feagi_evolutionary/genome/
parser.rs

1// Copyright 2025 Neuraville Inc.
2// SPDX-License-Identifier: Apache-2.0
3
4/*!
5Genome JSON parser.
6
7Parses FEAGI 2.1 genome JSON format into runtime data structures.
8
9## Genome Structure (v2.1)
10
11```json
12{
13  "genome_id": "...",
14  "genome_title": "...",
15  "version": "2.1",
16  "blueprint": {
17    "cortical_id": {
18      "cortical_name": "...",
19      "block_boundaries": [x, y, z],
20      "relative_coordinate": [x, y, z],
21      "cortical_type": "IPU/OPU/CUSTOM/CORE/MEMORY",
22      ...
23    }
24  },
25  "brain_regions": {
26    "root": {
27      "title": "...",
28      "parent_region_id": null,
29      "coordinate_3d": [x, y, z],
30      "areas": ["cortical_id1", ...],
31      "regions": ["child_region_id1", ...]
32    }
33  },
34  "neuron_morphologies": { ... },
35  "physiology": { ... }
36}
37```
38
39Copyright 2025 Neuraville Inc.
40Licensed under the Apache License, Version 2.0
41*/
42
43use serde::{Deserialize, Serialize};
44use serde_json::Value;
45use std::collections::HashMap;
46use tracing::warn;
47
48use crate::types::{EvoError, EvoResult};
49use feagi_structures::genomic::brain_regions::RegionID;
50use feagi_structures::genomic::cortical_area::CorticalID;
51use feagi_structures::genomic::cortical_area::{
52    CorticalArea, CorticalAreaDimensions as Dimensions,
53};
54use feagi_structures::genomic::descriptors::GenomeCoordinate3D;
55use feagi_structures::genomic::{BrainRegion, RegionType};
56
57/// Parsed genome data ready for ConnectomeManager
58#[derive(Debug, Clone)]
59pub struct ParsedGenome {
60    /// Genome metadata
61    pub genome_id: String,
62    pub genome_title: String,
63    pub version: String,
64
65    /// Cortical areas extracted from blueprint
66    pub cortical_areas: Vec<CorticalArea>,
67
68    /// Brain regions and hierarchy
69    pub brain_regions: Vec<(BrainRegion, Option<String>)>, // (region, parent_id)
70
71    /// Raw neuron morphologies (for later processing)
72    pub neuron_morphologies: HashMap<String, Value>,
73
74    /// Raw physiology data (for later processing)
75    pub physiology: Option<Value>,
76}
77
78/// Raw genome JSON structure for deserialization
79#[derive(Debug, Clone, Deserialize, Serialize)]
80pub struct RawGenome {
81    pub genome_id: Option<String>,
82    pub genome_title: Option<String>,
83    pub genome_description: Option<String>,
84    pub version: String,
85    pub blueprint: HashMap<String, RawCorticalArea>,
86    #[serde(default)]
87    pub brain_regions: HashMap<String, RawBrainRegion>,
88    #[serde(default)]
89    pub neuron_morphologies: HashMap<String, Value>,
90    #[serde(default)]
91    pub physiology: Option<Value>,
92    /// Root brain region ID (UUID string) - for O(1) root lookup
93    #[serde(default, skip_serializing_if = "Option::is_none")]
94    pub brain_regions_root: Option<String>,
95}
96
97/// Raw cortical area from blueprint
98#[derive(Debug, Clone, Deserialize, Serialize)]
99pub struct RawCorticalArea {
100    pub cortical_name: Option<String>,
101    pub block_boundaries: Option<Vec<u32>>,
102    pub relative_coordinate: Option<Vec<i32>>,
103    pub cortical_type: Option<String>,
104
105    // Optional properties
106    pub group_id: Option<String>,
107    pub sub_group_id: Option<String>,
108    pub per_voxel_neuron_cnt: Option<u32>,
109    pub cortical_mapping_dst: Option<Value>,
110
111    // Neural properties
112    pub synapse_attractivity: Option<f32>,
113    pub refractory_period: Option<u32>,
114    pub firing_threshold: Option<f32>,
115    pub firing_threshold_limit: Option<f32>,
116    pub firing_threshold_increment_x: Option<f32>,
117    pub firing_threshold_increment_y: Option<f32>,
118    pub firing_threshold_increment_z: Option<f32>,
119    pub leak_coefficient: Option<f32>,
120    pub leak_variability: Option<f32>,
121    pub neuron_excitability: Option<f32>,
122    pub postsynaptic_current: Option<f32>,
123    pub postsynaptic_current_max: Option<f32>,
