第 4 章 字符串操作
Handling Strings with R 和 R for Data Science 提供字符串入门介绍 ，Sara Stoudt 整理了 stringr 包与 Base R 正则表达式函数的对应表 https://stringr.tidyverse.org/articles/from-base.html
stringr 基于 stringi 包字符串处理包， re2r 包基于 Google 开发的 C++ 库 re2，Google 编程之夏项目提供了一份 正则表达式性能综述， stringdist Approximate String Matching and String Distance Functions 近似字符串匹配和字符串距离计算函数 [11]
- janitor
- Manipulating strings with the stringr package
- filesstrings 基于 stringr 操作字符串
- strex 一些没有包含在 stringr 或者 stringi 中的字符串操作函数
- tidytext Text mining using dplyr, ggplot2, and other tidy tools
stringdist stringfish stringb stringi stringr
字符和字符串类型的数据值得单独拿出来讲，不仅因为内容多，而且比较难，应用范围最广，特别是面对文本类型的数据时，几乎是避不开的！R 的前身是 S，S 的前身是一些 Fortran 和 C 子程序，最早在贝尔实验室是用于文本分析领域，因此在 R 基础包中提供了丰富的字符串处理函数，你可以在R控制台中执行如下一行命令查看
help.search(keyword = "character", package = "base")本章主要介绍 R 内置的字符串操作函数
4.1 字符数统计
nchar 函数统计字符串向量中每个元素的字符个数，注意与函数length 的差别，它统计向量中元素的个数，即向量的长度。
## [1] 5 5 1
R.version.string## [1] "R version 4.2.2 (2022-10-31)"
nchar(R.version.string)## [1] 28## [1] "function (x, ...) " "UseMethod(\"mean\")"## [1] 18 17一些特殊的情况
nchar("")## [1] 0
nchar(NULL)## integer(0)
nchar(0)## [1] 1
pi## [1] 3.141593
nchar(pi)## [1] 16
exp(1)## [1] 2.718282## [1] 16
nchar(NA)## [1] NA4.2 字符串翻译
tolower 将字符串或字符串向量中含有的大写字母全都转化为小写， toupper 函数正好与之相反.
## [1] "hello" "hello, r" "hello"## [1] "HELLO" "HELLO, R" "HELLO"4.3 字符串连接
paste 函数设置参数 sep 作为连接符，设置参数 collapse 可以将字符串拼接后连成一个字符串
paste("A", "B", sep = "")## [1] "AB"## [1] "A-1" "B-2" "C-3"## [1] "A-1;B-2;C-3"paste0 相当于 sep 设为空，没有连接符
paste0("A", "B")## [1] "AB"## [1] "A1" "B2" "C3"## [1] "A1;B2;C3"4.4 字符串拆分
strsplit(x, split, fixed = FALSE, perl = FALSE, useBytes = FALSE)strsplit 函数用于字符串拆分，参数 x 是被拆分的字符串向量，其每个元素都会被拆分，而参数 split 表示拆分的位置，可以用正则表达式来描述位置，拆分的结果是一个列表。
参数 fixed 默认设置 fixed = FALSE 表示正则表达式匹配，而 fixed = TRUE 表示正则表达式的精确匹配或者按文本字符的字面意思匹配，即按普通文本匹配。我们知道按普通文本匹配速度快。
当启用 perl = TRUE 时，由 PCRE_use_JIT 控制细节。perl 参数的设置与 Perl 软件版本有关，如果正则表达式很长，除了正确设置正则表达式，使用 perl = TRUE 可以提高运算速度
参数 useBytes 设置是否按照逐个字节地进行匹配，默认设置为 FALSE，即按照字符而不是字节进行匹配
x <- c(as = "asfef", qu = "qwerty", "yuiop[", "b", "stuff.blah.yech")
# 按字母 e 拆分字符串向量 x
strsplit(x, "e")## $as
## [1] "asf" "f"
##
## $qu
## [1] "qw" "rty"
##
## [[3]]
## [1] "yuiop["
##
## [[4]]
## [1] "b"
##
## [[5]]
## [1] "stuff.blah.y" "ch"参数 split 支持通过正则表达式的方式指明拆分位置
# 默认将点号 . 看作一个正则表达式，它是一个元字符，匹配任意字符
strsplit("a.b.c", ".")## [[1]]
## [1] "" "" "" "" ""
# 这才是按点号拆分
strsplit("a.b.c", ".", fixed = TRUE)## [[1]]
## [1] "a" "b" "c"
# 或者
strsplit("a.b.c", "[.]")## [[1]]
## [1] "a" "b" "c"
# 或者转义点号，去掉元字符的特殊意义
strsplit("a.b.c", "\\.")## [[1]]
## [1] "a" "b" "c"这里介绍一个将字符串逆序的函数 str_rev
str_rev <- function(x)
sapply(lapply(strsplit(x, NULL), rev), paste, collapse = "")
str_rev(c("abc", "Statistics"))## [1] "cba" "scitsitatS"为了加深理解，再举几个例子
## [[1]]
## [1] "" "a"
# 空字符只有有定义的时候才会产生
strsplit("", " ")## [[1]]
## character(0)
strsplit(" ", " ")## [[1]]
## [1] ""4.5 字符串匹配
agrep 和 agrepl 函数做近似（模糊）匹配 (Approximate Matching or Fuzzy Matching) ，对于匹配，考虑到参数 pattern 在参数 x 中匹配时，允许参数值x存在最小可能的插入、删除和替换，这种修改叫做Levenshtein 编辑距离，max.distance 控制其细节
agrep(pattern, x, max.distance = 0.1, costs = NULL,
ignore.case = FALSE, value = FALSE, fixed = TRUE,
useBytes = FALSE)
agrepl(pattern, x, max.distance = 0.1, costs = NULL,
ignore.case = FALSE, fixed = TRUE, useBytes = FALSE)agrep 函数返回 pattern 在 x 中匹配到的一个位置向量，agrepl 返回一个逻辑向量，这一点类似 grep 和 grepl 这对函数，下面举例子说明
agrep("lasy", "1 lazy 2")## [1] 1## [1] 2## [1] 1
# 返回匹配到值，而不是位置下标，类似 grep(..., value = TRUE) 的返回值
agrep("laysy", c("1 lazy", "1", "1 LAZY"), max = 2, value = TRUE)## [1] "1 lazy"## [1] 1 3
startsWith(x, prefix)
endsWith(x, suffix)startsWith 和 endsWith 函数用来匹配字符串的前缀和后缀，返回值是一个逻辑向量，参数 prefix 和 suffix 不要包含特殊的正则表达式字符，如点号.，举例子
# 字符串向量
search()## [1] ".GlobalEnv" "package:stats" "package:graphics"
## [4] "package:grDevices" "package:utils" "package:datasets"
## [7] "package:methods" "Autoloads" "package:base"
# 匹配以 package: 开头的字符串
startsWith(search(), "package:")## [1] FALSE TRUE TRUE TRUE TRUE TRUE TRUE FALSE TRUE## [1] FALSE TRUE TRUE TRUE TRUE TRUE TRUE FALSE TRUE当前目录下，列出扩展名为 .Rmd 的文件
