Ruby script to build a Dash docset for Gambit-C (http://gambitscheme.org)

build-gsc-docset

#!/usr/bin/env ruby

require 'fileutils'
require 'sqlite3'
require 'cgi'


HELP_TEXT = <<-EOS
  build-gsc-docset [options...]

  Options:
    -help       Displays this text and exits.

    -verbose    Log most things.

    -log-sql    Especially verbose -- logs SQL statements.
                (Implies -verbose.)

    -nop        Dry run. This still creates a DB in memory to ensure DB
                ops are correct.
                (Implies -verbose.)

    -rebuild    Forces a complete rebuild of the docset (in-place). This
                includes re-downloading the Gambit-C manual HTML unless
                -skip-dl is specified.

    -skip-dl    If specified, the manual download is skipped. This is
                only really useful if you've made changes to the script
                to categorize entry types but don't want to redownload
                the manual's HTML file. Highly recommended you use this
                when messing with things so you don't spam Gambit's over
                and over.

EOS


class String
  def min_indent
    each_line.select { |l| l.chomp.length > 0 }.map { |l| l[/^ */].length }.min
  end

  def unindented
    to_strip = 0 ... min_indent
    if to_strip.size == 0
      dup
    else
      each_line.map { |l, r| l.chomp.length > 0 ? l.dup.tap { |k| k[to_strip] = '' } : l }.join ''
    end
  end
end


if ARGV.include?('-help')
  puts HELP_TEXT.unindented
  exit 0
end

nop     = ARGV.include?('-nop')
log_sql = ARGV.include?('-log-sql')
verbose = log_sql || nop || ARGV.include?('-verbose')
rebuild = ARGV.include?('-rebuild')
skip_dl = ARGV.include?('-skip-dl')

if verbose
  def log(*args) ; $stderr.puts(*args) ; end
else
  def log(*) ; end
end

INDENT              = 4
GSC_MANUAL_URL      = 'http://www.iro.umontreal.ca/~gambit/doc/gambit-c.html'
GSC_MANUAL_FILE     = 'Gambit-C.docset/Contents/Resources/Documents/index.html'
GSC_INDEX_DB        = 'Gambit-C.docset/Contents/Resources/docSet.dsidx'
GSC_INFO_PLIST      = 'Gambit-C.docset/Contents/Info.plist'

INFO_PLIST_SOURCE   = <<-ENDPLIST.unindented
  <?xml version="1.0" encoding="UTF-8"?>
  <!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
  <plist version="1.0">
  <dict>
    <key>CFBundleIdentifier</key>
    <string>gambit</string>
    <key>CFBundleName</key>
    <string>Gambit-C</string>
    <key>DocSetPlatformFamily</key>
    <string>gambit</string>
    <key>isDashDocset</key>
    <true/>
  </dict>
  </plist>
ENDPLIST

DOWNLOAD_GSC_MANUAL = %Q[curl '#{GSC_MANUAL_URL}' > '#{GSC_MANUAL_FILE}']
DOCSET_DIRECTORIES  = %w[
  Gambit-C.docset/Contents/Resources/Documents
]


SQL_CREATE_TABLES = <<-SQL
  CREATE TABLE searchIndex(id INTEGER PRIMARY KEY, name TEXT, type TEXT, path TEXT);
  CREATE UNIQUE INDEX anchor ON searchIndex (name, type, path);
SQL

SQL_CREATE_ENTRY = <<-SQL
  INSERT OR IGNORE INTO searchIndex(name, type, path) VALUES (?, ?, ?);
SQL

def log_insert(*vars)
  vars.map! { |k| "'#{SQLite3::Database.quote(k)}'" }
  log "  INSERT OR IGNORE INTO searchIndex(name, type, path) VALUES (#{vars.join ', '});"
end

INDEX_ENTRY_REGEX   = %r[<td valign="top"><a href="#(?<link_id>index-[^"]+)"><code>(?<link_name>.+?)</code></a></td>]i
INDEX_SECTION_REGEX = %r[<tr><td align="left" valign="top"><a href="#(?<link_id>[^"]+)">\d+\.(\d+\.?)*\s*(?<link_name>.+?)</a></td><td>&nbsp;&nbsp;]i

log "Creating docset directories."
FileUtils.mkdir_p(DOCSET_DIRECTORIES, noop: nop)

if !skip_dl && (rebuild || !File.exists?(GSC_MANUAL_FILE))
  log DOWNLOAD_GSC_MANUAL

  if File.exists?(GSC_MANUAL_FILE)
    log "Deleting old manual for rebuild."
    FileUtils.rm(GSC_MANUAL_FILE) unless nop
  end

  log "Downloading Gambit-C manual."
  unless nop
    system DOWNLOAD_GSC_MANUAL
    puts $?
  end
else
  log "'#{GSC_MANUAL_FILE}' already exists"
end

have_manual = File.exists?(GSC_MANUAL_FILE)
unless nop || have_manual
  $stderr.puts "Unable to build docset index -- no manual file found"
  exit 1
end

entries = []

log "Scanning '#{GSC_MANUAL_FILE}' for entries."
File.open(GSC_MANUAL_FILE, 'rb') do |io|
  io.each_line do |line; link_name, link_id|
    case line
    when INDEX_ENTRY_REGEX
      link_name = $~[:link_name]
      link_id   = $~[:link_id]
      link_type = 'Function'

    when INDEX_SECTION_REGEX
      link_name = $~[:link_name]
      link_id   = $~[:link_id]
      link_type = 'Guide'

    else next
    end

    link_name.gsub!(%r[</?.+?/?>], '')
    link_name = CGI.unescapeHTML(link_name)

    if link_type == 'Function'
      case link_name
      when /^[,\^]/ then link_type = 'Command'
      when /^-/ then link_type = 'Option'
      end
    end

    entries << [link_name, link_type, "index.html##{link_id}"]
  end
end if have_manual

if rebuild || !File.exists?(GSC_INFO_PLIST)
  log "Building '#{GSC_INFO_PLIST}'."
  File.open(GSC_INFO_PLIST, 'wb') { |io| io.write(INFO_PLIST_SOURCE) } unless nop
end

if File.exists?(GSC_INDEX_DB)
  log "Removing old '#{GSC_INDEX_DB}'."
  FileUtils.rm(GSC_INDEX_DB) unless nop
end

log "Create DB '#{GSC_INDEX_DB}'"

db = SQLite3::Database.new(nop ? ':memory:' : GSC_INDEX_DB)
log "Create tables."
log SQL_CREATE_TABLES if log_sql
db.execute SQL_CREATE_TABLES

insert_stmt = db.prepare(SQL_CREATE_ENTRY)

log "Insert table entries."

db.transaction do
  entries.each do |entry|
    log_insert(*entry) if log_sql
    insert_stmt.execute(*entry)
  end
end

log "Done."

log "Closing DB."
insert_stmt.close
db.close