124    pub degeneration: Option<f32>,
125    pub psp_uniform_distribution: Option<bool>,
126    pub mp_charge_accumulation: Option<bool>,
127    pub mp_driven_psp: Option<bool>,
128    pub visualization: Option<bool>,
129    pub burst_engine_activation: Option<bool>,
130    #[serde(rename = "2d_coordinate")]
131    pub coordinate_2d: Option<Vec<i32>>,
132
133    // Memory properties
134    pub is_mem_type: Option<bool>,
135    pub longterm_mem_threshold: Option<u32>,
136    pub lifespan_growth_rate: Option<f32>,
137    pub init_lifespan: Option<u32>,
138    pub temporal_depth: Option<u32>,
139    pub consecutive_fire_cnt_max: Option<u32>,
140    pub snooze_length: Option<u32>,
141
142    // Allow any other properties (future-proofing)
143    #[serde(flatten)]
144    pub other: HashMap<String, Value>,
145}
146
147/// Raw brain region from genome
148#[derive(Debug, Clone, Deserialize, Serialize)]
149pub struct RawBrainRegion {
150    pub title: Option<String>,
151    pub description: Option<String>,
152    pub parent_region_id: Option<String>,
153    pub coordinate_2d: Option<Vec<i32>>,
154    pub coordinate_3d: Option<Vec<i32>>,
155    pub areas: Option<Vec<String>>,
156    pub regions: Option<Vec<String>>,
157    pub inputs: Option<Vec<String>>,
158    pub outputs: Option<Vec<String>>,
159    pub signature: Option<String>,
160}
161
162/// Convert cortical_mapping_dst keys from old format to base64
163///
164/// This ensures all destination cortical IDs in dstmap are stored in the new base64 format.
165fn convert_dstmap_keys_to_base64(dstmap: &Value) -> Value {
166    if let Some(dstmap_obj) = dstmap.as_object() {
167        let mut converted = serde_json::Map::new();
168
169        for (dest_id_str, mapping_value) in dstmap_obj {
170            // Convert destination cortical_id to base64 format
171            match string_to_cortical_id(dest_id_str) {
172                Ok(dest_cortical_id) => {
173                    converted.insert(dest_cortical_id.as_base_64(), mapping_value.clone());
174                }
175                Err(e) => {
176                    // If conversion fails, keep original and log warning
177                    tracing::warn!(
178                        "Failed to convert dstmap key '{}' to base64: {}, keeping original",
179                        dest_id_str,
180                        e
181                    );
182                    converted.insert(dest_id_str.clone(), mapping_value.clone());
183                }
184            }
185        }
186
187        Value::Object(converted)
188    } else {
189        // Not an object, return as-is
190        dstmap.clone()
191    }
192}
193
194/// Convert a string cortical_id to CorticalID
195/// Handles both old 6-char format and new base64 format
196/// CRITICAL: Uses feagi-data-processing types as single source of truth for core areas
197pub fn string_to_cortical_id(id_str: &str) -> EvoResult<CorticalID> {
198    use feagi_structures::genomic::cortical_area::CoreCorticalType;
199
200    // Try base64 first (new format)
201    if let Ok(cortical_id) = CorticalID::try_from_base_64(id_str) {
202        let mut bytes = [0u8; CorticalID::CORTICAL_ID_LENGTH];
203        cortical_id.write_id_to_bytes(&mut bytes);
204        if bytes == *b"___power" {
205            return Ok(CoreCorticalType::Power.to_cortical_id());
206        }
207        if bytes == *b"___death" {
208            return Ok(CoreCorticalType::Death.to_cortical_id());
209        }
210        if bytes == *b"___fatig" {
211            return Ok(CoreCorticalType::Fatigue.to_cortical_id());
212        }
213        return Ok(cortical_id);
214    }
215
216    // Handle legacy CORE area names (6-char format) - use proper types from feagi-data-processing
217    if id_str == "_power" {
218        return Ok(CoreCorticalType::Power.to_cortical_id());
219    }
220    // Legacy shorthand used by older FEAGI genomes: "___pwr" (6-char) refers to core Power.
