Module: Wp2txt::IndexCommands

Defined in:
lib/wp2txt/index_commands.rb

Overview

CLI command handlers for --build-index and --find-articles. Mixed into WpApp (bin/wp2txt); relies on CliUI helpers for output.

Instance Method Summary collapse

Instance Method Details

#build_fulltext_index(opts, multistream_path, stream_offsets, meta_db_path, num_processes) ⇒ Object

Build the FTS5 full-text index (Tier 2) after the metadata index



131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
 
# File 'lib/wp2txt/index_commands.rb', line 131

def build_fulltext_index(opts, multistream_path, stream_offsets, meta_db_path, num_processes)
  fts_db_path = FtsIndex.path_for(multistream_path, cache_dir: opts[:cache_dir])
  tokenizer = opts[:fts_tokenizer] || FtsIndex.default_tokenizer(multistream_path)

  fts = FtsIndex.new(fts_db_path, meta_db_path)
  if fts.valid_for?(multistream_path) && !opts[:update_cache]
    print_success("Full-text index is up to date: #{fts_db_path}")
    fts.close
    return CliUI::EXIT_SUCCESS
  end
  fts.close

  puts pastel.cyan("Building full-text index (tokenizer: #{tokenizer})...") unless quiet?
  time_start = Time.now
  ok = run_phase_isolated do
    builder = FtsIndexBuilder.new(
      multistream_path, stream_offsets,
      db_path: fts_db_path, meta_db_path: meta_db_path,
      tokenizer: tokenizer, num_processes: num_processes,
      optimize: !opts[:skip_fts_optimize]
    )

    last_report = Time.now
    built = builder.build do |done, total|
      now = Time.now
      if !quiet? && (now - last_report >= DEFAULT_PROGRESS_INTERVAL || done == total)
        last_report = now
        percent = (done.to_f / total * 100).round(1)
        elapsed = now - time_start
        eta = done.positive? ? (total - done) * (elapsed / done) : 0
        puts pastel.dim(format("  [%d/%d] %.1f%% | ETA: %s", done, total, percent, format_duration(eta)))
      end
    end

    puts unless quiet?
    print_success("Full-text index built in #{format_duration(Time.now - time_start)}")
    stats = built.stats
    print_info("Tokenizer", stats[:tokenizer].to_s)
    print_info("Sections", stats[:section_count].to_s)
    print_info("Size", format_size(stats[:db_size]))
    unless stats[:optimized]
      print_info_message("Index is unoptimized (built with --skip-fts-optimize). Run 'wp2txt --fts-optimize' later for best query speed.")
    end
    built.close
    true
  end
  ok ? CliUI::EXIT_SUCCESS : CliUI::EXIT_ERROR
end

#load_stream_offsets(index_path, opts) ⇒ Array(Array<Integer>, Integer)

Load stream offsets without holding the full title index in memory. The builders fork worker processes; a parent heap holding millions of index entries (~14GB for enwiki) gets duplicated through copy-on-write breakage in every child and OOMs the machine. Offsets are a small integer array, so prefer reading just them from the SQLite cache.

Returns:

  • (Array(Array<Integer>, Integer)) —

    [stream_offsets, entry_count]



113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
 
# File 'lib/wp2txt/index_commands.rb', line 113

def load_stream_offsets(index_path, opts)
  cache = IndexCache.new(index_path, cache_dir: opts[:cache_dir])
  if cache.valid?
    [cache.stream_offsets, cache.stats[:entry_count]]
  else
    # First run: parse the index (this also writes the SQLite cache),
    # then discard the entry hashes before any fork
    ms_index = MultistreamIndex.new(index_path, cache_dir: opts[:cache_dir], show_progress: !quiet?)
    offsets = ms_index.stream_offsets
    count = ms_index.size
    ms_index = nil
    GC.start
    GC.compact if GC.respond_to?(:compact)
    [offsets, count]
  end
end

