307 lines
11 KiB
Python
307 lines
11 KiB
Python
#!/usr/bin/env python3
|
|
"""Final comprehensive analysis — clean version with bug fixes."""
|
|
|
|
import json, re
|
|
from pathlib import Path
|
|
from collections import defaultdict
|
|
|
|
BASE = Path("/home/morr/work/uberwald/fvtt-chroniques-de-l-etrange")
|
|
PACKS = BASE / "packs-src"
|
|
pdf_text = (BASE / "regles.txt").read_text(encoding="utf-8")
|
|
|
|
WATERMARK_RE = re.compile(
|
|
r's\s*c\s*r\s*a\s*l\s*e\s*l|les\s+chroniqu|de\s+l.etrange|chr.niqu|hr\s+ng',
|
|
re.IGNORECASE)
|
|
|
|
def strip_html(html):
|
|
return re.sub(r'<[^>]+>', '', html or '').strip()
|
|
|
|
def has_watermark_bleed(text):
|
|
plain = strip_html(text)
|
|
return bool(WATERMARK_RE.search(plain))
|
|
|
|
def has_bad_newlines(text):
|
|
lines = text.split('\n')
|
|
if len(lines) <= 1:
|
|
return False
|
|
for line in lines:
|
|
s = line.strip()
|
|
if s and not re.match(r'^<[/a-zA-Z]', s) and not s.endswith('>') and len(s) > 3:
|
|
return True
|
|
return False
|
|
|
|
def looks_truncated(text):
|
|
"""Text appears cut off at the end."""
|
|
if not text:
|
|
return False
|
|
plain = strip_html(text).strip()
|
|
if not plain:
|
|
return False
|
|
# Remove watermark garbage from end before checking
|
|
plain_clean = re.sub(r'\s+[a-z]\s+[a-z]\s+[a-z]{1,2}\s+.*$', '', plain).strip()
|
|
if not plain_clean:
|
|
return False # FIX: was using original empty plain, now correctly returns False
|
|
last = plain_clean[-1]
|
|
return last.isalpha() or last in ',;:-('
|
|
|
|
def looks_missing_beginning(text):
|
|
"""Text starts mid-sentence (truly lowercase first char only)."""
|
|
if not text:
|
|
return False
|
|
plain = strip_html(text).strip()
|
|
if not plain:
|
|
return False
|
|
# Only flag if truly starts lowercase (not French articles/prepositions)
|
|
first_char = plain[0]
|
|
return first_char.islower()
|
|
|
|
def is_garbled_page_layout(text):
|
|
"""Multiple single-letter <p> tags = garbled PDF artifact."""
|
|
single_p = re.findall(r'<p>([a-zA-Z0-9])</p>', text)
|
|
return len(single_p) >= 5
|
|
|
|
def get_field(data, path):
|
|
parts = path.split('.')
|
|
cur = data
|
|
for p in parts:
|
|
if isinstance(cur, dict):
|
|
cur = cur.get(p)
|
|
else:
|
|
return None
|
|
if cur is None:
|
|
return None
|
|
return cur
|
|
|
|
def pdf_search_after(keyword_text, context=500):
|
|
"""Return PDF text after the given keyword."""
|
|
plain = strip_html(keyword_text)
|
|
plain = re.sub(r'\s+[a-z]\s+[a-z]\s+[a-z]{1,2}\s+.*$', '', plain).strip()
|
|
if len(plain) < 15:
|
|
return None
|
|
for suffix_len in [40, 30, 20, 15]:
|
|
suffix = re.sub(r'\s+', ' ', plain[-suffix_len:]).strip()
|
|
if len(suffix) < 10:
|
|
continue
|
|
idx = pdf_text.find(suffix)
|
|
if idx != -1:
|
|
snippet = pdf_text[idx:min(len(pdf_text), idx + len(suffix) + context)]
|
|
snippet = re.sub(r'\n+', ' ', snippet)
|
|
snippet = re.sub(r'\s{3,}', ' ', snippet)
|
|
return snippet[:600]
|
|
return None
|
|
|
|
def pdf_search_before(keyword_text, context=400):
|
|
"""Return PDF text before the given keyword."""