221    if id_str == "___pwr" {
222        return Ok(CoreCorticalType::Power.to_cortical_id());
223    }
224    // Legacy 8-char core names used in some BV caches
225    if id_str == "___power" {
226        return Ok(CoreCorticalType::Power.to_cortical_id());
227    }
228    // 8-char padded form of ___pwr (from 6-char padding in legacy flat genomes)
229    if id_str == "___pwr__" {
230        return Ok(CoreCorticalType::Power.to_cortical_id());
231    }
232    if id_str == "___death" {
233        return Ok(CoreCorticalType::Death.to_cortical_id());
234    }
235    if id_str == "___fatig" {
236        return Ok(CoreCorticalType::Fatigue.to_cortical_id());
237    }
238    if id_str == "_death" {
239        return Ok(CoreCorticalType::Death.to_cortical_id());
240    }
241    if id_str == "_fatigue" {
242        return Ok(CoreCorticalType::Fatigue.to_cortical_id());
243    }
244
245    // For non-core areas, use CorticalID's legacy ASCII parser (6-char and 8-char)
246    if id_str.len() == 6 || id_str.len() == 8 {
247        CorticalID::try_from_legacy_ascii(id_str).map_err(|e| {
248            EvoError::InvalidArea(format!("Failed to convert cortical_id '{}': {}", id_str, e))
249        })
250    } else {
251        Err(EvoError::InvalidArea(format!(
252            "Invalid cortical_id length: '{}' (expected 6 or 8 ASCII chars, or base64)",
253            id_str
254        )))
255    }
256}
257
258/// Genome parser
259pub struct GenomeParser;
260
261impl GenomeParser {
262    /// Parse a genome JSON string into a ParsedGenome
263    ///
264    /// # Arguments
265    ///
266    /// * `json_str` - JSON string of the genome
267    ///
268    /// # Returns
269    ///
270    /// Parsed genome ready for loading into ConnectomeManager
271    ///
272    /// # Errors
273    ///
274    /// Returns error if:
275    /// - JSON is malformed
276    /// - Required fields are missing
277    /// - Data types are invalid
278    ///
279    pub fn parse(json_str: &str) -> EvoResult<ParsedGenome> {
280        // Deserialize raw genome
281        let raw: RawGenome = serde_json::from_str(json_str)
282            .map_err(|e| EvoError::InvalidGenome(format!("Failed to parse JSON: {}", e)))?;
283
284        // Validate version - support 2.x and 3.x (3.0 is flat format with base64 IDs)
285        if !raw.version.starts_with("2.") && !raw.version.starts_with("3.") && raw.version != "3" {
286            return Err(EvoError::InvalidGenome(format!(
287                "Unsupported genome version: {}. Expected 2.x or 3.x",
288                raw.version
289            )));
290        }
291
292        // Parse cortical areas from blueprint
293        let cortical_areas = Self::parse_cortical_areas(&raw.blueprint)?;
294
295        // Parse brain regions
296        let brain_regions = Self::parse_brain_regions(&raw.brain_regions)?;
297
298        Ok(ParsedGenome {
299            genome_id: raw.genome_id.unwrap_or_else(|| "unknown".to_string()),
300            genome_title: raw.genome_title.unwrap_or_else(|| "Untitled".to_string()),
301            version: raw.version,
302            cortical_areas,
303            brain_regions,
304            neuron_morphologies: raw.neuron_morphologies,
305            physiology: raw.physiology,
306        })
307    }
308
309    /// Parse cortical areas from blueprint
310    fn parse_cortical_areas(
311        blueprint: &HashMap<String, RawCorticalArea>,
312    ) -> EvoResult<Vec<CorticalArea>> {
313        let mut areas = Vec::with_capacity(blueprint.len());
314
315        for (cortical_id_str, raw_area) in blueprint.iter() {
316            // Skip empty IDs
317            if cortical_id_str.is_empty() {
318                warn!(target: "feagi-evo","Skipping empty cortical_id");
319                continue;
320            }
321
322            // Convert string cortical_id to CorticalID (handles 6-char legacy and base64)
323            let cortical_id = match string_to_cortical_id(cortical_id_str) {
324                Ok(id) => id,
325                Err(e) => {
326                    warn!(target: "feagi-evo","Skipping invalid cortical_id '{}': {}", cortical_id_str, e);
327                    continue;
328                }
329            };
330
331            // Extract required fields
332            let name = raw_area
333                .cortical_name
334                .clone()
335                .unwrap_or_else(|| cortical_id_str.clone());
336
337            let dimensions = if let Some(boundaries) = &raw_area.block_boundaries {
338                if boundaries.len() != 3 {
339                    return Err(EvoError::InvalidArea(format!(
340                        "Invalid block_boundaries for {}: expected 3 values, got {}",
341                        cortical_id_str,
342                        boundaries.len()
343                    )));
344                }
345                Dimensions::new(boundaries[0], boundaries[1], boundaries[2])
346                    .map_err(|e| EvoError::InvalidArea(format!("Invalid dimensions: {}", e)))?