#run_build_index(opts) ⇒ Object

Build (or refresh) the local metadata index for a dump



16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
 
# File 'lib/wp2txt/index_commands.rb', line 16

def run_build_index(opts)
  multistream_path, index_path = resolve_dump_paths(opts, download: true)
  return CliUI::EXIT_ERROR unless multistream_path

  db_path = MetadataIndex.path_for(multistream_path, cache_dir: opts[:cache_dir])
  meta = MetadataIndex.new(db_path)

  if meta.valid_for?(multistream_path) && !opts[:update_cache]
    print_success("Metadata index is up to date: #{db_path}")
    print_index_stats(meta.stats)
    meta.close
    return CliUI::EXIT_SUCCESS unless opts[:fulltext]

    stream_offsets, = load_stream_offsets(index_path, opts)
    num_processes = opts[:num_procs] || MemoryMonitor.optimal_processes
    return build_fulltext_index(opts, multistream_path, stream_offsets, db_path, num_processes)
  end
  meta.close

  print_mode_banner("Build Metadata Index", {
    "Dump" => File.basename(multistream_path),
    "Output" => db_path
  })

  time_start = Time.now
  puts pastel.cyan("Loading multistream index...") unless quiet?
  stream_offsets, entry_count = load_stream_offsets(index_path, opts)
  if stream_offsets.empty?
    print_error("Multistream index is empty or unreadable: #{index_path}")
    return CliUI::EXIT_ERROR
  end

  num_processes = opts[:num_procs] || MemoryMonitor.optimal_processes
  puts pastel.cyan("Scanning #{entry_count} pages in #{stream_offsets.size} streams (#{num_processes} processes)...") unless quiet?

  ok = run_phase_isolated do
    builder = MetadataIndexBuilder.new(
      multistream_path, stream_offsets,
      db_path: db_path, num_processes: num_processes
    )

    last_report = Time.now
    built = builder.build do |done, total|
      now = Time.now
      if !quiet? && (now - last_report >= DEFAULT_PROGRESS_INTERVAL || done == total)
        last_report = now
        percent = (done.to_f / total * 100).round(1)
        elapsed = now - time_start
        eta = done.positive? ? (total - done) * (elapsed / done) : 0
        puts pastel.dim(format("  [%d/%d] %.1f%% | ETA: %s", done, total, percent, format_duration(eta)))
      end
    end

    puts unless quiet?
    print_success("Metadata index built in #{format_duration(Time.now - time_start)}")
    print_index_stats(built.stats)
    built.close
    true
  end
  return CliUI::EXIT_ERROR unless ok

  return build_fulltext_index(opts, multistream_path, stream_offsets, db_path, num_processes) if opts[:fulltext]

  CliUI::EXIT_SUCCESS
end

#run_find_articles(opts) ⇒ Object

Query the metadata index and print matching article titles



331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
 
# File 'lib/wp2txt/index_commands.rb', line 331

def run_find_articles(opts)
  multistream_path, = resolve_dump_paths(opts, download: false)
  return CliUI::EXIT_ERROR unless multistream_path

  db_path = MetadataIndex.path_for(multistream_path, cache_dir: opts[:cache_dir])
  meta = MetadataIndex.new(db_path)

  unless meta.built?
    print_error("Metadata index not found for this dump.")
    print_info_message("Build it first with: wp2txt --build-index #{opts[:lang] ? "-L #{opts[:lang]}" : "-i #{opts[:input]}"}")
    return CliUI::EXIT_ERROR
  end

  unless meta.valid_for?(multistream_path)
    print_warning("Metadata index was built from a different version of this dump. Consider re-running --build-index.")
  end

  filters = {
    category: opts[:in_category],
    depth: opts[:in_category] ? opts[:depth] : 0,
    has_section: opts[:has_section],
    use_aliases: !opts[:no_section_aliases],
    alias_file: opts[:alias_file],
    title_match: opts[:title_match]
  }

  total = meta.count_articles(**filters)
  titles = meta.find_articles(**filters, limit: opts[:limit])
  dump_name = meta.stats[:dump_name]
  meta.close

  if opts[:format].to_s.downcase == "json"
    puts JSON.generate({ dump: dump_name, total: total, returned: titles.size, titles: titles })
  else
    titles.each { |t| puts t }
    $stderr.puts pastel.dim("# #{titles.size} of #{total} matching articles (dump: #{dump_name})")
  end
  CliUI::EXIT_SUCCESS
end

