phanipavank/IIITB_Menu
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix csv decode to parse latin-1 space

Browse Source
This commit is contained in:
Phani Pavan K
2023-12-04 18:18:49 +05:30
parent b8ed730c34
commit 1e7197e17a
1 changed files with 6 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
scrapeToJson.py
View File

@@ -1,5 +1,6 @@
import pandas import pandas
import numpy import numpy
import numpy
import datetime import datetime
import json import json
import hashlib import hashlib
@@ -15,12 +16,15 @@ ALLDAYS = [
] ]
if __name__ == "__main__": if __name__ == "__main__":
#! TODO: Find a better XLSX import process. XLS IS ASS
# mainData = pandas.read_excel("menu.ods") # mainData = pandas.read_excel("menu.ods")
mainData = pandas.read_csv("menu.csv") mainData = pandas.read_csv("menu.csv")
mainData = mainData.drop("Unnamed: 9", axis="columns").drop("Unnamed: 10", axis="columns").drop(" ", axis="columns")
mainData[" .1"] = mainData[" .1"].replace("\xa0", numpy.nan)
breaksLoc = numpy.where( breaksLoc = numpy.where(
pandas.isna(mainData.drop(" ", axis="columns")).all(1).to_numpy() pandas.isna(mainData[" .1"]).to_numpy()
== True == True
)[0] )[0][2:]
print(breaksLoc) print(breaksLoc)
rowNames = mainData[" .1"].to_numpy() rowNames = mainData[" .1"].to_numpy()
# Item names: x[x["Unnamed: 1"]=="BREAKFAST 1"]["MONDAY"].item() # Item names: x[x["Unnamed: 1"]=="BREAKFAST 1"]["MONDAY"].item()