formualizer_eval/engine/
formula_ingest.rs

1use std::collections::BTreeMap;
2use std::sync::Arc;
3
4use super::FormulaPlaneMode;
5use super::arena::AstNodeId;
6use super::formula_source::{SourceFamilyId, SourceFormulaOrder};
7
8#[derive(Clone, Debug)]
9pub struct FormulaIngestRecord {
10    pub row: u32,
11    pub col: u32,
12    pub ast_id: AstNodeId,
13    pub formula_text: Option<Arc<str>>,
14    pub(crate) source_order: Option<SourceFormulaOrder>,
15    pub(crate) source_family: Option<SourceFamilyId>,
16    pub(crate) partition_owner: Option<SourceFamilyId>,
17    /// Load-time family grouping (Program 2): `Some(anchor)` (0-based)
18    /// when this cell is a relative copy of the formula at `anchor`, whose
19    /// arena AST is `ast_id`; the member's own AST was never interned.
20    pub(crate) member_anchor: Option<(u32, u32)>,
21}
22
23impl FormulaIngestRecord {
24    pub fn new(row: u32, col: u32, ast_id: AstNodeId, formula_text: Option<Arc<str>>) -> Self {
25        Self {
26            row,
27            col,
28            ast_id,
29            formula_text,
30            source_order: None,
31            source_family: None,
32            partition_owner: None,
33            member_anchor: None,
34        }
35    }
36
37    /// A relative copy of `template` (the formula at 0-based `anchor`).
38    pub(crate) fn member(row: u32, col: u32, template: AstNodeId, anchor: (u32, u32)) -> Self {
39        Self {
40            member_anchor: Some(anchor),
41            ..Self::new(row, col, template, None)
42        }
43    }
44
45    /// Whether this record is a grouped family member (its `ast_id` is the
46    /// family template, valid at the anchor cell, not at this cell).
47    pub fn is_family_member(&self) -> bool {
48        self.member_anchor.is_some()
49    }
50
51    pub(crate) fn with_source_proof(
52        mut self,
53        source_order: SourceFormulaOrder,
54        source_family: Option<SourceFamilyId>,
55        partition_owner: Option<SourceFamilyId>,
56    ) -> Self {
57        self.source_order = Some(source_order);
58        self.source_family = source_family;
59        self.partition_owner = partition_owner;
60        self
61    }
62}
63
64/// Load-time family grouping state for one sheet (Program 2, P2-M2):
65/// the last formula staged in each column and the one before the current
66/// cell in its row. A formula that is exactly the formula above it (or to
67/// its left) relocated becomes a member of that formula's family and is
68/// never interned; see `Engine::stage_formula_ast`.
69#[derive(Debug, Default)]
70pub struct FormulaFamilyGrouper {
71    /// col0 -> (row0 of the last formula staged in the column, its family)
72    pub(crate) by_col: rustc_hash::FxHashMap<u32, (u32, GroupedFamily)>,
73    /// (row0, col0, family) of the last formula staged.
74    pub(crate) last: Option<(u32, u32, GroupedFamily)>,
75    /// Members grouped so far.
76    pub(crate) members: u64,
77}
78
79impl FormulaFamilyGrouper {
80    pub fn new() -> Self {
81        Self::default()
82    }
83
84    /// Number of formulas staged as family members.
85    pub fn members(&self) -> u64 {
86        self.members
87    }
88
89    pub(crate) fn note(&mut self, row0: u32, col0: u32, family: GroupedFamily) {
90        self.by_col.insert(col0, (row0, family.clone()));
91        self.last = Some((row0, col0, family));
92    }
93}
94
95/// A family template candidate: the template's arena root, its 0-based
96/// anchor cell, and (computed on first use) per reference whether the
97/// template's text is its rendering; `Some(None)` when the template
98/// cannot have members (it needs a structural rewrite).