# list.files(path = ".", pattern = "\\.Rmd$")
# 而不是 endsWith(list.files(), "\\.Rmd")
endsWith(list.files(), ".Rmd")## [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE TRUE TRUE FALSE
## [13] FALSE FALSE FALSE FALSE FALSE TRUE FALSE TRUE FALSE FALSE TRUE FALSE
## [25] FALSE TRUE FALSE TRUE FALSE FALSE FALSE FALSE TRUE FALSE FALSE FALSE
## [37] TRUE FALSE FALSE FALSE TRUE FALSE FALSE FALSE TRUE FALSE TRUE FALSE
## [49] FALSE TRUE TRUE FALSE TRUE FALSE FALSE FALSE TRUE TRUE TRUE TRUE
## [61] FALSE TRUE FALSE FALSE FALSE FALSE TRUE FALSE FALSE TRUE TRUE FALSE
## [73] FALSE FALSE FALSE FALSE FALSE FALSE TRUE TRUE TRUE FALSE TRUE FALSE
# 或者
grepl("\\.Rmd$", list.files())## [1] FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE FALSE TRUE TRUE FALSE
## [13] FALSE FALSE FALSE FALSE FALSE TRUE FALSE TRUE FALSE FALSE TRUE FALSE
## [25] FALSE TRUE FALSE TRUE FALSE FALSE FALSE FALSE TRUE FALSE FALSE FALSE
## [37] TRUE FALSE FALSE FALSE TRUE FALSE FALSE FALSE TRUE FALSE TRUE FALSE
## [49] FALSE TRUE TRUE FALSE TRUE FALSE FALSE FALSE TRUE TRUE TRUE TRUE
## [61] FALSE TRUE FALSE FALSE FALSE FALSE TRUE FALSE FALSE TRUE TRUE FALSE
## [73] FALSE FALSE FALSE FALSE FALSE FALSE TRUE TRUE TRUE FALSE TRUE FALSE部分匹配(Partial String Matching)
match(x, table, nomatch = NA_integer_, incomparables = NULL)
x %in% table
charmatch(x, table, nomatch = NA_integer_)
pmatch(x, table, nomatch = NA_integer_, duplicates.ok = FALSE)这几个 match 函数的返回值都是一个向量，每个元素是参数x在参数table中第一次匹配到的位置，charmatch 与 pmatch(x, table, nomatch = NA_integer_, duplicates.ok = TRUE) 类似，所以 pmatch 在默认 duplicates.ok = FALSE 的情况下，若x在第二个参数table中有多次匹配就会返回 NA，因此，实际上 pmatch 只允许在第二个参数中匹配一次
## [1] 2## [1] TRUE FALSE TRUE FALSE TRUE FALSE FALSE FALSE TRUE FALSE
# charmatch 就比较奇怪，规则太多
charmatch("", "") # returns 1## [1] 1## [1] 0
# med 只在table参数值的第二个位置部分匹配到，所以返回2
charmatch("med", c("mean", "median", "mode", "quantile")) # returns 2## [1] 2
charmatch("xx", "xx")## [1] 1
charmatch("xx", "xxa")## [1] 1
charmatch("xx", "axx")## [1] NA
# 注意比较与 charmatch 的不同
pmatch("", "") # returns NA## [1] NA## [1] NA## [1] 24.6 字符串查询
grep(pattern, x,
ignore.case = FALSE, perl = FALSE, value = FALSE,
fixed = FALSE, useBytes = FALSE, invert = FALSE
)
grepl(pattern, x,
ignore.case = FALSE, perl = FALSE,
fixed = FALSE, useBytes = FALSE
)grep 和 grepl 是一对字符串查询函数，查看字符串向量 x 中是否包含正则表达式 pattern 描述的内容
-
ignore.case:TRUE表示忽略大小写，FALSE表示匹配的时候区分大小写 -
fixed = TRUE表示启用 literal regular expression 字面正则表达式，默认情况下fixed = FALSE -
grep函数返回匹配到的字符串向量x的元素的下标，如果value=TRUE则返回下标对应的值 -
grepl函数返回一个逻辑向量，检查字符串向量x中的每个元素是否匹配到，匹配到返回TRUE，没有匹配到返回FALSE
# 返回下标位置
grep("[a-z]", letters)## [1] 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
## [26] 26
# 返回查询到的值
grep("[a-z]", letters, value = TRUE)## [1] "a" "b" "c" "d" "e" "f" "g" "h" "i" "j" "k" "l" "m" "n" "o" "p" "q" "r" "s"
## [20] "t" "u" "v" "w" "x" "y" "z"继续举例子
## [1] 1 2## [1] 2## [1] "apple" "banana"## [1] "banana"关于 grepl 函数的使用例子
## [1] TRUE TRUE## [1] FALSE TRUE在 R 里面分别表示 a\\b 和 a\b
writeLines(c("a\\\\b", "a\\b"))## a\\b
## a\b下面在 R 里面分别匹配字符串 a\\b 和 a\b 中的 \\ 和 \
## [1] "a\\\\b" "a\\b"## [1] "a\\\\b" "a\\b"## [1] "a\\\\b"## [1] "a\\\\b"
# 匹配字符串中的两个反斜杠 \\
grepl(x = "a\\\\b", pattern = "\\\\\\\\", fixed = FALSE)## [1] TRUE
grepl(x = "a\\\\b", pattern = "\\\\\\\\", fixed = TRUE)## [1] FALSE
grepl(x = "a\\\\b", pattern = "\\\\", fixed = TRUE)## [1] TRUE
regexpr(pattern, text,
ignore.case = FALSE, perl = FALSE,
fixed = FALSE, useBytes = FALSE
)
gregexpr(pattern, text,
ignore.case = FALSE, perl = FALSE,
fixed = FALSE, useBytes = FALSE
)
regexec(pattern, text,
ignore.case = FALSE, perl = FALSE,
fixed = FALSE, useBytes = FALSE
) 当启用 perl=TRUE 时， 函数 regexpr 和 gregexpr 支持 Python 环境下的命名捕获(named captures)，但是不支持长向量的输入。如果一个分组被命名了，如 (?<first>[A-Z][a-z]+) 那么匹配到的位置按命名返回。函数 sub 不支持命名反向引用 (Named backreferences)
