Alex Lam S.L
a3cc3a9b87
make expect_stdout work on Node.js 0.12 ( #1623 )
...
That particular version of Node.js has messed up error messages, so provide a version-specific workaround.
Also fixed an formatting issue which would cause `expect_stdout` to fail if error message contains excerpts of input.
Apply `expect_stdout` to more applicable tests.
2017-03-19 12:00:32 +08:00
Alex Lam S.L
96f8befdd7
fix commit 88fb83a ( #1622 )
...
The following is wrong:
`a == (b ? a : c)` => `b`
Because:
- `b` may not be boolean
- `a` might have side effects
- `a == a` is not always `true` (think `NaN`)
- `a == c` is not always `false`
2017-03-19 11:59:42 +08:00
Alex Lam S.L
cd58635dcc
fix AST_Binary.lift_sequences() ( #1621 )
...
Commit eab99a1c fails to account for side effects from compound assignments.
2017-03-19 03:04:22 +08:00
Alex Lam S.L
274331d0ea
transform String.charAt() to index access ( #1620 )
...
Guarded by `unsafe` as `charAt()` can be overridden.
2017-03-19 02:17:15 +08:00
Alex Lam S.L
0489d6de64
handle runtime errors in expect_stdout ( #1618 )
...
allow test to pass if both `input` and `expect` throws the same kind of error
2017-03-18 02:33:51 +08:00
Alex Lam S.L
fb092839c2
fix top-level directives in compress tests ( #1615 )
...
`input` and `expect` are parsed as `AST_BlockStatement` which does not support `AST_Directive` by default.
Emulate that by transforming preceding `AST_SimpleStatement`s of `AST_String` into `AST_Directive`.
2017-03-18 01:56:15 +08:00
Alex Lam S.L
ac40301813
fix chained evaluation ( #1610 )
...
`reduce_vars` enables substitution of variables but did not clone the value's `AST_Node`.
This confuses `collapse_vars` and result in invalid AST and subsequent crash.
fixes #1609
2017-03-17 00:26:48 +08:00
Alex Lam S.L
3563d8c09e
extend test/run-tests.js to optionally execute uglified output ( #1604 )
...
fixes #1588
2017-03-16 23:20:06 +08:00
Alex Lam S.L
5ae04b3545
make collapse_vars consistent with toplevel ( #1608 )
...
fixes #1605
2017-03-16 13:22:26 +08:00
Alex Lam S.L
a80b228d8b
fix hoist_vars on reduce_vars ( #1607 )
...
`hoist_vars` converts variable declarations into plain assignments, which then confuses `reduce_vars`
fixes #1606
2017-03-16 12:03:30 +08:00
Alex Lam S.L
8223b2e0db
fix AST_Node.optimize() ( #1602 )
...
Liberal use of `Compressor.transform()` and `AST_Node.optimize()` presents an issue for look-up operations like `TreeWalker.in_boolean_context()` and `TreeWalker.parent()`.
This is an incremental fix such that `AST_Node.optimize()` would now contain the correct stack information when called correctly.
2017-03-15 18:44:13 +08:00
Alex Lam S.L
d9344f30b8
disallow parameter substitution for named IIFEs ( #1596 )
...
Self-referenced function has non-fixed values assigned to its parameters.
Let `unused` & `!keep_fnames` do the scanning, then apply `reduce_vars` only to unnamed functions.
fixes #1595
2017-03-11 03:34:55 +08:00
Alex Lam S.L
be80f7e706
support multi-line string in tests ( #1590 )
...
`expect_exact` sometimes have multiple lines and `\n` are hard to read.
Use array of strings to emulate line breaks and improve readability.
2017-03-10 11:27:30 +08:00
Alex Lam S.L
9e6b128374
fix catch variable reference in IE8 ( #1587 )
...
`AST_Scope.def_variable()` will overwrite `AST_Symbol.thedef`, so save a copy before calling.
fixes #1586
2017-03-10 03:15:21 +08:00
Alex Lam S.L
b633706ce4
fix & improve function argument compression ( #1584 )
...
- one-use function call => IIFE should take `eval()` & `arguments` into account
- if unused parameter cannot be eliminated, replace it with `0`
fixes #1583
2017-03-09 19:11:05 +08:00
Alex Lam S.L
7e465d4a01
scan RHS of dropped assignments ( #1581 )
...
- similar case as #1578 but against #1450 instead
- fix `this` binding in #1450 as well
closes #1580
2017-03-09 05:22:27 +08:00
Alex Lam S.L
711f88dcb4
scan assignment value in drop_unused() ( #1578 )
...
those were not optimised for `unused` before, which made it necessary for `reduce_vars` to have separate steps for `keep_fnames`
docs update by @kzc
closes #1577
2017-03-08 18:37:32 +08:00
Alex Lam S.L
c7cdcf06a6
fix function name eliminiation ( #1576 )
...
