Class: Wp2txt::MetadataIndex
- Inherits:
-
Object
- Object
- Wp2txt::MetadataIndex
- Defined in:
- lib/wp2txt/metadata_index.rb
Overview
Local metadata index (Tier 1) built from a multistream dump. Stores per-page categories, section headings, redirects, and the category hierarchy in SQLite, enabling offline exhaustive queries such as "all articles in category X that have a Plot section" without any API access.
Constant Summary collapse
- SCHEMA_VERSION =
2- CACHE_SUFFIX =
"_meta.sqlite3"- NS_ARTICLE =
0- NS_CATEGORY =
14
Instance Attribute Summary collapse
-
#db_path ⇒ Object
readonly
Returns the value of attribute db_path.
Class Method Summary collapse
-
.clean_heading(text) ⇒ Object
Remove wiki markup from a heading ('''bold''', [[link|label]], HTML tags).
-
.expand_section_names(name, alias_file: nil) ⇒ Object
Expand a section name to its full alias group (bidirectional): "Plot" => ["Plot", "Synopsis", ...]; "Synopsis" => same group.
-
.normalize_category(name) ⇒ Object
Normalize a category name the way MediaWiki treats titles: underscores to spaces, trimmed, first letter capitalized.
-
.path_for(multistream_path, cache_dir: nil) ⇒ Object
Default index location for a given multistream file (mirrors IndexCache naming).
Instance Method Summary collapse
-
#built? ⇒ Boolean
True if the index file exists and has a compatible schema.
-
#categories_of(title) ⇒ Array<String>?
Categories of one article (by exact title).
-
#category_tree(category, depth: 2) ⇒ Object
Subcategory tree starting at category, as [depth:, ...] (BFS order).
- #close ⇒ Object
-
#count_articles(category: nil, depth: 0, categories: nil, category_match: nil, has_section: nil, sections: nil, alias_set: nil, use_aliases: true, alias_file: nil, title_match: nil) ⇒ Object
Count articles matching the same filters as find_articles.
- #delete_alias_set(name) ⇒ Object
- #finalize_build!(source_path) ⇒ Object
-
#find_articles(category: nil, depth: 0, categories: nil, category_match: nil, has_section: nil, sections: nil, alias_set: nil, use_aliases: true, alias_file: nil, title_match: nil, limit: 0, offset: 0) ⇒ Array<String>
Find article titles matching the given filters.
-
#get_alias_set(name) ⇒ Hash?
{ name:, groups:, created_at: } or nil if not found.
-
#initialize(db_path) ⇒ MetadataIndex
constructor
A new instance of MetadataIndex.
-
#insert_batch(rows) ⇒ Object
Insert one scanned batch: categories:, sections:, hierarchy:.
- #list_alias_sets ⇒ Object
-
#prepare_build! ⇒ Object
Build into a sidecar file and atomically rename in finalize_build!, so a failed multi-hour rebuild never destroys a working index.
-
#save_alias_set(name, groups) ⇒ Object
Save a named alias set.
-
#section_cooccurrence(headings, category: nil, depth: 0) ⇒ Hash
Article counts, average positions, and pairwise co-occurrence for a set of headings.
-
#section_stats(category: nil, depth: 0, top_n: 50) ⇒ Object
Section heading frequencies across articles, optionally scoped to a category.
- #stats ⇒ Object
-
#valid_for?(multistream_path) ⇒ Boolean
True if the index was built from the given (unchanged) multistream file.
Constructor Details
#initialize(db_path) ⇒ MetadataIndex
Returns a new instance of MetadataIndex.
27 28 29 30 |
# File 'lib/wp2txt/metadata_index.rb', line 27 def initialize(db_path) @db_path = db_path @db = nil end |
Instance Attribute Details
#db_path ⇒ Object (readonly)
Returns the value of attribute db_path.
