chore: Rework References to do fewer loops

[stloyd]

Change Log

Added

Fixed

Changed

• Rework `References` to do fewer loops

Removed

Deprecated

Security

---

Description

While looking on this class I noticed quite extensive amount of loops and array operations, unfortunately I don't have Blackfire (or alternative) installed right now to provide numbers.

[github-actions]

Flow PHP - Benchmarks

^{_{Results of the benchmarks from this PR are compared with the results from 1.x branch.}}

Extractors

+-----------------------+-------------------+------+-----+-----------------+------------------+----------------+
| benchmark             | subject           | revs | its | mem_peak        | mode             | rstdev         |
+-----------------------+-------------------+------+-----+-----------------+------------------+----------------+
| CSVExtractorBench     | bench_extract_10k | 1    | 3   | 3.916mb +0.01%  | 508.279ms -0.35% | ±0.02% -98.08% |
| JsonExtractorBench    | bench_extract_10k | 1    | 3   | 3.949mb +0.01%  | 1.081s +0.90%    | ±0.78% +56.86% |
| ParquetExtractorBench | bench_extract_10k | 1    | 3   | 28.509mb +0.00% | 437.551ms +3.10% | ±0.34% -44.73% |
| TextExtractorBench    | bench_extract_10k | 1    | 3   | 3.676mb +0.01%  | 33.451ms -2.74%  | ±0.79% -41.46% |
| XmlExtractorBench     | bench_extract_10k | 1    | 3   | 3.622mb +0.01%  | 431.671ms +0.55% | ±0.36% -29.54% |
+-----------------------+-------------------+------+-----+-----------------+------------------+----------------+

Transformers

+-----------------------------+--------------------------+------+-----+------------------+-----------------+----------------+
| benchmark                   | subject                  | revs | its | mem_peak         | mode            | rstdev         |
+-----------------------------+--------------------------+------+-----+------------------+-----------------+----------------+
| RenameEntryTransformerBench | bench_transform_10k_rows | 1    | 3   | 115.968mb +0.00% | 60.627ms +1.09% | ±1.09% +64.49% |
+-----------------------------+--------------------------+------+-----+------------------+-----------------+----------------+

Loaders

+--------------------+----------------+------+-----+------------------+-----------------+------------------+
| benchmark          | subject        | revs | its | mem_peak         | mode            | rstdev           |
+--------------------+----------------+------+-----+------------------+-----------------+------------------+
| CSVLoaderBench     | bench_load_10k | 1    | 3   | 54.071mb +0.00%  | 85.615ms +2.91% | ±0.39% -81.50%   |
| JsonLoaderBench    | bench_load_10k | 1    | 3   | 106.503mb +0.00% | 51.763ms -0.09% | ±1.64% +1191.47% |
| ParquetLoaderBench | bench_load_10k | 1    | 3   | 123.785mb +0.00% | 1.236s +1.12%   | ±0.78% +172.74%  |
| TextLoaderBench    | bench_load_10k | 1    | 3   | 16.865mb +0.00%  | 44.582ms +2.73% | ±0.24% +72.77%   |
+--------------------+----------------+------+-----+------------------+-----------------+------------------+