|
|
plain = strip_html(keyword_text)
|
|
plain_start = re.sub(r'\s+', ' ', plain[:60]).strip()
|
|
if len(plain_start) < 15:
|
|
return None
|
|
for prefix_len in [50, 40, 30, 20]:
|
|
prefix = re.sub(r'\s+', ' ', plain_start[:prefix_len]).strip()
|
|
if len(prefix) < 10:
|
|
continue
|
|
idx = pdf_text.find(prefix)
|
|
if idx != -1:
|
|
start = max(0, idx - context)
|
|
snippet = pdf_text[start:idx + len(prefix)]
|
|
snippet = re.sub(r'\n+', ' ', snippet)
|
|
snippet = re.sub(r'\s{3,}', ' ', snippet)
|
|
return snippet[-400:]
|
|
return None
|
|
|
|
issues = []
|
|
all_files = sorted(PACKS.rglob("*.json"))
|
|
print(f"Scanning {len(all_files)} files...", flush=True)
|
|
|
|
HTML_FIELDS = [
|
|
"system.description",
|
|
"system.effects",
|
|
"system.examples",
|
|
"system.components",
|
|
"system.notes",
|
|
"system.techniques.technique1.technique",
|
|
"system.techniques.technique2.technique",
|
|
"system.techniques.technique3.technique",
|
|
]
|
|
PLAIN_FIELDS = ["system.style"]
|
|
|
|
for jf in sorted(all_files):
|
|
rel = str(jf.relative_to(PACKS))
|
|
try:
|
|
data = json.loads(jf.read_text(encoding="utf-8"))
|
|
except Exception as e:
|
|
issues.append({"file": rel, "field": "(file)", "issue": "json_error",
|
|
"item_name": "?", "current_text": str(e)})
|
|
continue
|
|
|
|
name = data.get("name", "?")
|
|
|
|
def add(field, issue_type, **kwargs):
|
|
issues.append({"file": rel, "field": field, "issue": issue_type,
|
|
"item_name": name, **kwargs})
|
|
|
|
for field in HTML_FIELDS + PLAIN_FIELDS:
|
|
val = get_field(data, field)
|
|
if not val or not isinstance(val, str) or not val.strip():
|
|
continue
|
|
plain = strip_html(val).strip()
|
|
|
|
# Garbled page layout (skip other checks)
|
|
if is_garbled_page_layout(val):
|
|
add(field, "garbled_page_layout",
|
|
current_text=val[:400],
|
|
note="Text broken into single-character <p> tags — PDF layout artifact")
|
|
continue
|
|
|
|
# Watermark bleeding
|
|
if has_watermark_bleed(val):
|
|
add(field, "bleeding_watermark",
|
|
current_text=val[:400],
|
|
plain_text=plain[:300],
|
|
pdf_context=pdf_search_after(val))
|
|
|
|
# Missing beginning (only truly lowercase-starting)
|
|
if looks_missing_beginning(val):
|
|
add(field, "missing_beginning",
|
|
current_start=plain[:150],
|
|
pdf_context_before=pdf_search_before(val))
|
|
|
|
# Truncation
|
|
if looks_truncated(val):
|
|
# Skip empty ingredient placeholders
|
|
is_ingredient = 'cde-ingredients' in rel
|
|
if is_ingredient:
|
|
# Only flag if there's actually short meaningful text
|
|
if plain and len(plain) < 30:
|
|
add(field, "empty_or_short_ingredient",
|
|
current_text=plain,
|
|
note="Short ingredient description — check if intentional")
|
|
else:
|
|
add(field, "truncated",
|
|
current_end=plain[-120:],
|
|
current_preview=plain[:200],
|
|
pdf_context=pdf_search_after(val))
|
|
|
|
# Unwanted newlines
|
|
if has_bad_newlines(val):
|
|
add(field, "unwanted_newlines",
|
|
current_text=val[:400],
|
|
plain_text=plain[:200])
|
|
|
|
# Technique cross-checks
|
|
for tkey in ['technique1', 'technique2', 'technique3']:
|
|
tech = get_field(data, f"system.techniques.{tkey}")
|
|
if not tech:
|
|
continue
|
|
t_text = tech.get("technique", "")
|
|
if not t_text:
|
|
continue
|
|
plain_t = strip_html(t_text)
|
|
activation_count = plain_t.count("Activation :")
|
|
if activation_count > 1:
|
|
add(f"system.techniques.{tkey}.technique",
|
|
"bleeding_multiple_techniques",
|
|
activation_count=activation_count,
|
|
current_text=t_text[:500],
|
|
note=f"{activation_count} 'Activation :' markers — multiple techniques merged")
|
|