99#[derive(Clone, Debug)]
100pub(crate) struct GroupedFamily {
101    pub(crate) template: AstNodeId,
102    pub(crate) anchor: (u32, u32),
103    pub(crate) rendered: Option<Option<Arc<[bool]>>>,
104}
105
106#[derive(Clone, Debug)]
107pub struct FormulaIngestBatch {
108    pub sheet_name: String,
109    pub formulas: Vec<FormulaIngestRecord>,
110}
111
112impl FormulaIngestBatch {
113    pub fn new(sheet_name: impl Into<String>, formulas: Vec<FormulaIngestRecord>) -> Self {
114        Self {
115            sheet_name: sheet_name.into(),
116            formulas,
117        }
118    }
119
120    pub fn len(&self) -> usize {
121        self.formulas.len()
122    }
123
124    pub fn is_empty(&self) -> bool {
125        self.formulas.is_empty()
126    }
127}
128
129#[derive(Debug, Clone, PartialEq, Eq)]
130pub struct FormulaIngestReport {
131    pub mode: FormulaPlaneMode,
132    pub formula_cells_seen: u64,
133    pub graph_formula_cells_materialized: u64,
134
135    pub shadow_candidate_cells: u64,
136    pub shadow_accepted_span_cells: u64,
137    pub shadow_fallback_cells: u64,
138    pub shadow_templates_interned: u64,
139    pub shadow_spans_created: u64,
140
141    pub graph_formula_vertices_avoided_shadow: u64,
142    pub ast_roots_avoided_shadow: u64,
143    pub edge_rows_avoided_shadow: u64,
144
145    pub graph_vertices_created: u64,
146    pub graph_edges_created: u64,
147
148    pub source_formula_events: u64,
149    pub source_formula_records_spooled: u64,
150    pub source_spool_encoded_bytes: u64,
151    pub source_spool_peak_memory_bytes: u64,
152    pub source_spool_spilled_bytes: u64,
153    pub source_spool_spill_files: u64,
154    pub source_spool_replays: u64,
155    pub source_ordinary_events: u64,
156    pub source_shared_anchor_events: u64,
157    pub source_shared_descendant_events: u64,
158    pub source_unknown_events: u64,
159    pub source_families_seen: u64,
160    pub source_family_cells_seen: u64,
161    pub source_family_shadow_eligible: u64,
162    pub source_family_shadow_eligible_cells: u64,
163    pub source_family_promoted: u64,
164    pub source_family_promoted_cells: u64,
165    pub source_family_fallback: u64,
166    pub source_family_fallback_cells: u64,
167    pub source_forward_descendants: u64,
168    pub source_evidence_limit_fallbacks: u64,
169    pub source_evidence_peak_bytes: u64,
170    pub source_anchor_parses: u64,
171    pub source_anchor_asts: u64,
172    pub source_anchor_analyses: u64,
173    pub source_descendant_strings_avoided: u64,
174    pub source_descendant_events_avoided: u64,
175    pub source_descendant_analyses_avoided: u64,
176    pub source_compressed_families_prepared: u64,
177    pub source_compressed_cells_prepared: u64,
178    pub source_partitioned_families_seen: u64,
179    pub source_partitioned_families_prepared: u64,
180    pub source_partitioned_families_rejected: u64,
181    pub source_partition_fragments_prepared: u64,
182    pub source_partition_span_cells_prepared: u64,
183    pub source_partition_fallback_cells: u64,
184    pub source_partition_analyses_reused: u64,
185    pub source_partition_function_semantics: u64,
186    pub source_partition_holes: u64,
187    pub source_partition_ordinary_exceptions: u64,
188    pub source_partition_failures: u64,
189    pub source_partition_surviving_cells: u64,
190
191    pub fallback_reasons: BTreeMap<String, u64>,
192}
193
194impl Default for FormulaIngestReport {
195    fn default() -> Self {
196        Self {
197            mode: FormulaPlaneMode::Off,
198            formula_cells_seen: 0,
199            graph_formula_cells_materialized: 0,
200            shadow_candidate_cells: 0,
201            shadow_accepted_span_cells: 0,
202            shadow_fallback_cells: 0,
203            shadow_templates_interned: 0,
204            shadow_spans_created: 0,
205            graph_formula_vertices_avoided_shadow: 0,
206            ast_roots_avoided_shadow: 0,
207            edge_rows_avoided_shadow: 0,
208            graph_vertices_created: 0,
209            graph_edges_created: 0,
210            source_formula_events: 0,
211            source_formula_records_spooled: 0,
212            source_spool_encoded_bytes: 0,
213            source_spool_peak_memory_bytes: 0,
214            source_spool_spilled_bytes: 0,
215            source_spool_spill_files: 0,
216            source_spool_replays: 0,
217            source_ordinary_events: 0,
218            source_shared_anchor_events: 0,
219            source_shared_descendant_events: 0,
220            source_unknown_events: 0,
221            source_families_seen: 0,
222            source_family_cells_seen: 0,
223            source_family_shadow_eligible: 0,
224            source_family_shadow_eligible_cells: 0,
225            source_family_promoted: 0,
226            source_family_promoted_cells: 0,
227            source_family_fallback: 0,
228            source_family_fallback_cells: 0,
229            source_forward_descendants: 0,
230            source_evidence_limit_fallbacks: 0,
231            source_evidence_peak_bytes: 0,
232            source_anchor_parses: 0,
233            source_anchor_asts: 0,
