pdftools example

Osaka_survey.R

library(pdftools)
library(tidyr)
library(dplyr)
library(ggplot2)
browseURL("https://www.jstage.jst.go.jp/article/jph/50/8/50_686/_article/-char/ja/")
text = pdf_text("~/Downloads/50_686.pdf")

tab2raw = unlist(strsplit(text[grep("表２", text)],"  "))
tab2raw = tab2raw[sapply(tab2raw, nchar) > 0]
tab2raw = tab2raw[16:145]

tab2 = tab2raw[-c(15:19, 27:29,
                  37:39, 47:49, 
                  57:61, 69:75,
                  83:87, 95:101,
                  109:112,120:124)]
tab2 = tab2[-grep("月",tab2)]
numtab2 = as.integer(gsub(" |\n","", tab2))

mat2 = matrix(numtab2,ncol=6, byrow = TRUE)
mat2 = mat2[,c(1,3,5)]
colnames(mat2) <- c("野宿生活者", "簡易生活者", "総数")
df_2 = as.data.frame(mat2[,-3]) %>% 
  mutate(month=1:12) %>% 
  pivot_longer(1:2)

ggplot(df_2, aes(x=factor(month), y=value, colour=name, group=name))+
  geom_area(alpha=0.1, position = "identity")+
  labs(x="発見月", y="人数", colour="",
       title="ホームレス者死亡の発見月（逢坂他, 2003）")+
  theme_classic(15,"osaka")

#ggsave("line.png")