if ("Style" in plain_t or "Orientation :" in plain_t) and len(plain_t) > 300:
|
|
already = any(i['file'] == rel and i['field'] == f"system.techniques.{tkey}.technique"
|
|
and i['issue'] == 'bleeding_style_or_orientation' for i in issues)
|
|
if not already:
|
|
add(f"system.techniques.{tkey}.technique",
|
|
"bleeding_style_or_orientation",
|
|
current_text=t_text[:500],
|
|
note="Contains 'Style' or 'Orientation' — extra text from PDF page layout")
|
|
|
|
print(f"Found {len(issues)} issues.", flush=True)
|
|
|
|
out_json = BASE / "compendium-issues.json"
|
|
out_txt = BASE / "compendium-issues.txt"
|
|
|
|
with open(out_json, 'w', encoding='utf-8') as f:
|
|
json.dump(issues, f, ensure_ascii=False, indent=2)
|
|
|
|
by_type = defaultdict(list)
|
|
by_file = defaultdict(list)
|
|
for iss in issues:
|
|
by_type[iss['issue']].append(iss)
|
|
by_file[iss['file']].append(iss)
|
|
|
|
PRIORITY_ORDER = [
|
|
'garbled_page_layout',
|
|
'missing_beginning',
|
|
'bleeding_watermark',
|
|
'bleeding_multiple_techniques',
|
|
'bleeding_style_or_orientation',
|
|
'truncated',
|
|
'unwanted_newlines',
|
|
'empty_or_short_ingredient',
|
|
]
|
|
|
|
with open(out_txt, 'w', encoding='utf-8') as f:
|
|
w = f.write
|
|
w("=" * 80 + "\n")
|
|
w("COMPENDIUM TEXT QUALITY REPORT\n")
|
|
w("Les Chroniques de l'Etrange — FoundryVTT\n")
|
|
w("=" * 80 + "\n\n")
|
|
w(f"Files scanned: {len(all_files)}\n")
|
|
w(f"Files with issues: {len(by_file)}\n")
|
|
w(f"Total issues: {len(issues)}\n\n")
|
|
|
|
w("SUMMARY BY ISSUE TYPE:\n")
|
|
for itype in PRIORITY_ORDER:
|
|
ilist = by_type.get(itype, [])
|
|
if ilist:
|
|
desc = {
|
|
'garbled_page_layout': 'text broken into single-char HTML tags (PDF artifact)',
|
|
'missing_beginning': 'field starts mid-word (lowercase start = truncated at front)',
|
|
'bleeding_watermark': '"Les Chroniques de l\'Étrange" watermark fragments in text',
|
|
'bleeding_multiple_techniques': 'multiple techniques merged into one field',
|
|
'bleeding_style_or_orientation': 'Style/Orientation text bled into technique field',
|
|
'truncated': 'field ends mid-sentence without proper punctuation',
|
|
'unwanted_newlines': 'raw newlines inside HTML string values',
|
|
'empty_or_short_ingredient': 'ingredient has empty or very short description',
|
|
}.get(itype, '')
|
|
w(f" {itype:45s} {len(ilist):3d} — {desc}\n")
|
|
|
|
w("\nFILES WITH ISSUES:\n")
|
|
for fpath in sorted(by_file.keys()):
|
|
types = sorted(set(i['issue'] for i in by_file[fpath]))
|
|
w(f" {fpath}\n [{', '.join(types)}]\n")
|
|
|
|
w("\n")
|
|
w("=" * 80 + "\n")
|
|
w("DETAILED ISSUES (by priority)\n")
|
|
w("=" * 80 + "\n")
|
|
|
|
for itype in PRIORITY_ORDER:
|
|
ilist = by_type.get(itype, [])
|
|
if not ilist:
|
|
continue
|
|
w(f"\n{'─'*80}\n")
|
|
w(f"ISSUE TYPE: {itype} ({len(ilist)} occurrences)\n")
|
|
w(f"{'─'*80}\n")
|
|
for iss in ilist:
|
|
w(f"\n File: {iss['file']}\n")
|
|
w(f" Item: {iss.get('item_name','?')}\n")
|
|
w(f" Field: {iss['field']}\n")
|
|
if iss.get('note'):
|
|
w(f" Note: {iss['note']}\n")
|
|
if iss.get('current_start'):
|
|
w(f" Starts: {iss['current_start'][:160]}\n")
|
|
if iss.get('current_end'):
|
|
w(f" Ends: ...{iss['current_end']}\n")
|
|
if iss.get('current_preview'):
|
|
w(f" Text: {iss['current_preview'][:200]}\n")
|
|
if iss.get('current_text'):
|
|
ct = iss['current_text']
|
|
w(f" Text: {ct[:300]}\n")
|
|
if iss.get('plain_text'):
|
|
w(f" Plain: {iss['plain_text'][:200]}\n")
|
|
if iss.get('pdf_context'):
|
|
w(f" PDF>>: {iss['pdf_context'][:400]}\n")
|
|
if iss.get('pdf_context_before'):
|
|
w(f" <<PDF: {iss['pdf_context_before'][:400]}\n")
|
|
|
|
print(f"Written: {out_json}\n {out_txt}", flush=True)
|