diff options
Diffstat (limited to 'src/lib/utils/batchSolrQueries.js')
| -rw-r--r-- | src/lib/utils/batchSolrQueries.js | 113 |
1 files changed, 113 insertions, 0 deletions
diff --git a/src/lib/utils/batchSolrQueries.js b/src/lib/utils/batchSolrQueries.js new file mode 100644 index 00000000..486f1778 --- /dev/null +++ b/src/lib/utils/batchSolrQueries.js @@ -0,0 +1,113 @@ +/** + * Batch utility functions for handling large product ID arrays in Solr queries + * Prevents URL length limit errors when querying with >100 product IDs + */ + +/** + * Split an array into chunks of specified size + * @param {Array} array - Array to split + * @param {number} size - Chunk size (default: 100) + * @returns {Array<Array>} Array of chunks + */ +export const chunkArray = (array, size = 100) => { + if (!Array.isArray(array) || array.length === 0) return []; + + const chunks = []; + for (let i = 0; i < array.length; i += size) { + chunks.push(array.slice(i, i + size)); + } + return chunks; +}; + +/** + * Build a product ID OR clause for Solr query + * @param {Array<string|number>} ids - Array of product IDs + * @returns {string} Formatted OR clause: "product_id_i:123 OR product_id_i:456..." + */ +export const buildProductIdOrClause = (ids) => { + if (!Array.isArray(ids) || ids.length === 0) { + return '*:*'; + } + return ids.map((id) => `product_id_i:${id}`).join(' OR '); +}; + +/** + * Validate query size to prevent exceeding HTTP limits + * @param {string} query - Query string to validate + * @param {number} maxSize - Maximum allowed size in characters (default: 6000) + * @returns {Object} { valid: boolean, message: string, size: number } + */ +export const validateQuerySize = (query, maxSize = 6000) => { + const size = query.length; + return { + valid: size <= maxSize, + message: + size > maxSize + ? `Query size ${size} exceeds limit of ${maxSize}` + : `Query size ${size} is within limit`, + size, + }; +}; + +/** + * Build batched Solr query parameters for large ID arrays + * Chunks IDs into groups and creates separate OR clauses + * @param {Array<string|number>} ids - Product IDs to query + * @param {number} chunkSize - How many IDs per chunk (default: 100) + * @returns {Array<string>} Array of OR clauses, one per chunk + */ +export const buildBatchedOrClauses = (ids, chunkSize = 100) => { + if (!Array.isArray(ids) || ids.length === 0) { + return ['*:*']; + } + + const chunks = chunkArray(ids, chunkSize); + + if (chunks.length === 1) { + // Single chunk, return standard OR clause + return [buildProductIdOrClause(ids)]; + } + + // Multiple chunks: return OR clauses wrapped with parentheses for combining + return chunks.map((chunk) => `(${buildProductIdOrClause(chunk)})`); +}; + +/** + * Combine multiple OR clauses into a single query (for Solr) + * @param {Array<string>} orClauses - Array of OR clauses + * @returns {string} Combined query with OR between clauses + */ +export const combineOrClauses = (orClauses) => { + if (!Array.isArray(orClauses) || orClauses.length === 0) { + return '*:*'; + } + if (orClauses.length === 1) { + return orClauses[0]; + } + return orClauses.join(' OR '); +}; + +/** + * Merge Solr response documents from multiple queries + * Removes duplicates based on product_id_i + * @param {Array<Array>} responseArrays - Array of response.docs arrays + * @returns {Array} Merged and deduplicated docs + */ +export const mergeSolrResults = (responseArrays) => { + const seen = new Set(); + const merged = []; + + responseArrays.forEach((docs) => { + if (Array.isArray(docs)) { + docs.forEach((doc) => { + const id = doc.product_id_i; + if (id && !seen.has(id)) { + seen.add(id); + merged.push(doc); + } + }); + } + }); + + return merged; +}; |