函数 regmatches 用来提取函数regexpr, gregexpr 和 regexec 匹配到的子字符串
useBytes = FALSE 匹配位置和长度默认是按照字符级别来的，如果 useBytes = TRUE 则是按照逐个字节的匹配结果
如果使用到了 命名捕获 则会返回更多的属性 “capture.start”，“capture.length” 和 “capture.names”，分别表示捕获的起始位置、捕获的长度和捕获的命名。
-
regexpr函数返回一个整型向量，第一次匹配的初始位置，-1 表示没有匹配到，返回的属性match.length表示匹配的字符数量，是一个整型向量，向量长度是匹配的文本的长度，-1 表示没有匹配到
## [1] 9 5 14
## attr(,"match.length")
## [1] 4 4 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
txt <- c(
"The", "licenses", "for", "most", "software", "are",
"designed", "to", "take", "away", "your", "freedom",
"to", "share", "and", "change", "it.",
"", "By", "contrast,", "the", "GNU", "General", "Public", "License",
"is", "intended", "to", "guarantee", "your", "freedom", "to",
"share", "and", "change", "free", "software", "--",
"to", "make", "sure", "the", "software", "is",
"free", "for", "all", "its", "users"
)
# gregexpr("en", txt)
regexpr("en", txt)## [1] -1 4 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 2 -1 4
## [26] -1 4 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1
## attr(,"match.length")
## [1] -1 2 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 2 -1 2
## [26] -1 2 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1 -1
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE-
gregexpr函数返回一个列表，返回列表的长度与字符串向量的长度一样，列表中每个元素的形式与regexpr的返回值一样, except that the starting positions of every (disjoint) match are given.
gregexpr("Adam", text)## [[1]]
## [1] 9
## attr(,"match.length")
## [1] 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
##
## [[2]]
## [1] 5
## attr(,"match.length")
## [1] 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
##
## [[3]]
## [1] 14
## attr(,"match.length")
## [1] 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE-
regexec函数返回一个列表，类似函数gregexpr的返回结果，长度与字符串向量的长度一样，如果没有匹配到就返回 -1，匹配到了就返回一个匹配的初值位置的整型序列，所有子字符串与括号分组的正则表达式的子表达式对应，属性 “match.length” 是一个表示匹配的长度的向量，如果是 -1 表示没有匹配到。位置、长度和属性的解释与regexpr一致
regexec("Adam", text)## [[1]]
## [1] 9
## attr(,"match.length")
## [1] 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
##
## [[2]]
## [1] 5
## attr(,"match.length")
## [1] 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
##
## [[3]]
## [1] 14
## attr(,"match.length")
## [1] 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE下面这个将链接分解的例子由 Luke Tierney 提供14
x <- "http://stat.umn.edu:80/xyz"
m <- regexec("^(([^:]+)://)?([^:/]+)(:([0-9]+))?(/.*)", x)
m## [[1]]
## [1] 1 1 1 8 20 21 23
## attr(,"match.length")
## [1] 26 7 4 12 3 2 4
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE这里 x 是一个字符串，所以函数 regexec 返回的列表长度为1，正则表达式 ^(([^:]+)://)?([^:/]+)(:([0-9]+))?(/.*) 括号分组匹配到了7次，第一次匹配整个字符串，所以起始位置是1，而匹配长度是26，即整个字符串的长度，读者可以调用函数 nchar(x) 算一下，如果你愿意手动数一下也可以哈！余下不一一介绍，可以根据返回结果和图4.1一起看，最后还可以调用regmatches函数抽取匹配到的结果
regmatches(x, m)## [[1]]
## [1] "http://stat.umn.edu:80/xyz" "http://"
## [3] "http" "stat.umn.edu"
## [5] ":80" "80"
## [7] "/xyz"我们可以在 https://regex101.com/ 上测试表达式，如图4.1所示，表达式 ^(([^:]+)://)?([^:/]+)(:([0-9]+))?(/.*) 包含7个组，每个组的匹配结果见图的右下角，这样我们不难理解，函数 regmatches 返回的第列表中，第3个位置是传输协议 protocol http ，第4个位置是主机 host stat.umn.edu， 第6个位置是端口 port 80 ，第7个位置是路径 path /xyz，所以函数 regmatches 的作用就是根据函数 regexec 匹配的结果抽取子字符串。
图 4.1: 正则表达式匹配结果
进一步，我们可以用 regmatches 函数抽取 URL 的部分内容，如前面提到的传输协议，主机等
URL_parts <- function(x) {
m <- regexec("^(([^:]+)://)?([^:/]+)(:([0-9]+))?(/.*)", x)
parts <- do.call(
rbind,
lapply(regmatches(x, m), `[`, c(3L, 4L, 6L, 7L))
# 3,4,6,7是索引位置
)
colnames(parts) <- c("protocol", "host", "port", "path")
parts
}
URL_parts(x)## protocol host port path
## [1,] "http" "stat.umn.edu" "80" "/xyz"目前还没有 gregexec 函数，但是可以模拟一个，首先用 gregexpr 函数返回匹配的位置，regmatches 抽取相应的值，然后用 regexec 作用到每一个提取的值，做再一次匹配和值的抽取，实现了全部的匹配。另一个例子
## There is no gregexec() yet, but one can emulate it by running
## regexec() on the regmatches obtained via gregexpr(). E.g.:
pattern <- "([[:alpha:]]+)([[:digit:]]+)"
s <- "Test: A1 BC23 DEF456"
gregexpr(pattern, s)## [[1]]