347            } else {
348                // Default to 1x1x1 if not specified (should not happen in valid genomes)
349                warn!(target: "feagi-evo","Cortical area {} missing block_boundaries, defaulting to 1x1x1", cortical_id_str);
350                Dimensions::new(1, 1, 1).map_err(|e| {
351                    EvoError::InvalidArea(format!("Invalid default dimensions: {}", e))
352                })?
353            };
354
355            let position = if let Some(coords) = &raw_area.relative_coordinate {
356                if coords.len() != 3 {
357                    return Err(EvoError::InvalidArea(format!(
358                        "Invalid relative_coordinate for {}: expected 3 values, got {}",
359                        cortical_id_str,
360                        coords.len()
361                    )));
362                }
363                GenomeCoordinate3D::new(coords[0], coords[1], coords[2])
364            } else {
365                // Default to origin if not specified
366                warn!(target: "feagi-evo","Cortical area {} missing relative_coordinate, defaulting to (0,0,0)", cortical_id_str);
367                GenomeCoordinate3D::new(0, 0, 0)
368            };
369
370            // Determine cortical type from cortical_id
371            let cortical_type = cortical_id.as_cortical_type().map_err(|e| {
372                EvoError::InvalidArea(format!(
373                    "Failed to determine cortical type from ID {}: {}",
374                    cortical_id_str, e
375                ))
376            })?;
377
378            // Create cortical area with CorticalID object (zero-copy, type-safe)
379            let mut area = CorticalArea::new(
380                cortical_id,
381                0, // cortical_idx will be assigned by ConnectomeManager
382                name,
383                dimensions,
384                position,
385                cortical_type,
386            )?;
387
388            // Store cortical_type as cortical_group for new type system
389            if let Some(ref cortical_type_str) = raw_area.cortical_type {
390                area.properties.insert(
391                    "cortical_group".to_string(),
392                    serde_json::json!(cortical_type_str),
393                );
394            }
395
396            // Store all properties in the properties HashMap
397            // Neural properties
398            if let Some(v) = raw_area.synapse_attractivity {
399                area.properties
400                    .insert("synapse_attractivity".to_string(), serde_json::json!(v));
401            }
402            if let Some(v) = raw_area.refractory_period {
403                area.properties
404                    .insert("refractory_period".to_string(), serde_json::json!(v));
405            }
406            if let Some(v) = raw_area.firing_threshold {
407                area.properties
408                    .insert("firing_threshold".to_string(), serde_json::json!(v));
409            }
410            if let Some(v) = raw_area.firing_threshold_limit {
411                area.properties
412                    .insert("firing_threshold_limit".to_string(), serde_json::json!(v));
413            }
414            if let Some(v) = raw_area.firing_threshold_increment_x {
415                area.properties.insert(
416                    "firing_threshold_increment_x".to_string(),
417                    serde_json::json!(v),
418                );
419            }
420            if let Some(v) = raw_area.firing_threshold_increment_y {
421                area.properties.insert(
422                    "firing_threshold_increment_y".to_string(),
423                    serde_json::json!(v),
424                );
425            }
426            if let Some(v) = raw_area.firing_threshold_increment_z {
427                area.properties.insert(
428                    "firing_threshold_increment_z".to_string(),
429                    serde_json::json!(v),
430                );
431            }
432            if let Some(v) = raw_area.leak_coefficient {
433                area.properties
434                    .insert("leak_coefficient".to_string(), serde_json::json!(v));
435            }
436            if let Some(v) = raw_area.leak_variability {
437                area.properties
438                    .insert("leak_variability".to_string(), serde_json::json!(v));
439            }
440            if let Some(v) = raw_area.neuron_excitability {
441                area.properties
442                    .insert("neuron_excitability".to_string(), serde_json::json!(v));
443            }
444            if let Some(v) = raw_area.postsynaptic_current {
445                area.properties
446                    .insert("postsynaptic_current".to_string(), serde_json::json!(v));
