@olib-ai/owl-browser-sdk 2.0.5 → 2.0.7

package/dist/extraction/pagination.d.ts

@@ -0,0 +1,29 @@
+/**
+ * Pagination detection and navigation.
+ *
+ * Detects pagination type from DOM heuristics: next links, page numbers,
+ * scroll markers, load-more buttons, rel="next".
+ */
+import type { PaginationConfig } from './types.js';
+import type { HTMLProcessor } from './html-processor.js';
+/**
+ * Auto-detect pagination type from the current page DOM.
+ *
+ * Uses a single browser evaluate() call to check all DOM-based patterns
+ * (rel="next", known selectors, text-based detection, load-more) at once,
+ * reducing round-trips from ~20 to 1.
+ */
+export declare function detectPagination(proc: HTMLProcessor): Promise<PaginationConfig | null>;
+/**
+ * Check if there's a next page available.
+ */
+export declare function hasNextPage(proc: HTMLProcessor, config: PaginationConfig, currentPage: number): Promise<boolean>;
+/**
+ * Navigate to the next page.
+ */
+export declare function goToNextPage(proc: HTMLProcessor, config: PaginationConfig, currentPage: number): Promise<boolean>;
+/**
+ * Resolve a URL pattern with a page number.
+ */
+export declare function resolvePageUrl(pattern: string, page: number): string;
+//# sourceMappingURL=pagination.d.ts.map

package/dist/extraction/pagination.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pagination.d.ts","sourceRoot":"","sources":["../../src/extraction/pagination.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAEH,OAAO,KAAK,EAAE,gBAAgB,EAAE,MAAM,YAAY,CAAC;AACnD,OAAO,KAAK,EAAE,aAAa,EAAE,MAAM,qBAAqB,CAAC;AAiCzD;;;;;;GAMG;AACH,wBAAsB,gBAAgB,CAAC,IAAI,EAAE,aAAa,GAAG,OAAO,CAAC,gBAAgB,GAAG,IAAI,CAAC,CA0L5F;AAED;;GAEG;AACH,wBAAsB,WAAW,CAC/B,IAAI,EAAE,aAAa,EACnB,MAAM,EAAE,gBAAgB,EACxB,WAAW,EAAE,MAAM,GAClB,OAAO,CAAC,OAAO,CAAC,CA0BlB;AAED;;GAEG;AACH,wBAAsB,YAAY,CAChC,IAAI,EAAE,aAAa,EACnB,MAAM,EAAE,gBAAgB,EACxB,WAAW,EAAE,MAAM,GAClB,OAAO,CAAC,OAAO,CAAC,CAkClB;AAED;;GAEG;AACH,wBAAgB,cAAc,CAAC,OAAO,EAAE,MAAM,EAAE,IAAI,EAAE,MAAM,GAAG,MAAM,CAEpE"}

package/dist/extraction/pagination.js

@@ -0,0 +1,323 @@
+/**
+ * Pagination detection and navigation.
+ *
+ * Detects pagination type from DOM heuristics: next links, page numbers,
+ * scroll markers, load-more buttons, rel="next".
+ */
+// Patterns for detecting "Next" buttons/links
+const NEXT_LINK_PATTERNS = [
+    'a[rel="next"]',
+    'link[rel="next"]',
+    '[class*="next"] a',
+    '[class*="pagination"] a[class*="next"]',
+    'a[aria-label*="next" i]',
+    'a[aria-label*="Next"]',
+    'button[aria-label*="next" i]',
+    'button[aria-label*="Next"]',
+    'nav[class*="pagination"] a:last-child',
+    '.pagination .next a',
+    '.pagination a.next',
+    '.pager .next a',
+    'a.next-page',
+    'a[class*="next-page"]',
+    'button.next-page',
+    'button[class*="next-page"]',
+];
+// Patterns for "Load more" buttons
+const LOAD_MORE_PATTERNS = [
+    'button[class*="load-more"]',
+    'button[class*="loadmore"]',
+    'a[class*="load-more"]',
+    '[class*="load-more"] button',
+    '[data-action="load-more"]',
+    'button[class*="show-more"]',
+    'a[class*="show-more"]',
+];
+/**
+ * Auto-detect pagination type from the current page DOM.