25 26 27 |
# File 'lib/wp2txt/metadata_index.rb', line 25 def db_path @db_path end |
Class Method Details
.clean_heading(text) ⇒ Object
Remove wiki markup from a heading ('''bold''', [[link|label]], HTML tags)
50 51 52 53 54 |
# File 'lib/wp2txt/metadata_index.rb', line 50 def self.clean_heading(text) t = text.gsub(/'{2,}/, "") t = t.gsub(/\[\[(?:[^\]|]*\|)?([^\]]*)\]\]/) { ::Regexp.last_match(1) } t.gsub(/<[^>]+>/, "").strip end |
.expand_section_names(name, alias_file: nil) ⇒ Object
Expand a section name to its full alias group (bidirectional): "Plot" => ["Plot", "Synopsis", ...]; "Synopsis" => same group
58 59 60 61 62 63 64 65 66 67 68 69 70 71 |
# File 'lib/wp2txt/metadata_index.rb', line 58 def self.(name, alias_file: nil) aliases = SectionExtractor::DEFAULT_ALIASES if alias_file custom = SectionExtractor.load_aliases_from_file(alias_file) aliases = aliases.merge(custom) unless custom.empty? end down = name.downcase aliases.each do |canonical, list| group = [canonical, *list] return group if group.any? { |g| g.downcase == down } end [name] end |
.normalize_category(name) ⇒ Object
Normalize a category name the way MediaWiki treats titles: underscores to spaces, trimmed, first letter capitalized
42 43 44 45 46 47 |
# File 'lib/wp2txt/metadata_index.rb', line 42 def self.normalize_category(name) n = name.to_s.tr("_", " ").strip.squeeze(" ") return n if n.empty? n[0].upcase + n[1..].to_s end |
.path_for(multistream_path, cache_dir: nil) ⇒ Object
Default index location for a given multistream file (mirrors IndexCache naming)
33 34 35 36 37 38 |
# File 'lib/wp2txt/metadata_index.rb', line 33 def self.path_for(multistream_path, cache_dir: nil) dir = cache_dir || File.("~/.wp2txt/cache") basename = File.basename(multistream_path, ".*").sub(/\.xml\z/, "") path_hash = Digest::MD5.hexdigest(multistream_path)[0, 8] File.join(dir, "#{basename}_#{path_hash}#{CACHE_SUFFIX}") end |
Instance Method Details
#built? ⇒ Boolean
True if the index file exists and has a compatible schema
78 79 80 81 82 83 84 85 |
# File 'lib/wp2txt/metadata_index.rb', line 78 def built? return false unless File.exist?(@db_path) = !.nil? && [:schema_version].to_i == SCHEMA_VERSION && ![:built_at].nil? rescue SQLite3::Exception false end |
#categories_of(title) ⇒ Array<String>?
Categories of one article (by exact title)
263 264 265 266 267 268 269 270 |
# File 'lib/wp2txt/metadata_index.rb', line 263 def categories_of(title) row = open_db.execute("SELECT page_id FROM pages WHERE title = ?", [title]).first return nil unless row open_db.execute( "SELECT category FROM page_categories WHERE page_id = ? ORDER BY category", [row[0]] ).map(&:first) end |
#category_tree(category, depth: 2) ⇒ Object
Subcategory tree starting at category, as [depth:, ...] (BFS order)
273 274 275 276 277 278 279 280 281 282 283 284 285 286 |
# File 'lib/wp2txt/metadata_index.rb', line 273 def category_tree(category, depth: 2) cat = self.class.normalize_category(category) sql = <<~SQL WITH RECURSIVE cat_tree(name, d) AS ( SELECT ?, 0 UNION SELECT ch.child, ct.d + 1 FROM category_hierarchy ch JOIN cat_tree ct ON ch.parent = ct.name WHERE ct.d < #{depth.to_i} ) SELECT name, MIN(d) FROM cat_tree GROUP BY name ORDER BY MIN(d), name SQL open_db.execute(sql, [cat]).map { |name, d| { name: name, depth: d } } end |
#close ⇒ Object
114 115 116 117 |
# File 'lib/wp2txt/metadata_index.rb', line 114 def close @db&.close @db = nil end |
#count_articles(category: nil, depth: 0, categories: nil, category_match: nil, has_section: nil, sections: nil, alias_set: nil, use_aliases: true, alias_file: nil, title_match: nil) ⇒ Object
