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    if id_str == "___death" {
229        return Ok(CoreCorticalType::Death.to_cortical_id());
230    }
231    if id_str == "___fatig" {
232        return Ok(CoreCorticalType::Fatigue.to_cortical_id());
233    }
234    if id_str == "_death" {
235        return Ok(CoreCorticalType::Death.to_cortical_id());
236    }
237    if id_str == "_fatigue" {
238        return Ok(CoreCorticalType::Fatigue.to_cortical_id());
239    }
240
241    // For non-core areas, handle 6-char and 8-char ASCII formats
242    if id_str.len() == 6 {
243        // Legacy 6-char non-core IDs: pad with underscores on the right
244        let mut bytes = [b'_'; 8];
245        bytes[..6].copy_from_slice(id_str.as_bytes());
246
247        CorticalID::try_from_bytes(&bytes).map_err(|e| {
248            EvoError::InvalidArea(format!("Failed to convert cortical_id '{}': {}", id_str, e))
249        })
250    } else if id_str.len() == 8 {
251        // Already 8 bytes - convert directly
252        let mut bytes = [0u8; 8];
253        bytes.copy_from_slice(id_str.as_bytes());
254
255        CorticalID::try_from_bytes(&bytes).map_err(|e| {
256            EvoError::InvalidArea(format!("Failed to convert cortical_id '{}': {}", id_str, e))
257        })
258    } else {
259        Err(EvoError::InvalidArea(format!(
260            "Invalid cortical_id length: '{}' (expected 6 or 8 ASCII chars, or base64)",
261            id_str
262        )))
263    }
264}
265
266/// Genome parser
267pub struct GenomeParser;
268
269impl GenomeParser {
270    /// Parse a genome JSON string into a ParsedGenome
271    ///
272    /// # Arguments
273    ///
274    /// * `json_str` - JSON string of the genome
275    ///
276    /// # Returns
277    ///
278    /// Parsed genome ready for loading into ConnectomeManager
279    ///
280    /// # Errors
281    ///
282    /// Returns error if:
283    /// - JSON is malformed
284    /// - Required fields are missing
285    /// - Data types are invalid
286    ///
287    pub fn parse(json_str: &str) -> EvoResult<ParsedGenome> {
288        // Deserialize raw genome
289        let raw: RawGenome = serde_json::from_str(json_str)
290            .map_err(|e| EvoError::InvalidGenome(format!("Failed to parse JSON: {}", e)))?;
291
292        // Validate version - support 2.x and 3.0 (3.0 is flat format with all IDs in base64)
293        if !raw.version.starts_with("2.") && raw.version != "3.0" {
294            return Err(EvoError::InvalidGenome(format!(
295                "Unsupported genome version: {}. Expected 2.x or 3.0",
296                raw.version
297            )));
298        }
299
300        // Parse cortical areas from blueprint
301        let cortical_areas = Self::parse_cortical_areas(&raw.blueprint)?;
302
303        // Parse brain regions
304        let brain_regions = Self::parse_brain_regions(&raw.brain_regions)?;
305
306        Ok(ParsedGenome {
307            genome_id: raw.genome_id.unwrap_or_else(|| "unknown".to_string()),
308            genome_title: raw.genome_title.unwrap_or_else(|| "Untitled".to_string()),
309            version: raw.version,
310            cortical_areas,
311            brain_regions,
312            neuron_morphologies: raw.neuron_morphologies,
313            physiology: raw.physiology,
314        })
315    }
316
317    /// Parse cortical areas from blueprint
318    fn parse_cortical_areas(
319        blueprint: &HashMap<String, RawCorticalArea>,
320    ) -> EvoResult<Vec<CorticalArea>> {
321        let mut areas = Vec::with_capacity(blueprint.len());
322
323        for (cortical_id_str, raw_area) in blueprint.iter() {
324            // Skip empty IDs
325            if cortical_id_str.is_empty() {
326                warn!(target: "feagi-evo","Skipping empty cortical_id");
327                continue;
328            }
329
330            // Convert string cortical_id to CorticalID (handles 6-char legacy and base64)
331            let cortical_id = match string_to_cortical_id(cortical_id_str) {
332                Ok(id) => id,
333                Err(e) => {
334                    warn!(target: "feagi-evo","Skipping invalid cortical_id '{}': {}", cortical_id_str, e);
335                    continue;
336                }
337            };
338
339            // Extract required fields
340            let name = raw_area
341                .cortical_name
342                .clone()
343                .unwrap_or_else(|| cortical_id_str.clone());
344
345            let dimensions = if let Some(boundaries) = &raw_area.block_boundaries {
346                if boundaries.len() != 3 {
347                    return Err(EvoError::InvalidArea(format!(
348                        "Invalid block_boundaries for {}: expected 3 values, got {}",
349                        cortical_id_str,
350                        boundaries.len()
351                    )));
352                }
353                Dimensions::new(boundaries[0], boundaries[1], boundaries[2])
354                    .map_err(|e| EvoError::InvalidArea(format!("Invalid dimensions: {}", e)))?
