booker 1.3.0 → 2.1.0

data/lib/booker/output.rb

@@ -0,0 +1,84 @@
+# frozen_string_literal: true
+
+# terminal width, colored strings, and exiting with a message: the presentation
+# helpers the rest of booker leans on
+
+require "io/console"
+
+module Booker
+  # int number of columns on screen, answered in process rather than by a `tput
+  # cols` subshell on every run. io/console reads /dev/tty rather than stdout,
+  # so a width still comes back through a pipe, and nil when there is no
+  # terminal at all - a completion subshell, or ci
+  module Term
+    def self.width
+      @width ||= IO.console&.winsize&.last&.nonzero? ||
+        ENV["COLUMNS"]&.to_i&.nonzero? ||
+        100
+    end
+  end
+
+  # compl. color codes space
+  CODEWIDTH = 16
+
+  # print a message and stop, for the cli and the installer both reporting a
+  # failure and exiting in one breath. stderr, not stdout: stdout is the data
+  # channel the completion scripts read
+  module Output
+    def pexit(msg, sig = 1)
+      warn msg
+      exit sig
+    end
+  end
+
+  # "Success: ".grn at every call site, but as a refinement rather than an open
+  # class: a gem claiming String#red for the whole process is taking a name it
+  # does not own. files that want these say `using Booker::Colors` at the top
+  module Colors
+    CODES = {red: 31, grn: 32, yel: 33, blu: 34, cyan: 36}.freeze
+
+    class << self
+      # nil means decide per call from the environment; true or false forces it,
+      # which is what the specs do since their $stdout is never a terminal
+      attr_writer :enabled
+
+      # honor NO_COLOR (no-color.org), and skip the escapes when stdout is not a
+      # terminal - otherwise `booker > out.txt` and `booker | less` fill up with
+      # raw \033[ sequences
+      def enabled?
+        return @enabled unless @enabled.nil?
+
+        ENV["NO_COLOR"].to_s.empty? && $stdout.tty?
+      end
+
+      # color by name, reachable without the refinement. a caller holding the
+      # color as data - the browser table's :yel, say - cannot go through the
+      # refined methods, because refinements are invisible to send
+      def paint(str, name)
+        return str unless enabled?
+
+        "\033[0;#{CODES.fetch(name)};49m#{str}\033[0;0m"
+      end
+    end
+
+    refine String do
+      def window(width)
+        if length >= width
+          self[0..width - 1]
+        else
+          ljust(width)
+        end
+      end
+
+      def blu = Colors.paint(self, :blu)
+
+      def cyan = Colors.paint(self, :cyan)
+
+      def yel = Colors.paint(self, :yel)
+
+      def grn = Colors.paint(self, :grn)
+
+      def red = Colors.paint(self, :red)
+    end
+  end
+end