Count articles matching the same filters as find_articles
246 247 248 249 250 251 252 253 254 255 256 257 258 259 |
# File 'lib/wp2txt/metadata_index.rb', line 246 def count_articles(category: nil, depth: 0, categories: nil, category_match: nil, has_section: nil, sections: nil, alias_set: nil, use_aliases: true, alias_file: nil, title_match: nil) cte, where, params = build_article_query( category: category, depth: depth, categories: categories, category_match: category_match, has_section: has_section, sections: sections, alias_set: alias_set, use_aliases: use_aliases, alias_file: alias_file, title_match: title_match ) sql = +"" sql << "WITH RECURSIVE #{cte} " if cte sql << "SELECT COUNT(*) FROM pages p WHERE #{where}" open_db.get_first_value(sql, params).to_i end |
#delete_alias_set(name) ⇒ Object
395 396 397 398 399 |
# File 'lib/wp2txt/metadata_index.rb', line 395 def delete_alias_set(name) ensure_alias_table open_db.execute("DELETE FROM alias_sets WHERE name = ?", [name]) nil end |
#finalize_build!(source_path) ⇒ Object
183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 |
# File 'lib/wp2txt/metadata_index.rb', line 183 def finalize_build!(source_path) db = open_db db.execute("CREATE INDEX IF NOT EXISTS idx_pages_title ON pages(title)") db.execute("CREATE INDEX IF NOT EXISTS idx_pc_category ON page_categories(category)") db.execute("CREATE INDEX IF NOT EXISTS idx_pc_page ON page_categories(page_id)") db.execute("CREATE INDEX IF NOT EXISTS idx_ps_heading ON page_sections(heading COLLATE NOCASE)") db.execute("CREATE INDEX IF NOT EXISTS idx_ps_page ON page_sections(page_id)") db.execute("CREATE INDEX IF NOT EXISTS idx_ch_parent ON category_hierarchy(parent)") stat = File.stat(source_path) dump_name = File.basename(source_path)[/\A[a-z0-9_\-]+?-\d{8}/] || File.basename(source_path) ( schema_version: SCHEMA_VERSION, wp2txt_version: Wp2txt::VERSION, source_path: source_path, source_size: stat.size, source_mtime: stat.mtime.to_i, dump_name: dump_name, built_at: Time.now.utc.iso8601 ) db.execute("ANALYZE") close if @build_path File.rename(@build_path, @db_path) FileUtils.rm_f(["#{@db_path}-wal", "#{@db_path}-shm"]) @build_path = nil end end |
#find_articles(category: nil, depth: 0, categories: nil, category_match: nil, has_section: nil, sections: nil, alias_set: nil, use_aliases: true, alias_file: nil, title_match: nil, limit: 0, offset: 0) ⇒ Array<String>
Find article titles matching the given filters.
228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 |
# File 'lib/wp2txt/metadata_index.rb', line 228 def find_articles(category: nil, depth: 0, categories: nil, category_match: nil, has_section: nil, sections: nil, alias_set: nil, use_aliases: true, alias_file: nil, title_match: nil, limit: 0, offset: 0) cte, where, params = build_article_query( category: category, depth: depth, categories: categories, category_match: category_match, has_section: has_section, sections: sections, alias_set: alias_set, use_aliases: use_aliases, alias_file: alias_file, title_match: title_match ) sql = +"" sql << "WITH RECURSIVE #{cte} " if cte sql << "SELECT p.title FROM pages p WHERE #{where} ORDER BY p.page_id" sql << " LIMIT #{limit.to_i}" if limit.to_i.positive? sql << " OFFSET #{offset.to_i}" if offset.to_i.positive? open_db.execute(sql, params).map { |row| row[0] } end |
#get_alias_set(name) ⇒ Hash?
Returns { name:, groups:, created_at: } or nil if not found.