+ *
+ * Uses a single browser evaluate() call to check all DOM-based patterns
+ * (rel="next", known selectors, text-based detection, load-more) at once,
+ * reducing round-trips from ~20 to 1.
+ */
+export async function detectPagination(proc) {
+    const url = await proc.getUrl();
+    // Single JS call that checks all patterns at once
+    const result = await proc.evaluate(`
+    (function() {
+      var result = { type: null, selector: null, href: null };
+
+      // 1. rel="next" link
+      var relNext = document.querySelector('a[rel="next"], link[rel="next"]');
+      if (relNext) {
+        result.type = 'rel-next';
+        result.href = relNext.href || null;
+        result.selector = 'a[rel="next"]';
+        return result;
+      }
+
+      // 2. Check known next-button selectors
+      var nextSelectors = ${JSON.stringify(NEXT_LINK_PATTERNS)};
+      for (var i = 0; i < nextSelectors.length; i++) {
+        try {
+          var el = document.querySelector(nextSelectors[i]);
+          if (el && el.offsetParent !== null) {
+            result.type = 'click-next';
+            result.selector = nextSelectors[i];
+            return result;
+          }
+        } catch(e) {}
+      }
+
+      // 3. Text-based "next" detection in pagination containers (with i18n)
+      var nextTexts = ['next', 'siguiente', 'suivant', 'weiter', 'volgende', 'avanti', '\\u0434\\u0430\\u043b\\u0435\\u0435', '\\u6b21\\u3078', '\\u4e0b\\u4e00\\u9875', '\\u45e4\\u44eb'];
+      var arrowTexts = ['>', '>>', '\\u203a', '\\u2192', '\\u276f'];
+      var containers = document.querySelectorAll(
+        'nav, [class*="pagination"], [class*="pager"], [role="navigation"]'
+      );
+      for (var i = 0; i < containers.length; i++) {
+        var clickables = containers[i].querySelectorAll('a, button');
+        for (var j = 0; j < clickables.length; j++) {
+          var el = clickables[j];
+          if (el.disabled) continue;
+          var text = (el.textContent || '').trim().toLowerCase();
+          var ariaLabel = (el.getAttribute('aria-label') || '').toLowerCase();
+
+          var isNext = false;
+          for (var k = 0; k < nextTexts.length; k++) {
+            if (text.indexOf(nextTexts[k]) === 0 || ariaLabel.indexOf(nextTexts[k]) === 0) {
+              isNext = true;
+              break;
+            }
+          }
+          if (!isNext) {
+            for (var k = 0; k < arrowTexts.length; k++) {
+              if (text === arrowTexts[k]) { isNext = true; break; }
+            }
+          }
+
+          if (isNext) {
+            // Build a precise selector
+            var tag = el.tagName.toLowerCase();
+            var sel = tag;
+            if (el.id) {
+              sel = tag + '#' + el.id;
+            } else if (el.getAttribute('aria-label')) {
+              sel = tag + '[aria-label="' + el.getAttribute('aria-label').replace(/"/g, '\\\\"') + '"]';
+            } else {
+              // Use classes + nth-child for precision
+              var cls = Array.from(el.classList).slice(0, 3).join('.');
+              if (cls) {
+                sel = tag + '.' + cls;
+                // Check uniqueness
+                var matches = document.querySelectorAll(sel);
+                if (matches.length > 1) {
+                  var parent = el.parentElement;
+                  if (parent) {
+                    var siblings = parent.querySelectorAll(':scope > ' + sel);
+                    for (var s = 0; s < siblings.length; s++) {
+                      if (siblings[s] === el) {
+                        sel = tag + '.' + cls + ':nth-child(' + (s + 1) + ')';
+                        break;
+                      }
+                    }
+                  }
+                }
+              } else {
+                // No classes — use parent context + nth-child
+                var parent = el.parentElement;
+                if (parent) {
+                  var siblings = parent.querySelectorAll(':scope > ' + tag);
+                  for (var s = 0; s < siblings.length; s++) {
+                    if (siblings[s] === el) {
+                      var parentSel = '';
