def main():
args = utils.get_args('generates .html diffs with deep clues for all puzzles in similar.tsv')
outf = utils.open_output()
similars = utils.parse_tsv('gxd/similar.tsv', 'Similar')
xdids_todo = [ parse_pathname(fn).base for fn in args.inputs ]
if not xdids_todo:
xdids_todo = [ xdid for xdid, matches in metadb.get_similar_grids().items() if matches ]
for mainxdid in xdids_todo:
progress(mainxdid)
mainxd = xdfile.get_xd(mainxdid)
if not mainxd:
continue
matches = metadb.get_similar_grids().get(mainxdid, [])
xddates = {}
xddates[mainxdid] = mainxd.date() # Dict to store XD dates for further sort
html_grids = {}
html_clues = {}
# these are added directly to similar.tsv
nstaleclues = 0
nstaleanswers = 0
ntotalclues = 0
poss_answers = [] # TODO:
pub_uses = {} # [pubid] -> set(ClueAnswer)
dcl_html = ''
deepcl_html = [] # keep deep clues to parse later - per row
for pos, mainclue, mainanswer in mainxd.iterclues():
deepcl_html = [] # Temporary to be replaced late
mainca = ClueAnswer(mainxdid, mainxd.date(), mainanswer, mainclue)
# 'grid position' column
deepcl_html.append('<td class="pos">%s.</td>' % pos)
# find other uses of this clue, and other answers, in a single pass
for clueans in find_clue_variants(mainclue):
if clueans.answer != mainanswer:
poss_answers.append(clueans)
if clueans.answer == mainanswer:
if clueans.pubid in pub_uses:
otherpubs = pub_uses[clueans.pubid]
else:
otherpubs = set() # set of ClueAnswer
pub_uses[clueans.pubid] = otherpubs
otherpubs.add(clueans)
# add 'other uses' to clues_html
stale = False
deepcl_html.append('<td class="other-uses">')
if len(pub_uses) > 0:
sortable_uses = []
for pubid, uses in pub_uses.items():
# show the earliest unboiled clue
for u in sorted(uses, key=lambda x: x.date or ""):
# only show those published earlier
if u.date and u.date <= mainxd.date():
if pubid == mainxdid and u.date == mainxd.date():
pass
else:
stale = True
sortable_uses.append((u.date, u, 1))
deepcl_html.append(html_select([ (clue, nuses) for dt, clue, nuses in sorted(sortable_uses, key=lambda x: x[0], reverse=True) ], top_option=mainclue))
else:
deepcl_html.append('<div class="original">%s</div>' % esc(mainclue))
deepcl_html.append('</td>')
# add 'other answers' to clues_html
deepcl_html.append('<td class="other-answers">')
deepcl_html.append(html_select_options(poss_answers, strmaker=lambda ca: ca.answer, force_top=mainca))
deepcl_html.append('</td>')
# add 'other clues' to clues_html
deepcl_html.append('<td class="other-clues">')
# bclues is all boiled clues for this particular answer: { [bc] -> #uses }
bclues = load_answers().get(mainanswer, [])
stale_answer = False
if bclues:
uses = []
for bc, nuses in bclues.items():
# then find all clues besides this one
clue_usages = [ ca for ca in load_clues().get(bc, []) if ca.answer == mainanswer and ca.date < mainxd.date() ]
if clue_usages:
stale_answer = True
if nuses > 1:
# only use one (the most recent) ClueAnswer per boiled clue
# but use the clue only (no xdid)
ca = sorted(clue_usages, key=lambda ca: ca.date or "z")[-1].clue
else:
ca = sorted(clue_usages, key=lambda ca: ca.date or "z")[-1]
uses.append((ca, nuses))
if uses:
deepcl_html.append(html_select(uses))
deepcl_html.append('</td>') # end 'other-clues'
if stale_answer:
nstaleanswers += 1
if stale:
nstaleclues += 1
ntotalclues += 1
# Quick and dirty - to be replaced
dcl_html += '<tr>' + ' '.join(deepcl_html) + '</tr>'