380 381 382 383 384 385 386 |
# File 'lib/wp2txt/metadata_index.rb', line 380 def get_alias_set(name) ensure_alias_table row = open_db.execute("SELECT name, groups, created_at FROM alias_sets WHERE name = ?", [name]).first return nil unless row { name: row[0], groups: JSON.parse(row[1]), created_at: row[2] } end |
#insert_batch(rows) ⇒ Object
Insert one scanned batch: categories:, sections:, hierarchy:
162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 |
# File 'lib/wp2txt/metadata_index.rb', line 162 def insert_batch(rows) db = open_db db.transaction do stmt = db.prepare("INSERT OR IGNORE INTO pages (page_id, title, namespace, redirect_to, text_length) VALUES (?, ?, ?, ?, ?)") rows[:pages].each { |r| stmt.execute(r) } stmt.close stmt = db.prepare("INSERT INTO page_categories (page_id, category) VALUES (?, ?)") rows[:categories].each { |r| stmt.execute(r) } stmt.close stmt = db.prepare("INSERT INTO page_sections (page_id, heading, level, ord) VALUES (?, ?, ?, ?)") rows[:sections].each { |r| stmt.execute(r) } stmt.close stmt = db.prepare("INSERT INTO category_hierarchy (child, parent) VALUES (?, ?)") rows[:hierarchy].each { |r| stmt.execute(r) } stmt.close end end |
#list_alias_sets ⇒ Object
388 389 390 391 392 393 |
# File 'lib/wp2txt/metadata_index.rb', line 388 def list_alias_sets ensure_alias_table open_db.execute("SELECT name, groups, created_at FROM alias_sets ORDER BY name").map do |name, groups, created_at| { name: name, group_count: JSON.parse(groups).size, created_at: created_at } end end |
#prepare_build! ⇒ Object
Build into a sidecar file and atomically rename in finalize_build!, so a failed multi-hour rebuild never destroys a working index
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 |
# File 'lib/wp2txt/metadata_index.rb', line 125 def prepare_build! FileUtils.mkdir_p(File.dirname(@db_path)) close @build_path = "#{@db_path}.building" FileUtils.rm_f([@build_path, "#{@build_path}-wal", "#{@build_path}-shm"]) db = open_db db.execute(<<~SQL) CREATE TABLE metadata (key TEXT PRIMARY KEY, value TEXT) SQL db.execute(<<~SQL) CREATE TABLE pages ( page_id INTEGER PRIMARY KEY, title TEXT, namespace INTEGER, redirect_to TEXT, text_length INTEGER ) SQL db.execute("CREATE TABLE page_categories (page_id INTEGER, category TEXT)") # ord: position of the section within the article, where 0 is the lead # text before the first heading. Lead rows are NOT stored here (they have # no heading), so ord starts at 1 — consistent with fts_map.ord in the # full-text DB, where the lead IS stored as ord 0. db.execute(<<~SQL) CREATE TABLE page_sections ( page_id INTEGER, heading TEXT, level INTEGER, -- ord: section position in the article; 0 = lead text (not stored in -- this table), so headings start at 1. Same semantics as fts_map.ord. ord INTEGER ) SQL db.execute("CREATE TABLE category_hierarchy (child TEXT, parent TEXT)") end |
#save_alias_set(name, groups) ⇒ Object
Save a named alias set. groups is an array of heading groups, e.g. [["あらすじ", "ストーリー", "物語"], ["脚注", "出典"]]
368 369 370 371 372 373 374 375 376 377 |
# File 'lib/wp2txt/metadata_index.rb', line 368 def save_alias_set(name, groups) raise ArgumentError, "groups must be a non-empty array of arrays" unless groups.is_a?(Array) && !groups.empty? && groups.all? { |g| g.is_a?(Array) && !g.empty? } ensure_alias_table open_db.execute( "INSERT OR REPLACE INTO alias_sets (name, groups, created_at) VALUES (?, ?, ?)", [name, JSON.generate(groups), Time.now.utc.iso8601] ) { name: name, groups: groups } end |
#section_cooccurrence(headings, category: nil, depth: 0) ⇒ Hash
Article counts, average positions, and pairwise co-occurrence for a set of headings. Synonymous headings almost never co-occur in the same article, so a high co-occurrence ratio is evidence AGAINST treating them as aliases.
