20  * A collection manages a group of related indexes that together provide a specific search use case.
140      * @param int $group Index group (0 for non-split, token length for split)
144     public function getTokenIndex(int $group = 0): AbstractIndex  argument
149         return new MemoryIndex($this->idxToken, $this->groupToSuffix($group), $this->isWritable);
153      * @param int $group Index group (0 for non-split, token length for split)
157     public function getFrequencyIndex(int $group = 0): AbstractIndex  argument
159 …      return new MemoryIndex($this->idxFrequency, $this->groupToSuffix($group), $this->isWritable);
182      * Convert a logical group number to the index file suffix
184      * Group 0 represents non-split indexes (suffix '') while positive integers
187      * @param int $group
188      * @return string The file suffix ('' for group 0, the group number as string otherwise)
189      * @throws IndexUsageException when group does not match the collection's split mode
191     protected function groupToSuffix(int $group): string  argument
193         if ($group === 0 && $this->splitByLength) {
194             throw new IndexUsageException('Group 0 is not valid for split-by-length collections');
196         if ($group !== 0 && !$this->splitByLength) {
197             throw new IndexUsageException("Group $group is not valid for non-split collections");
199         return $group === 0 ? '' : (string)$group;
205      * Given a set of token IDs from a specific index group, returns the entities
210      * @param int $group Index group (0 for non-split, token length for split)
214     public function resolveTokenFrequencies(int $group, array $tokenIds): array  argument
216         $freqIndex = $this->getFrequencyIndex($group);
235         foreach ($groups as $group) {
236             $freqIndex = $this->getFrequencyIndex($group);
267      * - token == frequency (per group, both keyed by token RID)
278         foreach ($groups as $group) {
279             $tokenIndex = $this->getTokenIndex($group);
280             $freqIndex = $this->getFrequencyIndex($group);
286                     "Group $group: missing " .
295                     "Group $group: token count ($tc) != frequency count ($fc)"
353      * Resolve raw tokens into the two-level structure [group => [tokenId => frequency]]
360      * @return array [group => [tokenId => frequency, ...], ...]
367         // group tokens by their index suffix
370             $group = $this->splitByLength ? Tokenizer::tokenLength($token) : 0;
371             $groups[$group][$token] = $freq;
376         foreach ($groups as $group => $tokenFreqs) {
377             $tokenIndex = $this->getTokenIndex($group);
378             $result[$group] = [];
381                 $result[$group][$tokenId] = $freq;
458 …* For split collections the format is "group*tokenId:group*tokenId:..." where group is the token l…
459      * For non-split collections the group prefix is omitted: "tokenId:tokenId:..."
463      * @return array [group => [tokenId => 0, ...], ...]
471             $group = (int)(array_pop($parts) ?? 0);
472             $result[$group][$tokenId] = 0;
480      * @param array $data [group => [tokenId => freq, ...], ...]
486         foreach ($data as $group => $tokens) {
487             $prefix = $group === 0 ? '' : "$group*";
498      * Iterates over the two-level structure [group => [tokenId => freq]] and updates the
499      * corresponding frequency index for each group. A frequency of 0 removes the entity
502      * @param array $data [group => [tokenId => frequency, ...], ...]
508         foreach ($data as $group => $tokens) {
509             $freqIndex = $this->getFrequencyIndex($group);