+                      if (parent.id) parentSel = '#' + parent.id;
+                      else if (parent.classList.length > 0) parentSel = parent.tagName.toLowerCase() + '.' + Array.from(parent.classList).slice(0, 2).join('.');
+                      else parentSel = parent.tagName.toLowerCase();
+                      sel = parentSel + ' > ' + tag + ':nth-child(' + (s + 1) + ')';
+                      break;
+                    }
+                  }
+                }
+              }
+            }
+            result.type = 'click-next';
+            result.selector = sel;
+            return result;
+          }
+        }
+      }
+
+      // 4. Load-more buttons
+      var loadMoreSelectors = ${JSON.stringify(LOAD_MORE_PATTERNS)};
+      for (var i = 0; i < loadMoreSelectors.length; i++) {
+        try {
+          var el = document.querySelector(loadMoreSelectors[i]);
+          if (el && el.offsetParent !== null) {
+            result.type = 'load-more';
+            result.selector = loadMoreSelectors[i];
+            return result;
+          }
+        } catch(e) {}
+      }
+
+      return null;
+    })()
+  `);
+    if (result) {
+        if (result.type === 'rel-next' && result.href) {
+            const pattern = detectUrlPattern(url, result.href);
+            if (pattern) {
+                return {
+                    type: 'url-pattern',
+                    urlPattern: pattern.pattern,
+                    startPage: pattern.currentPage,
+                };
+            }
+            return { type: 'click-next', nextSelector: result.selector };
+        }
+        if (result.type === 'click-next') {
+            return { type: 'click-next', nextSelector: result.selector };
+        }
+        if (result.type === 'load-more') {
+            return { type: 'load-more', nextSelector: result.selector };
+        }
+    }
+    // 5. Check for URL-based pagination (page= in current URL)
+    const urlMatch = /[?&](page|p)=(\d+)/.exec(url);
+    if (urlMatch) {
+        const param = urlMatch[1];
+        const currentPage = parseInt(urlMatch[2], 10);
+        return {
+            type: 'url-pattern',
+            urlPattern: url.replace(new RegExp(`([?&])${param}=\\d+`), `$1${param}={page}`),
+            startPage: currentPage,
+        };
+    }
+    // 6. Check for /page/N pattern in URL
+    const pathMatch = /\/page\/(\d+)/.exec(url);
+    if (pathMatch) {
+        return {
+            type: 'url-pattern',
+            urlPattern: url.replace(/\/page\/\d+/, '/page/{page}'),
+            startPage: parseInt(pathMatch[1], 10),
+        };
+    }
+    // 7. Check for offset-based pagination
+    const offsetMatch = /[?&](offset|start|skip)=(\d+)/.exec(url);
+    if (offsetMatch) {
+        const param = offsetMatch[1];
+        const currentOffset = parseInt(offsetMatch[2], 10);
+        // Try to detect limit param
+        const limitMatch = /[?&](limit|count|size)=(\d+)/.exec(url);
+        const limit = limitMatch ? parseInt(limitMatch[2], 10) : 20;
+        return {
+            type: 'url-pattern',
+            urlPattern: url.replace(new RegExp(`([?&])${param}=\\d+`), `$1${param}={page}`),
+            startPage: currentOffset,
+        };
+    }
+    return null;
+}
+/**
+ * Check if there's a next page available.
+ */
+export async function hasNextPage(proc, config, currentPage) {
+    if (config.type === 'url-list') {
+        return config.urls !== undefined && currentPage < config.urls.length;
+    }
+    if (config.type === 'url-pattern') {
+        // We can always try the next page — will be validated by the scraper
+        return true;
+    }
+    if (config.type === 'click-next' || config.type === 'load-more') {
+        if (!config.nextSelector)
+            return false;
+        const exists = await proc.evaluate(`
+      (function() {
+        var el = document.querySelector(${JSON.stringify(config.nextSelector)});
+        return el && el.offsetParent !== null && !el.disabled;
+      })()
+    `);
+        return exists;
+    }
+    if (config.type === 'infinite-scroll') {
+        return true; // Handled by scroll behavior
+    }
+    return false;
+}
+/**
+ * Navigate to the next page.
