Fri, 17 Mar 2023 11:02:12 +0000
A range of fixes for Lua 5.2 support
1 | 1 | --[[-------------------------------------------------------------------- |
2 | ||
3 | llex.lua: Lua 5.1 lexical analyzer in Lua | |
4 | This file is part of LuaSrcDiet, based on Yueliang material. | |
5 | ||
6 | Copyright (c) 2008 Kein-Hong Man <khman@users.sf.net> | |
7 | The COPYRIGHT file describes the conditions | |
8 | under which this software may be distributed. | |
9 | ||
10 | See the ChangeLog for more information. | |
11 | ||
12 | ----------------------------------------------------------------------]] | |
13 | ||
14 | --[[-------------------------------------------------------------------- | |
15 | -- NOTES: | |
16 | -- * This is a version of the native 5.1.x lexer from Yueliang 0.4.0, | |
17 | -- with significant modifications to handle LuaSrcDiet's needs: | |
18 | -- (1) llex.error is an optional error function handler | |
19 | -- (2) seminfo for strings include their delimiters and no | |
20 | -- translation operations are performed on them | |
21 | -- * ADDED shbang handling has been added to support executable scripts | |
22 | -- * NO localized decimal point replacement magic | |
23 | -- * NO limit to number of lines | |
24 | -- * NO support for compatible long strings (LUA_COMPAT_LSTR) | |
25 | -- * Please read technotes.txt for more technical details. | |
26 | ----------------------------------------------------------------------]] | |
27 | ||
28 | local base = _G | |
29 | local string = require "string" | |
99
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
30 | --module "llex" |
1 | 31 | |
32 | local find = string.find | |
33 | local match = string.match | |
34 | local sub = string.sub | |
35 | ||
99
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
36 | local z = '' -- source |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
37 | local sourceid = '' -- name of source |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
38 | local I = 1 -- lexer's position in source |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
39 | local ln = 1 -- line number |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
40 | local tok = {} -- lexed token list* |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
41 | local seminfo = {} -- lexed semantic information list* |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
42 | local tokln = {} -- line numbers for messages* |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
43 | local buff = '' |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
44 | -- NOTE: see init() for module variables (externally visible): |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
45 | -- tok, seminfo, tokln |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
46 | |
1 | 47 | ---------------------------------------------------------------------- |
48 | -- initialize keyword list, variables | |
49 | ---------------------------------------------------------------------- | |
50 | ||
51 | local kw = {} | |
52 | for v in string.gmatch([[ | |
53 | and break do else elseif end false for function if in | |
54 | local nil not or repeat return then true until while]], "%S+") do | |
55 | kw[v] = true | |
56 | end | |
57 | ||
58 | ||
59 | ---------------------------------------------------------------------- | |
60 | -- add information to token listing | |
61 | ---------------------------------------------------------------------- | |
62 | ||
63 | local function addtoken(token, info) | |
64 | local i = #tok + 1 | |
65 | tok[i] = token | |
66 | seminfo[i] = info | |
67 | tokln[i] = ln | |
68 | end | |
69 | ||
70 | ---------------------------------------------------------------------- | |
71 | -- handles line number incrementation and end-of-line characters | |
72 | ---------------------------------------------------------------------- | |
73 | ||
74 | local function inclinenumber(i, is_tok) | |
75 | local sub = sub | |
76 | local old = sub(z, i, i) | |
77 | i = i + 1 -- skip '\n' or '\r' | |
78 | local c = sub(z, i, i) | |
79 | if (c == "\n" or c == "\r") and (c ~= old) then | |
80 | i = i + 1 -- skip '\n\r' or '\r\n' | |
81 | old = old..c | |
82 | end | |
83 | if is_tok then addtoken("TK_EOL", old) end | |
84 | ln = ln + 1 | |
85 | I = i | |
86 | return i | |
87 | end | |
88 | ||
89 | ---------------------------------------------------------------------- | |
90 | -- initialize lexer for given source _z and source name _sourceid | |
91 | ---------------------------------------------------------------------- | |
92 | ||
93 | function init(_z, _sourceid) | |
94 | z = _z -- source | |
99
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
95 | sourceid = _sourceid -- name of source |
1 | 96 | -------------------------------------------------------------------- |
97 | -- initial processing (shbang handling) | |