355            } else {
356                // Default to 1x1x1 if not specified (should not happen in valid genomes)
357                warn!(target: "feagi-evo","Cortical area {} missing block_boundaries, defaulting to 1x1x1", cortical_id_str);
358                Dimensions::new(1, 1, 1).map_err(|e| {
359                    EvoError::InvalidArea(format!("Invalid default dimensions: {}", e))
360                })?
361            };
362
363            let position = if let Some(coords) = &raw_area.relative_coordinate {
364                if coords.len() != 3 {
365                    return Err(EvoError::InvalidArea(format!(
366                        "Invalid relative_coordinate for {}: expected 3 values, got {}",
367                        cortical_id_str,
368                        coords.len()
369                    )));
370                }
371                GenomeCoordinate3D::new(coords[0], coords[1], coords[2])
372            } else {
373                // Default to origin if not specified
374                warn!(target: "feagi-evo","Cortical area {} missing relative_coordinate, defaulting to (0,0,0)", cortical_id_str);
375                GenomeCoordinate3D::new(0, 0, 0)
376            };
377
378            // Determine cortical type from cortical_id
379            let cortical_type = cortical_id.as_cortical_type().map_err(|e| {
380                EvoError::InvalidArea(format!(
381                    "Failed to determine cortical type from ID {}: {}",
382                    cortical_id_str, e
383                ))
384            })?;
385
386            // Create cortical area with CorticalID object (zero-copy, type-safe)
387            let mut area = CorticalArea::new(
388                cortical_id,
389                0, // cortical_idx will be assigned by ConnectomeManager
390                name,
391                dimensions,
392                position,
393                cortical_type,
394            )?;
395
396            // Store cortical_type as cortical_group for new type system
397            if let Some(ref cortical_type_str) = raw_area.cortical_type {
398                area.properties.insert(
399                    "cortical_group".to_string(),
400                    serde_json::json!(cortical_type_str),
401                );
402            }
403
404            // Store all properties in the properties HashMap
405            // Neural properties
406            if let Some(v) = raw_area.synapse_attractivity {
407                area.properties
408                    .insert("synapse_attractivity".to_string(), serde_json::json!(v));
409            }
410            if let Some(v) = raw_area.refractory_period {
411                area.properties
412                    .insert("refractory_period".to_string(), serde_json::json!(v));
413            }
414            if let Some(v) = raw_area.firing_threshold {
415                area.properties
416                    .insert("firing_threshold".to_string(), serde_json::json!(v));
417            }
418            if let Some(v) = raw_area.firing_threshold_limit {
419                area.properties
420                    .insert("firing_threshold_limit".to_string(), serde_json::json!(v));
421            }
422            if let Some(v) = raw_area.firing_threshold_increment_x {
423                area.properties.insert(
424                    "firing_threshold_increment_x".to_string(),
425                    serde_json::json!(v),
426                );
427            }
428            if let Some(v) = raw_area.firing_threshold_increment_y {
429                area.properties.insert(
430                    "firing_threshold_increment_y".to_string(),
431                    serde_json::json!(v),
432                );
433            }
434            if let Some(v) = raw_area.firing_threshold_increment_z {
435                area.properties.insert(
436                    "firing_threshold_increment_z".to_string(),
437                    serde_json::json!(v),
438                );
439            }
440            if let Some(v) = raw_area.leak_coefficient {
441                area.properties
442                    .insert("leak_coefficient".to_string(), serde_json::json!(v));
443            }
444            if let Some(v) = raw_area.leak_variability {
445                area.properties
446                    .insert("leak_variability".to_string(), serde_json::json!(v));
447            }
448            if let Some(v) = raw_area.neuron_excitability {
449                area.properties
450                    .insert("neuron_excitability".to_string(), serde_json::json!(v));
451            }
452            if let Some(v) = raw_area.postsynaptic_current {