data/lib/booker/parsers.rb

@@ -0,0 +1,294 @@
+# frozen_string_literal: true
+
+# one parser per browser: chrome keeps bookmarks in json, firefox in sqlite, and
+# safari in a binary plist. each turns its own format into Booker::Bookmark
+# objects, so Bookmarks itself never has to care which browser it is reading.
+
+require "json"
+require "fileutils"
+require "open3"
+
+require_relative "output"
+
+using Booker::Colors
+
+module Booker
+  module Parsers
+    # Abstract base class for bookmark parsers
+    class Base
+      attr_reader :allurls
+
+      def initialize(file_path, search_term)
+        @file_path = file_path
+        # escaped, not interpolated: the term comes straight off ARGV, so
+        # `booker "c++"` used to raise RegexpError before it ever searched
+        @searching = Regexp.new(Regexp.escape(search_term), Regexp::IGNORECASE)
+        # an empty term is the completion scripts' bare `booker <TAB>`, and //
+        # matches every field of every bookmark - so answer it once here rather
+        # than running four guaranteed matches per node
+        @match_all = search_term.empty?
+        @allurls = []
+      end
+
+      # Abstract method - must be implemented by subclasses
+      def parse
+        raise NotImplementedError, "Subclasses must implement parse method"
+      end
+
+      def results = @allurls
+
+      protected
+
+      # match? rather than match: this runs on every node of every source, and
+      # the MatchData the latter allocates is thrown away unread
+      def matches_search?(values)
+        @match_all || values.any? { |v| v && @searching.match?(v.to_s) }
+      end
+    end
+
+    # Chrome/Chromium bookmark parser (JSON format)
+    class Chrome < Base
+      def parse
+        begin
+          local_bookmarks = JSON.parse(File.read(@file_path))
+          @chrome_bookmarks = local_bookmarks["roots"]["bookmark_bar"]["children"]
+        rescue
+          warn "Warning: ".yel + "Bookmarks file not found or invalid."
+          warn "Suggest: ".grn + "booker --install bookmarks"
+          return
+        end
+
+        parse_recursive
+      end
+
+      private
+
+      def parse_recursive(root = nil)
+        root = Folder.new(@chrome_bookmarks, "|") if root.nil?
+
+        root.json.each { |x| parse_link(root.title, x) }
+        root.json.each { |x| parse_folder(root, x) }
+      end
+
+      def parse_link(base, link)
+        return unless link["type"] == "url"
+        return unless matches_search?([base, link["name"], link["url"], link["id"]])
+
+        @allurls.push(Bookmark.new(
+          folder: base, title: link["name"], url: link["url"], id: link["id"]
+        ))
+      end
+
+      def parse_folder(base, link)
+        if link["type"] == "folder"
+          title = base.title + link["name"] + "/"
+          subdir = Folder.new(link["children"], title)
+          parse_recursive(subdir)
+        end
+      end
+    end
+
+    # Firefox bookmark parser (SQLite format)
+    class Firefox < Base
+      class << self
+        # "is firefox running" is a question about the machine, not about a
+        # profile, so a config naming three profiles asked pgrep three times -
+        # a fork each, and the slowest single thing booker did on that config
+        attr_writer :running
+
+        def running?
+          return @running unless @running.nil?
+
+          out, _err, _status = Open3.capture3("pgrep", "-x", "firefox")
+          @running = !out.strip.empty?
+        rescue
+          @running = false
+        end
+
+        # the memo outlives a single example, so the suite drops it between them
+        def reset! = @running = nil
+      end
+
+      def parse
+        begin
+          require "sqlite3"
+        rescue LoadError
+          warn "Error: ".red + "sqlite3 gem not installed"
+          warn "Run: ".grn + "gem install sqlite3"
+          return
+        end
+
+        # Copy database to temp file if Firefox is running (to avoid lock).
+        # Tempfile.create removes the copy when the block ends, so there is no
+        # unlink left to remember in an ensure further down
+        if firefox_running?
+          require "tempfile"
+          Tempfile.create(["places", ".sqlite"]) do |temp|
+            temp.close
+            FileUtils.cp(@file_path, temp.path)
+            query_places(temp.path)
+          end
+        else
+          query_places(@file_path)
+        end
+      end
+
+      private
+
+      def query_places(db_path)
+        # the block form closes the handle for us, on the error path too
+        SQLite3::Database.new(db_path) do |db|
+          db.results_as_hash = true
+
+          # Recursive CTE query to build folder paths
+          query = <<-SQL
+            WITH RECURSIVE bookmark_tree(id, parent, title, url, path) AS (
+              -- Start with bookmarks toolbar root
+              SELECT b.id, b.parent, b.title, p.url,