98 | -------------------------------------------------------------------- | |
99 | local p, _, q, r = find(z, "^(#[^\r\n]*)(\r?\n?)") | |
100 | if p then -- skip first line | |
101 | I = I + #q | |
102 | addtoken("TK_COMMENT", q) | |
103 | if #r > 0 then inclinenumber(I, true) end | |
104 | end | |
105 | end | |
106 | ||
107 | ---------------------------------------------------------------------- | |
108 | -- returns a chunk name or id, no truncation for long names | |
109 | ---------------------------------------------------------------------- | |
110 | ||
111 | function chunkid() | |
112 | if sourceid and match(sourceid, "^[=@]") then | |
113 | return sub(sourceid, 2) -- remove first char | |
114 | end | |
115 | return "[string]" | |
116 | end | |
117 | ||
118 | ---------------------------------------------------------------------- | |
119 | -- formats error message and throws error | |
120 | -- * a simplified version, does not report what token was responsible | |
121 | ---------------------------------------------------------------------- | |
122 | ||
123 | function errorline(s, line) | |
124 | local e = error or base.error | |
125 | e(string.format("%s:%d: %s", chunkid(), line or ln, s)) | |
126 | end | |
127 | local errorline = errorline | |
128 | ||
129 | ------------------------------------------------------------------------ | |
130 | -- count separators ("=") in a long string delimiter | |
131 | ------------------------------------------------------------------------ | |
132 | ||
133 | local function skip_sep(i) | |
134 | local sub = sub | |
135 | local s = sub(z, i, i) | |
136 | i = i + 1 | |
137 | local count = #match(z, "=*", i) -- note, take the length | |
138 | i = i + count | |
139 | I = i | |
140 | return (sub(z, i, i) == s) and count or (-count) - 1 | |
141 | end | |
142 | ||
143 | ---------------------------------------------------------------------- | |
144 | -- reads a long string or long comment | |
145 | ---------------------------------------------------------------------- | |
146 | ||
147 | local function read_long_string(is_str, sep) | |
148 | local i = I + 1 -- skip 2nd '[' | |
149 | local sub = sub | |
150 | local c = sub(z, i, i) | |
151 | if c == "\r" or c == "\n" then -- string starts with a newline? | |
152 | i = inclinenumber(i) -- skip it | |
153 | end | |
154 | local j = i | |
155 | while true do | |
156 | local p, q, r = find(z, "([\r\n%]])", i) -- (long range) | |
157 | if not p then | |
158 | errorline(is_str and "unfinished long string" or | |
159 | "unfinished long comment") | |
160 | end | |
161 | i = p | |
162 | if r == "]" then -- delimiter test | |
163 | if skip_sep(i) == sep then | |
164 | buff = sub(z, buff, I) | |
165 | I = I + 1 -- skip 2nd ']' | |
166 | return buff | |
167 | end | |
168 | i = I | |
169 | else -- newline | |
170 | buff = buff.."\n" | |
171 | i = inclinenumber(i) | |
172 | end | |
173 | end--while | |
174 | end | |
175 | ||
176 | ---------------------------------------------------------------------- | |
177 | -- reads a string | |
178 | ---------------------------------------------------------------------- | |
179 | ||
180 | local function read_string(del) | |
181 | local i = I | |
182 | local find = find | |
183 | local sub = sub | |
184 | while true do | |
185 | local p, q, r = find(z, "([\n\r\\\"\'])", i) -- (long range) | |
186 | if p then | |
187 | if r == "\n" or r == "\r" then | |
188 | errorline("unfinished string") | |
189 | end | |
190 | i = p | |
191 | if r == "\\" then -- handle escapes | |
192 | i = i + 1 | |
193 | r = sub(z, i, i) | |
194 | if r == "" then break end -- (EOZ error) | |
195 | p = find("abfnrtv\n\r", r, 1, true) | |
196 | ------------------------------------------------------ | |
197 | if p then -- special escapes | |
198 | if p > 7 then | |
199 | i = inclinenumber(i) | |
200 | else | |
201 | i = i + 1 | |
202 | end | |
203 | ------------------------------------------------------ | |
204 | elseif find(r, "%D") then -- other non-digits | |
205 | i = i + 1 | |
206 | ------------------------------------------------------ | |
207 | else -- \xxx sequence | |
208 | local p, q, s = find(z, "^(%d%d?%d?)", i) | |
209 | i = q + 1 | |
210 | if s + 1 > 256 then -- UCHAR_MAX | |
211 | errorline("escape sequence too large") | |
212 | end | |
213 | ------------------------------------------------------ | |
214 | end--if p | |
215 | else | |
216 | i = i + 1 | |
217 | if r == del then -- ending delimiter | |
218 | I = i | |
219 | return sub(z, buff, i - 1) -- return string | |
220 | end | |
221 | end--if r | |
222 | else | |
223 | break -- (error) | |
224 | end--if p | |
225 | end--while | |
226 | errorline("unfinished string") | |
227 | end | |
228 | ||
229 | ------------------------------------------------------------------------ | |
230 | -- main lexer function | |