447            }
448            if let Some(v) = raw_area.postsynaptic_current_max {
449                area.properties
450                    .insert("postsynaptic_current_max".to_string(), serde_json::json!(v));
451            }
452            if let Some(v) = raw_area.degeneration {
453                area.properties
454                    .insert("degeneration".to_string(), serde_json::json!(v));
455            }
456
457            // Boolean properties
458            if let Some(v) = raw_area.psp_uniform_distribution {
459                area.properties
460                    .insert("psp_uniform_distribution".to_string(), serde_json::json!(v));
461            }
462            if let Some(v) = raw_area.mp_charge_accumulation {
463                area.properties
464                    .insert("mp_charge_accumulation".to_string(), serde_json::json!(v));
465            }
466            if let Some(v) = raw_area.mp_driven_psp {
467                area.properties
468                    .insert("mp_driven_psp".to_string(), serde_json::json!(v));
469                tracing::info!(
470                    target: "feagi-evo",
471                    "[GENOME-LOAD] Loaded mp_driven_psp={} for area {}",
472                    v,
473                    cortical_id_str
474                );
475            } else {
476                tracing::debug!(
477                    target: "feagi-evo",
478                    "[GENOME-LOAD] mp_driven_psp not found in raw_area for {}, will use default=false",
479                    cortical_id_str
480                );
481            }
482            if let Some(v) = raw_area.visualization {
483                area.properties
484                    .insert("visualization".to_string(), serde_json::json!(v));
485                // Also store as "visible" for compatibility with getters
486                area.properties
487                    .insert("visible".to_string(), serde_json::json!(v));
488            }
489            if let Some(v) = raw_area.burst_engine_activation {
490                area.properties
491                    .insert("burst_engine_active".to_string(), serde_json::json!(v));
492            }
493            if let Some(v) = raw_area.is_mem_type {
494                area.properties
495                    .insert("is_mem_type".to_string(), serde_json::json!(v));
496            }
497
498            // Memory properties
499            if let Some(v) = raw_area.longterm_mem_threshold {
500                area.properties
501                    .insert("longterm_mem_threshold".to_string(), serde_json::json!(v));
502            }
503            if let Some(v) = raw_area.lifespan_growth_rate {
504                area.properties
505                    .insert("lifespan_growth_rate".to_string(), serde_json::json!(v));
506            }
507            if let Some(v) = raw_area.init_lifespan {
508                area.properties
509                    .insert("init_lifespan".to_string(), serde_json::json!(v));
510            }
511            if let Some(v) = raw_area.temporal_depth {
512                area.properties
513                    .insert("temporal_depth".to_string(), serde_json::json!(v));
514            }
515            if let Some(v) = raw_area.consecutive_fire_cnt_max {
516                area.properties
517                    .insert("consecutive_fire_cnt_max".to_string(), serde_json::json!(v));
518                // Also store as "consecutive_fire_limit" for getter compatibility
519                area.properties
520                    .insert("consecutive_fire_limit".to_string(), serde_json::json!(v));
521            }
522            if let Some(v) = raw_area.snooze_length {
523                area.properties
524                    .insert("snooze_period".to_string(), serde_json::json!(v));
525            }
526
527            // Other properties
528            if let Some(v) = &raw_area.group_id {
529                area.properties
530                    .insert("group_id".to_string(), serde_json::json!(v));
531            }
532            if let Some(v) = &raw_area.sub_group_id {
533                area.properties
534                    .insert("sub_group_id".to_string(), serde_json::json!(v));
535            }
536            // Store neurons_per_voxel in properties HashMap
537            if let Some(v) = raw_area.per_voxel_neuron_cnt {
538                area.properties
539                    .insert("neurons_per_voxel".to_string(), serde_json::json!(v));
540            }
541            if let Some(v) = &raw_area.cortical_mapping_dst {
542                // Convert dstmap keys from old format to base64
543                let converted_dstmap = convert_dstmap_keys_to_base64(v);