234            source_anchor_analyses: 0,
235            source_descendant_strings_avoided: 0,
236            source_descendant_events_avoided: 0,
237            source_descendant_analyses_avoided: 0,
238            source_compressed_families_prepared: 0,
239            source_compressed_cells_prepared: 0,
240            source_partitioned_families_seen: 0,
241            source_partitioned_families_prepared: 0,
242            source_partitioned_families_rejected: 0,
243            source_partition_fragments_prepared: 0,
244            source_partition_span_cells_prepared: 0,
245            source_partition_fallback_cells: 0,
246            source_partition_analyses_reused: 0,
247            source_partition_function_semantics: 0,
248            source_partition_holes: 0,
249            source_partition_ordinary_exceptions: 0,
250            source_partition_failures: 0,
251            source_partition_surviving_cells: 0,
252            fallback_reasons: BTreeMap::new(),
253        }
254    }
255}
256
257impl FormulaIngestReport {
258    pub(crate) fn with_mode(mode: FormulaPlaneMode) -> Self {
259        Self {
260            mode,
261            ..Self::default()
262        }
263    }
264
265    pub(crate) fn accumulate(&mut self, other: &Self) {
266        self.formula_cells_seen = self
267            .formula_cells_seen
268            .saturating_add(other.formula_cells_seen);
269        self.graph_formula_cells_materialized = self
270            .graph_formula_cells_materialized
271            .saturating_add(other.graph_formula_cells_materialized);
272        self.shadow_candidate_cells = self
273            .shadow_candidate_cells
274            .saturating_add(other.shadow_candidate_cells);
275        self.shadow_accepted_span_cells = self
276            .shadow_accepted_span_cells
277            .saturating_add(other.shadow_accepted_span_cells);
278        self.shadow_fallback_cells = self
279            .shadow_fallback_cells
280            .saturating_add(other.shadow_fallback_cells);
281        self.shadow_templates_interned = self
282            .shadow_templates_interned
283            .saturating_add(other.shadow_templates_interned);
284        self.shadow_spans_created = self
285            .shadow_spans_created
286            .saturating_add(other.shadow_spans_created);
287        self.graph_formula_vertices_avoided_shadow = self
288            .graph_formula_vertices_avoided_shadow
289            .saturating_add(other.graph_formula_vertices_avoided_shadow);
290        self.ast_roots_avoided_shadow = self
291            .ast_roots_avoided_shadow
292            .saturating_add(other.ast_roots_avoided_shadow);
293        self.edge_rows_avoided_shadow = self
294            .edge_rows_avoided_shadow
295            .saturating_add(other.edge_rows_avoided_shadow);
296        self.graph_vertices_created = self
297            .graph_vertices_created
298            .saturating_add(other.graph_vertices_created);
299        self.graph_edges_created = self
300            .graph_edges_created
301            .saturating_add(other.graph_edges_created);
302        self.source_formula_events = self
303            .source_formula_events
304            .saturating_add(other.source_formula_events);
305        self.source_formula_records_spooled = self
306            .source_formula_records_spooled
307            .saturating_add(other.source_formula_records_spooled);
308        self.source_spool_encoded_bytes = self
309            .source_spool_encoded_bytes
310            .saturating_add(other.source_spool_encoded_bytes);
311        self.source_spool_peak_memory_bytes = self
312            .source_spool_peak_memory_bytes
313            .max(other.source_spool_peak_memory_bytes);
314        self.source_spool_spilled_bytes = self
315            .source_spool_spilled_bytes
316            .saturating_add(other.source_spool_spilled_bytes);
317        self.source_spool_spill_files = self
318            .source_spool_spill_files
319            .saturating_add(other.source_spool_spill_files);
320        self.source_spool_replays = self
321            .source_spool_replays
322            .saturating_add(other.source_spool_replays);
323        self.source_ordinary_events = self
324            .source_ordinary_events
325            .saturating_add(other.source_ordinary_events);
326        self.source_shared_anchor_events = self
327            .source_shared_anchor_events
328            .saturating_add(other.source_shared_anchor_events);
329        self.source_shared_descendant_events = self
330            .source_shared_descendant_events
331            .saturating_add(other.source_shared_descendant_events);
332        self.source_unknown_events = self
333            .source_unknown_events
334            .saturating_add(other.source_unknown_events);
335        self.source_families_seen = self
336            .source_families_seen
337            .saturating_add(other.source_families_seen);
338        self.source_family_cells_seen = self
339            .source_family_cells_seen
340            .saturating_add(other.source_family_cells_seen);
341        self.source_family_shadow_eligible = self
