Marc Kupietz | 55fc316 | 2022-12-04 16:25:49 +0100 | [diff] [blame] | 1 | const chai = require('chai'); |
| 2 | const assert = chai.assert; |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 3 | |
| 4 | class KorAPRC { |
| 5 | korap_url = "" |
| 6 | |
| 7 | constructor(korap_url) { |
| 8 | this.korap_url = korap_url |
| 9 | } |
| 10 | |
| 11 | static new(korap_url) { |
| 12 | return new KorAPRC(korap_url) |
| 13 | } |
| 14 | |
| 15 | async login(page, username, password) { |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 16 | page.goto(this.korap_url); |
| 17 | await page.waitForNavigation({ waitUntil: 'networkidle2' }); |
Marc Kupietz | 55fc316 | 2022-12-04 16:25:49 +0100 | [diff] [blame] | 18 | if (this.login == "") return false; |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 19 | if (username == "") return false; |
| 20 | if (password == "") return false; |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 21 | |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 22 | await page.click('.dropdown-btn'); |
| 23 | await page.waitForSelector('input[name=handle_or_email]', { visible: true }); |
| 24 | const username_field = await page.$("input[name=handle_or_email]") |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 25 | if (username_field != null) { |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 26 | await username_field.focus(); |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 27 | await username_field.type(username); |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 28 | const password_field = await page.$("input[name=pwd]") |
| 29 | await password_field.focus() |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 30 | await page.keyboard.type(password) |
| 31 | await page.keyboard.press("Enter") |
Marc Kupietz | e533ca8 | 2022-12-06 07:53:22 +0100 | [diff] [blame] | 32 | } else { |
| 33 | return false |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 34 | } |
| 35 | |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 36 | await page.waitForNavigation({ waitUntil: 'networkidle2' }); |
| 37 | const logout = await page.$(".logout") |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 38 | if (logout == null) { |
| 39 | return false |
| 40 | } |
| 41 | |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 42 | return true |
| 43 | } |
| 44 | |
| 45 | async search(page, query) { |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 46 | const query_field = await page.$("#q-field"); |
| 47 | assert.notEqual(query_field, null, "Query field not found"); |
| 48 | |
| 49 | await page.waitForSelector("#q-field", { visible: true }); |
| 50 | await query_field.click({ clickCount: 3 }); |
| 51 | await page.keyboard.type(query); |
| 52 | await page.keyboard.press("Enter"); |
| 53 | |
| 54 | await page.waitForNavigation({ waitUntil: 'networkidle2' }); |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 55 | |
Marc Kupietz | 964e777 | 2025-06-03 15:02:30 +0200 | [diff] [blame] | 56 | // Wait for search results to be fully loaded |
| 57 | try { |
| 58 | await page.waitForSelector('ol li, #resultinfo, .result-item', { |
| 59 | visible: true, |
| 60 | timeout: 15000 |
| 61 | }); |
| 62 | // Give additional time for the results count to be populated |
| 63 | await new Promise(resolve => setTimeout(resolve, 2000)); |
| 64 | } catch (error) { |
| 65 | // Continue if timeout, fallback methods will handle it |
| 66 | } |
| 67 | |
| 68 | const resultsInfo = await page.evaluate(() => { |
| 69 | // Check common selectors for result counts |
| 70 | const selectors = [ |
| 71 | '#total-results', |
| 72 | '#resultinfo', |
| 73 | '.result-count', |
| 74 | '.total-results', |
| 75 | '[data-results]', |
| 76 | '.found' |
| 77 | ]; |
| 78 | |
| 79 | for (const selector of selectors) { |
| 80 | const element = document.querySelector(selector); |
| 81 | if (element) { |
| 82 | const text = element.textContent || element.innerText || ''; |
| 83 | const numbers = text.match(/\d+/g); |
| 84 | if (numbers && numbers.length > 0) { |
| 85 | return { |
| 86 | selector: selector, |
| 87 | numbers: numbers |
| 88 | }; |
| 89 | } |
| 90 | } |
| 91 | } |
| 92 | |
| 93 | // Look in the page title for results count |
| 94 | const title = document.title; |
| 95 | if (title) { |
| 96 | const numbers = title.match(/\d+/g); |
| 97 | if (numbers && numbers.length > 0) { |
| 98 | return { |
| 99 | selector: 'title', |
| 100 | numbers: numbers |
| 101 | }; |
| 102 | } |
| 103 | } |
| 104 | |
| 105 | // Count the actual result items as fallback |
| 106 | const resultItems = document.querySelectorAll('ol li'); |
| 107 | if (resultItems.length > 0) { |
| 108 | return { |
| 109 | selector: 'counted-items', |
| 110 | numbers: [resultItems.length.toString()] |
| 111 | }; |
| 112 | } |
| 113 | |
| 114 | return null; |
| 115 | }); |
| 116 | |
| 117 | if (!resultsInfo || !resultsInfo.numbers || resultsInfo.numbers.length === 0) { |
| 118 | // Final fallback: just count visible list items |
| 119 | const itemCount = await page.evaluate(() => { |