544                area.properties
545                    .insert("cortical_mapping_dst".to_string(), converted_dstmap);
546            }
547            if let Some(v) = &raw_area.coordinate_2d {
548                area.properties
549                    .insert("2d_coordinate".to_string(), serde_json::json!(v));
550            }
551
552            // Store any other custom properties
553            for (key, value) in &raw_area.other {
554                area.properties.insert(key.clone(), value.clone());
555            }
556
557            // Note: cortical_type parsing disabled - CorticalArea is now a minimal data structure
558            // CorticalAreaType information is stored in properties["cortical_group"] if needed
559
560            areas.push(area);
561        }
562
563        Ok(areas)
564    }
565
566    /// Parse brain regions
567    fn parse_brain_regions(
568        raw_regions: &HashMap<String, RawBrainRegion>,
569    ) -> EvoResult<Vec<(BrainRegion, Option<String>)>> {
570        let mut regions = Vec::with_capacity(raw_regions.len());
571
572        for (region_id_str, raw_region) in raw_regions.iter() {
573            let title = raw_region
574                .title
575                .clone()
576                .unwrap_or_else(|| region_id_str.clone());
577
578            // Convert string region_id to RegionID (UUID)
579            // For now, try to parse as UUID if it's already a UUID, otherwise generate new one
580            let region_id = match RegionID::from_string(region_id_str) {
581                Ok(id) => id,
582                Err(_) => {
583                    // If not a valid UUID, generate a new one
584                    // This handles legacy string-based region IDs
585                    RegionID::new()
586                }
587            };
588
589            let region_type = RegionType::Undefined; // Default to Undefined
590
591            let mut region = BrainRegion::new(region_id, title, region_type)?;
592
593            // Add cortical areas to region (using CorticalID directly)
594            if let Some(areas) = &raw_region.areas {
595                for area_id in areas {
596                    // Convert area_id to CorticalID
597                    match string_to_cortical_id(area_id) {
598                        Ok(cortical_id) => {
599                            region.add_area(cortical_id);
600                        }
601                        Err(e) => {
602                            warn!(target: "feagi-evo",
603                                "Failed to convert brain region area ID '{}' to CorticalID: {}. Skipping.",
604                                area_id, e);
605                        }
606                    }
607                }
608            }
609
610            // Store properties in HashMap
611            if let Some(desc) = &raw_region.description {
612                region.add_property("description".to_string(), serde_json::json!(desc));
613            }
614            if let Some(coord_2d) = &raw_region.coordinate_2d {
615                region.add_property("coordinate_2d".to_string(), serde_json::json!(coord_2d));
616            }
617            if let Some(coord_3d) = &raw_region.coordinate_3d {
618                region.add_property("coordinate_3d".to_string(), serde_json::json!(coord_3d));
619            }
620            // Store inputs/outputs as base64 strings
621            if let Some(inputs) = &raw_region.inputs {
622                let input_ids: Vec<String> = inputs
623                    .iter()
624                    .filter_map(|id| match string_to_cortical_id(id) {
625                        Ok(cortical_id) => Some(cortical_id.as_base_64()),
626                        Err(e) => {
627                            warn!(target: "feagi-evo",
628                                    "Failed to convert brain region input ID '{}': {}. Skipping.",
629                                    id, e);
630                            None
631                        }
632                    })
633                    .collect();
634                if !input_ids.is_empty() {
635                    region.add_property("inputs".to_string(), serde_json::json!(input_ids));
636                }
637            }
638            if let Some(outputs) = &raw_region.outputs {
639                let output_ids: Vec<String> = outputs
640                    .iter()
641                    .filter_map(|id| match string_to_cortical_id(id) {
642                        Ok(cortical_id) => Some(cortical_id.as_base_64()),
643                        Err(e) => {
644                            warn!(target: "feagi-evo",
645                                    "Failed to convert brain region output ID '{}': {}. Skipping.",
646                                    id, e);
