ruby CSV compatibility */
-      }
-      if (cur_cell->has_quotes) {
-        rb_funcall(new_string, gsub, 2, dquote, quote);
+    if (cur_row->cell_count > 1 || cur_cell->len) {
+      for (j = 0; j < cur_row->cell_count; j++) {
+        if (cur_cell->has_quotes) {
+          new_string = rb_str_new(cur_cell->start+1, cur_cell->len-2);
+          rb_funcall(new_string, gsub, 2, dquote, quote);
+        } else {
+          if (cur_cell->len)
+            new_string = rb_str_new(cur_cell->start, cur_cell->len);
+          else
+            new_string = Qnil; /* Empty, unquoted cells are nil, for default ruby CSV compatibility */
+        }
+        rb_ary_store(row, j, new_string);
+        cur_cell = cur_cell->next_cell;
       }
-      rb_ary_store(row, j, new_string);
-      cur_cell = cur_cell->next_cell;
     }
     cur_row = cur_row->next_row;
   }
@@ -90,9 +92,9 @@ VALUE build_matrix_from_pointer_tree(struct s_Row *first_row, int num_rows) {
 
 void finalize_cell(struct s_Cell *cell, char *cur, int quote_count) {
   if (*(cur-1) == '\r') 
-    cell->len = cur-(cell->start)-1;
+    cell->len = (int)(cur-(cell->start)-1);
   else
-    cell->len = cur-(cell->start);
+    cell->len = (int)(cur-(cell->start));
 
   if (quote_count) cell->has_quotes = 1;
 }
@@ -102,21 +104,25 @@ VALUE build_matrix(char *buf, int bufsize) {
   int num_rows = 1;
   int quote_count = 0, quotes_matched = 1;
 
-  struct s_Row *first_row = alloc_row();
+  struct s_Row *first_row = alloc_row(0);
   struct s_Row *cur_row = first_row;
-  struct s_Cell *cur_cell = alloc_cell();
-  first_row->first_cell = cur_cell;
+  struct s_Cell *cur_cell = cur_row->first_cell;
   cur_cell->start = buf;
 
   VALUE matrix;
 
   char *cur;
+
+  if (bufsize > 0 && *(buf+bufsize-1) == '\n') {
+    *(buf+bufsize-1) = 0;
+    --bufsize;
+  }
   
   for (cur = buf; cur < buf+bufsize; cur++) {
 
     if (*cur == '"') {
       if (0 == quote_count && cur_cell->start != cur) /* Quotes begin past opening of cell */
-        rb_raise(BAMFCSV_MalformedCSVError_class, "Illegal quoting on line %d, cell %d: Quoted cell must open with '\"'", num_rows, cur_row->cell_count+1);
+        rb_raise(BAMFCSV_MalformedCSVError_class, "Illegal quoting on line %d, cell %d: Quoted cell must open with '\"'", num_rows, cur_row->cell_count);
       else
         ++quote_count;
     }
@@ -128,50 +134,39 @@ VALUE build_matrix(char *buf, int bufsize) {
       if (*cur == ',') {
         
         if (quote_count && *(cur-1) != '"')
-          rb_raise(BAMFCSV_MalformedCSVError_class, "Unclosed quoted field on line %d, cell %d.", num_rows, cur_row->cell_count+1);
+          rb_raise(BAMFCSV_MalformedCSVError_class, "Unclosed quoted field on line %d, cell %d.", num_rows, cur_row->cell_count);
 
-        finalize_cell(cur_cell,cur,quote_count);
-        cur_cell->next_cell = alloc_cell();
-        cur_cell = cur_cell->next_cell;
+        finalize_cell(cur_cell, cur, quote_count);
+        cur_cell = alloc_cell(cur_row, cur_cell);
         cur_cell->start = cur+1;
-        cur_row->cell_count += 1;
         quote_count = 0;
 
-      }
-      
-      if (*cur == '\n') {
+      } else if (*cur == '\n') {
         
         if (quote_count && !(*(cur-1) == '"' || *(cur-1) == '\r' && *(cur-2) == '"'))
-            rb_raise(BAMFCSV_MalformedCSVError_class, "Unclosed quoted field on line %d, cell %d: EOL", num_rows, cur_row->cell_count+1);
+            rb_raise(BAMFCSV_MalformedCSVError_class, "Unclosed quoted field on line %d, cell %d: EOL", num_rows, cur_row->cell_count);
 