342            .source_family_shadow_eligible
343            .saturating_add(other.source_family_shadow_eligible);
344        self.source_family_shadow_eligible_cells = self
345            .source_family_shadow_eligible_cells
346            .saturating_add(other.source_family_shadow_eligible_cells);
347        self.source_family_promoted = self
348            .source_family_promoted
349            .saturating_add(other.source_family_promoted);
350        self.source_family_promoted_cells = self
351            .source_family_promoted_cells
352            .saturating_add(other.source_family_promoted_cells);
353        self.source_family_fallback = self
354            .source_family_fallback
355            .saturating_add(other.source_family_fallback);
356        self.source_family_fallback_cells = self
357            .source_family_fallback_cells
358            .saturating_add(other.source_family_fallback_cells);
359        self.source_forward_descendants = self
360            .source_forward_descendants
361            .saturating_add(other.source_forward_descendants);
362        self.source_evidence_limit_fallbacks = self
363            .source_evidence_limit_fallbacks
364            .saturating_add(other.source_evidence_limit_fallbacks);
365        self.source_evidence_peak_bytes = self
366            .source_evidence_peak_bytes
367            .max(other.source_evidence_peak_bytes);
368        self.source_anchor_parses = self
369            .source_anchor_parses
370            .saturating_add(other.source_anchor_parses);
371        self.source_anchor_asts = self
372            .source_anchor_asts
373            .saturating_add(other.source_anchor_asts);
374        self.source_anchor_analyses = self
375            .source_anchor_analyses
376            .saturating_add(other.source_anchor_analyses);
377        self.source_descendant_strings_avoided = self
378            .source_descendant_strings_avoided
379            .saturating_add(other.source_descendant_strings_avoided);
380        self.source_descendant_events_avoided = self
381            .source_descendant_events_avoided
382            .saturating_add(other.source_descendant_events_avoided);
383        self.source_descendant_analyses_avoided = self
384            .source_descendant_analyses_avoided
385            .saturating_add(other.source_descendant_analyses_avoided);
386        self.source_compressed_families_prepared = self
387            .source_compressed_families_prepared
388            .saturating_add(other.source_compressed_families_prepared);
389        self.source_compressed_cells_prepared = self
390            .source_compressed_cells_prepared
391            .saturating_add(other.source_compressed_cells_prepared);
392        self.source_partitioned_families_seen = self
393            .source_partitioned_families_seen
394            .saturating_add(other.source_partitioned_families_seen);
395        self.source_partitioned_families_prepared = self
396            .source_partitioned_families_prepared
397            .saturating_add(other.source_partitioned_families_prepared);
398        self.source_partitioned_families_rejected = self
399            .source_partitioned_families_rejected
400            .saturating_add(other.source_partitioned_families_rejected);
401        self.source_partition_fragments_prepared = self
402            .source_partition_fragments_prepared
403            .saturating_add(other.source_partition_fragments_prepared);
404        self.source_partition_span_cells_prepared = self
405            .source_partition_span_cells_prepared
406            .saturating_add(other.source_partition_span_cells_prepared);
407        self.source_partition_fallback_cells = self
408            .source_partition_fallback_cells
409            .saturating_add(other.source_partition_fallback_cells);
410        self.source_partition_analyses_reused = self
411            .source_partition_analyses_reused
412            .saturating_add(other.source_partition_analyses_reused);
413        self.source_partition_function_semantics = self
414            .source_partition_function_semantics
415            .saturating_add(other.source_partition_function_semantics);
416        self.source_partition_holes = self
417            .source_partition_holes
418            .saturating_add(other.source_partition_holes);
419        self.source_partition_ordinary_exceptions = self
420            .source_partition_ordinary_exceptions
421            .saturating_add(other.source_partition_ordinary_exceptions);
422        self.source_partition_failures = self
423            .source_partition_failures
424            .saturating_add(other.source_partition_failures);
425        self.source_partition_surviving_cells = self
426            .source_partition_surviving_cells
427            .saturating_add(other.source_partition_surviving_cells);
428        for (reason, count) in &other.fallback_reasons {
429            let total = self.fallback_reasons.entry(reason.clone()).or_default();
430            *total = total.saturating_add(*count);
431        }
432    }
433}
formualizer_eval/engine/formula_ingest.rs

formualizer_eval/engine/
formula_ingest.rs