231 | ------------------------------------------------------------------------ | |
232 | ||
233 | function llex() | |
234 | local find = find | |
235 | local match = match | |
236 | while true do--outer | |
237 | local i = I | |
238 | -- inner loop allows break to be used to nicely section tests | |
239 | while true do--inner | |
240 | ---------------------------------------------------------------- | |
241 | local p, _, r = find(z, "^([_%a][_%w]*)", i) | |
242 | if p then | |
243 | I = i + #r | |
244 | if kw[r] then | |
245 | addtoken("TK_KEYWORD", r) -- reserved word (keyword) | |
246 | else | |
247 | addtoken("TK_NAME", r) -- identifier | |
248 | end | |
249 | break -- (continue) | |
250 | end | |
251 | ---------------------------------------------------------------- | |
252 | local p, _, r = find(z, "^(%.?)%d", i) | |
253 | if p then -- numeral | |
254 | if r == "." then i = i + 1 end | |
255 | local _, q, r = find(z, "^%d*[%.%d]*([eE]?)", i) | |
256 | i = q + 1 | |
257 | if #r == 1 then -- optional exponent | |
258 | if match(z, "^[%+%-]", i) then -- optional sign | |
259 | i = i + 1 | |
260 | end | |
261 | end | |
262 | local _, q = find(z, "^[_%w]*", i) | |
263 | I = q + 1 | |
264 | local v = sub(z, p, q) -- string equivalent | |
265 | if not base.tonumber(v) then -- handles hex test also | |
266 | errorline("malformed number") | |
267 | end | |
268 | addtoken("TK_NUMBER", v) | |
269 | break -- (continue) | |
270 | end | |
271 | ---------------------------------------------------------------- | |
272 | local p, q, r, t = find(z, "^((%s)[ \t\v\f]*)", i) | |
273 | if p then | |
274 | if t == "\n" or t == "\r" then -- newline | |
275 | inclinenumber(i, true) | |
276 | else | |
277 | I = q + 1 -- whitespace | |
278 | addtoken("TK_SPACE", r) | |
279 | end | |
280 | break -- (continue) | |
281 | end | |
282 | ---------------------------------------------------------------- | |
283 | local r = match(z, "^%p", i) | |
284 | if r then | |
285 | buff = i | |
286 | local p = find("-[\"\'.=<>~", r, 1, true) | |
287 | if p then | |
288 | -- two-level if block for punctuation/symbols | |
289 | -------------------------------------------------------- | |
290 | if p <= 2 then | |
291 | if p == 1 then -- minus | |
292 | local c = match(z, "^%-%-(%[?)", i) | |
293 | if c then | |
294 | i = i + 2 | |
295 | local sep = -1 | |
296 | if c == "[" then | |
297 | sep = skip_sep(i) | |
298 | end | |
299 | if sep >= 0 then -- long comment | |
300 | addtoken("TK_LCOMMENT", read_long_string(false, sep)) | |
301 | else -- short comment | |
302 | I = find(z, "[\n\r]", i) or (#z + 1) | |
303 | addtoken("TK_COMMENT", sub(z, buff, I - 1)) | |
304 | end | |
305 | break -- (continue) | |
306 | end | |
307 | -- (fall through for "-") | |
308 | else -- [ or long string | |
309 | local sep = skip_sep(i) | |
310 | if sep >= 0 then | |
311 | addtoken("TK_LSTRING", read_long_string(true, sep)) | |
312 | elseif sep == -1 then | |
313 | addtoken("TK_OP", "[") | |
314 | else | |
315 | errorline("invalid long string delimiter") | |
316 | end | |
317 | break -- (continue) | |
318 | end | |
319 | -------------------------------------------------------- | |
320 | elseif p <= 5 then | |
321 | if p < 5 then -- strings | |
322 | I = i + 1 | |
323 | addtoken("TK_STRING", read_string(r)) | |
324 | break -- (continue) | |
325 | end | |
326 | r = match(z, "^%.%.?%.?", i) -- .|..|... dots | |
327 | -- (fall through) | |
328 | -------------------------------------------------------- | |
329 | else -- relational | |
330 | r = match(z, "^%p=?", i) | |
331 | -- (fall through) | |
332 | end | |
333 | end | |
334 | I = i + #r | |
335 | addtoken("TK_OP", r) -- for other symbols, fall through | |
336 | break -- (continue) | |
337 | end | |
338 | ---------------------------------------------------------------- | |
339 | local r = sub(z, i, i) | |
340 | if r ~= "" then | |
341 | I = i + 1 | |
342 | addtoken("TK_OP", r) -- other single-char tokens | |
343 | break | |
344 | end | |
345 | addtoken("TK_EOS", "") -- end of stream, | |
346 | return -- exit here | |
347 | ---------------------------------------------------------------- | |
348 | end--while inner | |
349 | end--while outer | |
350 | end | |
351 | ||
99
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
352 | return { |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
353 | llex = llex, |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
354 | init = init, |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
355 | chunkid = chunkid, |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
356 | tok = tok, |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
357 | tokln = tokln, |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
358 | seminfo = seminfo |
2b6416334a25
A range of fixes for Lua 5.2 support
Matthew Wild <mwild1@gmail.com>
parents:
85
diff
changeset
|
359 | } |