#!/usr/bin/env python3
"""Browsable HTML for Phase 2 review.
GOLD = original scan (truth, red). VEDA = vedabase (audited, green).
- Confidence tiers (alta/media/baja), sorted most-probable-first per book.
- Checkbox = SELECT (the good ones to revert). Marked rows are kept in the export.
- Per-book "Marcar todo" / "Quitar" buttons for bulk selection.
- Export = the SELECTED changes → vedabase_reversiones.md. localStorage persistence.
"""
import json, glob, os, re, difflib, html
BASE = os.path.dirname(os.path.abspath(__file__))
EXCLUDE = {"tlc", "ejop"}
GARBAGE = re.compile(r'[<>=~|{}\\^_]|[A-Za-z][0-9]|[0-9][A-Za-z]')
FUNC = set("the a an of to and in is was are be that this it as for on with by at "
"from or not but his her him he she they we you i my our their its "
"which who had has have been will would can could so if then than there "
"into out up no all any one two also did do does made after over only "
"such like most more new now me even some them when".split())
try:
WORDS = set(w.strip().lower() for w in open("/usr/share/dict/words"))
except Exception:
WORDS = set()
def clean(s): return ' '.join(str(s).split())
def is_garbled(s):
if GARBAGE.search(s): return True
if not re.findall(r'[A-Za-z]{2,}', s) and re.search(r'[A-Za-z]', s): return True
return False
def char_ratio(a, b):
a2 = re.sub(r'[^a-z]', '', a.lower()); b2 = re.sub(r'[^a-z]', '', b.lower())
if not a2 and not b2: return 1.0
return difflib.SequenceMatcher(None, a2, b2).ratio()
def classify(o, v):
o, v = clean(o), clean(v)
if not re.search(r'[A-Za-z]', o) and not re.search(r'[A-Za-z]', v): return "trivial"
if is_garbled(o): return "noise"
if char_ratio(o, v) >= 0.82: return "trivial"
return "real"
def confidence(o, v):
ot = re.findall(r"[A-Za-z']+", o); vt = re.findall(r"[A-Za-z']+", v)
def suspicious(toks):
return any(t.lower() not in WORDS and not t[0].isupper() and len(t) > 1 for t in toks)
if WORDS and (suspicious(ot) or suspicious(vt)):
return ("media", 55)
co = {t.lower() for t in ot if t.lower() not in FUNC}
cv = {t.lower() for t in vt if t.lower() not in FUNC}
if co ^ cv:
return ("alta", 92)
return ("baja", 30)
def esc(s): return html.escape(clean(s))
def main():
import sys
reports_dir = sys.argv[1] if len(sys.argv) > 1 else "reports"
out_name = sys.argv[2] if len(sys.argv) > 2 else "PHASE2_review.html"
books = []
for f in sorted(glob.glob(os.path.join(BASE, reports_dir, "*_diffs.json"))):
key = os.path.basename(f).replace("_diffs.json", "")
if key in EXCLUDE: continue
d = json.load(open(f))
reals = []
for x in d.get("differences", []):
if classify(x.get("original",""), x.get("vedabase","")) != "real":
continue
tier, score = confidence(clean(x.get("original","")), clean(x.get("vedabase","")))
reals.append((score, tier, x))
reals.sort(key=lambda r: -r[0])
if reals:
books.append((len(reals), key, d["book"], reals))
books.sort(reverse=True)
total = sum(b[0] for b in books)
n_alta = sum(1 for _,_,_,rs in books for s,t,_ in rs if t == "alta")
rows = "".join(
f'
'
for n, k, name, rs in books)
sections = []
for n, k, name, reals in books:
items = []
for i, (score, tier, x) in enumerate(reals):
cid = f"{k}_{i}"
cb = esc(x.get("context_before","")); ca = esc(x.get("context_after",""))
o = esc(x.get("original","")); v = esc(x.get("vedabase",""))
items.append(
f'')
sections.append(
f'