#run_fts_optimize(opts) ⇒ Object

Standalone optimize of an existing full-text index (--fts-optimize)



181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
 
# File 'lib/wp2txt/index_commands.rb', line 181

def run_fts_optimize(opts)
  multistream_path, = resolve_dump_paths(opts, download: false)
  return CliUI::EXIT_ERROR unless multistream_path

  fts = FtsIndex.new(
    FtsIndex.path_for(multistream_path, cache_dir: opts[:cache_dir]),
    MetadataIndex.path_for(multistream_path, cache_dir: opts[:cache_dir])
  )
  unless fts.built?
    print_error("Full-text index not found for this dump.")
    print_info_message("Build it first with: wp2txt --build-index --fulltext #{opts[:lang] ? "-L #{opts[:lang]}" : "-i #{opts[:input]}"}")
    return CliUI::EXIT_ERROR
  end

  if fts.optimized?
    print_success("Full-text index is already optimized.")
    fts.close
    return CliUI::EXIT_SUCCESS
  end

  print_info_message("Optimizing full-text index (single-threaded; can take 30-60+ minutes on large indexes)...")
  time_start = Time.now
  fts.optimize!
  print_success("Optimize complete in #{format_duration(Time.now - time_start)}")
  print_info("Size", format_size(File.size(fts.db_path)))
  fts.close
  CliUI::EXIT_SUCCESS
end

#run_import_langlinks(opts) ⇒ Object

Import the official langlinks dump (interlanguage links) into the metadata index. Version pinning: the langlinks file must carry the same dump date as the built index (enforced by LanglinksImporter, no override)



251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
 
# File 'lib/wp2txt/index_commands.rb', line 251

def run_import_langlinks(opts)
  manager = DumpManager.new(
    opts[:lang],
    cache_dir: opts[:cache_dir],
    dump_expiry_days: CLI.config.dump_expiry_days
  )
  multistream = manager.cached_multistream_path
  unless File.exist?(multistream)
    print_error("No cached dump found for '#{opts[:lang]}'.")
    print_info_message("Download and index it with: wp2txt --build-index -L #{opts[:lang]}")
    return CliUI::EXIT_ERROR
  end

  db_path = MetadataIndex.path_for(multistream, cache_dir: opts[:cache_dir])
  meta = MetadataIndex.new(db_path)
  unless meta.built?
    meta.close
    print_error("Metadata index not found for this dump.")
    print_info_message("Build it first with: wp2txt --build-index -L #{opts[:lang]}")
    return CliUI::EXIT_ERROR
  end

  dump_name = meta.stats[:dump_name]
  dump_date = dump_name[/\d{8}\z/]
  meta.close

  source = opts[:langlinks_file] || begin
    print_header("Downloading langlinks for '#{opts[:lang]}' (#{dump_date})")
    manager.download_langlinks(date: dump_date)
  end

  langs = opts[:langlinks_langs]&.split(",")&.map(&:strip)&.reject(&:empty?)
  langs = nil if langs&.empty?

