# pypy / pypy / doc / rlib.rst

 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Dario Bertini 5d78885 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 3523202 2011-04-26 David Malcolm 1e46012 2011-03-14 Dario Bertini 74b4a53 2011-04-25 David Malcolm 1e46012 2011-03-14 Carl Friedrich B… 07acca3 2011-04-25 Carl Friedrich B… 541308f 2011-04-27  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 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 ================================================= Generally Useful RPython Modules ================================================= .. _Python: http://www.python.org/dev/doc/maint24/ref/ref.html .. contents:: This page lists some of the modules in pypy/rlib_ together with some hints for what they can be used for. The modules here will make up some general library useful for RPython programs (since most of the standard library modules are not RPython). Most of these modules are somewhat rough still and are likely to change at some point. Usually it is useful to look at the tests in pypy/rlib/test_ to get an impression of how to use a module. listsort ============ The pypy/rlib/listsort.py_ module contains an implementation of the timsort sorting algorithm (the sort method of lists is not RPython). To use it, subclass from the listsort.TimSort class and override the lt method to change the comparison behaviour. The constructor of TimSort takes a list as an argument, which will be sorted in place when the sort method of the TimSort instance is called. **Warning:** currently only one type of list can be sorted using the listsort module in one program, otherwise the annotator will be confused. nonconst ============ The pypy/rlib/nonconst.py_ module is useful mostly for tests. The flow object space_ and the annotator_ do quite some constant folding, which is sometimes not desired in a test. To prevent constant folding on a certain value, use the NonConst class. The constructor of NonConst takes an arbitrary value. The instance of NonConst will behave during annotation like that value, but no constant folding will happen. .. _flow object space: objspace.html#the-flow-object-space .. _annotator: translation.html#the-annotation-pass objectmodel =============== The pypy/rlib/objectmodel.py_ module is a mixed bag of various functionality. Some of the more useful ones are: ComputedIntSymbolic: Instances of ComputedIntSymbolic are treated like integers of unknown value by the annotator. The value is determined by a no-argument function (which needs to be passed into the constructor of the class). When the backend emits code, the function is called to determine the value. CDefinedIntSymbolic: Instances of ComputedIntSymbolic are also treated like integers of unknown value by the annotator. When C code is emitted they will be represented by the attribute expr of the symbolic (which is also the first argument of the constructor). r_dict: An RPython dict-like object. The constructor of r_dict takes two functions: key_eq and key_hash which are used for comparing and hashing the entries in the dictionary. instantiate(cls): Instantiate class cls without calling __init__. we_are_translated(): This function returns False when run on top of CPython, but the annotator thinks its return value is True. Therefore it can be used to do different things on top of CPython than after translation. This should be used extremely sparingly (mostly for optimizations or debug code). cast_object_to_weakaddress(obj): Returns a sort of "weak reference" to obj, just without any convenience. The weak address that it returns is not invalidated if the object dies, so you need to take care yourself to know when the object dies. Use with extreme care. cast_weakadress_to_object(obj): Inverse of the previous function. If the object died then a segfault will ensue. UnboxedValue: This is a class which should be used as a base class for a class which carries exactly one integer field. The class should have __slots__ with