Function expression can be assigned to a variable and be given a name. Ensure function name is the reduced variable before clearing it out.
fixes #1573
fixes #1575
2017-03-08 12:39:57 +08:00
Alex Lam S.L
dedbeeff15
plan B for IE8 do-while semi-colon fix ( #1572 )
...
- omitting trailing semi-colon in do-while breaks non-browser parser, e.g. uglify-js 1.x
- trailing semi-colon only breaks IE8 if followed by `else` or `while`
- always use braces in do-while body to workaround 2nd case with no size loss in compression
fixes #1568
2017-03-08 05:07:05 +08:00
Alex Lam S.L
bd6dee52ab
fix return from recursive IIFE ( #1570 )
...
`side-effects` did not account for IIFEs being able to reference itself thus making its return value potentially significant
2017-03-08 03:31:51 +08:00
Alex Lam S.L
8a8a94a596
fix deep cloning of labels ( #1565 )
...
`AST_Label.references` get `.initialize()` to `[]` every time after `.clone()`
So walk down the tree to pick up the cloned `AST_LoopControl` pieces and put it back together.
2017-03-07 18:38:27 +08:00
Alex Lam S.L
8153b7bd8a
transform function calls to IIFEs ( #1560 )
...
- expose function body to call sites for potential optimisations
- suppress substitution of variable used within `AST_Defun`
2017-03-07 15:37:52 +08:00
Alex Lam S.L
d787d70127
avoid substitution of global variables ( #1557 )
...
- unless `toplevel` is enabled
- global `const` works as before
2017-03-07 03:11:03 +08:00
kzc
3ac2421932
collapse_vars: do not replace a constant in loop condition or init ( #1562 )
2017-03-07 01:42:33 +08:00
Alex Lam S.L
a9fc9ddc33
suppress semicolons after do/while ( #1556 )
...
- unless both `beautify` & `screw-ie8` are enabled
- deprecate workaround for if-do-while-else
fixes #186
2017-03-06 17:31:35 +08:00
Alex Lam S.L
067e5a5762
fixup for #1553 ( #1555 )
...
- `++a` is the one that is foldable
- transform `a++` into `++a` for better optimisation
2017-03-05 17:15:37 +08:00
Alex Lam S.L
35a849dc48
collapse assignment with adjacent subsequent usage ( #1553 )
...
- consolidate `cascade` optimisations
- support ++/-- postfixes
- remove redundant optimisation identified in #1460
fixes #368
2017-03-05 14:56:14 +08:00
Alex Lam S.L
b70591be1a
handle variable declaration within catch blocks ( #1546 )
...
accounts for IE8- scoping
2017-03-05 13:13:44 +08:00
Alex Lam S.L
b33e7f88e6
improve unsafe on undefined ( #1548 )
...
`unsafe` turns undefined keyword into a variable of the same name if found, but that interferes with other related optimisations.
Keep track of such transformations to ensure zero information loss in the process.
2017-03-05 13:09:27 +08:00
Alex Lam S.L
1f0333e9f1
stay safe with constants in IE8- ( #1547 )
...
- `undefined` etc. can be redefined at top-level for IE8-, so disable related optimisations
- fixed `--support-ie8` catch mangle bug
2017-03-05 12:51:11 +08:00
Alex Lam S.L
78d1bb92d4
fix a corner case in #1530 ( #1552 )
2017-03-05 12:12:59 +08:00
Alex Lam S.L
07accd2fbb
process code with implicit return statement ( #1522 )
...
Bookmarklet for instance implicitedly assumes a "completion value" without using `return`.
The `expression` option now supports such use cases.
Optimisations on IIFEs also enhanced.
fixes #354
fixes #543
fixes #625
fixes #628
fixes #640
closes #1293
2017-03-03 18:13:07 +08:00
Alex Lam S.L
18059cc94f
compress numerical expressions ( #1513 )
...
safe operations
- `a === b` => `a == b`
- `a + -b` => `a - b`
- `-a + b` => `b - a`
- `a+ +b` => `+b+a`
associative operations
(bit-wise operations are safe, otherwise `unsafe_math`)
- `a + (b + c)` => `(a + b) + c`
- `(n + 2) + 3` => `5 + n`
- `(2 * n) * 3` => `6 * n`
- `(a | 1) | (2 | d)` => `(3 | a) | b`
fixes #412
2017-03-03 18:04:32 +08:00
Alex Lam S.L
17b81350d4
fix chained assignment with unused ( #1540 )
...
