| 1 | #!/usr/bin/env bash
|
| 2 | #
|
| 3 | # Lexing / Parsing experiment
|
| 4 | #
|
| 5 | # Usage:
|
| 6 | # doctools/micro-syntax.sh <function name>
|
| 7 |
|
| 8 | # TODO:
|
| 9 | # - Rename to micro-syntax, from micro-grammars and uchex?
|
| 10 | # - micro-segmenting and lexing - comments, strings, and maybe { }
|
| 11 | # - micro-parsing: for indent/dedent
|
| 12 | #
|
| 13 | # - use GNU long flags, test them
|
| 14 |
|
| 15 | # C++
|
| 16 | #
|
| 17 | # - ANSI should cat all argv, and it should print line numbers
|
| 18 | # - HTML string can append with with netstrings!
|
| 19 | # - (path, html, path, html, ...) should be sufficient, though not fully general
|
| 20 | # - print SLOC at the top
|
| 21 | # - COALESCE tokens to save space
|
| 22 |
|
| 23 | # Then src-tree reads this stream
|
| 24 | # - actually it can take the filenames directly from here
|
| 25 | # - it can discard the big HTML!
|
| 26 |
|
| 27 | # Later: port some kind of parser combinator for
|
| 28 | # - def class, etc.
|
| 29 |
|
| 30 | set -o nounset
|
| 31 | set -o pipefail
|
| 32 | set -o errexit
|
| 33 |
|
| 34 | REPO_ROOT=$(cd "$(dirname $0)/.."; pwd) # tsv-lib.sh uses this
|
| 35 |
|
| 36 | #source build/dev-shell.sh # 're2c' in path
|
| 37 | source build/ninja-rules-cpp.sh
|
| 38 |
|
| 39 | my-re2c() {
|
| 40 | local in=$1
|
| 41 | local out=$2
|
| 42 |
|
| 43 | # Copied from build/py.sh, and added --tags
|
| 44 | re2c --tags -W -Wno-match-empty-string -Werror -o $out $in
|
| 45 | }
|
| 46 |
|
| 47 | readonly BASE_DIR=_tmp/micro-syntax
|
| 48 |
|
| 49 | build() {
|
| 50 | local variant=${1:-asan}
|
| 51 |
|
| 52 | case $variant in
|
| 53 | asan)
|
| 54 | cxxflags='-O0 -fsanitize=address'
|
| 55 | ;;
|
| 56 | opt)
|
| 57 | cxxflags='-O2'
|
| 58 | ;;
|
| 59 | *)
|
| 60 | die "Invalid variant $variant"
|
| 61 | ;;
|
| 62 | esac
|
| 63 |
|
| 64 | mkdir -p $BASE_DIR
|
| 65 |
|
| 66 | local cc=doctools/micro_syntax.cc
|
| 67 | local h=$BASE_DIR/micro_syntax.h
|
| 68 | local bin=$BASE_DIR/micro_syntax
|
| 69 |
|
| 70 | my-re2c doctools/micro_syntax.re2c.h $h
|
| 71 |
|
| 72 | # Note: with cc, you need gnu99 instead of c99 for fdopen() and getline()
|
| 73 |
|
| 74 | # g++ - otherwise virtual functions don't work!
|
| 75 |
|
| 76 | set -o xtrace
|
| 77 | g++ -std=c++11 -Wall -I $BASE_DIR $cxxflags \
|
| 78 | -o $bin $cc
|
| 79 | set +o xtrace
|
| 80 |
|
| 81 | strip -o $bin.stripped $bin
|
| 82 |
|
| 83 | log " CXX $cc"
|
| 84 |
|
| 85 | }
|
| 86 |
|
| 87 | readonly -a PY_TESTS=(
|
| 88 | 'abc' '""'
|
| 89 | '"dq \" backslash \\"' '"missing '
|
| 90 | "'sq \\' backslash \\\\'"
|
| 91 | '"line\n"' '"quote \" backslash \\ "'
|
| 92 | '"\n"'
|
| 93 | 'hi # comment'
|
| 94 | '"hi" # comment'
|
| 95 | '(r"raw dq")'
|
| 96 | "(r'raw \\' sq')"
|
| 97 |
|
| 98 | ' "L1" # first
|
| 99 | L2 # second'
|
| 100 |
|
| 101 | ' def f():
|
| 102 | """docstring
|
| 103 | with "quote"
|
| 104 | """
|
| 105 | pass'
|
| 106 |
|
| 107 | " def f():
|
| 108 | '''docstring
|
| 109 | with 'quote'
|
| 110 | '''
|
| 111 | pass"
|
| 112 |
|
| 113 | " print(r'''hello''')"
|
