bibi/src/readtable.jl

using FileIO
using CSV
using TidierData
using DataFrames
using TidierDates
using BenchmarkTools
using TidierStrings
using XLSX

df = CSV.read("TCSC_web2023.csv", DataFrame, header = 2)

# 比對檔案
srda = @chain pwd() begin
	joinpath("backend-DB/SRDA-BibData.csv")
	CSV.read(DataFrame, missingstring="")
end  

names(df)
l0 = uppercase.(df[!, :論著標題])
l0 = replace.(l0, r"\(.+\)$" => "")
l1 = uppercase.(skipmissing(srda[!, :原始題目]))
bol = l0 .∉ Ref(l1)
xx = df[bol, :]

CSV.write("crawled-data/tscs2023_checked.csv", xx)

ss = "高教擴張前後大學學歷對主觀社會地位的影響(A Causal Analysis of the Impact of College Expansion on Subjective Social Status)"
s2 = "臺灣社會的孝道變遷（1994-2021)：基於雙元孝道架構的年齡-時期-世代效應分析"