fyodor 0.2.1 → 0.2.6

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: f14c74a8744a5b6ecc59dbada1fb738913a965fac085dbdd6b1179fda4c4aaa1
-  data.tar.gz: dbdea2a6e45f4c8a90cf435cc879f5408838590308e56f540efbe566de8434dd
+  metadata.gz: dae694865dbdb0187906ad21b610ac7f9486819141f86392816a41fbe835f71c
+  data.tar.gz: 97c07e5ba51877015426221aed2158a20436f36d8afbca165b8433c20050e02a
 SHA512:
-  metadata.gz: 7e98fb7efcd144c55cebf5608d63143d6fc4f0e7cf734a482f918d653aa33a8368193ac50c860001c78006f2fb6332a35daa5647746a87b8d268a34318cf6fac
-  data.tar.gz: d8fb7b506b1f911be98005be170e3db68c526a7490462eb213a0f1d132d21f88aecd5374e529730c914d1fa84a4080e81eed5ff824bd9114821e9be5e0cbf415
+  metadata.gz: ec0b9492639f2b77ba2523106db931b3363c42731cab3b456f83e6c7e91ccc6ec196d4cbe9d8798f62a5c452eed391fec521d7e8da56c537be51a224e93963bc
+  data.tar.gz: 359197fe0aeb85d76f1d025c2e3517a7fb02c60712989f79dd14be2004f50f3892a312b4114a3d29245d26a8f2e72f59dd4a2925c7c800f3dc63b435d63e4952

data/README.md

@@ -6,7 +6,7 @@ Convert your Amazon Kindle highlights, notes and bookmarks into markdown files.
 
 This application parses `My Clippings.txt` from your Kindle and generates a markdown file for each book/document, in the format `#{Author} - #{Title}.md`. This way, your annotations are conveniently stored and easily managed.
 
-[For samples of the output, click here.](samples/)
+[For samples of the output, click here.](docs/output_demo)
 
 To read more about the motivation and what problem it tries to solve, [check this blog post](http://rafaelc.org/blog/export-all-your-kindle-highlights-and-notes/).
 
@@ -30,18 +30,26 @@ Install Ruby and run:
 $ gem install fyodor
 ```
 
+## Updating
+
+Run:
+
+```
+$ gem update fyodor
+```
+
 ## Configuration
 
-Fyodor has an optional configuration file, which is used for the following features.
+Fyodor has an optional configuration file, which is used for the following.
 
 ### Languages
 
-If your Kindle is not in English, you should tell Fyodor how some things are called by your `My Clippings.txt` (e.g. highlights, pages, etc). _Note that basic parsing should still work without configuration, but you won't take advantage of many features, resulting in a dirtier output._
+If your Kindle is not in English, you should tell Fyodor how some things are called by your `My Clippings.txt` (e.g. highlights, pages, etc). _Fyodor should still work without configuration, but you won't take advantage of many features, resulting in a dirtier output._
 
 1. Download the sample config to `~/.config/fyodor.toml` or `$XDG_CONFIG_HOME/fyodor.toml`:
 
