initial commit
This commit is contained in:
155
parser.py
Normal file
155
parser.py
Normal file
@@ -0,0 +1,155 @@
|
||||
import json
|
||||
|
||||
import xlrd
|
||||
|
||||
from coord import Coord, Merged
|
||||
import utils
|
||||
|
||||
|
||||
|
||||
class Parser:
|
||||
def __init__(self, sheet: "xlrd.sheet.Sheet"):
|
||||
self.sh: "xlrd.sheet.Sheet" = sheet
|
||||
self.groups = {}
|
||||
print("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")
|
||||
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)
|
||||
print(f"head={head}")
|
||||
self.groups = parse_groups(self.sh, head, monday, head_rx) # parse groups to self.groups
|
||||
print(f'self.groups={json.dumps(self.groups, indent=2, ensure_ascii=False)}')
|
||||
|
||||
print("\n\n\n")
|
||||
|
||||
for group in self.groups.values():
|
||||
print("\nSTART OF PROCESS GROUP\n")
|
||||
self.process_group(group, monday)
|
||||
print("\nEND OF PROCESS GROUP\n")
|
||||
|
||||
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": location, "leader": speaker, "name": 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'}
|
||||
"""
|
||||
print(f"process_group group={group}")
|
||||
group_name = group['name']
|
||||
print(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:
|
||||
print("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
|
||||
print("------")
|
||||
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_location = None
|
||||
parsed_leader = None
|
||||
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:
|
||||
if is_wide_maybe_potokoviy:
|
||||
ret = self.parse_potokoviy(merged)
|
||||
parsed_location = ret['loc']
|
||||
parsed_leader = ret['leader']
|
||||
dispname = ret['name']
|
||||
else:
|
||||
is_solid = pos_right in merged
|
||||
is_2pair = False
|
||||
may_prepod = merged.low.shift(down=2)
|
||||
if utils.border_bottom(self.sh, may_prepod) == 0 and utils.border_top(self.sh, may_prepod.shift(down=1)) == 0:
|
||||
next = 6
|
||||
is_2pair = True
|
||||
|
||||
dispname = cv
|
||||
dispname += (" SOLD" if is_solid else " SPLIT")
|
||||
dispname += (" [ДВУПАРНЫЙ]" if is_2pair else "")
|
||||
|
||||
parsed_uncotigorized = utils.parse_all_dirt(self.sh, merged.low, 2, next-1)
|
||||
|
||||
|
||||
if parsed_leader: dispname += f" [{parsed_leader}]"
|
||||
if parsed_location: dispname += f" [{parsed_location}]"
|
||||
dispname = dispname.replace("\n", "\\n")
|
||||
print(f"[{group_name}] row={row}; {pos} {pos_right} {pair} {weekday}: {'[ПОТОКОВЫЙ] ' if is_wide_maybe_potokoviy else ''}{dispname} {parsed_uncotigorized}")
|
||||
|
||||
# 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]
|
||||
print(f"while i={i} head[i]={x}")
|
||||
merged = utils.get_merged(sh, head_rx, i)
|
||||
if i > monday[1] + 1:
|
||||
if merged is None or x.value == "":
|
||||
break
|
||||
name = utils.unspace(x.value)
|
||||
groups[name] = {
|
||||
"name": name,
|
||||
"position": [head_rx, i],
|
||||
"position_human": utils.merged_humanize(merged)
|
||||
}
|
||||
|
||||
if merged is None:
|
||||
i += 1
|
||||
else:
|
||||
i += (merged[3] - merged[1] + 1)
|
||||
|
||||
return groups
|
||||
|
||||
def get_weekday_left(sh, rowx, colx):
|
||||
m = utils.get_merged(sh, rowx, colx)
|
||||
return sh.cell(m[0], m[1]).value
|
||||
Reference in New Issue
Block a user