-        finalize_cell(cur_cell,cur,quote_count);
-        cur_row->cell_count += 1;
-        cur_row->next_row = alloc_row();
-        cur_row = cur_row -> next_row;
-        cur_row->first_cell = alloc_cell();
+        finalize_cell(cur_cell, cur, quote_count);
+        cur_row = alloc_row(cur_row);
         cur_cell = cur_row->first_cell;
         cur_cell->start = cur+1;
         quote_count = 0;
         
         num_rows++;
 
-      }
+      } else if (quote_count && *cur != '\r' && *cur != '"')
+        rb_raise(BAMFCSV_MalformedCSVError_class, "Illegal quoting on line %d, cell %d", num_rows, cur_row->cell_count);
       
     }
 
   }
 
   if (!quotes_matched) /* Reached EOF without matching quotes */
-    rb_raise(BAMFCSV_MalformedCSVError_class, "Illegal quoting on line %d, cell %d: File ends without closing '\"'", num_rows, cur_row->cell_count+1);
-  else if (quote_count && *cur != '"')
-    rb_raise(BAMFCSV_MalformedCSVError_class, "Unclosed quoted field on line %d, cell %d: EOF", num_rows, cur_row->cell_count+1);
-
-  if (cur_row->cell_count == 0) { /* Ended with newline */
-    num_rows--;
-  } else { /* No newline before EOF */
-    finalize_cell(cur_cell, cur, quote_count);
-    cur_row->cell_count++;
-  }
+    rb_raise(BAMFCSV_MalformedCSVError_class, "Illegal quoting on line %d, cell %d: File ends without closing '\"'", num_rows, cur_row->cell_count);
+  else if (quote_count && *(cur-1) != '"') /* Quotes closed before end of final cell */
+    rb_raise(BAMFCSV_MalformedCSVError_class, "Unclosed quoted field on line %d, cell %d: EOF", num_rows, cur_row->cell_count);
+
+  finalize_cell(cur_cell, cur, quote_count);
 
   matrix = build_matrix_from_pointer_tree(first_row, num_rows);
 

data/lib/bamfcsv/table.rb

@@ -0,0 +1,46 @@
+module BAMFCSV
+  class Table
+    include Enumerable
+    def initialize(matrix)
+      @headers = matrix.shift
+      @matrix = matrix
+      @header_map = {}
+      @headers.each_with_index do |h, i|
+        @header_map[h] = i
+      end
+      @row_cache = []
+    end
+
+    def each
+      @matrix.size.times do |idx|
+        yield self[idx]
+      end
+    end
+
+    def [](idx)
+      idx += @matrix.size if idx < 0
+      return if idx < 0 || idx >= @matrix.size
+      @row_cache[idx] ||= Row.new(@header_map, @matrix[idx])
+    end
+
+    def inspect
+      "#<BAMFCSV::Table>"
+    end
+
+    private
+    def row_hash(row)
+      Hash[@headers.zip(row)]
+    end
+
+    class Row
+      def initialize(header_map, values)
+        @header_map = header_map
+        @values = values
+      end
+
+      def [](key)
+        @values[@header_map[key]]
+      end
+    end
+  end
+end

data/lib/bamfcsv/version.rb

@@ -1,3 +1,3 @@
 module BAMFCSV
-  VERSION = "0.1.0"
+  VERSION = "0.1.1"
 end

data/lib/bamfcsv.rb

@@ -1,13 +1,20 @@
 require 'bamfcsv/bamfcsv'
+require 'bamfcsv/table'
 
 module BAMFCSV
 
-  def self.read(thing_to_read)
-    __parse_string(File.read(thing_to_read))
+  def self.read(thing_to_read, opts={})
+    parse(File.read(thing_to_read), opts)
   end
 