exactly one entry defined. After translation, instances of this class won't be allocated but represented by *tagged pointers**, that is pointers that have the lowest bit set. rarithmetic =============== The pypy/rlib/rarithmetic.py_ module contains functionality to handle the small differences in the behaviour of arithmetic code in regular Python and RPython code. Most of them are already described in the coding guide_ .. _coding guide: coding-guide.html rbigint =========== The pypy/rlib/rbigint.py_ module contains a full RPython implementation of the Python long type (which itself is not supported in RPython). The rbigint class contains that implementation. To construct rbigint instances use the static methods fromint, frombool, fromfloat and fromdecimalstr. To convert back to other types use the methods toint, tobool, touint and tofloat. Since RPython does not support operator overloading, all the special methods of rbigint that would normally start and end with "__" have these underscores left out for better readability (so a.add(b) can be used to add two rbigint instances). rrandom =========== The pypy/rlib/rrandom.py_ module contains an implementation of the mersenne twister random number generator. It contains one class Random which most importantly has a random method which returns a pseudo-random floating point number between 0.0 and 1.0. rsocket =========== The pypy/rlib/rsocket.py_ module contains an RPython implementation of the functionality of the socket standard library with a slightly different interface. The difficulty with the Python socket API is that addresses are not "well-typed" objects: depending on the address family they are tuples, or strings, and so on, which is not suitable for RPython. Instead, rsocket contains a hierarchy of Address classes, in a typical static-OO-programming style. streamio ============ The pypy/rlib/streamio.py_ contains an RPython stream I/O implementation (which was started by Guido van Rossum as sio.py_ in the CPython sandbox as a prototype for the upcoming new file implementation in Python 3000). .. _sio.py: http://svn.python.org/view/sandbox/trunk/sio/sio.py unroll ========== The pypy/rlib/unroll.py_ module most importantly contains the function unrolling_iterable which wraps an iterator. Looping over the iterator in RPython code will not produce a loop in the resulting flow graph but will unroll the loop instead. parsing =========== The pypy/rlib/parsing/_ module is a still in-development module to generate tokenizers and parsers in RPython. It is still highly experimental and only really used by the Prolog interpreter_ (although in slightly non-standard ways). The easiest way to specify a tokenizer/grammar is to write it down using regular expressions and simple EBNF format. The regular expressions are implemented using finite automatons. The parsing engine uses packrat parsing_, which has O(n) parsing time but is more powerful than LL(n) and LR(n) grammars. .. _packrat parsing: http://pdos.csail.mit.edu/~baford/packrat/ Regular Expressions ------------------- The regular expression syntax is mostly a subset of the syntax of the re_ module. By default, non-special characters match themselves. If you concatenate regular expressions the result will match the concatenation of strings matched by the single regular expressions. | R|S matches any string that *either* matches R or matches S. * R* matches 0 or more repetitions of R. + R+ matches 1 or more repetitions of R. ? R? matches 0 or 1 repetition of R. (...) Parenthesis can be used to group regular expressions (note that in contrast to Python's re module you cannot later match the content of this group). {m} R{m} matches exactly m repetitions of R. {m, n} R{m, n} matches between m and n repetitions of R (including m and n). [] Matches a set of characters. The characters to be matched can be listed sequentially. A range of characters can be specified using -. For examples [ac-eg] matches the characters a, c, d, e and g. The whole