647                            None
648                        }
649                    })
650                    .collect();
651                if !output_ids.is_empty() {
652                    region.add_property("outputs".to_string(), serde_json::json!(output_ids));
653                }
654            }
655            if let Some(signature) = &raw_region.signature {
656                region.add_property("signature".to_string(), serde_json::json!(signature));
657            }
658
659            // Store parent_id for hierarchy construction
660            let parent_id = raw_region.parent_region_id.clone();
661            if let Some(ref parent_id_str) = parent_id {
662                // Store as property for serialization
663                region.add_property(
664                    "parent_region_id".to_string(),
665                    serde_json::json!(parent_id_str),
666                );
667            }
668
669            regions.push((region, parent_id));
670        }
671
672        Ok(regions)
673    }
674}
675
676#[cfg(test)]
677mod tests {
678    use super::*;
679
680    #[test]
681    fn test_parse_minimal_genome() {
682        // Test backward compatibility: parsing v2.1 genome with old 6-byte cortical ID
683        // Parser should convert old format to base64 for storage
684        let json = r#"{
685            "version": "2.1",
686            "blueprint": {
687                "_power": {
688                    "cortical_name": "Test Area",
689                    "block_boundaries": [10, 10, 10],
690                    "relative_coordinate": [0, 0, 0],
691                    "cortical_type": "CORE"
692                }
693            },
694            "brain_regions": {
695                "root": {
696                    "title": "Root",
697                    "parent_region_id": null,
698                    "areas": ["_power"]
699                }
700            }
701        }"#;
702
703        let parsed = GenomeParser::parse(json).unwrap();
704
705        assert_eq!(parsed.version, "2.1");
706        assert_eq!(parsed.cortical_areas.len(), 1);
707        // Input was "_power" (6 bytes), converted to "___power" (8 bytes, padded at start with underscores) then base64 encoded
708        assert_eq!(
709            parsed.cortical_areas[0].cortical_id.as_base_64(),
710            "X19fcG93ZXI="
711        );
712        assert_eq!(parsed.cortical_areas[0].name, "Test Area");
713        assert_eq!(parsed.brain_regions.len(), 1);
714
715        // Phase 2: Verify cortical_type_new is populated
716        // Note: cortical_type_new field removed - type is encoded in cortical_id
717        assert!(parsed.cortical_areas[0]
718            .cortical_id
719            .as_cortical_type()
720            .is_ok());
721    }
722
723    #[test]
724    fn test_parse_multiple_areas() {
725        // Test parsing multiple cortical areas with old format IDs
726        let json = r#"{
727            "version": "2.1",
728            "blueprint": {
729                "_power": {
730                    "cortical_name": "Area 1",
731                    "cortical_type": "CORE",
732                    "block_boundaries": [5, 5, 5],
733                    "relative_coordinate": [0, 0, 0]
734                },
735                "_death": {
736                    "cortical_name": "Area 2",
737                    "cortical_type": "CORE",
738                    "block_boundaries": [10, 10, 10],
739                    "relative_coordinate": [5, 0, 0]
740                }
741            }
742        }"#;
743
744        let parsed = GenomeParser::parse(json).unwrap();
745
746        assert_eq!(parsed.cortical_areas.len(), 2);
747
748        // Phase 2: Verify both areas have cortical_type_new populated
749        for area in &parsed.cortical_areas {
750            assert!(
751                area.cortical_id.as_cortical_type().is_ok(),
752                "Area {} should have cortical_type_new populated",
753                area.cortical_id
754            );
755        }
756    }
757
758    #[test]
759    fn test_string_to_cortical_id_legacy_power_shorthand() {
760        // Older FEAGI genomes may encode the power core area as "___pwr" (6-char shorthand).
761        // Migration must map this deterministically to the core Power cortical ID.
762        use feagi_structures::genomic::cortical_area::CoreCorticalType;
763        let id = string_to_cortical_id("___pwr").unwrap();
764        assert_eq!(
765            id.as_base_64(),
766            CoreCorticalType::Power.to_cortical_id().as_base_64()
767        );
768    }
769
770    #[test]
771    fn test_string_to_cortical_id_legacy_power_padded() {
772        // 8-char padded form ___pwr__ (from 6-char padding in legacy flat genomes).
