Archive 4 checkpoints (1-4)

paper-releases/pbisc-diffusion/updates/20261014T113716Z-figure1-overview-training-corpus/diffusion/dataset/server_1/model8/presentations/2026-09-30--paper-summary--r00/tools/build_figure1_master_20261014.py

@@ -0,0 +1,104 @@
+"""Replace Figure1a and insert the exact training-corpus Figure1b in the saved master."""
+from pathlib import Path
+from collections import defaultdict
+from datetime import datetime,timezone
+import copy
+import csv
+import hashlib
+import json
+import os
+import re
+import sys
+if os.environ.get('CUDA_VISIBLE_DEVICES') != '':raise SystemExit('CPU-only')
+D=Path(__file__).resolve().parents[1];A=D/'data/figure1_revision_20261014'
+PARENT=D/'data/strengthening_20261013/spec.json'
+OUT='PBISC-Diffusion_expanded_master_20261014_figure1'
+sys.path[:0]=[str(D),str(D/'tools')]
+import generate_deck as G
+from reorder_master_figures_20261013 import draw_card,draw_note
+from pptx import Presentation
+from pptx.util import Inches
+
+
+def sha(p):
+    with Path(p).open('rb') as f:return hashlib.file_digest(f,'sha256').hexdigest()
+
+
+def main():
+    parent=json.loads(PARENT.read_text())['slides'];assert len(parent)==154
+    inputs=[PARENT,Path(__file__),D/'generate_deck.py',D/'tools/reorder_master_figures_20261013.py']
+    bundles={name:json.loads((A/name/'result.json').read_text()) for name in ['overview','train_corpus']}
+    inputs.extend(A/name/'result.json' for name in bundles)
+    source_bindings={str(p):{'sha256':sha(p)} for p in inputs}
+    # Family proofs bind the exact plotted values and the source hash roster.
+    for name,b in bundles.items():
+        for r in b['sources']:
+            assert sha(r['path'])==r['sha256'],r['path']
+            source_bindings[r['path']]={'sha256':r['sha256']}
+    train_check=json.loads((A/'train_corpus/VERIFICATION.json').read_text())
+    assert train_check['status']=='PASS'
+    slides=[]
+    for n,ss in enumerate(parent,1):
+        s=copy.deepcopy(ss);s['parent_154_page']=n
+        if n==7:
+            replacement=copy.deepcopy(bundles['overview']['slides'][0])
+            s['superseded_overview_figure']=s['figures']
+            s.update(replacement);s['revision_key']='figure1a:organic_schematic'
+            for key in ['text_source','say','refs','refs_full']:s.pop(key,None)
+        if n==8:
+            train=copy.deepcopy(bundles['train_corpus']['slides'][0])
+            train.update(kind='content',deck_group='Figure 1',section='Figure 1',section_label='Fig. 1b',layout='full',
+                         addition_key='figure1b:training_corpus',keywords=[])
+            # Keep the long metadata roster in the source manifest, not speaker notes.
+            train['sources']='data/figure1_revision_20261014/train_corpus/result.json; data/figure1_revision_20261014/train_corpus/source_hashes.json; tools/figure1_train_corpus_20261014.py'
+            train.pop('source_refs',None);train.pop('source_bindings',None)
+            slides.append(train)
+            s['section_label']='Figure 1 support'
+            s['title']='Full source corpus: training, validation and test'
+        slides.append(s)
+    assert len(slides)==155
+    mapping={s['parent_154_page']:i for i,s in enumerate(slides,1) if 'parent_154_page' in s}
+    assert sorted(mapping)==list(range(1,155))
+    assert sorted(s['original_slide'] for s in slides if 'original_slide' in s)==list(range(1,77))
+    for s in slides:
+        if s.get('kind')=='plan_card':
+            s['rows']=[(h,b,re.sub(r'\d+',lambda m:str(mapping[int(m.group())]),loc) if re.match(r'^pp?\.',loc) else loc)
+                       for h,b,loc in s['rows']]
+    guide=next(s for s in slides if s.get('parent_154_page')==6)
+    guide['rows']=[('1a','Pretraining > paired adaptation > new-bulk generation','p. 7'),
+                   ('1b','Actual training corpus: condition and cell-type composition','p. 8'),
+                   ('Support','Full 41.4M source corpus; train/validation/test allocation','pp. 9, 10'),
+                   ('1c','K recipes and repeated training exposure','p. 11'),
+                   ('1d','Recorded noise-to-cell PCA stages','p. 13')]
+    guide['footnote']='Main corpus panel counts 33,141,380 unique training cells. The full 41,395,184-cell source pool remains support.'
+    guide['footnote_full']=guide['footnote']
+    slides[0]['subtitle']='155 slides | Organic experimental overview | Exact 33.14M training-corpus composition'
+    groups=defaultdict(list)
+    for i,s in enumerate(slides,1):groups[s['deck_group']].append(i)
+    prs=Presentation();prs.slide_width=Inches(G.SLIDE_W);prs.slide_height=Inches(G.SLIDE_H)
+    functions={'title':G.build_title,'agenda':G.build_agenda,'divider':G.build_divider,'content':G.build_content,'plan_card':draw_card,'evidence_note':draw_note}
+    for n,s in enumerate(slides,1):
+        fn=functions[s['kind']];fn(prs,s) if s['kind']=='title' else fn(prs,s,n)
+    prs.save(D/(OUT+'.pptx'))
+    spec={'deck':{'output_pptx':OUT+'.pptx','output_pdf':OUT+'.pdf'},'slides':slides}
+    (A/'spec.json').write_text(json.dumps(spec,indent=2)+'\n')
+    rows=[dict(page=n,parent_154_page=s.get('parent_154_page',''),original_76_page=s.get('original_slide',''),
+               group=s['deck_group'],panel=s.get('section_label',''),title=s['title']) for n,s in enumerate(slides,1)]
+    with (A/'page_map.csv').open('w',newline='',encoding='utf-8-sig') as f:
+        writer=csv.DictWriter(f,fieldnames=list(rows[0]));writer.writeheader();writer.writerows(rows)
+    oldm=json.loads((PARENT.parent/'manifest.json').read_text())
+    for s in parent:
+        for f in s.get('figures',[]):
+            p=str(D/f);assert sha(p)==oldm['sources'][p]['sha256']
+    for s in slides:
+        for f in s.get('figures',[]):source_bindings[str(D/f)]={'sha256':sha(D/f)}
+    protected={str(p):sha(p) for p in [PARENT,D/'PBISC-Diffusion_expanded_master_20261013_strengthened.pptx',D/'PBISC-Diffusion_expanded_master_20261013_strengthened.pdf']}
+    manifest={'schema':'pbisc.figure1-master.20261014.v1','created_at':datetime.now(timezone.utc).isoformat(),
+              'parent_spec':str(PARENT),'total_slides':155,'parent_pages':154,'groups':dict(groups),
+              'modified_parent_pages':[7],'newpages':[8],'sources':source_bindings,'protected_sources':protected,
+              'new_training':False,'new_inference':False,'output':{'pptx':{'path':str(D/(OUT+'.pptx')),'sha256':sha(D/(OUT+'.pptx'))}}}
+    (A/'manifest.json').write_text(json.dumps(manifest,indent=2)+'\n')
+    print(json.dumps({'slides':155,'overview':7,'training_corpus':8,'full_source_support':9,'source_bindings':len(source_bindings)}))
+
+
+if __name__=='__main__':main()