+ */
+export async function goToNextPage(proc, config, currentPage) {
+    try {
+        if (config.type === 'click-next' || config.type === 'load-more') {
+            if (!config.nextSelector)
+                return false;
+            await proc.click(config.nextSelector);
+            await proc.wait(config.waitAfter ?? 1000);
+            return true;
+        }
+        if (config.type === 'url-pattern') {
+            const nextPage = currentPage + 1;
+            const url = resolvePageUrl(config.urlPattern, nextPage);
+            await proc.goto(url);
+            await proc.wait(config.waitAfter ?? 1000);
+            return true;
+        }
+        if (config.type === 'url-list') {
+            if (!config.urls || currentPage >= config.urls.length)
+                return false;
+            await proc.goto(config.urls[currentPage]);
+            await proc.wait(config.waitAfter ?? 1000);
+            return true;
+        }
+        if (config.type === 'infinite-scroll') {
+            await proc.scrollToBottom();
+            await proc.wait(config.waitAfter ?? 1500);
+            return true;
+        }
+        return false;
+    }
+    catch {
+        return false;
+    }
+}
+/**
+ * Resolve a URL pattern with a page number.
+ */
+export function resolvePageUrl(pattern, page) {
+    return pattern.replace('{page}', String(page));
+}
+// ==================== Internal ====================
+function detectUrlPattern(currentUrl, nextUrl) {
+    try {
+        const current = new URL(currentUrl);
+        const next = new URL(nextUrl);
+        // Check query params
+        for (const [key, value] of next.searchParams) {
+            const currentValue = current.searchParams.get(key);
+            const nextNum = parseInt(value, 10);
+            const currentNum = currentValue ? parseInt(currentValue, 10) : NaN;
+            if (!isNaN(nextNum) && !isNaN(currentNum) && nextNum === currentNum + 1) {
+                const pattern = currentUrl.replace(new RegExp(`([?&])${key}=${currentNum}`), `$1${key}={page}`);
+                return { pattern, currentPage: currentNum };
+            }
+        }
+        // Check path pattern /page/N
+        const currentMatch = /\/page\/(\d+)/.exec(current.pathname);
+        const nextMatch = /\/page\/(\d+)/.exec(next.pathname);
+        if (currentMatch && nextMatch) {
+            const currentPage = parseInt(currentMatch[1], 10);
+            return {
+                pattern: currentUrl.replace(/\/page\/\d+/, '/page/{page}'),
+                currentPage,