453                area.properties
454                    .insert("postsynaptic_current".to_string(), serde_json::json!(v));
455            }
456            if let Some(v) = raw_area.postsynaptic_current_max {
457                area.properties
458                    .insert("postsynaptic_current_max".to_string(), serde_json::json!(v));
459            }
460            if let Some(v) = raw_area.degeneration {
461                area.properties
462                    .insert("degeneration".to_string(), serde_json::json!(v));
463            }
464
465            // Boolean properties
466            if let Some(v) = raw_area.psp_uniform_distribution {
467                area.properties
468                    .insert("psp_uniform_distribution".to_string(), serde_json::json!(v));
469            }
470            if let Some(v) = raw_area.mp_charge_accumulation {
471                area.properties
472                    .insert("mp_charge_accumulation".to_string(), serde_json::json!(v));
473            }
474            if let Some(v) = raw_area.mp_driven_psp {
475                area.properties
476                    .insert("mp_driven_psp".to_string(), serde_json::json!(v));
477                tracing::info!(
478                    target: "feagi-evo",
479                    "[GENOME-LOAD] Loaded mp_driven_psp={} for area {}",
480                    v,
481                    cortical_id_str
482                );
483            } else {
484                tracing::debug!(
485                    target: "feagi-evo",
486                    "[GENOME-LOAD] mp_driven_psp not found in raw_area for {}, will use default=false",
487                    cortical_id_str
488                );
489            }
490            if let Some(v) = raw_area.visualization {
491                area.properties
492                    .insert("visualization".to_string(), serde_json::json!(v));
493                // Also store as "visible" for compatibility with getters
494                area.properties
495                    .insert("visible".to_string(), serde_json::json!(v));
496            }
497            if let Some(v) = raw_area.burst_engine_activation {
498                area.properties
499                    .insert("burst_engine_active".to_string(), serde_json::json!(v));
500            }
501            if let Some(v) = raw_area.is_mem_type {
502                area.properties
503                    .insert("is_mem_type".to_string(), serde_json::json!(v));
504            }
505
506            // Memory properties
507            if let Some(v) = raw_area.longterm_mem_threshold {
508                area.properties
509                    .insert("longterm_mem_threshold".to_string(), serde_json::json!(v));
510            }
511            if let Some(v) = raw_area.lifespan_growth_rate {
512                area.properties
513                    .insert("lifespan_growth_rate".to_string(), serde_json::json!(v));
514            }
515            if let Some(v) = raw_area.init_lifespan {
516                area.properties
517                    .insert("init_lifespan".to_string(), serde_json::json!(v));
518            }
519            if let Some(v) = raw_area.temporal_depth {
520                area.properties
521                    .insert("temporal_depth".to_string(), serde_json::json!(v));
522            }
523            if let Some(v) = raw_area.consecutive_fire_cnt_max {
524                area.properties
525                    .insert("consecutive_fire_cnt_max".to_string(), serde_json::json!(v));
526                // Also store as "consecutive_fire_limit" for getter compatibility
527                area.properties
528                    .insert("consecutive_fire_limit".to_string(), serde_json::json!(v));
529            }
530            if let Some(v) = raw_area.snooze_length {
531                area.properties
532                    .insert("snooze_period".to_string(), serde_json::json!(v));
533            }
534
535            // Other properties
536            if let Some(v) = &raw_area.group_id {
537                area.properties
538                    .insert("group_id".to_string(), serde_json::json!(v));
539            }
540            if let Some(v) = &raw_area.sub_group_id {
541                area.properties
542                    .insert("sub_group_id".to_string(), serde_json::json!(v));
543            }
544            // Store neurons_per_voxel in properties HashMap
545            if let Some(v) = raw_area.per_voxel_neuron_cnt {
546                area.properties
547                    .insert("neurons_per_voxel".to_string(), serde_json::json!(v));
548            }