  print_mode_banner("Import Langlinks", {
    "Source" => File.basename(source),
    "Metadata DB" => db_path,
    "Languages" => langs ? langs.join(",") : "all"
  })

  importer = LanglinksImporter.new(db_path, cache_dir: opts[:cache_dir])
  time_start = Time.now
  last_report = Time.now
  result = importer.import!(source, langs: langs, force: opts[:update_cache]) do |rows|
    now = Time.now
    if !quiet? && now - last_report >= DEFAULT_PROGRESS_INTERVAL
      last_report = now
      puts pastel.dim(format("  [%s] %d rows imported", now.strftime("%H:%M:%S"), rows))
    end
  end

  if result[:status] == :already_imported
    print_success("Langlinks already imported (at #{result[:imported_at]}, #{result[:row_count]} rows).")
    print_info_message("Use -U/--update-cache to re-import.")
  else
    print_success("Langlinks imported: #{result[:row_count]} rows in #{format_duration(Time.now - time_start)}")
    prov = result[:provenance]
    print_info("Source", prov[:source].to_s)
    print_info("Languages", prov[:lang_filter].to_s)
    if result[:skipped_invalid].to_i.positive?
      print_warning("Skipped #{result[:skipped_invalid]} rows containing invalid UTF-8 bytes (recorded as langlinks_skipped_invalid)")
    end
    (result[:sanity] || []).each do |check|
      msg = format("join check ll_lang=%s: %d/%d titles found in %s (%.1f%%)",
                   check[:lang], check[:matched], check[:sampled],
                   check[:against], check[:match_rate] * 100)
      if check[:warning]
        print_warning("#{msg} — below 90%; title normalization may mismatch")
      else
        print_info_message(msg)
      end
    end
  end
  CliUI::EXIT_SUCCESS
rescue ArgumentError => e
  print_error(e.message)
  CliUI::EXIT_ERROR
end

#run_phase_isolated(&block) ⇒ Boolean

Run a build phase in a forked child process. Each phase pumps its whole dataset through the parent's heap (Marshal.load of every worker batch), leaving a multi-GB high-water mark; a later phase forking workers from that bloated parent duplicates it per worker via copy-on-write breakage and OOMs the machine (observed on enwiki: 18GB per FTS worker). A child process confines each phase's high-water mark to that phase.

Returns:

  • (Boolean) —

    whether the phase succeeded



89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
 
# File 'lib/wp2txt/index_commands.rb', line 89

def run_phase_isolated(&block)
  return block.call unless Process.respond_to?(:fork)

  pid = Process.fork do
    status = 1
    begin
      status = block.call ? 0 : 1
    rescue StandardError => e
      warn "#{e.class}: #{e.message}"
    end
    $stdout.flush
    $stderr.flush
    exit!(status) # skip at_exit handlers; they belong to the parent
  end
  _, status = Process.waitpid2(pid)
  status.success?
end

#run_search(opts) ⇒ Object

Full-text search from the CLI (--search)



211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
 
# File 'lib/wp2txt/index_commands.rb', line 211

def run_search(opts)
  multistream_path, = resolve_dump_paths(opts, download: false)
  return CliUI::EXIT_ERROR unless multistream_path

  corpus = Corpus.new(
    multistream_path: multistream_path,
    index_path: resolve_dump_paths(opts, download: false)[1],
    cache_dir: opts[:cache_dir]
  )

  unless corpus.fts.built?
    print_error("Full-text index not found for this dump.")
    print_info_message("Build it first with: wp2txt --build-index --fulltext #{opts[:lang] ? "-L #{opts[:lang]}" : "-i #{opts[:input]}"}")
    return CliUI::EXIT_ERROR
  end

  result = corpus.search_text(
    opts[:search],
    sections: opts[:has_section] ? [opts[:has_section]] : nil,
    category: opts[:in_category],
    depth: opts[:in_category] ? opts[:depth] : 0,
    limit: opts[:limit].positive? ? opts[:limit] : 20,
    count: "exact"
  )

  if opts[:format].to_s.downcase == "json"
    puts JSON.generate(result)
  else
    result[:hits].each do |hit|
      puts "#{hit[:section_path]}: #{hit[:snippet]}"
    end
    $stderr.puts pastel.dim("# #{result[:returned]} of #{result[:total]} matches (dump: #{result[:dump]})")
  end
  corpus.close
  CliUI::EXIT_SUCCESS
end