wikibase-cli 15.16.8 → 16.0.0

package/bin/wb-search

@@ -5,6 +5,7 @@ program
 .option('-p, --properties <properties>', 'request additional properties (separated by a comma) (implies verbose mode)')
 .option('-t, --type <type>', 'customize type: i|item, p|property, l|lexeme, f|form, s|sense (Default: item)')
 .option('-n, --limit <num>', 'set a custom limit (defaults to 10)')
+.option('--cirrus', 'use Cirrus search')
 .process('search')
 
 // Accept several words without requiring those to be joined by quotes
@@ -12,7 +13,7 @@ const search = program.args.join(' ')
 
 if (!(search && search.length > 0)) program.helpAndExit(0)
 
-let { lang: language, json, verbose, properties, limit, type } = program
+let { lang: language, json, verbose, properties, limit, type, cirrus } = program
 limit = limit || 20
 
 if (properties) program.verbose = verbose = true
@@ -25,40 +26,63 @@ if (!valide.lang(language)) {
 }
 
 require('../lib/exit_on_missing').instance(program.instance)
-const { searchEntities } = require('../lib/wbk')(program)
+const { searchEntities, cirrusSearchPages } = require('../lib/wbk')(program)
 const { get } = require('../lib/request')
 const parseEntityType = require('../lib/parse_entity_type')
 const { padEnd } = require('lodash')
 
 type = parseEntityType(type)
 
-const url = searchEntities({
-  search,
-  language,
-  // Fetch more in case we need to filter-out some results
-  limit: limit + 10,
-  type
-})
+let url
+if (cirrus) {
+  url = cirrusSearchPages({
+    search,
+    srprop: 'snippet|titlesnippet|sectiontitle',
+    limit,
+  })
+} else {
+  url = searchEntities({
+    search,
+    language,
+    // Fetch more in case we need to filter-out some results
+    limit: limit + 10,
+    type
+  })
+}
+
+const dropMarkup = str => {
+  return str
+  .replace(/<span class="searchmatch">/g, '')
+  .replace(/<\/span>/g, '')
+}
 
 get(url)
 .then(body => {
-  const { error, search } = body
+  const { error, search, query } = body
 
   if (error) {
     console.error('API error response: ' + error.info)
     process.exit(1)
   }
 
-  let results = search
+  let results = search || query.search
 
   results = results.slice(0, limit)
 
+  if (cirrus) {
+    results.forEach(result => {
+      result.id = result.title
+      result.label = dropMarkup(result.titlesnippet)
+      result.description = dropMarkup(result.snippet.split('\n')[0])
+    })
+  }
+
   if (json) {
     console.log(JSON.stringify(results, null, 2))
     return
   }
 
-  const ids = results.map(result => result.id)
+  const ids = results.map(result => result.id || result.title)
 
   if (ids.length === 0) {
     console.error('no result')
@@ -74,9 +98,22 @@ get(url)
     // in the summary verbose format, only in the normal, short, summary format
     program.verbose = false
     entityDataParser({ ids, parser: summaryParser })
+  } else if (cirrus) {
+    results
+    .forEach(result => {
+      let { id, label, description } = result
+      // Known case: when search with haswbstatement
+      if (!label) {
+        label = description
+        description = ''
+      }
+      let text = padEnd(id, 11) + label
+      if (description && description.length > 0) text += ' ' + grey(description)
+      console.log(text)
+    })
   } else {
     results
-    .forEach((result, index) => {
+    .forEach(result => {
       const { id, label, description } = result
       let text = padEnd(id, 11) + label
       if (description && description.length > 0) text += ' ' + grey(description)

package/lib/get_sitelink_url.js

@@ -5,7 +5,7 @@ const errors_ = require('../lib/errors')
 module.exports = program => {
   exitOnMissing.instance(program.instance)
   const { getEntities } = require('../lib/wbk')(program)
-  return (id, lang, sitelinkSuffix) => {
+  return async (id, lang, sitelinkSuffix) => {
     if (id[0] === 'P') {
       console.error("Wikibase properties don't have sitelinks")
       process.exit(1)
@@ -13,18 +13,30 @@ module.exports = program => {
 
     const url = getEntities({ ids: id, props: [ 'sitelinks' ] })
 