## [1] 7 10 15
## attr(,"match.length")
## [1] 2 4 6
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
regmatches(s, gregexpr(pattern, s))## [[1]]
## [1] "A1" "BC23" "DEF456"
lapply(
regmatches(s, gregexpr(pattern, s)),
function(e) regmatches(e, regexec(pattern, e))
)## [[1]]
## [[1]][[1]]
## [1] "A1" "A" "1"
##
## [[1]][[2]]
## [1] "BC23" "BC" "23"
##
## [[1]][[3]]
## [1] "DEF456" "DEF" "456"4.7 字符串替换
chartr 支持正则表达式的替换，chartr 是对应字符的替换操作
x <- "MiXeD cAsE 123"
# 将字符 iXs 替换为 why
chartr("iXs", "why", x)## [1] "MwheD cAyE 123"
# 将字符串 a-cX 中的字符挨个对应地替换为 D-Fw
chartr("a-cX", "D-Fw", x)## [1] "MiweD FAsE 123"两个 *sub 函数的区别：sub 替换第一次匹配到的结果，gsub 替换所有匹配的结果
sub(" .*", "", extSoftVersion()["PCRE"])## PCRE
## "10.39"参数 replacement 的值是正则表达式，其包含反向引用的用法， \\1 即引用表达式 ([ab])
gsub(pattern = "([ab])", replacement = "\\1_\\1_", x = "abc and ABC")## [1] "a_a_b_b_c a_a_nd ABC"4.8 字符串提取
substr 和 substring 函数通过位置进行字符串的拆分和提取，它们本身不使用正则表达式，结合其他正则表达式函数regexpr, gregexpr 和 regexec，可以很方便地从大量文本中提取所需的信息。作用类似之前提到的 regmatches 函数
参数设置基本相同
- x/text 是要拆分的字符串向量
- start/first 截取的起始位置向量
- stop/last 截取的终止位置向量
返回值有差别
-
substr返回的字串个数等于第一个参数 x 的长度 -
substring返回字串个数等于三个参数中最长向量长度，短向量循环使用。
## [1] "234"## [1] "234" "45" "2345678"
substr("abcdef", 2, 4)## [1] "bcd"
substring("abcdef", 1:6, 1:6)## [1] "a" "b" "c" "d" "e" "f"因为 x 的向量长度为1，所以 substr 获得的结果只有1个字串，即第2和第3个参数向量只用了第一个组合：起始位置2，终止位置4。而 substring 的语句三个参数中最长的向量为 c(4,5,8)，执行时按短向量循环使用的规则第一个参数事实上就是c(x,x,x)，第二个参数就成了c(2,4,2)，最终截取的字串起始位置组合为：2-4, 4-5和2-8。
## [1] "234" "de"## [1] "234" "de" "2345678"更加高级的字符串抽取
# 从字符串中抽取固定模式的文本，替代 stringr::str_extract
# 只抽取一个匹配的
extract_str <- function(text, pattern) regmatches(text, regexpr(pattern, text))
# 符合模式的全部抽取
gextract_str <- function(text, pattern) regmatches(text, gregexpr(pattern, text))举例子，抽取连续的数字
# 两个例子
extract_str(text = "abd123da345das", pattern = "(\\d+){3}")## [1] "123"
gextract_str(text = "abd123da345das", pattern = "(\\d+){3}")## [[1]]
## [1] "123" "345"4.9 命名捕捉
函数 regexpr(..., perl = TRUE) 和 gregexpr(..., perl = TRUE) 支持命名捕捉
## named capture
notables <- c(" Ben Franklin and Jefferson Davis",
"\tMillard Fillmore")
# name groups 'first' and 'last'
name.rex <- "(?<first>[[:upper:]][[:lower:]]+) (?<last>[[:upper:]][[:lower:]]+)"
(parsed <- regexpr(name.rex, notables, perl = TRUE))## [1] 3 2
## attr(,"match.length")
## [1] 12 16
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
## attr(,"capture.start")
## first last
## [1,] 3 7
## [2,] 2 10
## attr(,"capture.length")
## first last
## [1,] 3 8
## [2,] 7 8
## attr(,"capture.names")
## [1] "first" "last"
attr(parsed, 'capture.names')## [1] "first" "last"
regmatches(notables, parsed)## [1] "Ben Franklin" "Millard Fillmore"希望返回一个 data.frame，列名是指定的 named group 名字
# 有多个结果
(idx <- gregexpr(name.rex, notables, perl = TRUE))## [[1]]
## [1] 3 20
## attr(,"match.length")
## [1] 12 15
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
## attr(,"capture.start")
## first last
## [1,] 3 7
## [2,] 20 30
## attr(,"capture.length")
## first last
## [1,] 3 8
## [2,] 9 5
## attr(,"capture.names")
## [1] "first" "last"
##
## [[2]]
## [1] 2
## attr(,"match.length")
## [1] 16
## attr(,"index.type")
## [1] "chars"
## attr(,"useBytes")
## [1] TRUE
## attr(,"capture.start")
## first last
## [1,] 2 10
## attr(,"capture.length")
## first last
## [1,] 7 8
## attr(,"capture.names")
## [1] "first" "last"
regmatches(notables, idx)## [[1]]
## [1] "Ben Franklin" "Jefferson Davis"
##
## [[2]]
## [1] "Millard Fillmore"
attr(idx[[1]], 'capture.names')## [1] "first" "last"
library(magrittr)
data.frame(notable = notables) %>%
tidyr::extract(
notable, c("first", "last"), name.rex,
remove = FALSE
)## notable first last
## 1 Ben Franklin and Jefferson Davis Ben Franklin
## 2 \tMillard Fillmore Millard Fillmore4.11 模糊匹配
近似字符串匹配 (Approximate String Matching) 也叫模糊匹配 (Fuzzy Matching)
agrep() agrepl() aregexec() adist()
agrep(pattern = "lasy", x = "1 lazy 2")## [1] 1## [1] 2## [1] 1## [1] "1 lazy"## [1] 1 3
