228 lines
9.3 KiB
Python
228 lines
9.3 KiB
Python
import json
|
||
|
||
import xlrd
|
||
|
||
import aigenerated
|
||
from coord import Coord, Merged
|
||
import utils
|
||
|
||
LOGGING = True
|
||
|
||
def pprint(*args, **kwargs):
|
||
if LOGGING:
|
||
print(*args, **kwargs)
|
||
|
||
|
||
class Parser:
|
||
def __init__(self, sheet: "xlrd.sheet.Sheet"):
|
||
self.sh: "xlrd.sheet.Sheet" = sheet
|
||
self.groups = {}
|
||
self.teachers = set()
|
||
self.places = set()
|
||
pprint("Parser created for '{0}': size: {1}x{2}".format(self.sh.name, self.sh.nrows, self.sh.ncols))
|
||
|
||
def parse(self):
|
||
monday = utils.find(self.sh, "ПОНЕДЕЛЬНИК")
|
||
if monday is None:
|
||
print(" -- Failed parse! -- ")
|
||
print("ПОНЕДЕЛЬНИК НЕ НАЙДЕН!")
|
||
return
|
||
|
||
head_rx = monday[0] - 1 # выше первого понидельника
|
||
if head_rx < 0:
|
||
raise Exception("head_rx < 0: Программа пыталась найти 'ПОНЕДЕЛЬНИК', но по всей видимости не нашла.")
|
||
|
||
head = self.sh.row(head_rx) # get all ROW (months, groups)
|
||
pprint(f"head={head}")
|
||
self.groups = parse_groups(self.sh, head, monday, head_rx) # parse groups to self.groups
|
||
pprint(f'self.groups={json.dumps(self.groups, indent=2, ensure_ascii=False)}')
|
||
|
||
pprint("\n\n\n")
|
||
|
||
for group in self.groups.values():
|
||
pprint("\nSTART OF PROCESS GROUP\n")
|
||
self.process_group(group, monday)
|
||
pprint("\nEND OF PROCESS GROUP\n")
|
||
|
||
pprint(self.teachers)
|
||
|
||
def parse_potokoviy(self, merged: Merged):
|
||
speaker = None
|
||
location = None
|
||
|
||
# speaker
|
||
low = merged.low
|
||
speaker_pos = low.shift(down=merged.height())
|
||
speaker = speaker_pos.cell(self.sh).value
|
||
|
||
# location
|
||
location = merged.high.shift(down=1).cell(self.sh).value
|
||
|
||
return {"loc": str(location), "leader": str(speaker), "name": str(merged.cell(self.sh).value)}
|
||
|
||
def process_group(self, group, monday):
|
||
"""
|
||
Обработать группы, выполняется для каждой группы, после того как они распарены (parse_groups)
|
||
group = {'name': 'ИВТ-260', 'position': [5, 6], 'position_human': 'G6:J6'}
|
||
"""
|
||
pprint(f"process_group group={group}")
|
||
group_name = group['name']
|
||
pprint(group_name)
|
||
row = group['position'][0] + 1 # counter for while, +1 for shift down; также номер строки в таблице (вроде с нуля)
|
||
weeknum = 1 # номер недели, щёлкнет +1 при каком-то условии.
|
||
while row < self.sh.nrows: # maybe условие чтобы не уйти ниже чем есть строк
|
||
pos = Coord(row, group['position'][1]) # текущая позиция, верхний правый угол (=low)
|
||
pos_right = pos.shift(right=3)
|
||
pair_pos = pos.replace(col=5)
|
||
weekday_pos = pos.replace(col=4)
|
||
merged = utils.get_merged_coord(self.sh, pos)
|
||
right_cell = pos_right.cell(self.sh)
|
||
merged_cell = merged.cell(self.sh)
|
||
cv = merged_cell.value
|
||
# В конце (12 пара:>) название группы, можно использовать как якорь
|
||
if utils.unspace(cv) == group_name:
|
||
pprint("Lesson == group name; ending group loop.")
|
||
break
|
||
|
||
weekday_mr = utils.get_merged_coord(self.sh, weekday_pos)
|
||
weekday = utils.unspace(weekday_mr.cell(self.sh).value)
|
||
pair_mr = utils.get_merged_coord(self.sh, pair_pos)
|
||
pair = utils.unspace(pair_mr.cell(self.sh).value)
|
||
|
||
skip = 0
|
||
if weekday == "":
|
||
if weeknum == 1:
|
||
weeknum += 1
|
||
pprint("------")
|
||
skip = 1
|
||
row += 1
|
||
else:
|
||
break
|
||
if not skip:
|
||
next = 3 # на сколько пыгнуть для следующего шага?
|
||
|
||
is_empty_lesson = right_cell.ctype in utils.EMPTY_CTYPES and merged_cell.ctype in utils.EMPTY_CTYPES
|
||
dispname = ""
|
||
parsed_discipline_name = None
|
||
parsed_location = None
|
||
parsed_leader = None
|
||
is_2pair = False
|
||
is_solid = pos_right in merged
|
||
parsed_uncotigorized = []
|
||
is_wide_maybe_potokoviy = merged.width() > 4 # потоковая ли лекция (занимает несколько групп.)