| 114 | ' print(r"""hi there""")'
|
| 115 |
|
| 116 | '"hi" # comment'
|
| 117 | )
|
| 118 |
|
| 119 | readonly -a CPP_TESTS=(
|
| 120 | '#if 0'
|
| 121 | 'not prepreproc #ifdef 0'
|
| 122 | "// comment can't "
|
| 123 | "f(); // comment isn't "
|
| 124 |
|
| 125 | # Char literal in C
|
| 126 | "'\\''"
|
| 127 |
|
| 128 | 'void f(); /* multi-line
|
| 129 | comment
|
| 130 | */
|
| 131 | void g(int x);'
|
| 132 |
|
| 133 | '#include "foo.h"'
|
| 134 | '#include <foo.h> // comment'
|
| 135 |
|
| 136 | '#define X 3 // comment
|
| 137 | int g();'
|
| 138 |
|
| 139 | '// hello
|
| 140 | #include <stdio.h>
|
| 141 | #define SUM(x, y) \
|
| 142 | (x) + \
|
| 143 | (y) // comment
|
| 144 | void f();'
|
| 145 |
|
| 146 | '#undef x'
|
| 147 |
|
| 148 | '#define F(x) x##name'
|
| 149 |
|
| 150 | 'char* s = f(R"(one
|
| 151 | two
|
| 152 | three)");
|
| 153 | '
|
| 154 |
|
| 155 | 'char* s = f(R"zzXX(hi
|
| 156 | world
|
| 157 | )zzX" (not the end)
|
| 158 | )zzXX");
|
| 159 | '
|
| 160 |
|
| 161 | 'char* unclosed = f(R"zzXX(hi
|
| 162 | world
|
| 163 | )oops");
|
| 164 | '
|
| 165 | )
|
| 166 |
|
| 167 | readonly -a SHELL_TESTS=(
|
| 168 | "echo $'multi \\n
|
| 169 | sq \\' line'"
|
| 170 |
|
| 171 | # Quoted backslash
|
| 172 | "echo hi \\' there"
|
| 173 |
|
| 174 | 'echo one#two'
|
| 175 | 'echo $(( 16#ff ))'
|
| 176 |
|
| 177 | '# comment'
|
| 178 | '### comment'
|
| 179 |
|
| 180 | 'echo one # comment'
|
| 181 |
|
| 182 | 'cat <<EOF
|
| 183 | hello $world
|
| 184 | EOF'
|
| 185 |
|
| 186 | 'cat <<- "EOF"
|
| 187 | $3.99
|
| 188 | EOF '
|
| 189 |
|
| 190 | 'cat <<- \_ACAWK
|
| 191 | $3.99
|
| 192 | more
|
| 193 | _ACAWK
|
| 194 | echo yo'
|
| 195 |
|
| 196 | 'echo multiple << EOF1 << EOF2 > out
|
| 197 | one
|
| 198 | EOF1
|
| 199 | ...
|
| 200 | two
|
| 201 | EOF2
|
| 202 | echo done'
|
| 203 | )
|
| 204 |
|
| 205 | readonly -a R_TESTS=(
|
| 206 | 'f() # hello'
|
| 207 | 'x = f("1
|
| 208 | 2 \"quote\"
|
| 209 | 3")'
|
| 210 |
|
| 211 | "x = f('1
|
| 212 | 2
|
| 213 | 3')"
|
| 214 | )
|
| 215 |
|
| 216 | readonly -a HTML_TESTS=(
|
| 217 | '<p>hi</p>'
|
| 218 | 'hi <br/>'
|
| 219 | '<img src="foo"/>'
|
| 220 | '<a href=foo>link</a>'
|
| 221 | )
|
| 222 |
|
| 223 | run-cases() {
|
| 224 | local lang=$1
|
| 225 | shift
|
| 226 |
|
| 227 | local bin=$BASE_DIR/micro_syntax
|
| 228 |
|
| 229 | for s in "$@"; do
|
| 230 | echo "==== $s"
|
| 231 | echo "$s" | $bin -l $lang
|
| 232 | echo
|
| 233 | done
|
| 234 | }
|
| 235 |
|
| 236 | test-shell() {
|
| 237 | build # TODO: use Ninja
|
| 238 | run-cases shell "${SHELL_TESTS[@]}"
|
| 239 | }
|
| 240 |
|
| 241 | test-cpp() {
|
| 242 | build
|
| 243 | run-cases cpp "${CPP_TESTS[@]}"
|
| 244 | }
|
| 245 |
|
| 246 | test-py() {
|
| 247 | build
|
| 248 | run-cases py "${PY_TESTS[@]}"
|
| 249 | }
|
| 250 |
|
| 251 | test-R() {
|
| 252 | build
|
| 253 | run-cases R "${R_TESTS[@]}"
|
| 254 | }
|
| 255 |
|
| 256 | test-html() {
|
| 257 | build
|
| 258 | run-cases html "${HTML_TESTS[@]}"
|
| 259 | }
|
| 260 |
|
| 261 |
|
| 262 | run-tests() {
|
| 263 | local bin=$BASE_DIR/micro_syntax
|
| 264 |
|