agrepl(pattern = "lasy", x = "1 lazy 2")## [1] TRUE
## Cf. the examples for agrep.
x <- c("1 lazy", "1", "1 LAZY")
aregexec("laysy", x, max.distance = 2)## [[1]]
## [1] 3
## attr(,"match.length")
## [1] 4
##
## [[2]]
## [1] -1
## attr(,"match.length")
## [1] -1
##
## [[3]]
## [1] -1
## attr(,"match.length")
## [1] -1
aregexec("(lay)(sy)", x, max.distance = 2)## [[1]]
## [1] 3 3 5
## attr(,"match.length")
## [1] 4 2 2
##
## [[2]]
## [1] -1
## attr(,"match.length")
## [1] -1
##
## [[3]]
## [1] -1
## attr(,"match.length")
## [1] -1
aregexec("(lay)(sy)", x, max.distance = 2, ignore.case = TRUE)## [[1]]
## [1] 3 3 6
## attr(,"match.length")
## [1] 4 3 1
##
## [[2]]
## [1] -1
## attr(,"match.length")
## [1] -1
##
## [[3]]
## [1] 3 3 6
## attr(,"match.length")
## [1] 4 3 1
m <- aregexec("(lay)(sy)", x, max.distance = 2)
regmatches(x, m)## [[1]]
## [1] "lazy" "la" "zy"
##
## [[2]]
## character(0)
##
## [[3]]
## character(0)
## Cf. https://en.wikipedia.org/wiki/Levenshtein_distance
adist("kitten", "sitting")## [,1]
## [1,] 3
## To see the transformation counts for the Levenshtein distance:
drop(attr(adist("kitten", "sitting", counts = TRUE), "counts"))## ins del sub
## 1 0 2
## To see the transformation sequences:
attr(adist(c("kitten", "sitting"), counts = TRUE), "trafos")## [,1] [,2]
## [1,] "MMMMMM" "SMMMSMI"
## [2,] "SMMMSMD" "MMMMMMM"
## Cf. the examples for agrep:
adist("lasy", "1 lazy 2")## [,1]
## [1,] 5
## For a "partial approximate match" (as used for agrep):
adist("lasy", "1 lazy 2", partial = TRUE)## [,1]
## [1,] 1案例
4.12 高级的替换
相比于 sprintf() 格式化输出字符串的方式替换，它的优势在于提示性，或者说代码的可读性
glue_data <- function(param, text) {
idx <- gregexpr('\\{[^}]*\\}', text)[[1L]]
keys <- substring(text, idx, idx + attr(idx, 'match.length') - 1L)
for (key in keys) {
text <- gsub(key, param[[gsub('[{}]', '', key)]], text, fixed = TRUE)
}
text
}
cat(glue_data(
param = list(table = 'flights', origin = 'JFK'),
text = "
select count(*) as n
from {table}
where origin = '{origin}'
"
))##
## select count(*) as n
## from flights
## where origin = 'JFK'
## 4.13 高级的提取
从 text 中抽取给定模式 pattern 的字符串
str_extract <- function(text, pattern, ...) regmatches(text, regexpr(pattern, text, ...))举个栗子，比如提取数字
shopping_list <- c("apples x4", "bag of flour", "bag of sugar", "milk x2")
stringr::str_extract(shopping_list, "\\d")## [1] "4" NA NA "2"
# 注意二者的差别
str_extract(shopping_list, "\\d")## [1] "4" "2"提取所有符合匹配模式的字符串
str_extract_all <- function(text, pattern, ...) regmatches(text, gregexpr(pattern, text, ...))举个栗子，提取其中的英文字母
str_extract_all(shopping_list, "[a-z]+")## [[1]]
## [1] "apples" "x"
##
## [[2]]
## [1] "bag" "of" "flour"
##
## [[3]]
## [1] "bag" "of" "sugar"
##
## [[4]]
## [1] "milk" "x"
stringr::str_extract_all(shopping_list, "[a-z]+")## [[1]]
## [1] "apples" "x"
##
## [[2]]
## [1] "bag" "of" "flour"
##
## [[3]]
## [1] "bag" "of" "sugar"
##
## [[4]]
## [1] "milk" "x"4.14 其它操作
4.14.1 strwrap
strwrap(x, width = 0.9 * getOption("width"), indent = 0,
exdent = 0, prefix = "", simplify = TRUE, initial = prefix)该函数把一个字符串当成一个段落的文字（不管字符串中是否有换行符），按照段落的格式（缩进和长度）和断字方式进行分行，每一行是结果中的一个字符串。
# 读取一段文本
x <- paste(readLines(file.path(R.home("doc"), "THANKS")), collapse = "\n")
## 将文本拆分为段落，且移除前三段
x <- unlist(strsplit(x, "\n[ \t\n]*\n"))[-(1:3)]
# 每一段换两行
x <- paste(x, collapse = "\n\n")
# 每一行的宽度设定为60个字符
writeLines(strwrap(x, width = 60))## J. D. Beasley, David J. Best, Richard Brent, Kevin Buhr,