 ```
-$ wget https://raw.githubusercontent.com/rccavalcanti/fyodor/master/fyodor.toml.sample -O ~/.config/fyodor.toml
+$ wget https://raw.githubusercontent.com/rccavalcanti/fyodor/master/docs/fyodor.toml.sample -O ~/.config/fyodor.toml
 ```
 
 2. Open both the configuration and your `My Clippings.txt` in your preferred editor. Change the values in the `[parser]` section to mirror what you get in `My Clippings.txt`.
@@ -74,8 +82,8 @@ Where:
 - `CLIPPINGS_FILE` is the path for `My Clippings.txt`.
 - `OUTPUT_DIR` is the directory to write the markdown files. If none is supplied, this will be `fyodor_output` under the current directory.
 
-## LICENSE
+## License
 
-Released under [GNU GPL v3](LICENSE).
+Licensed under [GPLv3](LICENSE)
 
-Copyright 2019 Rafael Cavalcanti <hi@rafaelc.org>
+Copyright (C) 2019-2020 [Rafael Cavalcanti](https://rafaelc.org/)

data/bin/fyodor

@@ -1,15 +1,59 @@
 #!/usr/bin/env ruby
 
-require "fyodor/cli"
-
-# Pass --trace to get the stack trace
-if ARGV.include?("--trace")
-  ARGV.delete("--trace")
-  Fyodor::CLI.new.main
-else
-  begin
-    Fyodor::CLI.new.main
-  rescue StandardError => e
-    abort(e.message)
+require "fyodor"
+require "fyodor/config_getter"
+require "fyodor/stats_printer"
+require "fyodor/version"
+require "optimist"
+require "pathname"
+
+module Fyodor
+  class CLI
+    def main
+      parse_args
+      @config = ConfigGetter.new.config
+
+      if @cli_opts.trace
+        run
+      else
+        begin
+          run
+        rescue StandardError => e
+          abort(e.message)
+        end
+      end
+    end
+
+    private
+
+    def get_path(path)
+      Pathname.new(path).expand_path
+    end
+
+    def default_output_dir
+      Pathname.new(Dir.pwd) + "fyodor_output"
+    end
+
+    def run
+      library = Library.new
+      ClippingsParser.new(@clippings_path, @config["parser"]).parse(library)
+      StatsPrinter.new(library).print
+      OutputWriter.new(library, @output_dir, @config["output"]).write_all
+    end
+
+    def parse_args
+      @cli_opts = Optimist::options do
+        version "Fyodor v#{Fyodor::VERSION}"
+        synopsis "Usage: fyodor [options] my_clippings_path [output_dir]"
+
+        opt :trace, "Show backtrace", :default => false
+      end
+
+      Optimist::die "Wrong number of arguments." if ARGV.count != 1 && ARGV.count != 2
+      @clippings_path = get_path(ARGV[0])
+      @output_dir = ARGV[1].nil? ? default_output_dir : get_path(ARGV[1])
+    end
   end
 end
+
+Fyodor::CLI.new.main

data/lib/fyodor/clippings_parser.rb

@@ -3,7 +3,6 @@ require_relative "entry_parser"
 module Fyodor
   class ClippingsParser
     SEPARATOR = /^==========\r?\n$/
-    ENTRY_SIZE = 5
 
     def initialize(clippings_path, parser_config)
       @path = clippings_path
@@ -25,9 +24,7 @@ module Fyodor
     private
 
     def end_entry?(lines)
-      return false if lines.size < ENTRY_SIZE
-      return true if lines.size == ENTRY_SIZE && lines.last =~ SEPARATOR
-      raise "MyClippings is badly formatted"
+      lines.last =~ SEPARATOR
     end
 
     def parse_entry(lines)

data/lib/fyodor/entry.rb

@@ -1,23 +1,35 @@
 module Fyodor
   class Entry
-    TYPE = { note: "note",
-            highlight: "highlight",
-            bookmark: "bookmark",
-            clip: "clip" }
-
-    attr_reader :book_title, :book_author, :text, :desc, :type, :loc, :loc_start, :page, :time
-
-    def initialize(attrs)
-      @book_title = attrs[:book_title]
-      @book_author = attrs[:book_author]
-      @text = attrs[:text]
-      @desc = attrs[:desc]
-      @type = attrs[:type]
-      @loc = attrs[:loc]
-      # This is our comparable, we need it as a number.
-      @loc_start = attrs[:loc_start].to_i
-      @page = attrs[:page]
-      @time = attrs[:time]
+    TYPE = {
+      note: "note",
+      highlight: "highlight",
+      bookmark: "bookmark",
+      clip: "clip"
+    }
+
+    ATTRS = [
+      :book_title,
+      :book_author,
+      :text,
+      :desc,