| 265 | build
|
| 266 |
|
| 267 | run-cases shell "${SHELL_TESTS[@]}"
|
| 268 | run-cases cpp "${CPP_TESTS[@]}"
|
| 269 | run-cases py "${PY_TESTS[@]}"
|
| 270 | run-cases R "${R_TESTS[@]}"
|
| 271 | run-cases html "${HTML_TESTS[@]}"
|
| 272 |
|
| 273 | # No language specified
|
| 274 | echo '==== No language'
|
| 275 | head $0 | $bin
|
| 276 | echo
|
| 277 |
|
| 278 | echo '/dev/null'
|
| 279 | $bin < /dev/null
|
| 280 | }
|
| 281 |
|
| 282 | cpp-self() {
|
| 283 | build
|
| 284 | cat doctools/micro_syntax.{re2c.h,cc} | $BASE_DIR/micro_syntax -l cpp | less -r
|
| 285 | }
|
| 286 |
|
| 287 | sh-self() {
|
| 288 | build
|
| 289 | #$BASE_DIR/micro_syntax -l shell < doctools/micro_syntax.sh | less -r
|
| 290 |
|
| 291 | $BASE_DIR/micro_syntax -l shell doctools/micro-syntax.sh
|
| 292 | }
|
| 293 |
|
| 294 | lexer-def() {
|
| 295 | ### Test on a hard Python file
|
| 296 |
|
| 297 | build
|
| 298 | $BASE_DIR/micro_syntax -l py < frontend/lexer_def.py | less -r
|
| 299 | }
|
| 300 |
|
| 301 | git-comp() {
|
| 302 | ### Test on a hard shell file
|
| 303 |
|
| 304 | # Exposes nested double quote issue
|
| 305 | build
|
| 306 | $BASE_DIR/micro_syntax -l shell < testdata/completion/git | less -r
|
| 307 | }
|
| 308 |
|
| 309 | mycpp-runtime() {
|
| 310 | build
|
| 311 | cat mycpp/gc_str.* | $BASE_DIR/micro_syntax -l cpp | less -r
|
| 312 | }
|
| 313 |
|
| 314 | count() {
|
| 315 | wc -l doctools/micro_syntax*
|
| 316 | echo
|
| 317 | wc -l $BASE_DIR/*.h
|
| 318 | echo
|
| 319 | ls -l --si -h $BASE_DIR
|
| 320 | }
|
| 321 |
|
| 322 | test-usage() {
|
| 323 | build
|
| 324 |
|
| 325 | # help
|
| 326 | $BASE_DIR/micro_syntax -h
|
| 327 |
|
| 328 | echo 'ANSI'
|
| 329 | echo 'echo "hi $name"' | $BASE_DIR/micro_syntax -l shell
|
| 330 | echo
|
| 331 |
|
| 332 | echo 'WEB'
|
| 333 | echo 'echo "hi $name"' | $BASE_DIR/micro_syntax -l shell -w
|
| 334 | echo
|
| 335 |
|
| 336 | set -x
|
| 337 | echo 'TSV'
|
| 338 | echo 'echo "hi $name"' | $BASE_DIR/micro_syntax -l shell -t
|
| 339 |
|
| 340 | echo
|
| 341 | echo
|
| 342 | echo '"dq"' | $BASE_DIR/micro_syntax -l shell
|
| 343 |
|
| 344 | $BASE_DIR/micro_syntax -l shell configure | wc -l
|
| 345 |
|
| 346 | # TODO: need a nicer pattern for this test
|
| 347 | set +o errexit
|
| 348 | $BASE_DIR/micro_syntax -l shell _nonexistent_ZZ
|
| 349 | local status=$?
|
| 350 | if test $status -ne 1; then
|
| 351 | die 'Expected status 1'
|
| 352 | fi
|
| 353 | set -o errexit
|
| 354 | }
|
| 355 |
|
| 356 | soil-run() {
|
| 357 | test-usage
|
| 358 | echo
|
| 359 |
|
| 360 | run-tests
|
| 361 | }
|
| 362 |
|
| 363 | ### Shell Tests
|
| 364 |
|
| 365 | here-doc-syntax() {
|
| 366 | ### Test here doc syntax with $0 sh-self
|
| 367 |
|
| 368 | echo 42 > _tmp/42.txt
|
| 369 |
|
| 370 | # _tmp/42 and - are arguments to cat! Vim doesn't understand
|
| 371 | # and >_tmp/here.txt is not part of the here doc
|
| 372 |
|
| 373 | cat <<EOF _tmp/42.txt - >_tmp/here.txt
|
| 374 | x
|
| 375 | short
|
| 376 | hello there
|
| 377 | EOF
|
| 378 |
|
| 379 | cat _tmp/here.txt
|
| 380 | }
|
| 381 |
|
| 382 | "$@"
|
| 383 |
|