set can be inverted by starting it with ^. So [^a] matches anything except a. To parse a regular expression and to get a matcher for it, you can use the function make_runner(s) in the pypy.rlib.parsing.regexparse module. It returns a object with a recognize(input) method that returns True or False depending on whether input matches the string or not. .. _re: http://docs.python.org/library/re.html EBNF ---- To describe a tokenizer and a grammar the pypy.rlib.parsing.ebnfparse defines a syntax for doing that. The syntax file contains a sequence or rules. Every rule either describes a regular expression or a grammar rule. Regular expressions rules have the form:: NAME: "regex"; NAME is the name of the token that the regular expression produces (it has to consist of upper-case letters), regex is a regular expression with the syntax described above. One token name is special-cased: a token called IGNORE will be filtered out of the token stream before being passed on to the parser and can thus be used to match comments or non-significant whitespace. Grammar rules have the form:: name: expansion_1 | expansion_2 | ... | expansion_n; Where expansion_i is a sequence of nonterminal or token names:: symbol_1 symbol_2 symbol_3 ... symbol_n This means that the nonterminal symbol name (which has to consist of lower-case letters) can be expanded into any of the expansions. The expansions can consist of a sequence of token names, nonterminal names or literals, which are strings in quotes that are matched literally. An example to make this clearer:: IGNORE: " "; DECIMAL: "0|[1-9][0-9]*"; additive: multitive "+" additive | multitive; multitive: primary "*" multitive | primary; primary: "(" additive ")" | DECIMAL; This grammar describes the syntax of arithmetic impressions involving addition and multiplication. The tokenizer produces a stream of either DECIMAL tokens or tokens that have matched one of the literals "+", "*", "(" or ")". Any space will be ignored. The grammar produces a syntax tree that follows the precedence of the operators. For example the expression 12 + 4 * 5 is parsed into the following tree: .. graphviz:: digraph G{ "-1213931828" [label="additive"]; "-1213931828" -> "-1213951956"; "-1213951956" [label="multitive"]; "-1213951956" -> "-1213949172"; "-1213949172" [label="primary"]; "-1213949172" -> "-1213949812"; "-1213949812" [shape=box,label="DECIMAL\l'12'"]; "-1213931828" -> "-1213935220"; "-1213935220" [shape=box,label="__0_+\l'+'"]; "-1213931828" -> "-1213951316"; "-1213951316" [label="additive"]; "-1213951316" -> "-1213948180"; "-1213948180" [label="multitive"]; "-1213948180" -> "-1213951380"; "-1213951380" [label="primary"]; "-1213951380" -> "-1213951508"; "-1213951508" [shape=box,label="DECIMAL\l'4'"]; "-1213948180" -> "-1213948788"; "-1213948788" [shape=box,label="__1_*\l'*'"]; "-1213948180" -> "-1213951060"; "-1213951060" [label="multitive"]; "-1213951060" -> "-1213948980"; "-1213948980" [label="primary"]; "-1213948980" -> "-1213950420"; "-1213950420" [shape=box,label="DECIMAL\l'5'"]; } Parse Trees ----------- The parsing process builds up a tree consisting of instances of Symbol and Nonterminal, the former corresponding to tokens, the latter to nonterminal symbols. Both classes live in the pypy/rlib/parsing/tree.py_ module. You can use the view() method Nonterminal instances to get a pygame view of the parse tree. Symbol instances have the following attributes: symbol, which is the name of the token and additional_info which is the matched source. Nonterminal instances have the following attributes: symbol is the name of the nonterminal and children which is a list of the children attributes. Visitors ++++++++ To write tree visitors for the parse trees that are RPython, there is a special baseclass RPythonVisitor in pypy/rlib/parsing/tree.py_ to use. If your class uses this, it will grow a dispatch(node) method, that calls an appropriate visit_ method, depending on the node argument. Here the is replaced by the symbol