549            if let Some(v) = &raw_area.cortical_mapping_dst {
550                // Convert dstmap keys from old format to base64
551                let converted_dstmap = convert_dstmap_keys_to_base64(v);
552                area.properties
553                    .insert("cortical_mapping_dst".to_string(), converted_dstmap);
554            }
555            if let Some(v) = &raw_area.coordinate_2d {
556                area.properties
557                    .insert("2d_coordinate".to_string(), serde_json::json!(v));
558            }
559
560            // Store any other custom properties
561            for (key, value) in &raw_area.other {
562                area.properties.insert(key.clone(), value.clone());
563            }
564
565            // Note: cortical_type parsing disabled - CorticalArea is now a minimal data structure
566            // CorticalAreaType information is stored in properties["cortical_group"] if needed
567
568            areas.push(area);
569        }
570
571        Ok(areas)
572    }
573
574    /// Parse brain regions
575    fn parse_brain_regions(
576        raw_regions: &HashMap<String, RawBrainRegion>,
577    ) -> EvoResult<Vec<(BrainRegion, Option<String>)>> {
578        let mut regions = Vec::with_capacity(raw_regions.len());
579
580        for (region_id_str, raw_region) in raw_regions.iter() {
581            let title = raw_region
582                .title
583                .clone()
584                .unwrap_or_else(|| region_id_str.clone());
585
586            // Convert string region_id to RegionID (UUID)
587            // For now, try to parse as UUID if it's already a UUID, otherwise generate new one
588            let region_id = match RegionID::from_string(region_id_str) {
589                Ok(id) => id,
590                Err(_) => {
591                    // If not a valid UUID, generate a new one
592                    // This handles legacy string-based region IDs
593                    RegionID::new()
594                }
595            };
596
597            let region_type = RegionType::Undefined; // Default to Undefined
598
599            let mut region = BrainRegion::new(region_id, title, region_type)?;
600
601            // Add cortical areas to region (using CorticalID directly)
602            if let Some(areas) = &raw_region.areas {
603                for area_id in areas {
604                    // Convert area_id to CorticalID
605                    match string_to_cortical_id(area_id) {
606                        Ok(cortical_id) => {
607                            region.add_area(cortical_id);
608                        }
609                        Err(e) => {
610                            warn!(target: "feagi-evo",
611                                "Failed to convert brain region area ID '{}' to CorticalID: {}. Skipping.",
612                                area_id, e);
613                        }
614                    }
615                }
616            }
617
618            // Store properties in HashMap
619            if let Some(desc) = &raw_region.description {
620                region.add_property("description".to_string(), serde_json::json!(desc));
621            }
622            if let Some(coord_2d) = &raw_region.coordinate_2d {
623                region.add_property("coordinate_2d".to_string(), serde_json::json!(coord_2d));
624            }
625            if let Some(coord_3d) = &raw_region.coordinate_3d {
626                region.add_property("coordinate_3d".to_string(), serde_json::json!(coord_3d));
627            }
628            // Store inputs/outputs as base64 strings
629            if let Some(inputs) = &raw_region.inputs {
630                let input_ids: Vec<String> = inputs
631                    .iter()
632                    .filter_map(|id| match string_to_cortical_id(id) {
633                        Ok(cortical_id) => Some(cortical_id.as_base_64()),
634                        Err(e) => {
635                            warn!(target: "feagi-evo",
636                                    "Failed to convert brain region input ID '{}': {}. Skipping.",
637                                    id, e);
638                            None
639                        }
640                    })
641                    .collect();
642                if !input_ids.is_empty() {
643                    region.add_property("inputs".to_string(), serde_json::json!(input_ids));
644                }
645            }
646            if let Some(outputs) = &raw_region.outputs {
647                let output_ids: Vec<String> = outputs
648                    .iter()
649                    .filter_map(|id| match string_to_cortical_id(id) {