+                     CAST(b.title as TEXT) as path
+              FROM moz_bookmarks b
+              LEFT JOIN moz_places p ON b.fk = p.id
+              WHERE b.parent = (
+                SELECT id FROM moz_bookmarks
+                WHERE guid = 'toolbar_____'
+              )
+
+              UNION ALL
+
+              -- Recursively get children
+              SELECT b.id, b.parent, b.title, p.url,
+                     CAST(bt.path || '/' || b.title as TEXT) as path
+              FROM moz_bookmarks b
+              INNER JOIN bookmark_tree bt ON b.parent = bt.id
+              LEFT JOIN moz_places p ON b.fk = p.id
+            )
+            SELECT id, title, path, url
+            FROM bookmark_tree
+            WHERE url IS NOT NULL
+            ORDER BY path, title
+          SQL
+
+          db.execute(query).each do |row|
+            folder_path = row["path"].to_s.split("/")[0..-2].join("/") + "/"
+            folder_path = "|" + Folder.normalize(folder_path)
+
+            values = [folder_path, row["title"], row["url"], row["id"].to_s]
+            if matches_search?(values)
+              @allurls << Bookmark.new(
+                folder: folder_path,
+                title: row["title"] || "",
+                url: row["url"] || "",
+                id: row["id"].to_s
+              )
+            end
+          end
+        end
+      rescue SQLite3::Exception => e
+        warn "Error: ".red + "Could not read Firefox bookmarks database"
+        warn e.message
+      end
+
+      def firefox_running? = File.exist?(@file_path) && self.class.running?
+    end
+
+    # Safari bookmark parser (binary plist format)
+    class Safari < Base
+      def parse
+        xml = convert_plist_to_xml(@file_path)
+        return if xml.nil?
+
+        begin
+          require "rexml/document"
+          doc = REXML::Document.new(xml)
+          plist_el = doc.root
+          root_node = plist_el.elements.to_a.first
+          root = parse_node(root_node)
+        rescue => e
+          warn "Warning: ".yel + "Could not parse Safari bookmarks: #{e.message}"
+          return
+        end
+
+        walk(root, "|")
+      end
+
+      private
+
+      def convert_plist_to_xml(path)
+        # sniff the head rather than reading the whole file: on the binary path
+        # the contents are thrown away and plutil reads it again from disk, and
+        # a real Bookmarks.plist grows with the user's bookmarks
+        head = File.binread(path, 512).to_s
+
+        # Already an XML plist (fixtures, non-macOS). Binary plists need plutil
+        # (macOS-only); the JSON converter rejects <data>/<date> fields, so xml1.
+        return File.read(path) if head.start_with?("<?xml") || head.include?("<plist")
+
+        # capture3 hands back the status directly, rather than leaving it in the
+        # $? global for the next line to read
+        output, _err, status = Open3.capture3("plutil", "-convert", "xml1", "-o", "-", path)
+        return output if status.success?
+
+        warn "Warning: ".yel + "Could not read Safari bookmarks file."
+        warn "Suggest: ".grn + "grant terminal 'Full Disk Access' in System Settings"
+        nil
+      rescue Errno::ENOENT
+        warn "Warning: ".yel + "Safari bookmarks file not found."
+        nil
+      end
+
+      def parse_node(el)
+        case el.name
+        when "dict"
+          result = {}
+          children = el.elements.to_a
+          children.each_slice(2) do |key_el, val_el|
+            next if key_el.nil? || val_el.nil?
+            result[key_el.text.to_s] = parse_node(val_el)
+          end
+          result
+        when "array"
+          el.elements.map { |c| parse_node(c) }
+        when "string"
+          el.text.to_s
+        when "integer"
+          el.text.to_i
+        when "real"
+          el.text.to_f
+        when "true"
+          true
+        when "false"
+          false
+        when "data", "date"
+          el.text.to_s
+        end
+      end
+
+      def walk(node, folder_title)
+        children = node["Children"]
+        return unless children.is_a?(Array)
+
+        children.each do |child|
+          case child["WebBookmarkType"]
+          when "WebBookmarkTypeLeaf"
+            parse_leaf(folder_title, child)
+          when "WebBookmarkTypeList"
+            sub_title = folder_title + Folder.normalize(child["Title"].to_s) + "/"
+            walk(child, sub_title)
+          end
+        end
+      end
+
+      def parse_leaf(folder, leaf)
+        uri = leaf["URIDictionary"] || {}
+        title = (uri["title"] || "").to_s
+        url = (leaf["URLString"] || "").to_s
+        id = (leaf["WebBookmarkUUID"] || "").to_s
+
+        values = [folder, title, url, id]
+        if matches_search?(values)
+          @allurls << Bookmark.new(folder:, title:, url:, id:)
+        end
+      end
+    end
+  end
+end