+      :type,
+      :loc,
+      :loc_start,
+      :page,
+      :page_start,
+      :time
+    ]
+
+    attr_reader *ATTRS
+
+    def initialize(args)
+      ATTRS.each do |attr|
+        instance_variable_set("@#{attr}", args[attr])
+      end
+
+      # These are our comparables. Let's make sure they are numbers.
+      @loc_start = @loc_start.to_i
+      @page_start = @page_start.to_i
 
       raise ArgumentError, "Invalid Entry type" unless TYPE.value?(@type) || @type.nil?
     end
@@ -31,11 +43,13 @@ module Fyodor
     end
 
     def desc_parsed?
-      @loc_start != 0 && ! @type.nil?
+      ! @type.nil? && (@loc_start != 0 || @page_start != 0)
     end
 
     # Override this method to use a SortedSet.
     def <=>(other)
+      return @page_start <=> other.page_start if @loc_start == 0
+
       @loc_start <=> other.loc_start
     end
 
@@ -44,7 +58,7 @@ module Fyodor
       return false if @type != other.type || @text != other.text
 
       if desc_parsed? && other.desc_parsed?
-        @loc == other.loc
+        @loc == other.loc && @page == other.page
       else
         @desc == other.desc
       end
@@ -54,7 +68,7 @@ module Fyodor
 
     def hash
       if desc_parsed?
-        @text.hash ^ @type.hash ^ @loc.hash
+        @text.hash ^ @type.hash ^ @loc.hash ^ @page.hash
       else
         @text.hash ^ @desc.hash
       end

data/lib/fyodor/entry_parser.rb

@@ -9,26 +9,37 @@ module Fyodor
     end
 
     def entry
-      Entry.new({book_title: book[:title],
-                book_author: book[:author],
-                text: text,
-                desc: desc,
-                type: type,
-                loc: loc,
-                loc_start: loc_start,
-                page: page,
-                time: time})
+      Entry.new({
+        book_title: book[:title],
+        book_author: book[:author],
+        text: text,
+        desc: desc,
+        type: type,
+        loc: loc,
+        loc_start: loc_start,
+        page: page,
+        page_start: page_start,
+        time: time
+      })
     end
 
 
     private
 
     def book
-      title, author = @lines[0].scan(regex_cap(:title_author)).first
+      regex = /^(.*) \((.*)\)\r?\n$/
+
+      title, author = @lines[0].scan(regex).first
       # If book has no author, regex fails.
-      title = @lines[0] if title.nil?
+      if title.nil?
+        title = @lines[0]
+        author = ""
+      end
+
+      title = rm_zero_width_chars(title).strip
+      author = rm_zero_width_chars(author).strip
 
-      {title: title.strip, author: author.to_s.strip}
+      {title: title, author: author}
     end
 
     def desc
@@ -36,58 +47,61 @@ module Fyodor
     end
 
     def type
-      Entry::TYPE.values.find { |type| @lines[1] =~ regex_type(type) }
+      Entry::TYPE.values.find do |type|
+        keyword = Regexp.quote(@config[type])
+        regex = /^- #{keyword}/i
+
+        @lines[1] =~ regex
+      end
     end
 
     def loc
-      @lines[1][regex_cap(:loc), 1]
+      keyword = Regexp.quote(@config["loc"])
+      regex = /#{keyword} (\S+)/i
+
+      @lines[1][regex, 1]
     end
 
     def loc_start
-      @lines[1][regex_cap(:loc_start), 1].to_i
+      keyword = Regexp.quote(@config["loc"])
+      regex = /#{keyword} (\d+)(-\d+)?/i
+
+      @lines[1][regex, 1].to_i
     end
 
     def page
-      @lines[1][regex_cap(:page), 1]
-    end
+      keyword = Regexp.quote(@config["page"])
+      regex = /#{keyword} (\S+)/i
 
-    def time
-      @lines[1][regex_cap(:time), 1]
+      @lines[1][regex, 1]
     end
 
-    def text
-      @lines[3].strip
+    def page_start
+      keyword = Regexp.quote(@config["page"])
+      regex = /#{keyword} (\d+)(-\d+)?/i
+
+      @lines[1][regex, 1].to_i
     end
 
-    def regex_type(type)
-      s = Regexp.quote(@config[type])
-      /^- #{s}/
+    def time
+      keyword = Regexp.quote(@config["time"])
+      regex = /#{keyword} (.*)\r?\n$/i
+
+      @lines[1][regex, 1]
     end
 
