easy_sheet_io 0.1.4 → 0.2.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: 791e3662406be26d6afb90f394f0d585b7c9e956d4cdc200db484be820af4eb8
-  data.tar.gz: 7796eacac61646ec19c3d452b9ff738de37546b2a7360374a9f9d5fd9fabda8a
+  metadata.gz: 6de67291cc58a0ab7ea017efd79e847351f71e6790f5f2e4e3fb4462c2ff5f47
+  data.tar.gz: aa0dca360cfa89666b291d8ddeea149399068fbf57f12c422e615c5da03ac71e
 SHA512:
-  metadata.gz: 991506127b66bfdd211f99ec00554dcca3ad521274d3c712389d3f346cf35f4d0b1c5e0478817f9ead1a915aa89c34a88abf49502ee054d30a41a02f19068510
-  data.tar.gz: 6eb892fa842a1a522e939c432e66c0aef9384b37b4bae3b454bb9a82628a2dbd3b3c1cb3d13192b71cf915a7b886858fbb31331b5fa6f03a148baacefd5a611b
+  metadata.gz: 6cadf3040934873c1a9182000d6b90a9fbad20056e1bd0c17e2fe21929643a2a438832e4fdcdb5519b1def845925526cc3c4df460346affe397a8c05bddf95ce
+  data.tar.gz: 3d7df3a901737bdca5ee7cf8e9399b2d958d51134328ea5ee33ac5e17126847c54ff3e20617ad26b073029afbc790e31e10bef66d18b2d74429f36db417ddef9

data/Gemfile

@@ -16,3 +16,4 @@ gem "rover-df"
 gem "smarter_csv"
 gem "roo-xls"
 gem "spreadsheet"
+gem "ast"

data/Gemfile.lock

@@ -1,7 +1,7 @@
 PATH
   remote: .
   specs:
-    easy_sheet_io (0.1.3)
+    easy_sheet_io (0.2.1)
 
 GEM
   remote: https://rubygems.org/
@@ -72,6 +72,7 @@ PLATFORMS
   x86_64-linux
 
 DEPENDENCIES
+  ast
   daru
   easy_sheet_io!
   rake (~> 13.0)

data/easy_sheet_io.gemspec

@@ -9,7 +9,7 @@ Gem::Specification.new do |spec|
   spec.email         = ["shun_yamaguchi_tc@live.jp"]
 
   spec.summary       = "A simple way to Open .csv, .xls, .xlsx files."
-  spec.description   = "A simple way to Open .csv, .xls, .xlsx files. You can convert it to 2D Array, Hash, Dataframe."
+  spec.description   = "A simple way to Open .csv, .xls, .xlsx files. You can convert it to 2D array, hash, data frame."
   spec.homepage      = "https://github.com/show-o-atakun/easy_sheet_io"
   spec.license       = "MIT"
   spec.required_ruby_version = Gem::Requirement.new(">= 2.3.0")

data/lib/easy_sheet_io/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module EasySheetIo
-  VERSION = "0.1.4"
+  VERSION = "0.2.2"
 end

data/lib/easy_sheet_io.rb

@@ -17,22 +17,29 @@ module EasySheetIo
 		return /csv$/ === path ? read_csv(path, **opt) : read_excel(path, **opt)
 	end
 	
-	# ##Generate DF from CSV File
+	# ##Generate Array from CSV File, and convert it to Hash or DataFrame.
 	# **opt candidate= line_from: 1, header: 0
-	def read_csv(path, format: nil, encoding: "utf-8", **opt)
-		csv = CSV.parse(File.open path, encoding: encoding, &:read) # Get 2D Array
+	def read_csv(path, format: nil, encoding: "utf-8", col_sep: ",", **opt)
+		# Get 2D Array
+		begin
+			csv = CSV.parse(File.open(path, encoding: encoding, &:read), col_sep: col_sep)
+		rescue Encoding::InvalidByteSequenceError
+			# Try Another Encoding
+			puts "Fail Encoding #{encoding}. Trying cp932..."
+			csv = CSV.parse(File.open(path, encoding: "cp932", &:read), col_sep: col_sep)
+		end
+		
 		return csv if format.nil?
 