|
||
if is_empty_lesson:
|
||
dispname = "<no lesson>"
|
||
|
||
if not is_empty_lesson:
|
||
may_prepod = merged.low.shift(down=2)
|
||
if utils.has_no_bottom_border(self.sh, may_prepod):
|
||
next = 6
|
||
is_2pair = True
|
||
|
||
if is_wide_maybe_potokoviy:
|
||
ret = self.parse_potokoviy(merged)
|
||
parsed_location = ret['loc']
|
||
parsed_leader = ret['leader']
|
||
parsed_discipline_name = ret['name']
|
||
parsed_uncotigorized = list(utils.parse_all_dirt(self.sh, merged.low, merged.width(), next))
|
||
|
||
|
||
else:
|
||
if (is_solid):
|
||
parsed_discipline_name = cv
|
||
|
||
dispname = cv
|
||
dispname += (" SOLD" if is_solid else " SPLIT")
|
||
dispname += (" [ДВУПАРНЫЙ]" if is_2pair else "")
|
||
|
||
parsed_uncotigorized = list(utils.parse_all_dirt(self.sh, merged.low, 4, next))
|
||
|
||
|
||
if parsed_leader: dispname += f" [{parsed_leader}]"
|
||
if parsed_location: dispname += f" [{parsed_location}]"
|
||
dispname = dispname.replace("\n", "\\n")
|
||
pprint(f"[{group_name}] row={row}; {pos} {pos_right} {pair} {weekday}: {'[ПОТОКОВЫЙ] ' if is_wide_maybe_potokoviy else ''}{dispname} {parsed_uncotigorized}")
|
||
|
||
# пытаемся из некотегорезированных данных выцепить место и лидера (препода)
|
||
prepods = set()
|
||
if parsed_leader is not None: prepods.add(aigenerated.extract_last_name(parsed_leader))
|
||
|
||
locations = set()
|
||
if parsed_location is not None: locations.add(parsed_location.replace(" ", "").replace("-", ""))
|
||
|
||
for x in list(parsed_uncotigorized):
|
||
if aigenerated.is_surname_string(x):
|
||
prepods.add(aigenerated.extract_last_name(x))
|
||
|
||
if aigenerated.is_room_number(x):
|
||
locations.add(x.replace(" ", "").replace("-", "") if x is not None else None)
|
||
|
||
# оставшееся в дисциплину (костыль)
|
||
if parsed_discipline_name is None:
|
||
parsed_discipline_name = " ".join(parsed_uncotigorized)
|
||
|
||
prepods.discard(None)
|
||
prepods.discard("")
|
||
locations.discard(None)
|
||
locations.discard("")
|
||
if not is_empty_lesson:
|
||
slots = group['slots']
|
||
w = weekday + ("_1" if weeknum == 1 else "_2")
|
||
if w not in slots.keys():
|
||
slots[w] = {}
|
||
|
||
today = slots[w]
|
||
today[pair] = {
|
||
"pos": str(pos),
|
||
"discipline": parsed_discipline_name,
|
||
"locations": list(locations),
|
||
"leads": list(prepods),
|
||
"is_solid": is_solid,
|
||
"is_2pair": is_2pair,
|
||
"is_flow": is_wide_maybe_potokoviy,
|
||
"raw": parsed_uncotigorized,
|
||
"weeday": utils.weekday_to_num(weekday),
|
||
"weeknum": weeknum
|
||
}
|
||
|
||
|
||
self.teachers.add(aigenerated.extract_last_name(parsed_leader))
|
||
|
||
|
||
# INCREMENT на next и конец цикла.
|
||
row += next
|
||
|
||
|
||
|
||
def parse_groups(sh, head, monday, head_rx):
|
||
"""Распознать список групп и метаданные к ним, по сути получить список названий группы и координат её верхнего header-а (AQ6:AT6)"""
|
||
groups = {}
|
||
i = 0
|
||
while i < len(head):
|
||
x = head[i]
|
||
pprint(f"while i={i} head[i]={x}")
|
||
merged = utils.get_merged_coord(sh, Coord(head_rx, i))
|
||
if i > monday[1] + 1:
|
||
if merged is None or x.value == "":
|
||
break
|
||
|
||
if merged.width() != 4:
|
||
pprint(f"WARNING: group header witdh !=4 (found: {merged.width()}); blocks !=4 not supported by parser.")
|
||
break
|
||
|
||
name = utils.unspace(x.value)
|
||
groups[name] = {
|
||
"name": name,
|
||
"position": [head_rx, i],
|
||
"position_human": utils.merged_humanize(merged.as_numbers()),
|
||
"slots": {}
|
||
}
|
||
|
||
if merged is None:
|
||
i += 1
|
||
else:
|
||
i += merged.width()
|
||
|
||
return groups
|
||
|