+            };
+        }
+    }
+    catch {
+        // Invalid URLs
+    }
+    return null;
+}
+//# sourceMappingURL=pagination.js.map

package/dist/extraction/pagination.js.map

@@ -0,0 +1 @@
+{"version":3,"file":"pagination.js","sourceRoot":"","sources":["../../src/extraction/pagination.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAKH,8CAA8C;AAC9C,MAAM,kBAAkB,GAAG;IACzB,eAAe;IACf,kBAAkB;IAClB,mBAAmB;IACnB,wCAAwC;IACxC,yBAAyB;IACzB,uBAAuB;IACvB,8BAA8B;IAC9B,4BAA4B;IAC5B,uCAAuC;IACvC,qBAAqB;IACrB,oBAAoB;IACpB,gBAAgB;IAChB,aAAa;IACb,uBAAuB;IACvB,kBAAkB;IAClB,4BAA4B;CAC7B,CAAC;AAEF,mCAAmC;AACnC,MAAM,kBAAkB,GAAG;IACzB,4BAA4B;IAC5B,2BAA2B;IAC3B,uBAAuB;IACvB,6BAA6B;IAC7B,2BAA2B;IAC3B,4BAA4B;IAC5B,uBAAuB;CACxB,CAAC;AAEF;;;;;;GAMG;AACH,MAAM,CAAC,KAAK,UAAU,gBAAgB,CAAC,IAAmB;IACxD,MAAM,GAAG,GAAG,MAAM,IAAI,CAAC,MAAM,EAAE,CAAC;IAEhC,kDAAkD;IAClD,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,QAAQ,CAAC;;;;;;;;;;;;;;4BAcT,IAAI,CAAC,SAAS,CAAC,kBAAkB,CAAC;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;gCA4F9B,IAAI,CAAC,SAAS,CAAC,kBAAkB,CAAC;;;;;;;;;;;;;;GAc/D,CAAmE,CAAC;IAErE,IAAI,MAAM,EAAE,CAAC;QACX,IAAI,MAAM,CAAC,IAAI,KAAK,UAAU,IAAI,MAAM,CAAC,IAAI,EAAE,CAAC;YAC9C,MAAM,OAAO,GAAG,gBAAgB,CAAC,GAAG,EAAE,MAAM,CAAC,IAAI,CAAC,CAAC;YACnD,IAAI,OAAO,EAAE,CAAC;gBACZ,OAAO;oBACL,IAAI,EAAE,aAAa;oBACnB,UAAU,EAAE,OAAO,CAAC,OAAO;oBAC3B,SAAS,EAAE,OAAO,CAAC,WAAW;iBAC/B,CAAC;YACJ,CAAC;YACD,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,CAAC,QAAQ,EAAE,CAAC;QAC/D,CAAC;QAED,IAAI,MAAM,CAAC,IAAI,KAAK,YAAY,EAAE,CAAC;YACjC,OAAO,EAAE,IAAI,EAAE,YAAY,EAAE,YAAY,EAAE,MAAM,CAAC,QAAQ,EAAE,CAAC;QAC/D,CAAC;QAED,IAAI,MAAM,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;YAChC,OAAO,EAAE,IAAI,EAAE,WAAW,EAAE,YAAY,EAAE,MAAM,CAAC,QAAQ,EAAE,CAAC;QAC9D,CAAC;IACH,CAAC;IAED,2DAA2D;IAC3D,MAAM,QAAQ,GAAG,oBAAoB,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAChD,IAAI,QAAQ,EAAE,CAAC;QACb,MAAM,KAAK,GAAG,QAAQ,CAAC,CAAC,CAAE,CAAC;QAC3B,MAAM,WAAW,GAAG,QAAQ,CAAC,QAAQ,CAAC,CAAC,CAAE,EAAE,EAAE,CAAC,CAAC;QAC/C,OAAO;YACL,IAAI,EAAE,aAAa;YACnB,UAAU,EAAE,GAAG,CAAC,OAAO,CAAC,IAAI,MAAM,CAAC,SAAS,KAAK,OAAO,CAAC,EAAE,KAAK,KAAK,SAAS,CAAC;YAC/E,SAAS,EAAE,WAAW;SACvB,CAAC;IACJ,CAAC;IAED,sCAAsC;IACtC,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAC5C,IAAI,SAAS,EAAE,CAAC;QACd,OAAO;YACL,IAAI,EAAE,aAAa;YACnB,UAAU,EAAE,GAAG,CAAC,OAAO,CAAC,aAAa,EAAE,cAAc,CAAC;YACtD,SAAS,EAAE,QAAQ,CAAC,SAAS,CAAC,CAAC,CAAE,EAAE,EAAE,CAAC;SACvC,CAAC;IACJ,CAAC;IAED,uCAAuC;IACvC,MAAM,WAAW,GAAG,+BAA+B,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;IAC9D,IAAI,WAAW,EAAE,CAAC;QAChB,MAAM,KAAK,GAAG,WAAW,CAAC,CAAC,CAAE,CAAC;QAC9B,MAAM,aAAa,GAAG,QAAQ,CAAC,WAAW,CAAC,CAAC,CAAE,EAAE,EAAE,CAAC,CAAC;QACpD,4BAA4B;QAC5B,MAAM,UAAU,GAAG,8BAA8B,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;QAC5D,MAAM,KAAK,GAAG,UAAU,CAAC,CAAC,CAAC,QAAQ,CAAC,UAAU,CAAC,CAAC,CAAE,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,EAAE,CAAC;QAC7D,OAAO;YACL,IAAI,EAAE,aAAa;YACnB,UAAU,EAAE,GAAG,CAAC,OAAO,CAAC,IAAI,MAAM,CAAC,SAAS,KAAK,OAAO,CAAC,EAAE,KAAK,KAAK,SAAS,CAAC;YAC/E,SAAS,EAAE,aAAa;SACzB,CAAC;IACJ,CAAC;IAED,OAAO,IAAI,CAAC;AACd,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,WAAW,CAC/B,IAAmB,EACnB,MAAwB,EACxB,WAAmB;IAEnB,IAAI,MAAM,CAAC,IAAI,KAAK,UAAU,EAAE,CAAC;QAC/B,OAAO,MAAM,CAAC,IAAI,KAAK,SAAS,IAAI,WAAW,GAAG,MAAM,CAAC,IAAI,CAAC,MAAM,CAAC;IACvE,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,KAAK,aAAa,EAAE,CAAC;QAClC,qEAAqE;QACrE,OAAO,IAAI,CAAC;IACd,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,KAAK,YAAY,IAAI,MAAM,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;QAChE,IAAI,CAAC,MAAM,CAAC,YAAY;YAAE,OAAO,KAAK,CAAC;QACvC,MAAM,MAAM,GAAG,MAAM,IAAI,CAAC,QAAQ,CAAC;;0CAEG,IAAI,CAAC,SAAS,CAAC,MAAM,CAAC,YAAY,CAAC;;;KAGxE,CAAY,CAAC;QACd,OAAO,MAAM,CAAC;IAChB,CAAC;IAED,IAAI,MAAM,CAAC,IAAI,KAAK,iBAAiB,EAAE,CAAC;QACtC,OAAO,IAAI,CAAC,CAAC,6BAA6B;IAC5C,CAAC;IAED,OAAO,KAAK,CAAC;AACf,CAAC;AAED;;GAEG;AACH,MAAM,CAAC,KAAK,UAAU,YAAY,CAChC,IAAmB,EACnB,MAAwB,EACxB,WAAmB;IAEnB,IAAI,CAAC;QACH,IAAI,MAAM,CAAC,IAAI,KAAK,YAAY,IAAI,MAAM,CAAC,IAAI,KAAK,WAAW,EAAE,CAAC;YAChE,IAAI,CAAC,MAAM,CAAC,YAAY;gBAAE,OAAO,KAAK,CAAC;YACvC,MAAM,IAAI,CAAC,KAAK,CAAC,MAAM,CAAC,YAAY,CAAC,CAAC;YACtC,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;YAC1C,OAAO,IAAI,CAAC;QACd,CAAC;QAED,IAAI,MAAM,CAAC,IAAI,KAAK,aAAa,EAAE,CAAC;YAClC,MAAM,QAAQ,GAAG,WAAW,GAAG,CAAC,CAAC;YACjC,MAAM,GAAG,GAAG,cAAc,CAAC,MAAM,CAAC,UAAW,EAAE,QAAQ,CAAC,CAAC;YACzD,MAAM,IAAI,CAAC,IAAI,CAAC,GAAG,CAAC,CAAC;YACrB,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;YAC1C,OAAO,IAAI,CAAC;QACd,CAAC;QAED,IAAI,MAAM,CAAC,IAAI,KAAK,UAAU,EAAE,CAAC;YAC/B,IAAI,CAAC,MAAM,CAAC,IAAI,IAAI,WAAW,IAAI,MAAM,CAAC,IAAI,CAAC,MAAM;gBAAE,OAAO,KAAK,CAAC;YACpE,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,IAAI,CAAC,WAAW,CAAE,CAAC,CAAC;YAC3C,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;YAC1C,OAAO,IAAI,CAAC;QACd,CAAC;QAED,IAAI,MAAM,CAAC,IAAI,KAAK,iBAAiB,EAAE,CAAC;YACtC,MAAM,IAAI,CAAC,cAAc,EAAE,CAAC;YAC5B,MAAM,IAAI,CAAC,IAAI,CAAC,MAAM,CAAC,SAAS,IAAI,IAAI,CAAC,CAAC;YAC1C,OAAO,IAAI,CAAC;QACd,CAAC;QAED,OAAO,KAAK,CAAC;IACf,CAAC;IAAC,MAAM,CAAC;QACP,OAAO,KAAK,CAAC;