-    def regex_cap(item)
-      case item
-      when :title_author
-        /^(.*) \((.*)\)\r?\n$/
-      when :loc
-        s = Regexp.quote(@config["loc"])
-        /#{s} (\S+)/
-      when :loc_start
-        s = Regexp.quote(@config["loc"])
-        /#{s} (\d+)(-\d+)?/
-      when :page
-        s = Regexp.quote(@config["page"])
-        /#{s} (\S+)/
-      when :time
-        s = Regexp.quote(@config["time"])
-        /#{s} (.*)\r?\n$/
-      end
+    def text
+      @lines[3..-2].join.strip
     end
 
     def format_check
-      raise "Entry must have five lines" unless @lines.size == 5
       raise "Entry is badly formatted" if @lines[0].strip.empty?
       raise "Entry is badly formatted" if @lines[1].strip.empty?
       raise "Entry is badly formatted" unless @lines[2].strip.empty?
     end
+
+    def rm_zero_width_chars(str)
+      str.gsub(/[\u200B-\u200D\uFEFF]/, '')
+    end
   end
 end

data/lib/fyodor/md_generator.rb

@@ -52,25 +52,24 @@ module Fyodor
       output = "---\n\n"
       output += "## #{title}\n\n" unless title.nil?
       entries.each do |entry|
-        output += "#{entry_text(entry)}\n\n"
-        output += "<p style=\"text-align: right;\"><sup>#{entry_desc(entry)}</sup></p>\n\n"
+        output += "#{item_text(entry)}\n\n"
+        output += "<p style=\"text-align: right;\"><sup>#{item_desc(entry)}</sup></p>\n\n"
       end
       output
     end
 
-    def entry_text(entry)
+    def item_text(entry)
       case entry.type
       when Entry::TYPE[:bookmark]
         "* #{page(entry)}"
       when Entry::TYPE[:note]
-
-        "> _#{text(entry)}_"
+        "* _Note:_\n#{entry.text.strip}"
       else
-        "> #{text(entry)}"
+        "* #{entry.text.strip}"
       end
     end
 
-    def entry_desc(entry)
+    def item_desc(entry)
       return entry.desc unless entry.desc_parsed?
 
       case entry.type
@@ -93,10 +92,5 @@ module Fyodor
     def type(entry)
       SINGULAR[entry.type]
     end
-
-    def text(entry)
-      # Markdown needs no white space between text and formatters
-      entry.text.strip
-    end
   end
 end

data/lib/fyodor/version.rb

@@ -1,3 +1,3 @@
 module Fyodor
-  VERSION = "0.2.1".freeze
+  VERSION = "0.2.6".freeze
 end

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: fyodor
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.2.6
 platform: ruby
 authors:
 - Rafael Cavalcanti
 autorequire: 
 bindir: bin
 cert_chain: []
-date: 2020-01-08 00:00:00.000000000 Z
+date: 2020-06-30 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: toml
@@ -36,7 +36,6 @@ files:
 - bin/fyodor
 - lib/fyodor.rb
 - lib/fyodor/book.rb
-- lib/fyodor/cli.rb
 - lib/fyodor/clippings_parser.rb
 - lib/fyodor/config_getter.rb
 - lib/fyodor/core_extensions/hash/merging.rb
@@ -68,7 +67,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.0.6
+rubygems_version: 3.1.2
 signing_key: 
 specification_version: 4
 summary: Kindle clippings parser

data/lib/fyodor/cli.rb

@@ -1,42 +0,0 @@
-require "fyodor/config_getter"
-require "fyodor/stats_printer"
-require "fyodor/library"
-require "fyodor/clippings_parser"
-require "fyodor/output_writer"
-require "pathname"
-
-module Fyodor
-  class CLI
-    USAGE = "Usage: #{File.basename($0)} my_clippings_path [output_dir]"
-
-    def initialize
-      get_args
-      @config = ConfigGetter.new.config
-    end
-
-    def main
-      library = Library.new
-      ClippingsParser.new(@clippings_path, @config["parser"]).parse(library)
-      StatsPrinter.new(library).print
-      OutputWriter.new(library, @output_dir, @config["output"]).write_all
-    end
-
-
-    private
-
-    def get_args
-      abort(USAGE) if ARGV.count != 1 && ARGV.count != 2
-
-      @clippings_path = get_path(ARGV[0])
-      @output_dir = ARGV[1].nil? ? default_output_dir : get_path(ARGV[1])
-    end
-
-    def get_path(path)
-      Pathname.new(path).expand_path
-    end
-
-    def default_output_dir
-      Pathname.new(Dir.pwd) + "fyodor_output"
-    end
-  end
-end