## Michael A. Covington, Bill Cleveland, Robert Cleveland,, G.
## W. Cran, C. G. Ding, Ulrich Drepper, Paul Eggert, J. O.
## Evans, David M. Gay, H. Frick, G. W. Hill, Richard H.
## Jones, Eric Grosse, Shelby Haberman, Bruno Haible, John
## Hartigan, Andrew Harvey, Trevor Hastie, Min Long Lam,
## George Marsaglia, K. J. Martin, Gordon Matzigkeit, C. R.
## Mckenzie, Jean McRae, Cyrus Mehta, Fionn Murtagh, John C.
## Nash, Finbarr O'Sullivan, R. E. Odeh, William Patefield,
## Nitin Patel, Alan Richardson, D. E. Roberts, Patrick
## Royston, Russell Lenth, Ming-Jen Shyu, Richard C.
## Singleton, S. G. Springer, Supoj Sutanthavibul, Irma
## Terpenning, G. E. Thomas, Rob Tibshirani, Wai Wan Tsang,
## Berwin Turlach, Gary V. Vaughan, Michael Wichura, Jingbo
## Wang, M. A. Wong, and the Free Software Foundation (for
## autoconf code and utilities). See also files under
## src/extras.
##
## Many more, too numerous to mention here, have contributed
## by sending bug reports and suggesting various improvements.
##
## Simon Davies whilst at the University of Auckland wrote the
## original version of glm().
##
## Julian Harris and Wing Kwong (Tiki) Wan whilst at the
## University of Auckland assisted Ross Ihaka with the
## original Macintosh port.
##
## R was inspired by the S environment which has been
## principally developed by John Chambers, with substantial
## input from Douglas Bates, Rick Becker, Bill Cleveland,
## Trevor Hastie, Daryl Pregibon and Allan Wilks.
##
## A special debt is owed to John Chambers who has graciously
## contributed advice and encouragement in the early days of R
## and later became a member of the core team.
##
## Stefano Iacus (up to 2014, a former member of R Core) and
## Simon Urbanek developed the macOS port, including the R.app
## GUI, toolchains and packaging.
##
## The Windows port was originally developed by Guido
## Masarotto (for a while a member of R Core) and Brian
## Ripley, then further by Duncan Murdoch (a former member of
## R Core) and then Jeroen Ooms (base) and Uwe Ligges
## (packages). Tomas Kalibera is the current main developer
## of the Windows port and provides assistance with package
## porting.
##
## Tomas Kalibera's work has been sponsored by Jan Vitek and
## funded by his European Research Council grant "Evolving
## Language Ecosystems (ELE)".
##
## Computing support (including hardware, hosting and
## infrastructure) has been provided/funded by the R
## Foundation, employers of R-Core members (notably WU Wien,
## ETH Zurich, U Oxford and U Iowa) and by Northeastern
## University and the University of Kent.
##
## Distributions of R contain the recommended packages, whose
## authors/contributors are listed in their DESCRIPTION files.
# 每一段的段首缩进5个字符
writeLines(strwrap(x, width = 60, indent = 5))## J. D. Beasley, David J. Best, Richard Brent, Kevin
## Buhr, Michael A. Covington, Bill Cleveland, Robert
## Cleveland,, G. W. Cran, C. G. Ding, Ulrich Drepper, Paul
## Eggert, J. O. Evans, David M. Gay, H. Frick, G. W. Hill,
## Richard H. Jones, Eric Grosse, Shelby Haberman, Bruno
## Haible, John Hartigan, Andrew Harvey, Trevor Hastie, Min
## Long Lam, George Marsaglia, K. J. Martin, Gordon
## Matzigkeit, C. R. Mckenzie, Jean McRae, Cyrus Mehta, Fionn
## Murtagh, John C. Nash, Finbarr O'Sullivan, R. E. Odeh,
## William Patefield, Nitin Patel, Alan Richardson, D. E.
## Roberts, Patrick Royston, Russell Lenth, Ming-Jen Shyu,
## Richard C. Singleton, S. G. Springer, Supoj Sutanthavibul,
## Irma Terpenning, G. E. Thomas, Rob Tibshirani, Wai Wan
## Tsang, Berwin Turlach, Gary V. Vaughan, Michael Wichura,
## Jingbo Wang, M. A. Wong, and the Free Software Foundation
## (for autoconf code and utilities). See also files under
## src/extras.
##
## Many more, too numerous to mention here, have
## contributed by sending bug reports and suggesting various
## improvements.
##
## Simon Davies whilst at the University of Auckland
## wrote the original version of glm().
##
## Julian Harris and Wing Kwong (Tiki) Wan whilst at the
## University of Auckland assisted Ross Ihaka with the
## original Macintosh port.
##
## R was inspired by the S environment which has been
## principally developed by John Chambers, with substantial
## input from Douglas Bates, Rick Becker, Bill Cleveland,
## Trevor Hastie, Daryl Pregibon and Allan Wilks.
##
## A special debt is owed to John Chambers who has
## graciously contributed advice and encouragement in the
## early days of R and later became a member of the core team.
##
## Stefano Iacus (up to 2014, a former member of R Core)
## and Simon Urbanek developed the macOS port, including the
## R.app GUI, toolchains and packaging.
##
## The Windows port was originally developed by Guido
## Masarotto (for a while a member of R Core) and Brian
## Ripley, then further by Duncan Murdoch (a former member of
## R Core) and then Jeroen Ooms (base) and Uwe Ligges
## (packages). Tomas Kalibera is the current main developer
## of the Windows port and provides assistance with package
## porting.
##
## Tomas Kalibera's work has been sponsored by Jan Vitek
## and funded by his European Research Council grant "Evolving
## Language Ecosystems (ELE)".
##
## Computing support (including hardware, hosting and
## infrastructure) has been provided/funded by the R
## Foundation, employers of R-Core members (notably WU Wien,
## ETH Zurich, U Oxford and U Iowa) and by Northeastern
## University and the University of Kent.
##
## Distributions of R contain the recommended packages,
## whose authors/contributors are listed in their DESCRIPTION
## files.
# 除了段首，每一段的余下诸行都缩进5个字符
writeLines(strwrap(x, width = 60, exdent = 5))## J. D. Beasley, David J. Best, Richard Brent, Kevin Buhr,