317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 |
# File 'lib/wp2txt/metadata_index.rb', line 317 def section_cooccurrence(headings, category: nil, depth: 0) scope_cte = nil scope_cond = "p.namespace = #{NS_ARTICLE} AND p.redirect_to IS NULL" scope_params = [] if category scope_cte, cond, scope_params = category_condition(category, depth) scope_cond += " AND #{cond}" end db = open_db with = scope_cte ? "WITH RECURSIVE #{scope_cte} " : "" # Placeholders bind positionally: with a recursive CTE the category `?` # sits inside the CTE (before any heading `?`); without one it sits # inside scope_cond (after the heading `?`) cte_params = scope_cte ? scope_params : [] cond_params = scope_cte ? [] : scope_params heading_stats = headings.map do |h| row = db.execute( "#{with}SELECT COUNT(DISTINCT ps.page_id), AVG(ps.ord) FROM page_sections ps " \ "JOIN pages p ON p.page_id = ps.page_id " \ "WHERE ps.heading COLLATE NOCASE = ? AND #{scope_cond}", cte_params + [h] + cond_params ).first { heading: h, articles: row[0].to_i, avg_position: row[1]&.round(2) } end counts = heading_stats.to_h { |s| [s[:heading], s[:articles]] } pairs = headings.combination(2).map do |a, b| both = db.get_first_value( "#{with}SELECT COUNT(*) FROM (" \ "SELECT ps.page_id FROM page_sections ps JOIN pages p ON p.page_id = ps.page_id " \ "WHERE ps.heading COLLATE NOCASE = ? AND #{scope_cond} " \ "INTERSECT " \ "SELECT ps.page_id FROM page_sections ps JOIN pages p ON p.page_id = ps.page_id " \ "WHERE ps.heading COLLATE NOCASE = ? AND #{scope_cond})", cte_params + [a] + cond_params + [b] + cond_params ).to_i min = [counts[a], counts[b]].min { a: a, b: b, both: both, cooccurrence_ratio: min.positive? ? (both.to_f / min).round(3) : 0.0 } end { headings: heading_stats, pairs: pairs } end |
#section_stats(category: nil, depth: 0, top_n: 50) ⇒ Object
Section heading frequencies across articles, optionally scoped to a category
289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 |
# File 'lib/wp2txt/metadata_index.rb', line 289 def section_stats(category: nil, depth: 0, top_n: 50) conds = ["p.namespace = #{NS_ARTICLE}", "p.redirect_to IS NULL"] params = [] cte = nil if category cte, cond, cat_params = category_condition(category, depth) conds << cond params.concat(cat_params) end sql = +"" sql << "WITH RECURSIVE #{cte} " if cte sql << <<~SQL SELECT ps.heading, COUNT(*) AS cnt FROM page_sections ps JOIN pages p ON p.page_id = ps.page_id WHERE #{conds.join(' AND ')} GROUP BY ps.heading ORDER BY cnt DESC, ps.heading LIMIT #{top_n.to_i} SQL open_db.execute(sql, params) end |
#stats ⇒ Object
97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 |
# File 'lib/wp2txt/metadata_index.rb', line 97 def stats return nil unless File.exist?(@db_path) = || {} { db_path: @db_path, db_size: File.size(@db_path), dump_name: [:dump_name], built_at: [:built_at], built_with: [:wp2txt_version], page_count: count_scalar("SELECT COUNT(*) FROM pages"), article_count: count_scalar("SELECT COUNT(*) FROM pages WHERE namespace = #{NS_ARTICLE} AND redirect_to IS NULL"), category_count: count_scalar("SELECT COUNT(DISTINCT category) FROM page_categories"), section_count: count_scalar("SELECT COUNT(*) FROM page_sections") } end |
#valid_for?(multistream_path) ⇒ Boolean
True if the index was built from the given (unchanged) multistream file
88 89 90 91 92 93 94 95 |
# File 'lib/wp2txt/metadata_index.rb', line 88 def valid_for?(multistream_path) return false unless built? return false unless File.exist?(multistream_path) = stat = File.stat(multistream_path) [:source_size].to_i == stat.size && [:source_mtime].to_i == stat.mtime.to_i end |