| 120 | return document.querySelectorAll('ol li').length; |
| 121 | }); |
| 122 | |
| 123 | if (itemCount > 0) { |
| 124 | return itemCount; |
| 125 | } |
| 126 | |
| 127 | throw new Error("Cannot find any results count on the page"); |
| 128 | } |
| 129 | |
| 130 | // Extract the largest number found (likely the total results) |
| 131 | const hits = Math.max(...resultsInfo.numbers.map(n => parseInt(n, 10))); |
Marc Kupietz | 9e0f519 | 2025-03-09 12:12:16 +0100 | [diff] [blame] | 132 | return hits; |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 133 | } |
| 134 | |
| 135 | async logout(page) { |
Marc Kupietz | 964e777 | 2025-06-03 15:02:30 +0200 | [diff] [blame] | 136 | try { |
| 137 | // Direct navigation to logout URL - most reliable method |
| 138 | const currentUrl = await page.url(); |
| 139 | const logoutUrl = currentUrl.replace(/\/$/, '') + '/logout'; |
| 140 | |
| 141 | await page.goto(logoutUrl, { waitUntil: 'domcontentloaded', timeout: 10000 }); |
| 142 | |
| 143 | // Navigate back to main page to ensure clean state for subsequent tests |
| 144 | await page.goto(this.korap_url, { waitUntil: 'domcontentloaded', timeout: 10000 }); |
| 145 | |
| 146 | return true; |
| 147 | } catch (error) { |
| 148 | return false; |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 149 | } |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 150 | } |
| 151 | |
| 152 | async assure_glimpse_off(page) { |
| 153 | const glimpse = await page.$("input[name=cutoff]") |
| 154 | const glimpse_value = await (await glimpse.getProperty('checked')).jsonValue() |
| 155 | if (glimpse_value) { |
| 156 | await page.click("#glimpse") |
| 157 | } |
Marc Kupietz | c8ffb2b | 2025-06-12 16:44:23 +0200 | [diff] [blame] | 158 | } |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 159 | |
Marc Kupietz | c8ffb2b | 2025-06-12 16:44:23 +0200 | [diff] [blame] | 160 | async check_corpus_statistics(page, minTokenThreshold = 1000) { |
| 161 | try { |
| 162 | // Navigate to the corpus view if not already there |
| 163 | await page.goto(this.korap_url, { waitUntil: 'networkidle2' }); |
| 164 | |
| 165 | // Click the vc-choose element to open corpus selection |
| 166 | await page.waitForSelector('#vc-choose', { visible: true, timeout: 10000 }); |
| 167 | await page.click('#vc-choose'); |
| 168 | |
| 169 | // Wait a moment for the UI to respond |
| 170 | await new Promise(resolve => setTimeout(resolve, 1000)); |
| 171 | |
| 172 | // Click the statistic element |
| 173 | await page.waitForSelector('.statistic', { visible: true, timeout: 10000 }); |
| 174 | await page.click('.statistic'); |
| 175 | |
| 176 | // Wait for statistics to load |
| 177 | await new Promise(resolve => setTimeout(resolve, 3000)); |
| 178 | |
| 179 | // Look for the tokens count in a dd element that follows an element with title "tokens" |
| 180 | const tokenCount = await page.evaluate((minThreshold) => { |
| 181 | // Find the element with title "tokens" |
| 182 | const tokenTitleElements = document.querySelectorAll('[title="tokens"], [title*="token"]'); |
| 183 | |
| 184 | for (const element of tokenTitleElements) { |
| 185 | // Look for the next dd element |
| 186 | let nextElement = element.nextElementSibling; |
| 187 | while (nextElement) { |
| 188 | if (nextElement.tagName.toLowerCase() === 'dd') { |
| 189 | const text = nextElement.textContent || nextElement.innerText || ''; |
| 190 | // Remove number separators (commas and periods) and extract number |
| 191 | const cleanedText = text.replace(/[,\.]/g, ''); |
| 192 | const numbers = cleanedText.match(/\d+/g); |
| 193 | if (numbers && numbers.length > 0) { |
| 194 | return parseInt(numbers[0], 10); |
| 195 | } |
| 196 | } |
| 197 | nextElement = nextElement.nextElementSibling; |
| 198 | } |
| 199 | } |
| 200 | |
| 201 | // Alternative approach: look for dd elements that contain large numbers |
| 202 | const ddElements = document.querySelectorAll('dd'); |
| 203 | for (const dd of ddElements) { |
| 204 | const text = dd.textContent || dd.innerText || ''; |
| 205 | // Remove separators and check if it's a large number (likely token count) |
| 206 | const cleanedText = text.replace(/[,\.]/g, ''); |
| 207 | const numbers = cleanedText.match(/\d+/g); |
| 208 | if (numbers && numbers.length > 0) { |
| 209 | const num = parseInt(numbers[0], 10); |
| 210 | // Use the provided threshold instead of hardcoded value |
| 211 | if (num > minThreshold) { |
| 212 | return num; |
| 213 | } |
| 214 | } |
| 215 | } |
| 216 | |
| 217 | return null; |
| 218 | }, minTokenThreshold); |
| 219 | |
| 220 | if (tokenCount === null) { |
| 221 | throw new Error("Could not find token count in corpus statistics"); |
| 222 | } |
| 223 | |
| 224 | return tokenCount; |
| 225 | |
| 226 | } catch (error) { |
| 227 | throw new Error(`Failed to check corpus statistics: ${error.message}`); |
| 228 | } |
Marc Kupietz | 5e45a2f | 2022-12-03 15:32:40 +0100 | [diff] [blame] | 229 | } |
| 230 | } |
| 231 | |
| 232 | module.exports = KorAPRC |