650                        Ok(cortical_id) => Some(cortical_id.as_base_64()),
651                        Err(e) => {
652                            warn!(target: "feagi-evo",
653                                    "Failed to convert brain region output ID '{}': {}. Skipping.",
654                                    id, e);
655                            None
656                        }
657                    })
658                    .collect();
659                if !output_ids.is_empty() {
660                    region.add_property("outputs".to_string(), serde_json::json!(output_ids));
661                }
662            }
663            if let Some(signature) = &raw_region.signature {
664                region.add_property("signature".to_string(), serde_json::json!(signature));
665            }
666
667            // Store parent_id for hierarchy construction
668            let parent_id = raw_region.parent_region_id.clone();
669            if let Some(ref parent_id_str) = parent_id {
670                // Store as property for serialization
671                region.add_property(
672                    "parent_region_id".to_string(),
673                    serde_json::json!(parent_id_str),
674                );
675            }
676
677            regions.push((region, parent_id));
678        }
679
680        Ok(regions)
681    }
682}
683
684#[cfg(test)]
685mod tests {
686    use super::*;
687
688    #[test]
689    fn test_parse_minimal_genome() {
690        // Test backward compatibility: parsing v2.1 genome with old 6-byte cortical ID
691        // Parser should convert old format to base64 for storage
692        let json = r#"{
693            "version": "2.1",
694            "blueprint": {
695                "_power": {
696                    "cortical_name": "Test Area",
697                    "block_boundaries": [10, 10, 10],
698                    "relative_coordinate": [0, 0, 0],
699                    "cortical_type": "CORE"
700                }
701            },
702            "brain_regions": {
703                "root": {
704                    "title": "Root",
705                    "parent_region_id": null,
706                    "areas": ["_power"]
707                }
708            }
709        }"#;
710
711        let parsed = GenomeParser::parse(json).unwrap();
712
713        assert_eq!(parsed.version, "2.1");
714        assert_eq!(parsed.cortical_areas.len(), 1);
715        // Input was "_power" (6 bytes), converted to "___power" (8 bytes, padded at start with underscores) then base64 encoded
716        assert_eq!(
717            parsed.cortical_areas[0].cortical_id.as_base_64(),
718            "X19fcG93ZXI="
719        );
720        assert_eq!(parsed.cortical_areas[0].name, "Test Area");
721        assert_eq!(parsed.brain_regions.len(), 1);
722
723        // Phase 2: Verify cortical_type_new is populated
724        // Note: cortical_type_new field removed - type is encoded in cortical_id
725        assert!(parsed.cortical_areas[0]
726            .cortical_id
727            .as_cortical_type()
728            .is_ok());
729    }
730
731    #[test]
732    fn test_parse_multiple_areas() {
733        // Test parsing multiple cortical areas with old format IDs
734        let json = r#"{
735            "version": "2.1",
736            "blueprint": {
737                "_power": {
738                    "cortical_name": "Area 1",
739                    "cortical_type": "CORE",
740                    "block_boundaries": [5, 5, 5],
741                    "relative_coordinate": [0, 0, 0]
742                },
743                "_death": {
744                    "cortical_name": "Area 2",
745                    "cortical_type": "CORE",
746                    "block_boundaries": [10, 10, 10],
747                    "relative_coordinate": [5, 0, 0]
748                }
749            }
750        }"#;
751
752        let parsed = GenomeParser::parse(json).unwrap();
753
754        assert_eq!(parsed.cortical_areas.len(), 2);
755
756        // Phase 2: Verify both areas have cortical_type_new populated
757        for area in &parsed.cortical_areas {
758            assert!(
759                area.cortical_id.as_cortical_type().is_ok(),
760                "Area {} should have cortical_type_new populated",
761                area.cortical_id
762            );
763        }
764    }
765
766    #[test]
767    fn test_string_to_cortical_id_legacy_power_shorthand() {
768        // Older FEAGI genomes may encode the power core area as "___pwr" (6-char shorthand).
769        // Migration must map this deterministically to the core Power cortical ID.