Building Blocks

+-------------------------+----------------------------+------+-----+------------------+------------------+------------------------------+
| benchmark               | subject                    | revs | its | mem_peak         | mode             | rstdev                       |
+-------------------------+----------------------------+------+-----+------------------+------------------+------------------------------+
| TypeDetectorBench       | bench_type_detector        | 1    | 3   | 59.697mb +0.00%  | 431.042ms -0.69% | ±0.55% -7.82%                |
| TypeDetectorBench       | bench_type_detector        | 1    | 3   | 14.236mb +0.00%  | 85.679ms +0.91%  | ±0.41% -34.38%               |
| NativeEntryFactoryBench | bench_entry_factory        | 1    | 3   | 116.518mb +0.00% | 488.588ms -1.10% | ±1.81% +31.28%               |
| NativeEntryFactoryBench | bench_entry_factory        | 1    | 3   | 59.997mb +0.00%  | 250.755ms +2.38% | ±1.49% +211.87%              |
| NativeEntryFactoryBench | bench_entry_factory        | 1    | 3   | 14.930mb +0.00%  | 53.676ms -1.87%  | ±0.16% -90.42%               |
| RowsBench               | bench_chunk_10_on_10k      | 2    | 3   | 86.788mb +0.00%  | 3.704ms +14.86%  | ±2.18% -2.44%                |
| RowsBench               | bench_diff_left_1k_on_10k  | 2    | 3   | 102.386mb +0.00% | 189.434ms +1.12% | ±0.27% -82.44%               |
| RowsBench               | bench_diff_right_1k_on_10k | 2    | 3   | 85.106mb +0.00%  | 18.808ms -1.87%  | ±0.79% -54.89%               |
| RowsBench               | bench_drop_1k_on_10k       | 2    | 3   | 88.028mb +0.00%  | 1.964ms +17.83%  | ±3.10% +303.66%              |
| RowsBench               | bench_drop_right_1k_on_10k | 2    | 3   | 88.028mb +0.00%  | 2.008ms +19.28%  | ±1.47% +119.03%              |
| RowsBench               | bench_entries_on_10k       | 2    | 3   | 85.140mb +0.00%  | 2.764ms +8.77%   | ±1.30% +106.37%              |
| RowsBench               | bench_filter_on_10k        | 2    | 3   | 85.669mb +0.00%  | 18.903ms +19.79% | ±2.08% +365.45%              |
| RowsBench               | bench_find_on_10k          | 2    | 3   | 85.669mb +0.00%  | 17.895ms +11.99% | ±1.70% +144.78%              |
| RowsBench               | bench_find_one_on_10k      | 10   | 3   | 83.573mb +0.00%  | 1.894μs +11.41%  | ±2.53% +0.00%                |
| RowsBench               | bench_first_on_10k         | 10   | 3   | 83.573mb +0.00%  | 0.300μs 0.00%    | ±0.00% 0.00%                 |
| RowsBench               | bench_flat_map_on_1k       | 2    | 3   | 92.923mb +0.00%  | 12.301ms +2.10%  | ±0.43% -66.14%               |
| RowsBench               | bench_map_on_10k           | 2    | 3   | 122.294mb +0.00% | 60.562ms +0.52%  | ±0.47% -68.78%               |
| RowsBench               | bench_merge_1k_on_10k      | 2    | 3   | 86.189mb +0.00%  | 1.668ms +37.18%  | ±2.54% +2908.81%             |
| RowsBench               | bench_partition_by_on_10k  | 2    | 3   | 89.536mb -0.00%  | 58.445ms +1.85%  | ±0.10% -85.78%               |
| RowsBench               | bench_remove_on_10k        | 2    | 3   | 88.290mb +0.00%  | 4.343ms +11.38%  | ±0.52% -31.96%               |
| RowsBench               | bench_sort_asc_on_1k       | 2    | 3   | 83.717mb +0.00%  | 39.199ms +0.54%  | ±1.32% +589.00%              |
| RowsBench               | bench_sort_by_on_1k        | 2    | 3   | 83.718mb +0.00%  | 39.925ms +0.80%  | ±1.02% -61.53%               |
| RowsBench               | bench_sort_desc_on_1k      | 2    | 3   | 83.717mb +0.00%  | 39.560ms +1.75%  | ±0.77% -40.28%               |
| RowsBench               | bench_sort_entries_on_1k   | 2    | 3   | 86.015mb +0.00%  | 7.390ms +1.28%   | ±0.52% -68.68%               |
| RowsBench               | bench_sort_on_1k           | 2    | 3   | 83.573mb +0.00%  | 29.260ms +3.32%  | ±1.14% +93.02%               |
| RowsBench               | bench_take_1k_on_10k       | 10   | 3   | 83.573mb +0.00%  | 14.000μs +5.26%  | ±0.58% +4366643765918200.00% |
| RowsBench               | bench_take_right_1k_on_10k | 10   | 3   | 83.573mb +0.00%  | 15.906μs +1.83%  | ±0.30% -63.07%               |
| RowsBench               | bench_unique_on_1k         | 2    | 3   | 102.387mb +0.00% | 193.378ms +0.72% | ±1.18% +89.73%               |
+-------------------------+----------------------------+------+-----+------------------+------------------+------------------------------+

[norberttech]

You might want to try to profile it with https://github.com/NoiseByNorthwest/php-spx but it looks good even without profiling, thanks!