TopolearnText processing is a daily task, and Lua patterns look like regex but behave differently.
All string functions live in the string table, and strings have it as their metatable, so s:upper() is the same as string.upper(s). The colon passes s as the first argument.
Indexes are 1-based and negative indexes count from the end: s:sub(-3) is the last three bytes. string.format works like C's printf (%d, %5.2f, %s, %q for a quoted literal). s:rep(n), s:reverse() and s:byte()/string.char round out the basics.
Patterns are Lua's lightweight alternative to regular expressions. Classes are written with %: %d digit, %a letter, %s space, %w alphanumeric, and the uppercase version negates. Quantifiers are *, +, ? and - (lazy *). Captures use parentheses. There is no alternation (|) and no {n,m}.
s:find(p)returns the start and end index, plus captures.s:match(p)returns the captures (or the whole match).s:gmatch(p)iterates over all matches in aforloop.s:gsub(p, repl)replaces;replcan be a string with%1, a table or a function.
Because % escapes magic characters, write %. for a literal dot. Pass true as the fourth argument of find for a plain substring search.
Strings are bytes: for UTF-8 text use the utf8 library (utf8.len, utf8.codes).
Example
local s = "Lua 5.4 was released in 2020"
print(s:upper():sub(1, 3)) --> LUA
print(s:find("%d+")) --> 5 5
print(s:match("(%d+)%.(%d+)")) --> 5 4
for word in s:gmatch("%a+") do io.write(word, ",") end
print() --> Lua,was,released,in,
local out, n = s:gsub("%d+", function(d) return "<" .. d .. ">" end)
print(out, n) --> Lua <5>.<4> was released in <2020> 3
print(("key = value"):match("^(%w+)%s*=%s*(%w+)$")) --> key value
print(string.format("%-5s|%6.2f|%q", "ab", math.pi, "hi")) --> ab | 3.14|"hi"
print(#"héllo", utf8.len("héllo")) --> 6 5Remember
s:method()callsstring.method(s)- Patterns use
%classes (%d,%a,%s) and have no alternation gsubaccepts a string, table or function as the replacement- Use
utf8functions for character counts in UTF-8 text
Watch out for
- Using
.or-unescaped when you mean the literal character; write%.and%-.