IACf,CAAC;AACH,CAAC;AAED;;GAEG;AACH,MAAM,UAAU,cAAc,CAAC,OAAe,EAAE,IAAY;IAC1D,OAAO,OAAO,CAAC,OAAO,CAAC,QAAQ,EAAE,MAAM,CAAC,IAAI,CAAC,CAAC,CAAC;AACjD,CAAC;AAED,qDAAqD;AAErD,SAAS,gBAAgB,CACvB,UAAkB,EAClB,OAAe;IAEf,IAAI,CAAC;QACH,MAAM,OAAO,GAAG,IAAI,GAAG,CAAC,UAAU,CAAC,CAAC;QACpC,MAAM,IAAI,GAAG,IAAI,GAAG,CAAC,OAAO,CAAC,CAAC;QAE9B,qBAAqB;QACrB,KAAK,MAAM,CAAC,GAAG,EAAE,KAAK,CAAC,IAAI,IAAI,CAAC,YAAY,EAAE,CAAC;YAC7C,MAAM,YAAY,GAAG,OAAO,CAAC,YAAY,CAAC,GAAG,CAAC,GAAG,CAAC,CAAC;YACnD,MAAM,OAAO,GAAG,QAAQ,CAAC,KAAK,EAAE,EAAE,CAAC,CAAC;YACpC,MAAM,UAAU,GAAG,YAAY,CAAC,CAAC,CAAC,QAAQ,CAAC,YAAY,EAAE,EAAE,CAAC,CAAC,CAAC,CAAC,GAAG,CAAC;YAEnE,IAAI,CAAC,KAAK,CAAC,OAAO,CAAC,IAAI,CAAC,KAAK,CAAC,UAAU,CAAC,IAAI,OAAO,KAAK,UAAU,GAAG,CAAC,EAAE,CAAC;gBACxE,MAAM,OAAO,GAAG,UAAU,CAAC,OAAO,CAChC,IAAI,MAAM,CAAC,SAAS,GAAG,IAAI,UAAU,EAAE,CAAC,EACxC,KAAK,GAAG,SAAS,CAClB,CAAC;gBACF,OAAO,EAAE,OAAO,EAAE,WAAW,EAAE,UAAU,EAAE,CAAC;YAC9C,CAAC;QACH,CAAC;QAED,6BAA6B;QAC7B,MAAM,YAAY,GAAG,eAAe,CAAC,IAAI,CAAC,OAAO,CAAC,QAAQ,CAAC,CAAC;QAC5D,MAAM,SAAS,GAAG,eAAe,CAAC,IAAI,CAAC,IAAI,CAAC,QAAQ,CAAC,CAAC;QACtD,IAAI,YAAY,IAAI,SAAS,EAAE,CAAC;YAC9B,MAAM,WAAW,GAAG,QAAQ,CAAC,YAAY,CAAC,CAAC,CAAE,EAAE,EAAE,CAAC,CAAC;YACnD,OAAO;gBACL,OAAO,EAAE,UAAU,CAAC,OAAO,CAAC,aAAa,EAAE,cAAc,CAAC;gBAC1D,WAAW;aACZ,CAAC;QACJ,CAAC;IACH,CAAC;IAAC,MAAM,CAAC;QACP,eAAe;IACjB,CAAC;IACD,OAAO,IAAI,CAAC;AACd,CAAC"}

package/dist/extraction/pattern-detector.d.ts

@@ -0,0 +1,16 @@
+/**
+ * Zero-config auto pattern discovery via pure DOM analysis.
+ *
+ * Finds repeating DOM structures without any AI — uses tag+class frequency,
+ * child consistency scoring, and semantic field inference.
+ */
+import type { DetectedPattern, DetectOptions, ExtractedRecord } from './types.js';
+/**
+ * Detect repeating patterns on the page.
+ */
+export declare function detect(html: string, options?: DetectOptions): DetectedPattern[];
+/**
+ * Detect patterns and immediately extract the best one.
+ */
+export declare function detectAndExtract(html: string, options?: DetectOptions): ExtractedRecord[];
+//# sourceMappingURL=pattern-detector.d.ts.map

package/dist/extraction/pattern-detector.d.ts.map

@@ -0,0 +1 @@
+{"version":3,"file":"pattern-detector.d.ts","sourceRoot":"","sources":["../../src/extraction/pattern-detector.ts"],"names":[],"mappings":"AAAA;;;;;GAKG;AAIH,OAAO,KAAK,EAAE,eAAe,EAAE,aAAa,EAAE,eAAe,EAAa,MAAM,YAAY,CAAC;AAI7F;;GAEG;AACH,wBAAgB,MAAM,CACpB,IAAI,EAAE,MAAM,EACZ,OAAO,CAAC,EAAE,aAAa,GACtB,eAAe,EAAE,CAoDnB;AAED;;GAEG;AACH,wBAAgB,gBAAgB,CAC9B,IAAI,EAAE,MAAM,EACZ,OAAO,CAAC,EAAE,aAAa,GACtB,eAAe,EAAE,CAMnB"}