+		# Convert Hash or DataFrame
 		ans = to_hash(csv, **opt)
 		return format==:hash || format=="hash" ? ans : to_df(ans, format: format)
 	end
 
-	# ##Generate DF from Excel File
+	# ##Generate Array from EXCEL File, and convert it to Hash or DataFrame.
 	# **opt candidate= line_from: 1, header: 0)
-	# !encoding parameter is not allowed yet
-	# !(Finally, I want to make it automatically recognize encoding of file).
-	def read_excel(path, sheet_i: 0, format: nil, **opt)
-		a2d = open_excel(path, sheet_i) # Get 2D Array
+	def read_excel(path, sheet_i: 0, format: nil, encoding: "utf-8", **opt)
+		a2d = open_excel(path, sheet_i, encoding: encoding) # Get 2D Array
 		return a2d if format.nil?
 		
 		ans = to_hash(a2d, **opt)
@@ -41,11 +48,31 @@ module EasySheetIo
 	
 	# Convert 2d Array to Hash
 	# ##header: nil -> Default Headers(:column1, column2,...) are generated.
-	def to_hash(array2d, line_from: 1, line_until: -1, header: 0)
-		output = array2d[line_from..line_until]
-		hd = header.nil? ? [*0...(output.longest_line)].map{"column#{_1}"} : array2d[header]
-		output_transpose = output[0].zip(*output[1..])
+	# line_until=nil means the data are picked up until the end line.
+	def to_hash(array2d, line_from: 1, line_until: nil, header: 0, symbol_header: false)
+		
+		# Define Read Range------------		
+		lfrom, luntil = line_from, line_until
+		lf_reg, lu_reg = line_from.kind_of?(Regexp), line_until.kind_of?(Regexp)
 		
+		if lf_reg || lu_reg
+			lines_ary = array2d.map{ _1.join "," }
+			lfrom = lines_ary.find_index{ line_from === _1 } if lf_reg
+			luntil = (lines_ary.length-1) - lines_ary.reverse.find_index{ line_until === _1 } if lu_reg
+		end
+		# -----------------------------
+			
+		# Define Data Array------------
+		output = array2d[lfrom...luntil]
+		output_transpose = output[0].zip(*output[1..])
+		# -----------------------------
+
+		# Define Header----------------
+		hd = header.nil? ? [*0...(output.longest_line)].map{"column#{_1}"} : check_header(array2d[header])
+		hd = hd.map { _1.intern } if symbol_header
+		# -----------------------------
+
+		# Make Hash(Header => Data Array)  
 		return hd.each_with_object({}).with_index {|(hdr, hash), i| hash[hdr]=output_transpose[i]}
 	end
 	
@@ -58,19 +85,27 @@ module EasySheetIo
 		end
 	end
 	
-	# ##Genarate Hash from excel file
-	def open_excel(path, sheet_i)
-		begin
-			book = /xlsx$/ === path ? Roo::Excelx.new(path) : Roo::Excel.new(path)
+	# ##Genarate Array from excel file
+	def open_excel(path, sheet_i, encoding: "utf-8")
+		if /xlsx$/ === path
+			puts "Sorry, encoding option is not supported yet for xlsx file." if encoding != "utf-8"
+
+			book = Roo::Excelx.new(path)
 			s = book.sheet(sheet_i)
 			
-			## bottole neck===
+			## bottole neck
 			return s.to_a
-			
-		rescue Encoding::InvalidByteSequenceError
-		
-			Spreadsheet.client_encoding="Windows-31J"
-			ss = Spreadsheet.open(path)
+
+		# xls
+		else
+			begin
+				Spreadsheet.client_encoding = encoding
+				ss = Spreadsheet.open(path)
+			rescue Encoding::InvalidByteSequenceError
+				puts "Fail Encoding #{encoding}. Trying Windows-31J..."
+				Spreadsheet.client_encoding = "Windows-31J"
+				ss = Spreadsheet.open(path)
+			end
 
 			a2d = []
 			ss.worksheets[sheet_i].rows.each do |row|
@@ -81,6 +116,28 @@ module EasySheetIo
 
 			return a2d
 		end
+	end
+
+	# Fix blank or duplicated header
+	def check_header(header_array)
+		ans = header_array.map.with_index do |item, i|
+			if item.nil?
+				"column#{i}"
+			elsif item.kind_of?(String)
+				/^\s*$/ === item ? "column#{i}" : item.gsub(/\s+/, "")
+			else
+				item
+			end
+		end 
+
+		dup_check = (0...(header_array.length)).group_by {|i| ans[i]}
+		dup_check.each do |item, i_s|
+			if i_s.length > 1
+				i_s.each_with_index {|i, index_in_i_s| ans[i] = "#{ans[i]}_#{index_in_i_s}"}
+			end
+		end
 
+		return ans
 	end
+
 end

metadata

@@ -1,17 +1,17 @@
 --- !ruby/object:Gem::Specification
 name: easy_sheet_io
 version: !ruby/object:Gem::Version
-  version: 0.1.4
+  version: 0.2.2
 platform: ruby
 authors:
 - show-o-atakun
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2022-03-22 00:00:00.000000000 Z
+date: 2022-03-23 00:00:00.000000000 Z
 dependencies: []
 description: A simple way to Open .csv, .xls, .xlsx files. You can convert it to 2D
-  Array, Hash, Dataframe.
+  array, hash, data frame.
 email:
 - shun_yamaguchi_tc@live.jp
 executables: []