## Michael A. Covington, Bill Cleveland, Robert
## Cleveland,, G. W. Cran, C. G. Ding, Ulrich Drepper,
## Paul Eggert, J. O. Evans, David M. Gay, H. Frick, G.
## W. Hill, Richard H. Jones, Eric Grosse, Shelby
## Haberman, Bruno Haible, John Hartigan, Andrew Harvey,
## Trevor Hastie, Min Long Lam, George Marsaglia, K. J.
## Martin, Gordon Matzigkeit, C. R. Mckenzie, Jean McRae,
## Cyrus Mehta, Fionn Murtagh, John C. Nash, Finbarr
## O'Sullivan, R. E. Odeh, William Patefield, Nitin
## Patel, Alan Richardson, D. E. Roberts, Patrick
## Royston, Russell Lenth, Ming-Jen Shyu, Richard C.
## Singleton, S. G. Springer, Supoj Sutanthavibul, Irma
## Terpenning, G. E. Thomas, Rob Tibshirani, Wai Wan
## Tsang, Berwin Turlach, Gary V. Vaughan, Michael
## Wichura, Jingbo Wang, M. A. Wong, and the Free
## Software Foundation (for autoconf code and utilities).
## See also files under src/extras.
##
## Many more, too numerous to mention here, have contributed
## by sending bug reports and suggesting various
## improvements.
##
## Simon Davies whilst at the University of Auckland wrote the
## original version of glm().
##
## Julian Harris and Wing Kwong (Tiki) Wan whilst at the
## University of Auckland assisted Ross Ihaka with the
## original Macintosh port.
##
## R was inspired by the S environment which has been
## principally developed by John Chambers, with
## substantial input from Douglas Bates, Rick Becker,
## Bill Cleveland, Trevor Hastie, Daryl Pregibon and
## Allan Wilks.
##
## A special debt is owed to John Chambers who has graciously
## contributed advice and encouragement in the early days
## of R and later became a member of the core team.
##
## Stefano Iacus (up to 2014, a former member of R Core) and
## Simon Urbanek developed the macOS port, including the
## R.app GUI, toolchains and packaging.
##
## The Windows port was originally developed by Guido
## Masarotto (for a while a member of R Core) and Brian
## Ripley, then further by Duncan Murdoch (a former
## member of R Core) and then Jeroen Ooms (base) and Uwe
## Ligges (packages). Tomas Kalibera is the current main
## developer of the Windows port and provides assistance
## with package porting.
##
## Tomas Kalibera's work has been sponsored by Jan Vitek and
## funded by his European Research Council grant
## "Evolving Language Ecosystems (ELE)".
##
## Computing support (including hardware, hosting and
## infrastructure) has been provided/funded by the R
## Foundation, employers of R-Core members (notably WU
## Wien, ETH Zurich, U Oxford and U Iowa) and by
## Northeastern University and the University of Kent.
##
## Distributions of R contain the recommended packages, whose
## authors/contributors are listed in their DESCRIPTION
## files.
# 在输出的每一行前面添加前缀
writeLines(strwrap(x, prefix = "THANKS> "))## THANKS> J. D. Beasley, David J. Best, Richard Brent, Kevin Buhr,
## THANKS> Michael A. Covington, Bill Cleveland, Robert Cleveland,, G. W.
## THANKS> Cran, C. G. Ding, Ulrich Drepper, Paul Eggert, J. O. Evans,
## THANKS> David M. Gay, H. Frick, G. W. Hill, Richard H. Jones, Eric
## THANKS> Grosse, Shelby Haberman, Bruno Haible, John Hartigan, Andrew
## THANKS> Harvey, Trevor Hastie, Min Long Lam, George Marsaglia, K. J.
## THANKS> Martin, Gordon Matzigkeit, C. R. Mckenzie, Jean McRae, Cyrus
## THANKS> Mehta, Fionn Murtagh, John C. Nash, Finbarr O'Sullivan, R. E.
## THANKS> Odeh, William Patefield, Nitin Patel, Alan Richardson, D. E.
## THANKS> Roberts, Patrick Royston, Russell Lenth, Ming-Jen Shyu, Richard
## THANKS> C. Singleton, S. G. Springer, Supoj Sutanthavibul, Irma
## THANKS> Terpenning, G. E. Thomas, Rob Tibshirani, Wai Wan Tsang, Berwin
## THANKS> Turlach, Gary V. Vaughan, Michael Wichura, Jingbo Wang, M. A.
## THANKS> Wong, and the Free Software Foundation (for autoconf code and
## THANKS> utilities). See also files under src/extras.
## THANKS>
## THANKS> Many more, too numerous to mention here, have contributed by
## THANKS> sending bug reports and suggesting various improvements.
## THANKS>
## THANKS> Simon Davies whilst at the University of Auckland wrote the
## THANKS> original version of glm().
## THANKS>
## THANKS> Julian Harris and Wing Kwong (Tiki) Wan whilst at the
## THANKS> University of Auckland assisted Ross Ihaka with the original
## THANKS> Macintosh port.
## THANKS>
## THANKS> R was inspired by the S environment which has been principally
## THANKS> developed by John Chambers, with substantial input from Douglas
## THANKS> Bates, Rick Becker, Bill Cleveland, Trevor Hastie, Daryl
## THANKS> Pregibon and Allan Wilks.
## THANKS>
## THANKS> A special debt is owed to John Chambers who has graciously
## THANKS> contributed advice and encouragement in the early days of R and
## THANKS> later became a member of the core team.
## THANKS>
## THANKS> Stefano Iacus (up to 2014, a former member of R Core) and Simon
## THANKS> Urbanek developed the macOS port, including the R.app GUI,
## THANKS> toolchains and packaging.
## THANKS>
## THANKS> The Windows port was originally developed by Guido Masarotto
## THANKS> (for a while a member of R Core) and Brian Ripley, then further
## THANKS> by Duncan Murdoch (a former member of R Core) and then Jeroen
## THANKS> Ooms (base) and Uwe Ligges (packages). Tomas Kalibera is the
## THANKS> current main developer of the Windows port and provides
## THANKS> assistance with package porting.
## THANKS>
## THANKS> Tomas Kalibera's work has been sponsored by Jan Vitek and
## THANKS> funded by his European Research Council grant "Evolving
## THANKS> Language Ecosystems (ELE)".
## THANKS>
## THANKS> Computing support (including hardware, hosting and
## THANKS> infrastructure) has been provided/funded by the R Foundation,
## THANKS> employers of R-Core members (notably WU Wien, ETH Zurich, U
## THANKS> Oxford and U Iowa) and by Northeastern University and the
## THANKS> University of Kent.
## THANKS>
## THANKS> Distributions of R contain the recommended packages, whose
## THANKS> authors/contributors are listed in their DESCRIPTION files.再举一个烧脑的例子
x <- paste(sapply(
sample(10, 100, replace = TRUE), # 从1-10个数字中有放回的随机抽取100个数
function(x) substring("aaaaaaaaaa", 1, x)
), collapse = " ")
sapply(
10:40,
function(m)
c(target = m, actual = max(nchar(strwrap(x, m))))
)## [,1] [,2] [,3] [,4] [,5] [,6] [,7] [,8] [,9] [,10] [,11] [,12] [,13]
## target 10 11 12 13 14 15 16 17 18 19 20 21 22
## actual 10 10 11 12 13 14 15 16 17 18 19 20 21
## [,14] [,15] [,16] [,17] [,18] [,19] [,20] [,21] [,22] [,23] [,24] [,25]
## target 23 24 25 26 27 28 29 30 31 32 33 34
## actual 22 23 24 25 26 27 28 29 30 31 32 33
## [,26] [,27] [,28] [,29] [,30] [,31]
## target 35 36 37 38 39 40
## actual 34 35 36 37 38 394.14.2 strtrim
strtrim(x, width)strtrim 函数将字符串x修剪到特定的显示宽度，返回的字符串向量的长度等于字符串向量 x 的长度，如果 width 的参数值（它是一个整型向量）的长度小于 x 的，就循环补齐。
## [1] "a" "abcde" "abcdef"4.14.3 strrep
strrep(x, times)以给定的次数重复字符串向量中每个元素的个数，并连接字符串的各个副本
strrep("ABC", 2)## [1] "ABCABC"## [1] "A" "BB" "CCC"
# 创建一个字符串向量，指定每个元素中空格的数量
strrep(" ", 1 : 5)## [1] " " " " " " " " " "4.14.4 trimws
trimws 函数用于移除字符串中的空格，这种空格可以来自制表符、回车符和换行符，位置可以位于字符串的开头或者结尾，which 参数指定空格的大致位置。举例如下
x <- " Some text. "
x## [1] " Some text. "
trimws(x)## [1] "Some text."
trimws(x, "l")## [1] "Some text. "
trimws(x, "r")## [1] " Some text."
shopping_list <- c("apples x4", "bag of flour", "bag of sugar", "milk x2")
stringr::str_replace(string = shopping_list, pattern = "\\d", replacement = "aa")## [1] "apples xaa" "bag of flour" "bag of sugar" "milk xaa"
# https://github.com/hadley/stringb/issues/5
# x is vector
str_replace <- function(x, pattern, fun, ...) {
loc <- gregexpr(pattern, text = x, perl = TRUE)
matches <- regmatches(x, loc)
out <- lapply(matches, fun, ...)