# Store in list to make further formatting as html table easier
mainxd = xdfile.get_xd(mainxdid)
if mainxd:
html_grids[mainxdid] = grid_diff_html(mainxd)
# Add for main XD
diff_l = []
for pos, mainclue, mainanswer in mainxd.iterclues():
diff_h = mktag('div','fullgrid main') + '%s. ' %pos
diff_h += mainclue
diff_h += mktag('span', tagclass='main', inner=' ~ ' + mainanswer.upper())
diff_l.append(diff_h)
html_clues[mainxdid] = diff_l
# Process for all matches
for xdid in matches:
xd = xdfile.get_xd(xdid)
if not xd:
continue
xddates[xdid] = xd.date()
# output each grid
html_grids[xdid] = grid_diff_html(xd, compare_with=mainxd)
diff_l = []
# output comparison of each set of clues
for pos, clue, answer in xd.iterclues():
diff_h = mktag('div','fullgrid') + '%s. ' %pos
# Sometimes can return clue == None
sm = difflib.SequenceMatcher(lambda x: x == ' ', mainxd.get_clue(pos) or '', clue)
if sm.ratio() < 0.50:
diff_h += clue
else:
# Compare based on op codes
for opcode in sm.get_opcodes():
c, a1, a2, b1, b2 = opcode
if c == 'equal':
diff_h += '<span class="match">%s</span>' % clue[b1:b2]
else:
diff_h += '<span class="diff">%s</span>' % clue[b1:b2]
diff_h += mktag('span', tagclass=(answer == mainxd.get_answer(pos)) and 'match' or 'diff', inner=' ~ ' + answer.upper())
diff_h += mktag('/div')
diff_l.append(diff_h)
html_clues[xdid] = diff_l
# Wrap into table
diff_h = mktag('table') + mktag('tr')
# Sort by date
sortedkeys = sorted(xddates.items(), key=operator.itemgetter(1))
for w, dt in sortedkeys:
# Wrap into table
diff_h += mktag('td') + html_grids[w] + mktag('/td')
diff_h += mktag('/tr')
for i, clue in enumerate(html_clues[sortedkeys[0][0]]):
diff_h += mktag('tr')
for w, dt in sortedkeys:
if i < len(html_clues[w]):
diff_h += mktag('td') + html_clues[w][i] + mktag('/td')
diff_h += mktag('/tr')
# Process deepclues
diff_h += mktag('table') + dcl_html + mktag('/table')
diff_h += mktag('/table')
outf.write_html('pub/deep/%s/index.html' % mainxdid, diff_h,
title='Deep clue comparison for ' + mainxdid)
def main():
args = utils.get_args('generate pub-years data')
outf = utils.open_output()
weekdays = [ 'Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun' ]
pubyears = {} # set()
for xd in xdfile.corpus():
puby = (xd.publication_id(), xd.year())
if puby not in pubyears:
pubyears[puby] = []
pubyears[puby].append(xd)
if pubyears:
metadb.delete_stats()
for puby, xdlist in sorted(pubyears.items()):
pubid, year = puby
npublic = 0
# TODO: SELECT FROM publications
nexisting = 0
# organize by day-of-week
byweekday = {}
byweekday_similar = {}
for w in weekdays:
byweekday[w] = []
byweekday_similar[w] = []
for xd in xdlist:
dow = dow_from_date(xd.get_header('Date'))
if dow: # Might be empty date or only a year
byweekday[dow].append(xd)
for r in metadb.xd_similar(pubid + str(year)):
if r.match_pct < 25:
continue
xd = xdfile.get_xd(r.xdid)
if xd:
dt = xd.get_header('Date')
if dt:
dow = dow_from_date(dt)
if dow: # Might be empty date or only a year
byweekday_similar[dow].append(r)
else:
debug("Date not set for: %s" % xd)
# tally stats
for weekday in weekdays:
copyrights = Counter() # [copyright_text] -> number of xd
editors = Counter() # [editor_name] -> number of xd
formats = Counter() # ["15x15 RS"] -> number of xd
# todo
nexisting = 0
nxd = len(byweekday[weekday])
public_xdids = [] # Empty for now
for xd in byweekday[weekday]:
xdid = xd.xdid()
if (year.isdigit() and int(year) <= 1965) or xdid in public_xdids:
npublic += 1
editor = xd.get_header('Editor').strip()
if editor:
editors[editor] += 1
sizestr = xd.sizestr()
if sizestr:
formats[sizestr] += 1
copyright = xd.get_header('Copyright').strip()
if copyright:
copyrights[copyright] += 1
# debug("ME: %s MCPR: %s MF: %s" % (list(editors), list(copyrights), list(formats)))
def process_counter(count, comp_value):
# Process counter comparing with comp_value
if count:
item, num = count.most_common(1)[0]
if num != comp_value:
item += " (%s)" % num
else:
item = ''
return item
#
maineditor = process_counter(editors, nxd)
maincopyright = process_counter(copyrights, nxd)
mainformat = process_counter(formats, nxd)
reprints = 0
touchups = 0
redones = 0
copies = 0
themecopies = 0
for r in byweekday_similar[weekday]:
xd1 = xdfile.get_xd(r.xdid)
xd2 = xdfile.get_xd(r.match_xdid)
if xd1 is None:
info("%s: similar puzzle %s not in corpus" % (r.match_xdid, r.xdid))
continue
if xd2 is None:
info("%s: similar puzzle %s not in corpus" % (r.xdid, r.match_xdid))
continue
dt1 = xd1.get_header('Date')
dt2 = xd2.get_header('Date')
aut1 = xd1.get_header('Author').lower()
aut2 = xd2.get_header('Author').lower()
pct = int(r.match_pct)
if dt2 < dt1: # only capture the later one
##deduce_similarity_type
if diff_authors(aut1, aut2): # suspicious
if pct >= 50:
copies += 1
elif pct >= 30:
themecopies += 1
else:
if pct == 100:
reprints += 1
elif pct >= 50:
touchups += 1
elif pct >= 30:
themecopies += 1
metadb.append_row("pub/stats",
(pubid, year, weekday,
mainformat, maineditor, maincopyright,
nexisting, nxd, npublic,
reprints, touchups, redones,
copies, themecopies))
def main():
args = utils.get_args('generate pub-years data')
outf = utils.open_output()
weekdays = ['Mon', 'Tue', 'Wed', 'Thu', 'Fri', 'Sat', 'Sun']
pubyears = {} # set()
for xd in xdfile.corpus():
puby = (xd.publication_id(), xd.year())
if puby not in pubyears:
pubyears[puby] = []
pubyears[puby].append(xd)
if pubyears:
metadb.delete_stats()
for puby, xdlist in sorted(pubyears.items()):
pubid, year = puby
npublic = 0
# TODO: SELECT FROM publications
nexisting = 0
# organize by day-of-week
byweekday = {}
byweekday_similar = {}
for w in weekdays:
byweekday[w] = []
byweekday_similar[w] = []
for xd in xdlist:
dow = dow_from_date(xd.get_header('Date'))
if dow: # Might be empty date or only a year
byweekday[dow].append(xd)
for r in metadb.xd_similar(pubid + str(year)):
if r.match_pct < 25:
continue
xd = xdfile.get_xd(r.xdid)
if xd:
dt = xd.get_header('Date')
if dt:
dow = dow_from_date(dt)
if dow: # Might be empty date or only a year
byweekday_similar[dow].append(r)
else:
debug("Date not set for: %s" % xd)
# tally stats
for weekday in weekdays:
copyrights = Counter() # [copyright_text] -> number of xd
editors = Counter() # [editor_name] -> number of xd
formats = Counter() # ["15x15 RS"] -> number of xd
# todo
nexisting = 0
nxd = len(byweekday[weekday])
public_xdids = [] # Empty for now
for xd in byweekday[weekday]:
xdid = xd.xdid()
if (year.isdigit()
and int(year) <= 1965) or xdid in public_xdids:
npublic += 1
editor = xd.get_header('Editor').strip()
if editor:
editors[editor] += 1
sizestr = xd.sizestr()
if sizestr:
formats[sizestr] += 1
copyright = xd.get_header('Copyright').strip()
if copyright:
copyrights[copyright] += 1