-  def self.parse(csv_str)
-    __parse_string(csv_str)
+  def self.parse(csv_str, opts={})
+    return [] if csv_str.empty?
+    matrix = __parse_string(csv_str)
+    if opts[:headers]
+      Table.new(matrix)
+    else
+      matrix
+    end
   end
 
 end

data/spec/lib/bamfcsv_spec.rb

@@ -57,7 +57,7 @@ describe BAMFCSV do
     end
 
     it 'correctly escaptes ""' do
-      BAMFCSV.parse("1,\"\"2\"\"\n").should == [["1", '"2"']]
+      BAMFCSV.parse(%Q|1,"""2"""\n|).should == [["1", '"2"']]
     end
 
     it "parses unquoted empty cells as nil" do
@@ -68,6 +68,12 @@ describe BAMFCSV do
       BAMFCSV.parse("1,\"\",2").should == [["1","","2"]]
     end
 
+    it "parses a single cell not followed by a newline correctly" do
+      BAMFCSV.parse("1").should == [["1"]]
+      BAMFCSV.parse("1\n2").should == [["1"],["2"]]
+      BAMFCSV.parse("1\r\n2").should == [["1"],["2"]]
+    end
+
     describe "default CSV module compatibility" do
       it "adds a nil cell after a trailing comma with no newline" do
         BAMFCSV.parse("1,2,").should == [["1","2",nil]]
@@ -114,6 +120,41 @@ describe BAMFCSV do
         expect { BAMFCSV.parse("\"\" \r\n") }.should raise_error(BAMFCSV::MalformedCSVError)
         expect { BAMFCSV.parse('1,"" ,2') }.should raise_error(BAMFCSV::MalformedCSVError)
       end
+
+      it "raises BAMFCSV::MalformedCSVError when unescaped quotes appear in a quoted cell" do
+        expect { BAMFCSV.parse('"a"b"c"') }.should raise_error(BAMFCSV::MalformedCSVError)
+        expect { BAMFCSV.parse('"a"b"c",2') }.should raise_error(BAMFCSV::MalformedCSVError)
+        expect { BAMFCSV.parse(%Q("a"b"c"\n)) }.should raise_error(BAMFCSV::MalformedCSVError)
+        expect { BAMFCSV.parse(%Q("a"b"c"\r\n)) }.should raise_error(BAMFCSV::MalformedCSVError)
+      end
+
+      ['', "\n", "\r\n", ",", "\n\n", "\r\n\r\n"].each do |csv|
+        csv_result = CSV.parse(csv)
+        it "BAMFCSV parses #{csv.inspect} the same way as CSV (#{csv_result.inspect})" do
+          BAMFCSV.parse(csv).should == csv_result
+        end
+      end
+    end
+  end
+
+  describe "generating a Table" do
+    describe "with only a header" do
+      let(:header_only) { BAMFCSV.parse("1,2,3", :headers => true) }
+      it "has no body rows" do
+        header_only.first.should be_nil
+      end
+
+      it "does nothing when iterating" do
+        expect { header_only.each { |x| raise "Oh dang!" } }.should_not raise_error
+      end
+    end
+
+    describe "with body rows" do
+      it "maps the headers the the values of each row" do
+        table = BAMFCSV.parse("a,b,c\r\n1,2,3\r\nx,y,z", :headers => true)
+        table.first["a"].should == "1"
+        table[1]["c"].should == "z"
+      end
     end
   end
 end

data/spec/spec_helper.rb

@@ -4,6 +4,7 @@ $:.unshift(File.join(root_path, "ext"))
 require 'bundler'
 require 'bamfcsv'
 Bundler.require(:development)
+require 'csv'
 
 RSpec.configure do |config|
   config.filter_run :focused => true

metadata

@@ -2,7 +2,7 @@
 name: bamfcsv
 version: !ruby/object:Gem::Version 
   prerelease: 
-  version: 0.1.0
+  version: 0.1.1
 platform: ruby
 authors: 
 - Jon Distad
@@ -11,7 +11,7 @@ autorequire:
 bindir: bin
 cert_chain: []
 
-date: 2011-04-03 00:00:00 -04:00
+date: 2011-04-08 00:00:00 -04:00
 default_executable: 
 dependencies: 
 - !ruby/object:Gem::Dependency 
@@ -69,6 +69,7 @@ files:
 - ext/bamfcsv/bamfcsv_ext.h
 - ext/bamfcsv/extconf.rb
 - lib/bamfcsv.rb
+- lib/bamfcsv/table.rb
 - lib/bamfcsv/version.rb
 - spec/fixtures/bamf-comma-comma.csv
 - spec/fixtures/double-quotes.csv