770        use feagi_structures::genomic::cortical_area::CoreCorticalType;
771        let id = string_to_cortical_id("___pwr").unwrap();
772        assert_eq!(
773            id.as_base_64(),
774            CoreCorticalType::Power.to_cortical_id().as_base_64()
775        );
776    }
777
778    #[test]
779    fn test_parse_with_properties() {
780        let json = r#"{
781            "version": "2.1",
782            "blueprint": {
783                "mem001": {
784                    "cortical_name": "Memory Area",
785                    "block_boundaries": [8, 8, 8],
786                    "relative_coordinate": [0, 0, 0],
787                    "cortical_type": "MEMORY",
788                    "is_mem_type": true,
789                    "firing_threshold": 50.0,
790                    "leak_coefficient": 0.9
791                }
792            }
793        }"#;
794
795        let parsed = GenomeParser::parse(json).unwrap();
796
797        assert_eq!(parsed.cortical_areas.len(), 1);
798        let area = &parsed.cortical_areas[0];
799
800        // Old type system (deprecated)
801        use feagi_structures::genomic::cortical_area::CorticalAreaType;
802        assert!(matches!(area.cortical_type, CorticalAreaType::Memory(_)));
803
804        // Properties stored correctly
805        assert!(area.properties.contains_key("is_mem_type"));
806        assert!(area.properties.contains_key("firing_threshold"));
807        assert!(area.properties.contains_key("cortical_group"));
808
809        // NEW: cortical_type should be derivable from cortical_id (Phase 2)
810        assert!(
811            area.cortical_id.as_cortical_type().is_ok(),
812            "cortical_id should be parseable to cortical_type"
813        );
814        if let Ok(cortical_type) = area.cortical_id.as_cortical_type() {
815            use feagi_structures::genomic::cortical_area::CorticalAreaType;
816            assert!(
817                matches!(cortical_type, CorticalAreaType::Memory(_)),
818                "Should be classified as MEMORY type"
819            );
820        }
821    }
822
823    #[test]
824    fn test_invalid_version() {
825        let json = r#"{
826            "version": "1.0",
827            "blueprint": {}
828        }"#;
829
830        let result = GenomeParser::parse(json);
831        assert!(result.is_err());
832    }
833
834    #[test]
835    fn test_malformed_json() {
836        let json = r#"{ "version": "2.1", "blueprint": { malformed"#;
837
838        let result = GenomeParser::parse(json);
839        assert!(result.is_err());
840    }
841
842    #[test]
843    fn test_cortical_type_new_population() {
844        // Test that cortical_type_new field is populated during parsing (Phase 2)
845        // This tests that parsing works with valid cortical IDs and populates types correctly
846        use feagi_structures::genomic::cortical_area::CoreCorticalType;
847        let power_id = CoreCorticalType::Power.to_cortical_id().as_base_64();
848        let json = format!(
849            r#"{{
850            "version": "2.1",
851            "blueprint": {{
852                "cvision1": {{
853                    "cortical_name": "Test Custom Vision",
854                    "cortical_type": "CUSTOM",
855                    "block_boundaries": [10, 10, 1],
856                    "relative_coordinate": [0, 0, 0]
857                }},
858                "cmotor01": {{
859                    "cortical_name": "Test Custom Motor",
860                    "cortical_type": "CUSTOM",
861                    "block_boundaries": [5, 5, 1],
862                    "relative_coordinate": [0, 0, 0]
863                }},
864                "{}": {{
865                    "cortical_name": "Test Core",
866                    "cortical_type": "CORE",
867                    "block_boundaries": [1, 1, 1],
868                    "relative_coordinate": [0, 0, 0]
869                }}
870            }}
871        }}"#,
872            power_id
873        );
874
875        let parsed = GenomeParser::parse(&json).unwrap();
876        assert_eq!(parsed.cortical_areas.len(), 3);
877
878        // Verify all areas have cortical_type_new populated
879        for area in &parsed.cortical_areas {
880            assert!(
881                area.cortical_id.as_cortical_type().is_ok(),
882                "Area {} should have cortical_type_new populated",
883                area.cortical_id
884            );
885
886            // Verify cortical_group property is also set
887            assert!(
888                area.properties.contains_key("cortical_group"),
889                "Area {} should have cortical_group property",
890                area.cortical_id
891            );
892
893            // Verify cortical group is consistent (avoid depending on feagi-brain-development)
894            if let Some(prop_group) = area
895                .properties
896                .get("cortical_group")
897                .and_then(|v| v.as_str())
898            {
899                assert!(
900                    !prop_group.is_empty(),
901                    "Area {} should have non-empty cortical_group property",
902                    area.cortical_id.as_base_64()
903                );
904            }
905        }
906    }
907}
feagi_evolutionary/genome/parser.rs

feagi_evolutionary/genome/
parser.rs
