xsv 0.2.1 → 0.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: ace53a58655a50f4de2f10c4a2d68819774e3e74625595353290502299630b30
-  data.tar.gz: 8278ea7b26ac261781ae71328762104db9d14de0cbc05be1b0551a3cb876ea35
+  metadata.gz: 2b097ec44b4cee4a7f6331a3a94f15e24f33c88258d375923f965dafe88ee13e
+  data.tar.gz: c2da68b26d04cb6bd28496d810f7721e327a655c1c96b695d9b091e569767d21
 SHA512:
-  metadata.gz: 31ccd6261073893e0a0873997d7befa39fc0f8088f5f4d9d1ec3f97b7d90f48d1153714cff4f124a4be853976ce714326a6ec8634e646f15ecf6cf428c245a39
-  data.tar.gz: eed43d77052870dc45bc009719b130096a05a6add11576a349ca027ff93b428bfe64813dd83a8ea58a1a7da94aa8b490b4394efa65bbb175ae09dd74c77ee20a
+  metadata.gz: b0a7b30c258e16d391990881f4574577eeee6ec3424524cdc360901ffa62033102c518b005d3b181d0b6e7887f52afa7dfc9415cb2cd762f3f0779d9801aaf5a
+  data.tar.gz: 32ab543df470d94eb4a81e2b271a18ba9295c0d7d9fd63cf918a18baa02b6c01e6bcfb35e6d6e42021d36f773affce5ace52fb07dd7d2ea4110b4a0cbe22fc9c

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    xsv (0.2.0)
+    xsv (0.2.1)
       nokogiri (~> 1.10)
       rubyzip (~> 2.2)
 

data/lib/xsv/helpers.rb

@@ -32,14 +32,14 @@ module Xsv
       47 => "mm:ss.0",
       48 => "##0.0E+0",
       49 => "@",
-    }
+    }.freeze
 
-    MINUTE = 60
-    HOUR = 3600
+    MINUTE = 60.freeze
+    HOUR = 3600.freeze
 
     # Return the index number for the given Excel column name
     def column_index(col)
-      col = col.scan(/^[A-Z]+/).first
+      col = col[/^[A-Z]+/]
 
       val = 0
       while col.length > 0

data/lib/xsv/sheet.rb

@@ -14,24 +14,14 @@ module Xsv
       @mode = :array
       @row_skip = 0
 
-      dimension = xml.css("dimension").first
+      @has_cells = !xml.at_css("sheetData c").nil?
 
-      if dimension
-        _firstCell, lastCell = dimension["ref"].split(":")
-      end
-
-      if lastCell
-        # Assume the dimension reflects the content
-        @column_count = column_index(lastCell) + 1
+      if @has_cells
+        @column_count, @last_row = get_sheet_dimensions
       else
-        # Find the last cell in every row that has a value
-        rightmost_cells = @xml.xpath("//xmlns:row/xmlns:c[*[local-name() = 'v']][last()]").map { |c| column_index(c["r"]) }
-        @column_count = rightmost_cells.max + 1
-
+        @column_count = 0
+        @last_row = 0
       end
-
-      # Find the last row that contains actual values
-      @last_row = @xml.xpath("//xmlns:row[*[xmlns:v]][last()]").first["r"].to_i
     end
 
     def inspect
@@ -69,7 +59,7 @@ module Xsv
 
     # Get row by number, starting at 0
     def [](number)
-      row_xml = xml.css("sheetData row[r=#{number + @row_skip + 1}]").first
+      row_xml = xml.at_css("sheetData row[r=#{number + @row_skip + 1}]")
 
       if row_xml
         parse_row(row_xml)
@@ -81,18 +71,24 @@ module Xsv
     # Load headers in the top row of the worksheet. After parsing of headers
     # all methods return hashes instead of arrays
     def parse_headers!
-      @mode = :array
       @headers = parse_headers
-
       @mode = :hash
 
       true
     end
 
+    def headers
+      if @headers.any?
+        @headers
+      else
+        parse_headers
+      end
+    end
+
     private
 
     def parse_headers
-      parse_row(@xml.css("sheetData row")[@row_skip])
+      parse_row(@xml.css("sheetData row")[@row_skip], :array)
     end
 