# debug("ME: %s MCPR: %s MF: %s" % (list(editors), list(copyrights), list(formats)))
def process_counter(count, comp_value):
# Process counter comparing with comp_value
if count:
item, num = count.most_common(1)[0]
if num != comp_value:
item += " (%s)" % num
else:
item = ''
return item
#
maineditor = process_counter(editors, nxd)
maincopyright = process_counter(copyrights, nxd)
mainformat = process_counter(formats, nxd)
reprints = 0
touchups = 0
redones = 0
copies = 0
themecopies = 0
for r in byweekday_similar[weekday]:
xd1 = xdfile.get_xd(r.xdid)
xd2 = xdfile.get_xd(r.match_xdid)
if xd1 is None:
info("%s: similar puzzle %s not in corpus" %
(r.match_xdid, r.xdid))
continue
if xd2 is None:
info("%s: similar puzzle %s not in corpus" %
(r.xdid, r.match_xdid))
continue
dt1 = xd1.get_header('Date')
dt2 = xd2.get_header('Date')
aut1 = xd1.get_header('Author').lower()
aut2 = xd2.get_header('Author').lower()
pct = int(r.match_pct)
if dt2 < dt1: # only capture the later one
##deduce_similarity_type
if diff_authors(aut1, aut2): # suspicious
if pct >= 50:
copies += 1
elif pct >= 30:
themecopies += 1
else:
if pct == 100:
reprints += 1
elif pct >= 50:
touchups += 1
elif pct >= 30:
themecopies += 1
metadb.append_row(
"pub/stats", (pubid, year, weekday, mainformat, maineditor,
maincopyright, nexisting, nxd, npublic, reprints,
touchups, redones, copies, themecopies))
def main():
args = utils.get_args("generates .html diffs for all puzzles in similar.tsv")
outf = utils.open_output()
similars = utils.parse_tsv("gxd/similar.tsv", "Similar")
xdids_todo = args.inputs or [xdid for xdid, matches in metadb.get_similar_grids().items() if matches]
for mainxdid in xdids_todo:
progress(mainxdid)
mainxd = xdfile.get_xd(mainxdid)
if not mainxd:
continue
matches = metadb.get_similar_grids().get(mainxdid, [])
xddates = {}
xddates[mainxdid] = mainxd.date() # Dict to store XD dates for further sort
html_grids = {}
html_clues = {}
# Store in list to make further formatting as html table easier
html_grids[mainxdid] = grid_diff_html(xdfile.get_xd(mainxdid))
# Add for main XD
diff_l = []
for pos, mainclue, mainanswer in mainxd.iterclues():
diff_h = mktag("div", "fullgrid main") + "%s. " % pos
diff_h += mainclue
diff_h += mktag("span", tagclass="main", inner=" ~ " + mainanswer.upper())
diff_l.append(diff_h)
html_clues[mainxdid] = diff_l
# Process for all matches
for xdid in matches:
xd = xdfile.get_xd(xdid)
# Continue if can't load xdid
if not xd:
continue
xddates[xdid] = xd.date()
# output each grid
html_grids[xdid] = grid_diff_html(xd, compare_with=mainxd)
diff_l = []
# output comparison of each set of clues
for pos, clue, answer in xd.iterclues():
diff_h = mktag("div", "fullgrid") + "%s. " % pos
# Sometimes can return clue == None
mainclue = mainxd.get_clue_for_answer(answer)
sm = difflib.SequenceMatcher(lambda x: x == " ", mainclue or "", clue)
debug("MCLUE: %s [%s]" % (mainclue, sm.ratio()))
if mainclue is None or sm.ratio() < 0.40:
diff_h += clue
else:
# Compare based on op codes
for opcode in sm.get_opcodes():
c, a1, a2, b1, b2 = opcode
if c == "equal":
diff_h += '<span class="match">%s</span>' % clue[b1:b2]
else:
diff_h += '<span class="diff">%s</span>' % clue[b1:b2]
tagclass = "match" if mainclue or answer == mainxd.get_answer(pos) else "diff"
diff_h += mktag("span", tagclass=tagclass, inner=" ~ " + answer.upper())
diff_h += mktag("/div")
diff_l.append(diff_h)
html_clues[xdid] = diff_l