attribute of the visited node. For the visitor to be RPython, the return values of all the visit methods need to be of the same type. Tree Transformations -------------------- As the tree of arithmetic example above shows, by default the parse tree contains a lot of nodes that are not really conveying useful information. To get rid of some of them, there is some support in the grammar format to automatically create a visitor that transforms the tree to remove the additional nodes. The simplest such transformation just removes nodes, but there are more complex ones. The syntax for these transformations is to enclose symbols in expansions of a nonterminal by [...], <...> or >...<. [symbol_1 symbol_2 ... symbol_n] ++++++++++++++++++++++++++++++++ This will produce a transformer that completely removes the enclosed symbols from the tree. Example:: IGNORE: " "; n: "A" [","] n | "A"; Parsing the string "A, A, A" gives the tree: .. graphviz:: digraph G{ "-1213678004" [label="n"]; "-1213678004" -> "-1213681108"; "-1213681108" [shape=box,label="__0_A\n'A'"]; "-1213678004" -> "-1213681332"; "-1213681332" [shape=box,label="__1_,\n','"]; "-1213678004" -> "-1213837780"; "-1213837780" [label="n"]; "-1213837780" -> "-1213837716"; "-1213837716" [shape=box,label="__0_A\n'A'"]; "-1213837780" -> "-1213839476"; "-1213839476" [shape=box,label="__1_,\n','"]; "-1213837780" -> "-1213839956"; "-1213839956" [label="n"]; "-1213839956" -> "-1213840948"; "-1213840948" [shape=box,label="__0_A\n'A'"]; } After transformation the tree has the "," nodes removed: .. graphviz:: digraph G{ "-1219325716" [label="n"]; "-1219325716" -> "-1219325844"; "-1219325844" [shape=box,label="__0_A\n'A'"]; "-1219325716" -> "-1219324372"; "-1219324372" [label="n"]; "-1219324372" -> "-1219325524"; "-1219325524" [shape=box,label="__0_A\n'A'"]; "-1219324372" -> "-1219324308"; "-1219324308" [label="n"]; "-1219324308" -> "-1219325492"; "-1219325492" [shape=box,label="__0_A\n'A'"]; } ++++++++ This will replace the parent with symbol. Every expansion can contain at most one symbol that is enclosed by <...>, because the parent can only be replaced once, obviously. Example:: IGNORE: " "; n: "a" "b" "c" m; m: "(" ")" | "d"; Parsing the string "a b c (a b c d)" gives the tree: .. graphviz:: digraph G{ "-1214029460" [label="n"]; "-1214029460" -> "-1214026452"; "-1214026452" [shape=box,label="__0_a\n'a'"]; "-1214029460" -> "-1214028276"; "-1214028276" [shape=box,label="__1_b\n'b'"]; "-1214029460" -> "-1214027316"; "-1214027316" [shape=box,label="__2_c\n'c'"]; "-1214029460" -> "-1214026868"; "-1214026868" [label="m"]; "-1214026868" -> "-1214140436"; "-1214140436" [shape=box,label="__3_(\n'('"]; "-1214026868" -> "-1214143508"; "-1214143508" [label="n"]; "-1214143508" -> "-1214141364"; "-1214141364" [shape=box,label="__0_a\n'a'"]; "-1214143508" -> "-1214141748"; "-1214141748" [shape=box,label="__1_b\n'b'"]; "-1214143508" -> "-1214140756"; "-1214140756" [shape=box,label="__2_c\n'c'"]; "-1214143508" -> "-1214144468"; "-1214144468" [label="m"]; "-1214144468" -> "-1214414868"; "-1214414868" [shape=box,label="__5_d\n'd'"]; "-1214026868" -> "-1214141492"; "-1214141492" [shape=box,label="__4_)\n')'"]; } After transformation the tree looks like this: .. graphviz:: digraph G{ "-1219949908" [label="n"]; "-1219949908" -> "-1214026452"; "-1214026452" [shape=box,label="__0_a\n'a'"]; "-1219949908" -> "-1214028276"; "-1214028276" [shape=box,label="__1_b\n'b'"]; "-1219949908" -> "-1214027316"; "-1214027316" [shape=box,label="__2_c\n'c'"]; "-1219949908" -> "-1219949876"; "-1219949876" [label="n"]; "-1219949876" -> "-1214141364"; "-1214141364" [shape=box,label="__0_a\n'a'"]; "-1219949876" -> "-1214141748"; "-1214141748" [shape=box,label="__1_b\n'b'"]; "-1219949876" -> "-1214140756"; "-1214140756" [shape=box,label="__2_c\n'c'"]; "-1219949876" -> "-1219949748"; "-1219949748" [label="m"]; "-1219949748" -> "-1214414868"; "-1214414868" [shape=box,label="__5_d\n'd'"]; } >nonterminal_1 nonterminal_2 ... nonterminal_n< +++++++++++++++++++++++++++++++++++++++++++++++ This replaces the nodes nonterminal_1 to nonterminal_n by their children. Example:: IGNORE: " "; DECIMAL: "0|[1-9][0-9]*"; list: DECIMAL >list< | DECIMAL; Parsing the string "1 2" gives the tree: .. graphviz:: digraph G{ "-1213518708" [label="list"]; "-1213518708" -> "-1213518196"; "-1213518196" [shape=box,label="DECIMAL\n'1'"]; "-1213518708" -> "-1213518260"; "-1213518260" [label="list"]; "-1213518260" -> "-1213520308"; "-1213520308" [shape=box,label="DECIMAL\n'2'"]; } after the transformation the tree looks like: .. graphviz:: digraph G{ "-1219505652" [label="list"]; "-1219505652" -> "-1213518196"; "-1213518196" [shape=box,label="DECIMAL\n'1'"]; "-1219505652" -> "-1213520308"; "-1213520308" [shape=box,label="DECIMAL\n'2'"]; } Note that the transformation works recursively. That means that the following also works: if the string "1 2 3 4 5" is parsed the tree at first looks like this: .. graphviz:: digraph G{ "-1213611892" [label="list"]; "-1213611892" -> "-1213608980"; "-1213608980" [shape=box,label="DECIMAL\n'1'"]; "-1213611892" -> "-1213623476"; "-1213623476" [label="list"]; "-1213623476" -> "-1213623380"; "-1213623380" [shape=box,label="DECIMAL\n'2'"]; "-1213623476" -> "-1213442868"; "-1213442868" [label="list"]; "-1213442868" -> "-1213441652"; "-1213441652" [shape=box,label="DECIMAL\n'3'"]; "-1213442868" -> "-1213441332"; "-1213441332" [label="list"]; "-1213441332" -> "-1213441620"; "-1213441620" [shape=box,label="DECIMAL\n'4'"]; "-1213441332" -> "-1213443060"; "-1213443060" [label="list"]; "-1213443060" -> "-1213442100"; "-1213442100" [shape=box,label="DECIMAL\n'5'"]; } But after transformation the whole thing collapses to one node with a lot of children: .. graphviz:: digraph G{ "-1219430228" [label="list"]; "-1219430228" -> "-1213608980"; "-1213608980" [shape=box,label="DECIMAL\n'1'"]; "-1219430228" -> "-1213623380"; "-1213623380" [shape=box,label="DECIMAL\n'2'"]; "-1219430228" -> "-1213441652"; "-1213441652" [shape=box,label="DECIMAL\n'3'"]; "-1219430228" -> "-1213441620"; "-1213441620" [shape=box,label="DECIMAL\n'4'"]; "-1219430228" -> "-1213442100"; "-1213442100" [shape=box,label="DECIMAL\n'5'"]; } Extensions to the EBNF grammar format ------------------------------------- There are some extensions to the EBNF grammar format that are really only syntactic sugar but make writing grammars less tedious. These are: symbol?: matches 0 or 1 repetitions of symbol symbol*: matches 0 or more repetitions of symbol. After the tree transformation all these repetitions are children of the current symbol. symbol+: matches 1 or more repetitions of symbol. After the tree transformation all these repetitions are children of the current symbol. These are implemented by adding some more rules to the grammar in the correct way. Examples: the grammar:: s: a b? c; is transformed to look like this:: s: a >_maybe_symbol_0_< c | a c; _maybe_symbol_0_: b; The grammar:: s: a b* c; is transformed to look like this:: s: a >_star_symbol_0< c | a c; _star_symbol_0: b >_symbol_star_0< | b; The grammar:: s: a b+ c; is transformed to look like this:: s: a >_plus_symbol_0< c; _plus_symbol_0: b >_plus_symbol_0< | b; Full Example ------------ A semi-complete parser for the json format`_:: STRING: "\\"[^\\\\"]*\\""; NUMBER: "\-?(0|[1-9][0-9]*)(\.[0-9]+)?([eE][\+\-]?[0-9]+)?"; IGNORE: " |\n"; value: | |
Tip: Filter by directory path e.g. /media app.js to search for public/media/app.js.
Tip: Use camelCasing e.g. ProjME to search for ProjectModifiedEvent.java.
Tip: Filter by extension type e.g. /repo .js to search for all .js files in the /repo directory.
Tip: Separate your search with spaces e.g. /ssh pom.xml to search for src/ssh/pom.xml.
Tip: Use ↑ and ↓ arrow keys to navigate and return to view the file.
Tip: You can also navigate files with Ctrl+j (next) and Ctrl+k (previous) and view the file with Ctrl+o.
Tip: You can also navigate files with Alt+j (next) and Alt+k (previous) and view the file with Alt+o.