     def empty_row
@@ -104,7 +100,8 @@ module Xsv
       end
     end
 
-    def parse_row(xml)
+    def parse_row(xml, mode = nil)
+      mode ||= @mode
       row = empty_row
 
       xml.css("c").first(@column_count).each do |c_xml|
@@ -116,7 +113,7 @@ module Xsv
           when "e" # N/A
             nil
           when nil
-            v = c_xml.css("v").first
+            v = c_xml.at_css("v")
 
             if v.nil?
               nil
@@ -147,7 +144,7 @@ module Xsv
         # Determine column position and pad row with nil values
         col_index = column_index(c_xml["r"])
 
-        case @mode
+        case mode
         when :array
           row[col_index] = value
         when :hash
@@ -157,5 +154,28 @@ module Xsv
 
       row
     end
+
+    # Read or estimate outer bounds of sheet
+    def get_sheet_dimensions
+      dimension = xml.at_css("dimension")
+
+      if dimension
+        _firstCell, lastCell = dimension["ref"].split(":")
+      end
+
+      if lastCell
+        # Assume the dimension reflects the content
+        column_count = column_index(lastCell) + 1
+      else
+        # Find the last cell in every row that has a value
+        rightmost_cells = @xml.xpath("//xmlns:row/xmlns:c[*[local-name() = 'v']][last()]").map { |c| column_index(c["r"]) }
+        column_count = rightmost_cells.max + 1
+      end
+
+      # Find the last row that contains actual values
+      last_row = @xml.at_xpath("//xmlns:row[*[xmlns:v]][last()]")["r"].to_i
+
+      return [column_count, last_row]
+    end
   end
 end

data/lib/xsv/version.rb

@@ -1,3 +1,3 @@
 module Xsv
-  VERSION = "0.2.1"
+  VERSION = "0.2.2"
 end

data/lib/xsv/workbook.rb

@@ -17,7 +17,7 @@ module Xsv
 
       @sheets = []
       @xfs = []
-      @numFmts = Xsv::Helpers::BUILT_IN_NUMBER_FORMATS
+      @numFmts = Xsv::Helpers::BUILT_IN_NUMBER_FORMATS.dup
 
       fetch_shared_strings
       fetch_styles
@@ -33,11 +33,11 @@ module Xsv
     def fetch_shared_strings
       stream = @zip.glob("xl/sharedStrings.xml").first.get_input_stream
       xml = Nokogiri::XML(stream)
-      expected_count = xml.css("sst").first["uniqueCount"].to_i
-      @shared_strings = xml.css("sst si t").map(&:inner_text)
+      expected_count = xml.at_css("sst")["uniqueCount"].to_i
+      @shared_strings = xml.css("sst si").map { |si| si.css("t").map(&:inner_text).join }
 
       if @shared_strings.count != expected_count
-        raise Xsv::Error, "Mismatch in shared strings count! #{expected_count} <> #{@shared_strings.count}"
+        raise Xsv::AssertionFailed, "Mismatch in shared strings count! #{expected_count} <> #{@shared_strings.count}"
       end
 
       stream.close
@@ -58,7 +58,7 @@ module Xsv
 
     def fetch_sheets
       @zip.glob("xl/worksheets/sheet*.xml").sort do |a, b|
-        a.name.scan(/\d+/).first.to_i <=> b.name.scan(/\d+/).first.to_i
+        a.name[/\d+/].to_i <=> b.name[/\d+/].to_i
       end.each do |entry|
         @sheets << Xsv::Sheet.new(self, Nokogiri::XML(entry.get_input_stream))
       end

data/lib/xsv.rb

@@ -7,4 +7,7 @@ require "xsv/workbook"
 
 module Xsv
   class Error < StandardError; end
+  # An AssertionFailed error indicates an unexpected condition, meaning a bug
+  # or misinterpreted .xlsx document
+  class AssertionFailed < StandardError; end
 end

metadata

@@ -1,7 +1,7 @@
 --- !ruby/object:Gem::Specification
 name: xsv
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.2.2
 platform: ruby
 authors:
 - Martijn Storck