regmatches(x, loc) <- out
x
}
loc <- gregexpr(pattern = "\\d", text = shopping_list, perl = TRUE)
matches = regmatches(x = shopping_list, loc)
matches
out <- lapply(matches, transform, "aa")
regmatches(x = shopping_list, loc) <- out
shopping_list
str_replace(shopping_list, pattern = "\\\\d", replace = "aa")4.14.5 tolower
tolower 和 toupper 是一对，将大写转小写，小写转大写
simpleCap <- function(x) {
x <- tolower(x)
s <- strsplit(x, " ")[[1]]
paste(toupper(substring(s, 1, 1)), substring(s, 2),
sep = "", collapse = " "
)
}
# 参考文献条目里需要将每个英文单词的首字母大写
simpleCap(x = "THE USE OF MULTIPLE MEASUREMENTS IN TAXONOMIC PROBLEMS")## [1] "The Use Of Multiple Measurements In Taxonomic Problems"4.15 字符串加密
字符串编码加密， openssl 包提供了 sha1 函数 15
library(openssl)
encode_mobile <- function(phone_number) paste("*", paste(toupper(sha1(sha1(charToRaw(paste(phone_number, "$1$mobile$", sep = ""))))), collapse = ""), sep = "")
# 随意模拟两个手机号
mobile_vec <- c("18601013453", "13811674545")
sapply(mobile_vec, encode_mobile)## 18601013453
## "*B1D46D1D62C7280137F0E14249EE500865247B7B"
## 13811674545
## "*0554DA6E403491F58F1567DF2EDEB19186B77173"4.16 处理性能
当你对一个很长的字符串进行大量的正则表达式匹配的时候，你需要考虑性能问题了，这时候该考虑启用合适的选项，一般来讲， PCRE 比默认的正则表达式引擎快，fixed=TRUE 可以继续加快匹配速度，特别是当每个模式只匹配少量次数时。
连接字符串，paste/c/bfile/bracket 函数性能比较
https://wch.github.io/string_builder/index.html
R 内置的默认正则表达式匹配方式是基于 PCRE 的匹配，options 控制 PCRE 默认的三个选项 PCRE_limit_recursion=NA 、PCRE_study=10 和 PCRE_use_JIT=TRUE，当前系统环境下 PCRE 的支持情况
## UTF-8 Unicode properties JIT stack
## TRUE TRUE TRUE FALSE查看R环境的 PCRE 配置
## PCRE_limit_recursion PCRE_study PCRE_use_JIT
## NA FALSE TRUE4.17 网络爬虫
用 R 语言写爬虫 curl、httr、 xml2、XML 和 rvest 解析网页16
# 查看 libcurl 库的版本
libcurlVersion()## [1] "7.81.0"
## attr(,"ssl_version")
## [1] "OpenSSL/3.0.2"
## attr(,"libssh_version")
## [1] "libssh/0.9.6/openssl/zlib"
## attr(,"protocols")
## [1] "dict" "file" "ftp" "ftps" "gopher" "gophers" "http"
## [8] "https" "imap" "imaps" "ldap" "ldaps" "mqtt" "pop3"
## [15] "pop3s" "rtmp" "rtsp" "scp" "sftp" "smb" "smbs"
## [22] "smtp" "smtps" "telnet" "tftp"于主编利用 tidyRSS 包 抓取解析博客站点的订阅信息，并将此设置为定时任务，创建自动更新内容的博客聚合网站 Daily R
抓取地震台信息
4.18 文本挖掘
How did Axios rectangle Trump’s PDF schedule? A try with R 使用 pdftools 和 magick 处理表格，这两个 R 包分别依赖 Poppler C++ 和 ImageMagick++，在 Ubuntu 上安装 pdftools 和 magick 包
sudo apt-get install libpoppler-cpp-dev libmagick++-dev
install.packages(c("pdftools", "magick"))除了 pdftools 包外，PDF 文档中表格抽取工具还有 tabulizer。扫描版 PDF 文档需要OCR识别技术支持的 tesseract 包
4.19 运行环境
xfun::session_info()## R version 4.2.2 (2022-10-31)
## Platform: x86_64-pc-linux-gnu (64-bit)
## Running under: Ubuntu 22.04.1 LTS
##
## Locale:
## LC_CTYPE=en_US.UTF-8 LC_NUMERIC=C
## LC_TIME=en_US.UTF-8 LC_COLLATE=en_US.UTF-8
## LC_MONETARY=en_US.UTF-8 LC_MESSAGES=en_US.UTF-8
## LC_PAPER=en_US.UTF-8 LC_NAME=C
## LC_ADDRESS=C LC_TELEPHONE=C
## LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C
##
## Package version:
## askpass_1.1 assertthat_0.2.1 base64enc_0.1.3 bookdown_0.26
## brio_1.1.3 bslib_0.3.1 cachem_1.0.6 cli_3.3.0
## compiler_4.2.2 cpp11_0.4.2 crayon_1.5.1 curl_4.3.2
## DBI_1.1.2 desc_1.4.1 digest_0.6.29 downlit_0.4.0
## dplyr_1.0.9 ellipsis_0.3.2 evaluate_0.15 fansi_1.0.3
## fastmap_1.1.0 fs_1.5.2 generics_0.1.2 glue_1.6.2
## graphics_4.2.2 grDevices_4.2.2 highr_0.9 htmltools_0.5.2
## jquerylib_0.1.4 jsonlite_1.8.0 knitr_1.39 lifecycle_1.0.1
## magrittr_2.0.3 memoise_2.0.1 methods_4.2.2 openssl_2.0.1
## pillar_1.7.0 pkgconfig_2.0.3 purrr_0.3.4 R6_2.5.1
## rappdirs_0.3.3 rlang_1.0.2 rmarkdown_2.14 rprojroot_2.0.3
## sass_0.4.1 stats_4.2.2 stringi_1.7.6 stringr_1.4.0
## sys_3.4 sysfonts_0.8.8 tibble_3.1.7 tidyr_1.2.0
## tidyselect_1.1.2 tinytex_0.39 tools_4.2.2 utf8_1.2.2
## utils_4.2.2 vctrs_0.4.1 xfun_0.31 xml2_1.3.3
## yaml_2.3.5参考文献
[11]
M. P. J. van der Loo, “The stringdist package for approximate string matching,” The R Journal, vol. 6, pp. 111–122, 2014,Available: https://CRAN.R-project.org/package=stringdist