data/lib/booker/picker.rb

@@ -0,0 +1,142 @@
+# frozen_string_literal: true
+
+# an external fuzzy finder (fzf by default) as booker's selection ui: candidate
+# lines in on stdin, the chosen one back on stdout, with the finder opening
+# /dev/tty itself for the interface. it knows nothing about bookmarks - lines
+# in, first tab separated field out - which keeps it testable against `head`
+# and `cat` rather than against a real finder.
+
+require "shellwords"
+
+module Booker
+  module Picker
+    # --delimiter/--with-nth hide the id column: it is noise on screen, and
+    # hiding it also takes it out of the match, so typing "2_" does not pull up
+    # every bookmark from the second source. fzf ranks by match score, so a
+    # literal "sofa" beats s..o..f..a scattered across a long url, and
+    # --tiebreak=index keeps booker's ordering among equal scores (and for the
+    # empty query, which fzf never sorts).
+    #
+    # deliberately no --multi: under it fzf reads tab as "mark this and move
+    # down" - the one key booker's own completion trained everybody to press -
+    # so a couple of taps to scroll would mark bookmarks silently and open them
+    # all on the next return.
+    #
+    # a plain array literal, not %w[]: neither the real tab in the delimiter nor
+    # the prompt's trailing space survives %w[]
+    DEFAULT = [
+      "fzf",
+      "--height=40%",
+      "--reverse",
+      "--tiebreak=index",
+      "--delimiter=\t",
+      "--with-nth=2..",
+      "--prompt=bookmark> "
+    ].freeze
+
+    class << self
+      # nil means decide from the environment; true or false forces it. the
+      # specs set this, and so does --list - same shape as Colors.enabled=
+      attr_writer :enabled
+
+      # #enabled? and #select both ask for the command, so an interactive run
+      # worked it out twice; both memos outlive an example, so the suite drops
+      # them between them
+      def reset!
+        @enabled = nil
+        @command = nil
+        @command_known = false
+      end
+
+      def enabled?
+        return @enabled unless @enabled.nil?
+
+        # tty? first: it is a pair of syscalls, where #command reads and parses
+        # ~/.booker.yml, and a run that is not interactive never needs to know
+        # whether a finder is installed
+        tty? && !command.nil?
+      end
+
+      # both ends, not just stdout: `booker foo | less` still writes to a
+      # terminal, but there is no keyboard on the other side of the pipe
+      def tty? = $stdin.tty? && $stdout.tty?
+
+      # the configured picker, or fzf when it is on PATH. nil when neither is
+      # installed, which is the signal to stay on booker's pre-picker behavior
+      def command
+        return @command if @command_known
+
+        @command_known = true
+        @command = resolve_command
+      end
+
+      # no subshell to `which`: this runs before every interactive invocation,
+      # and the answer is a PATH scan booker can do itself. a name carrying a
+      # separator is a path, exactly as execvp treats it
+      def which(bin)
+        return executable?(bin) ? bin : nil if bin.include?(File::SEPARATOR)
+
+        # lazy: the answer is usually in the first few entries, and eagerly
+        # joining all forty PATH directories to throw them away is the one
+        # thing a hand written which is here to avoid
+        ENV.fetch("PATH", "").split(File::PATH_SEPARATOR).lazy
+          .map { |dir| File.join(dir, bin) }
+          .find { |path| executable?(path) }
+      end
+
+      def executable?(path) = File.executable?(path) && !File.directory?(path)
+
+      # candidates in, the id of what was picked out - or nil for every other
+      # outcome: cancelled, nothing matched, no usable finder. one id and not a
+      # list, since dropping --multi above only covers the default finder and
+      # not one named in ~/.booker.yml
+      def select(lines)
+        argv = command
+        return nil if argv.nil?
+
+        io = IO.popen(argv, "r+")
+        feed(io, lines)
+        chosen = io.read
+        io.close
+
+        # fzf: 0 selected, 1 no match, 2 error, 130 interrupted. a selection
+        # that came back empty counts as a cancel too - that pairs with the
+        # finder exiting 0 the moment it is closed, before it has read anything
+        return nil unless Process.last_status&.exitstatus&.zero?
+
+        id = chosen.lines.first.to_s.split("\t", 2).first.to_s.strip
+        id unless id.empty?
+      rescue Errno::ENOENT, Interrupt
+        # the binary went away between the PATH check and the spawn, or the
+        # user interrupted booker itself rather than the finder
+        nil
+      end
+
+      private
+
+      def resolve_command
+        configured = Config.default.picker
+
+        # yaml reads a bare `:picker: false` as the boolean, and that means
+        # "leave the picker off" - otherwise it looks like nothing was
+        # configured and quietly starts fzf instead. blank says the same thing
+        return nil if configured == false
+
+        argv = configured ? Shellwords.split(configured.to_s) : DEFAULT
+        (!argv.empty? && which(argv.first)) ? argv : nil
+      end
+
+      # one write, not one per bookmark. IO.popen hands back a synced IO, so
+      # `puts` per line is an unbuffered syscall each - measured 10x slower
+      # than a single write at a few thousand bookmarks
+      def feed(io, lines)
+        io.write(lines.join("\n") + "\n") unless lines.empty?
+        io.close_write
+      rescue Errno::EPIPE
+        # expected, not exceptional: the finder exits the instant escape is
+        # pressed, and with a few hundred bookmarks booker is usually still
+        # writing. the exit status above is what says what happened
+      end
+    end
+  end
+end

data/lib/booker/version.rb

@@ -0,0 +1,5 @@
+# frozen_string_literal: true
+
+module Booker
+  VERSION = "2.1.0"
+end