-    return get(url)
-    .then(res => {
-      const entity = res.entities[id]
-      if (!entity || entity.missing === '') throw errors_.new('entity not found', { id, lang, sitelinkSuffix })
-      const sitelink = entity.sitelinks[`${lang}${sitelinkSuffix}`]
-      if (!sitelink) throw errors_.new('sitelink not found', { id, lang, sitelinkSuffix })
-      const domain = sitelinkDomain[sitelinkSuffix]
-      return `https://${lang}.${domain}/wiki/${sitelink.title}`
-    })
+    const { entities } = await get(url)
+    const entity = entities[id]
+    if (!entity || entity.missing === '') throw errors_.new('entity not found', { id, lang, sitelinkSuffix })
+    const sitelink = entity.sitelinks[`${lang}${sitelinkSuffix}`]
+    if (!sitelink) throw errors_.new('sitelink not found', { id, lang, sitelinkSuffix })
+    const domain = sitelinkDomain[sitelinkSuffix]
+    const formattedTitle = formatTitle(sitelink.title)
+    return `https://${lang}.${domain}/wiki/${formattedTitle}`
   }
 }
 
 const sitelinkDomain = {
   wiki: 'wikipedia.org'
 }
+
+const formatTitle = title => {
+  title = title.replace(/\s/g, '_')
+  return fixedEncodeURIComponent(title)
+}
+
+// encodeURIComponent ignores !, ', (, ), and *
+// cf https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/encodeURIComponent#Description
+const fixedEncodeURIComponent = str => {
+  return encodeURIComponent(str).replace(/[!'()*]/g, encodeCharacter)
+}
+
+const encodeCharacter = c => '%' + c.charCodeAt(0).toString(16)

package/metadata/search.js

@@ -15,9 +15,10 @@ module.exports = {
     { args: 'Harry Potter --limit 25', comment: 'displays up to 25 results matching "Harry Potter"' },
     { args: 'Harry Potter --verbose', comment: 'display rich results (aka summaries)' },
     { args: 'Harry Potter --properties P577,P935', comment: 'request additional properties (separated by a comma) to be added to the results summaries' },
-    { args: 'date --type property', comment: "Search properties (but `wb props` might be doing a better job)" },
-    { args: 'date --type lexeme', comment: "Search lexemes" },
-    { args: 'code --type form', comment: "Search forms" },
-    { args: 'test --type sense', comment: "Searching senses doesn't seem to work currently (2020-04-17)" }
+    { args: 'date --type property', comment: 'Search properties (but `wb props` might be doing a better job)' },
+    { args: 'date --type lexeme', comment: 'Search lexemes' },
+    { args: 'code --type form', comment: 'Search forms' },
+    { args: 'test --type sense', comment: "Searching senses doesn't seem to work currently (2020-04-17)" },
+    { args: '--cirrus "porte haswbstatement:P31=Q5"', comment: 'Use Cirrus search to find humans (Q5) matching "porte"' },
   ]
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "wikibase-cli",
-  "version": "15.16.8",
+  "version": "16.0.0",
   "description": "A command-line interface to Wikibase",
   "main": "index.js",
   "bin": {
@@ -50,8 +50,8 @@
     "shell-quote": "^1.7.3",
     "split": "^1.0.1",
     "through": "^2.3.8",
-    "wikibase-edit": "^4.16.1",
-    "wikibase-sdk": "^7.14.4",
+    "wikibase-edit": "^5.0.0",
+    "wikibase-sdk": "^8.0.0",
     "wikidata-lang": "^2.0.11"
   },
   "devDependencies": {
@@ -67,6 +67,6 @@
     "should": "^13.2.3"
   },
   "engines": {
-    "node": ">= 7.6.0"
+    "node": ">= 10.0.0"
   }
 }