Dmitry Kokorin
99e889d206
libgen re-sweep of EN gaps in sections 01-06 (user request; improved sweep: file-level objects[]=f + full 100-row dumps) — 75 candidates, 7 rescues verified: - sec03 #15 Witches Ogres (Shambhala 1992, 236pp, cover OK) - sec03 #19 Hermes and His Children (Daimon 2020 2nd ed., epub ISBN in file) - sec03 #31 European Folktale (Indiana 1986, 204pp, cover OK) - sec03 #32 Fairytale as Art Form (Indiana 1987, 232pp, title OK) - sec03 #33 Once Upon a Time (Ungar 1976 1st ed., 204pp, title OK; RAW 'Indiana 1970' = ISAP typo) - sec03 #36 Irresistible Fairy Tale (Princeton 2012, epub OK) - sec03 #47 Heart of the Hunter (Vintage 2010, mobi) + sec02 #16 A Little Course in Dreams (Shambhala 1986, 137pp, verified) Canonical location fix: Eliade 'Myths, Dreams and Mysteries' files moved sec05 -> sec03 (first appearance); sec03 #25 card ❌ -> ✅ (RU was found in sec05 rescue 2026-09-21, card was stale). Still downloading (queue tools/lgdl-en0106-queue.sh): sec03 #50 (Atlas x4), #55, #57, sec04 #04 (Art of C.G. Jung 2018, 298MB), #17, sec01 #42 (Freud Unconscious Penguin). Rejected: 04#23 Encyclopedia (JSTOR article 467KB), 05#16 Singer 2023 (different book).
138 lines
6.7 KiB
Python
138 lines
6.7 KiB
Python
#!/usr/bin/env python3
|
|
"""tools/en_sweep_0106.py — re-sweep EN gaps of sections 01-06 (2026-09-28, user request).
|
|
|
|
Round 1: soft libgen queries (author + 1-2 title words), whole-word semantics.
|
|
Saves data/en-sweep-0106.tsv: sec,item,query,n,top3 (first lines).
|
|
Polite: 4s between queries. Resume-safe (skips items already in TSV).
|
|
"""
|
|
import os, re, sys, time
|
|
sys.path.insert(0, os.path.dirname(os.path.abspath(__file__)))
|
|
from lg import fetch, UA
|
|
|
|
OUT = os.path.join(os.path.dirname(os.path.dirname(os.path.abspath(__file__))), "data", "en-sweep-0106.tsv")
|
|
|
|
CANDIDATES = [
|
|
# (sec, item, query, note)
|
|
("01","13","Franz inferior function","Shambhala 1993"),
|
|
("01","16","Henderson shadow self","Chiron 1990"),
|
|
("01","38","Douglas woman mirror","Sigo 1989"),
|
|
("01","42","Freud unconscious","SE 14"),
|
|
("01","48","Meier consciousness","Sigo 1989"),
|
|
("01","50","Molton eternal women","Fisher King 2011"),
|
|
("01","52","Stevens on jung","Princeton 1990"),
|
|
("01","54","Ulanov transforming sexuality","Shambhala 1994"),
|
|
("02","16","little course dreams","Shambhala 1988"),
|
|
("02","20","way of dream","Windrose 1988"),
|
|
("02","23","clinical uses dreams","Hall 1977"),
|
|
("02","42","dream tending","2009"),
|
|
("03","06","Dieckmann twice told tales","Chiron"),
|
|
("03","12","archetypal symbols profane","Franz vol1 2021"),
|
|
("03","13","archetypal symbols hero","Franz vol2 2021"),
|
|
("03","14","archetypal symbols maiden","Franz vol3 2021"),
|
|
("03","15","witches ogres","Jacoby/Kast/Riedel 1992"),
|
|
("03","17","kast emotions maturity","Fromm 1993"),
|
|
("03","19","Lopez-Pedraza hermes","Daimon 1989"),
|
|
("03","21","psyche stories","Stein/Corbett Chiron"),
|
|
("03","31","Luthi european folktale","Indiana 1986"),
|
|
("03","32","Luthi art form","Indiana"),
|
|
("03","33","Luthi once upon","Indiana"),
|
|
("03","36","Zipes irresistible","Princeton 2012"),
|
|
("03","47","Post heart hunter","Van der Post 1961"),
|
|
("03","49","animal life myth dreams","Caspari/Robbins Chiron 2003"),
|
|
("03","50","historical atlas mythology","Caspari/Robbins"),
|
|
("03","53","folklore mythology legend","Leach/Fried"),
|
|
("03","54","herder dictionary symbols","Chiron 1993"),
|
|
("03","55","teutonic myth legend","MacKenzie 2008"),
|
|
("03","57","primal myths","Sproul 1979"),
|
|
("03","58","ariadne clue","Stevens 1998"),
|
|
("03","59","motif index folklore","Thompson"),
|
|
("03","61","american indian mythology","Burland 1996"),
|
|
("03","63","animal within","Federer 2002"),
|
|
("03","64","hill fairy tale drama","Jung and Moreno 2014"),
|
|
("03","65","gilgamesh significance","Kluger 1991"),
|
|
("03","67","celtic heritage","Rees 1998"),