# Wrap into table
diff_h = mktag("table") + mktag("tr")
# Sort by date
sortedkeys = sorted(xddates.items(), key=operator.itemgetter(1))
for w, dt in sortedkeys:
# Wrap into table
diff_h += mktag("td") + html_grids[w] + mktag("/td")
diff_h += mktag("/tr")
for i, clue in enumerate(html_clues[sortedkeys[0][0]]):
diff_h += mktag("tr")
for w, dt in sortedkeys:
if i < len(html_clues[w]):
diff_h += mktag("td") + html_clues[w][i] + mktag("/td")
diff_h += mktag("/tr")
diff_h += mktag("/table")
outf.write_html("pub/%s/index.html" % mainxdid, diff_h, title="Comparison for " + mainxdid)
def main():
args = utils.get_args('generates .html diffs for all puzzles in similar.tsv')
outf = utils.open_output()
similars = utils.parse_tsv('gxd/similar.tsv', 'Similar')
xdids_todo = {}
for row in metadb.xd_similar_all():
if row.xdid not in xdids_todo:
xdids_todo[row.xdid] = []
xdids_todo[row.xdid].append(row)
for mainxdid in xdids_todo:
progress(mainxdid)
mainxd = xdfile.get_xd(mainxdid)
if not mainxd:
warn('%s not in corpus' % mainxdid)
continue
matches = xdids_todo[mainxdid]
info('generating diffs for %s (%d matches)' % (mainxdid, len(matches)))
xddates = {}
xddates[mainxdid] = mainxd.date() # Dict to store XD dates for further sort
html_grids = {}
html_clues = {}
# Store in list to make further formatting as html table easier
html_grids[mainxdid] = grid_diff_html(xdfile.get_xd(mainxdid))
# Add for main XD
diff_l = []
for pos, mainclue, mainanswer in mainxd.iterclues():
if not mainclue:
continue
diff_h = mktag('div','fullgrid main') + '%s. ' %pos
diff_h += mainclue
diff_h += mktag('span', tagclass='main', inner=' ~ ' + mainanswer.upper())
diff_l.append(diff_h)
html_clues[mainxdid] = diff_l
# Process for all matches
for row in matches:
xdid = row.match_xdid
xd = xdfile.get_xd(xdid)
# Continue if can't load xdid
if not xd:
continue
xddates[xdid] = xd.date()
# output each grid
html_grids[xdid] = grid_diff_html(xd, compare_with=mainxd)
diff_l = []
# output comparison of each set of clues
for pos, clue, answer in xd.iterclues():
diff_h = mktag('div','fullgrid') + '%s. ' %pos
if not clue:
continue
# Sometimes can return clue == None
mainclue = mainxd.get_clue_for_answer(answer)
sm = difflib.SequenceMatcher(lambda x: x == ' ', mainclue or '', clue)
debug('MCLUE: %s [%s]' % (mainclue, sm.ratio()))
if mainclue is None or sm.ratio() < 0.40:
diff_h += clue
else:
# Compare based on op codes
for opcode in sm.get_opcodes():
c, a1, a2, b1, b2 = opcode
if c == 'equal':
diff_h += '<span class="match">%s</span>' % clue[b1:b2]
else:
diff_h += '<span class="diff">%s</span>' % clue[b1:b2]
tagclass = 'match' if mainclue or answer == mainxd.get_answer(pos) else 'diff'
diff_h += mktag('span', tagclass=tagclass, inner=' ~ ' + answer.upper())
diff_h += mktag('/div')
diff_l.append(diff_h)
html_clues[xdid] = diff_l
# Wrap into table
diff_h = mktag('table') + mktag('tr')
# Sort by date
sortedkeys = sorted(xddates.items(), key=operator.itemgetter(1))
for w, dt in sortedkeys:
# Wrap into table
diff_h += mktag('td') + html_grids[w] + mktag('/td')
diff_h += mktag('/tr')
for i, clue in enumerate(html_clues[sortedkeys[0][0]]):
diff_h += mktag('tr')
for w, dt in sortedkeys:
if i < len(html_clues[w]):
diff_h += mktag('td') + html_clues[w][i] + mktag('/td')
diff_h += mktag('/tr')
diff_h += mktag('/table')
outf.write_html('pub/%s/index.html' % mainxdid, diff_h, title='Comparison for ' + mainxdid)