773        use feagi_structures::genomic::cortical_area::CoreCorticalType;
774        let id = string_to_cortical_id("___pwr__").unwrap();
775        assert_eq!(
776            id.as_base_64(),
777            CoreCorticalType::Power.to_cortical_id().as_base_64()
778        );
779    }
780
781    #[test]
782    fn test_parse_with_properties() {
783        let json = r#"{
784            "version": "2.1",
785            "blueprint": {
786                "mem001": {
787                    "cortical_name": "Memory Area",
788                    "block_boundaries": [8, 8, 8],
789                    "relative_coordinate": [0, 0, 0],
790                    "cortical_type": "MEMORY",
791                    "is_mem_type": true,
792                    "firing_threshold": 50.0,
793                    "leak_coefficient": 0.9
794                }
795            }
796        }"#;
797
798        let parsed = GenomeParser::parse(json).unwrap();
799
800        assert_eq!(parsed.cortical_areas.len(), 1);
801        let area = &parsed.cortical_areas[0];
802
803        // Old type system (deprecated)
804        use feagi_structures::genomic::cortical_area::CorticalAreaType;
805        assert!(matches!(area.cortical_type, CorticalAreaType::Memory(_)));
806
807        // Properties stored correctly
808        assert!(area.properties.contains_key("is_mem_type"));
809        assert!(area.properties.contains_key("firing_threshold"));
810        assert!(area.properties.contains_key("cortical_group"));
811
812        // NEW: cortical_type should be derivable from cortical_id (Phase 2)
813        assert!(
814            area.cortical_id.as_cortical_type().is_ok(),
815            "cortical_id should be parseable to cortical_type"
816        );
817        if let Ok(cortical_type) = area.cortical_id.as_cortical_type() {
818            use feagi_structures::genomic::cortical_area::CorticalAreaType;
819            assert!(
820                matches!(cortical_type, CorticalAreaType::Memory(_)),
821                "Should be classified as MEMORY type"
822            );
823        }
824    }
825
826    #[test]
827    fn test_invalid_version() {
828        let json = r#"{
829            "version": "1.0",
830            "blueprint": {}
831        }"#;
832
833        let result = GenomeParser::parse(json);
834        assert!(result.is_err());
835    }
836
837    #[test]
838    fn test_malformed_json() {
839        let json = r#"{ "version": "2.1", "blueprint": { malformed"#;
840
841        let result = GenomeParser::parse(json);
842        assert!(result.is_err());
843    }
844
845    #[test]
846    fn test_cortical_type_new_population() {
847        // Test that cortical_type_new field is populated during parsing (Phase 2)
848        // This tests that parsing works with valid cortical IDs and populates types correctly
849        use feagi_structures::genomic::cortical_area::CoreCorticalType;
850        let power_id = CoreCorticalType::Power.to_cortical_id().as_base_64();
851        let json = format!(
852            r#"{{
853            "version": "2.1",
854            "blueprint": {{
855                "cvision1": {{
856                    "cortical_name": "Test Custom Vision",
857                    "cortical_type": "CUSTOM",
858                    "block_boundaries": [10, 10, 1],
859                    "relative_coordinate": [0, 0, 0]
860                }},
861                "cmotor01": {{
862                    "cortical_name": "Test Custom Motor",
863                    "cortical_type": "CUSTOM",
864                    "block_boundaries": [5, 5, 1],
865                    "relative_coordinate": [0, 0, 0]
866                }},
867                "{}": {{
868                    "cortical_name": "Test Core",
869                    "cortical_type": "CORE",
870                    "block_boundaries": [1, 1, 1],
871                    "relative_coordinate": [0, 0, 0]
872                }}
873            }}
874        }}"#,
875            power_id
876        );
877
878        let parsed = GenomeParser::parse(&json).unwrap();
879        assert_eq!(parsed.cortical_areas.len(), 3);
880
881        // Verify all areas have cortical_type_new populated
882        for area in &parsed.cortical_areas {
883            assert!(
884                area.cortical_id.as_cortical_type().is_ok(),
885                "Area {} should have cortical_type_new populated",
886                area.cortical_id
887            );
888
889            // Verify cortical_group property is also set
890            assert!(
891                area.properties.contains_key("cortical_group"),
892                "Area {} should have cortical_group property",
893                area.cortical_id
894            );
895
896            // Verify cortical group is consistent (avoid depending on feagi-brain-development)
897            if let Some(prop_group) = area
898                .properties
899                .get("cortical_group")
900                .and_then(|v| v.as_str())
901            {
902                assert!(
903                    !prop_group.is_empty(),
904                    "Area {} should have non-empty cortical_group property",
905                    area.cortical_id.as_base_64()
906                );
907            }
908        }
909    }
910}
feagi_evolutionary/genome/parser.rs

feagi_evolutionary/genome/
parser.rs