|
|
("04","04","art of jung","Norton 2018"),
|
|
("04","06","acton paintings","Routledge 2008"),
|
|
("04","07","acton modern art","Routledge 2004"),
|
|
("04","08","patterns creative imagination","Brutsche Chiron 2018"),
|
|
("04","09","creative fingerprint","Carter/Staples 2017"),
|
|
("04","11","dynamics of symbols","Schaverien 1992"),
|
|
("04","12","sandplay vulnerable","Pattis Zoja 2011"),
|
|
("04","13","soul meets matter","Pattis Zoja 2018"),
|
|
("04","14","painting therapy","Riedel/Henzler Daimon 2022"),
|
|
("04","15","psyche and the arts","Rowland 2008"),
|
|
("04","16","revealing image","Schaverien 1999"),
|
|
("04","17","art psychotherapy psychosis","Routledge 1997"),
|
|
("04","18","healing transformation sandplay","Ammann 1991"),
|
|
("04","19","spontaneous paintings","Geigy 1969"),
|
|
("04","20","life paints span","Bach Daimon 1990"),
|
|
("04","21","personality tree drawings","Bolander 1977"),
|
|
("04","23","encyclopedia archetypal symbolism","Shambhala (v1 Moon 1997 / v2 Elder 1996)"),
|
|
("04","26","secret world drawings","Peiry Sigo 2002"),
|
|
("04","27","images meanings connection","Goldstein Daimon 1999"),
|
|
("04","29","horizons sandplay","Kalff T&F 2021"),
|
|
("04","31","tree test","Huber 1952"),
|
|
("04","33","imagination in action","McNiff Shambhala 2015"),
|
|
("04","34","depth psychology of art","C.C. Thomas 1999"),
|
|
("04","35","integrating arts therapy","C.C. Thomas 2004"),
|
|
("05","04","buhrmann","Chiron 1993"),
|
|
("05","05","burleson","1981"),
|
|
("05","14","thresholds of initiation","Henderson (ia RESTRICTED)"),
|
|
("05","15","kirsch","SUNY 1997"),
|
|
("05","16","cultural complex","Singer"),
|
|
("05","23","lewis-williams dowson","1988"),
|
|
("06","04","idea of the numinous","Chiron 2006"),
|
|
("06","07","passion of perpetua","von Franz"),
|
|
("06","08","niklaus von flue","2022 EN"),
|
|
("06","10","jung white letters",""),
|
|
("06","15","jung treatment christianity","Stein (ia CDL)"),
|
|
("06","21","satan old testament","(ia CDL)"),
|
|
("06","22","religion greeks romans","(ia CDL)"),
|
|
]
|
|
|
|
def parse_rows(html):
|
|
rows = re.findall(r"<tr[^>]*>\s*<td[^>]*>.*?</tr>", html, re.S)
|
|
out = []
|
|
for r in rows:
|
|
tds = re.findall(r"<td[^>]*>(.*?)</td>", r, re.S)
|
|
tds = [re.sub(r"<[^>]+>", "", t).replace("&", "&").strip() for t in tds]
|
|
tds = [re.sub(r"\s+", " ", t) for t in tds]
|
|
if len(tds) >= 6:
|
|
md5s = re.findall(r"ads\.php\?md5=([0-9a-f]{32})", r)
|
|
ids = re.findall(r"edition\.php\?id=(\d+)", r)
|
|
out.append((len(tds), tds, md5s[0] if md5s else "", ids[0] if ids else ""))
|
|
return out
|
|
|
|
def main():
|
|
done = set()
|
|
if os.path.exists(OUT):
|
|
for ln in open(OUT):
|
|
p = ln.split("\t")
|
|
if len(p) > 1:
|
|
done.add((p[0], p[1]))
|
|
new = [c for c in CANDIDATES if (c[0], c[1]) not in done]
|
|
print(f"{len(new)} candidates to sweep ({len(done)} already done)", flush=True)
|
|
fh = open(OUT, "a", encoding="utf-8")
|
|
for i, (sec, item, q, note) in enumerate(new, 1):
|
|
url = ("https://libgen.vg/index.php?req=" + urllib.parse.quote(q) +
|
|
"&res=100&columns%5B%5D=t&columns%5B%5D=a&columns%5B%5D=s&columns%5B%5D=y&columns%5B%5D=p&columns%5B%5D=i"
|
|
"&objects%5B%5D=f&objects%5B%5D=e&objects%5B%5D=s&objects%5B%5D=a&objects%5B%5D=p&objects%5B%5D=w"
|
|
"&topics%5B%5D=l&topics%5B%5D=c&topics%5B%5D=f&topics%5B%5D=a&topics%5B%5D=m&topics%5B%5D=r&topics%5B%5D=s")
|
|
try:
|
|
html = fetch(url, tries=2)
|
|
rows = parse_rows(html)
|
|
top = " || ".join(" / ".join(r[1][:6]) + (f" [md5={r[2]}]" if r[2] else "") for r in rows[:4])
|
|
status = f"HITS={len(rows)}"
|
|
except Exception as e:
|
|
status, top = f"ERR {e}", ""
|
|
fh.write(f"{sec}\t{item}\t{q}\t{status}\t{top[:600]}\n{note}\n")
|
|
fh.flush()
|
|
print(f"[{i}/{len(new)}] {sec}#{item} '{q}' -> {status}", flush=True)
|
|
time.sleep(4)
|
|
fh.close()
|
|
print("SWEEP DONE", flush=True)
|
|
|
|
import urllib.parse
|
|
|
|
if __name__ == "__main__":
|
|
main()
|