When #1450 optimises `a=b=42`, it stops after the first variable even if both are unused.
fixes #1539
2017-03-03 04:45:20 +08:00
kzc
4d63d4f5b3
collapse_vars should not replace constant in for-in init section ( #1538 )
...
fixes #1537
2017-03-03 03:51:15 +08:00
Alex Lam S.L
70d72ad806
properly cover all cases of for-in loop variables ( #1536 )
2017-03-03 02:39:57 +08:00
Alex Lam S.L
fe9227a41b
fix reference marking in for-in loops ( #1535 )
...
fixes #1533
2017-03-03 00:56:06 +08:00
Alex Lam S.L
b49e142a26
disable do{...}while(false) optimisation ( #1534 )
...
- fails to handle `break` in body
fixes #1532
2017-03-03 00:54:41 +08:00
kzc
ee3b39b909
optimize trivial IIFEs returning constants ( #1530 )
2017-03-02 15:11:40 +08:00
Alex Lam S.L
9699ffb1af
trim unused invocation parameters ( #1526 )
2017-03-02 11:33:59 +08:00
Alex Lam S.L
fdc9b9413b
minor improvement to string optimisation ( #1514 )
...
- "" + "a" => "a"
- "" + a + "b" => a + "b"
- "a" + "" => "a" (improving on #45 )
2017-03-02 11:31:39 +08:00
Alex Lam S.L
7aa69117e1
fix corner cases in reduce_vars ( #1524 )
...
Avoid variable substitution in the following cases:
- use of variable before declaration
- declaration within conditional code blocks
- declaration within loop body
fixes #1518
fixes #1525
2017-03-02 00:20:53 +08:00
Alex Lam S.L
b34fa11a13
fix evaluate on object getter & setter ( #1515 )
2017-03-01 02:03:47 +08:00
Alex Lam S.L
858e6c78a4
warn & drop #__PURE__ iff IIFE is dropped ( #1511 )
...
- consolidate `side-effects` optimisations
- improve string `+` optimisation
- enhance literal & `conditionals` optimisations
2017-02-28 02:25:44 +08:00
Alex Lam S.L
16cd5d57a5
consolidate evaluate & reduce_vars ( #1505 )
...
- improve marking efficiency
- apply smarter `const` replacement to `var`
fixes #1501
2017-02-26 00:40:33 +08:00
alexlamsl
4e49302916
enable collapse_vars & reduce_vars by default
...
- fix corner cases in `const` optimisation
- deprecate `/*@const*/`
fixes #1497
closes #1498
2017-02-24 01:46:57 +08:00
kzc
1e51586996
Support marking a call as pure
...
A function call or IIFE with an immediately preceding comment
containing `@__PURE__` or `#__PURE__` is deemed to be a
side-effect-free pure function call and can potentially be
dropped.
Depends on `side_effects` option.
`[#@]__PURE__` hint will be removed from comment when pure
call is dropped.
fixes #1261
closes #1448
2017-02-21 14:24:18 +08:00
alexlamsl
26fbeece1c
fix pure_funcs & improve side_effects
...
- only drops side-effect-free arguments
- drop side-effect-free parts with discarded value from `AST_Seq` & `AST_SimpleStatement`
closes #1494
2017-02-21 13:31:59 +08:00
alexlamsl
8898b8a0fe
clean up max_line_len
...
- never exceed specified limit
- otherwise warning is shown
- enabled only for final output
closes #1496
2017-02-21 13:29:58 +08:00
alexlamsl
ec64acd2c8
introduce unsafe_proto
...
- `Array.prototype.slice` => `[].slice`
closes #1491
2017-02-21 13:29:58 +08:00
Anthony Van de Gejuchte
c06a50f338
Add .gitattributes to checkout lf eol style
...
closes #1487
2017-02-21 13:29:58 +08:00
alexlamsl
e275148998
enhance global_defs
...
- support arrays, objects & AST_Node
- support `"a.b":1` on both cli & API
- emit warning if variable is modified
- override top-level variables
fixes #1416
closes #1198
closes #1469
2017-02-21 13:29:58 +08:00
alexlamsl
974247c8c0
evaluate AST_SymbolRef as parameter
...
fix invalid boolean conversion now exposed in `make_node_from_constant()`
closes #1477
2017-02-21 13:29:58 +08:00
alexlamsl
a0f4fd390a
improve reduce_vars and fix a bug
...
- update modified flag between compress() passes
- support IIFE arguments
- fix corner case with multiple definitions
closes #1473
2017-02-21 13:29:58 +08:00
alexlamsl
b8b133d91a
improve keep_fargs & keep_fnames
...
- utilise in_use_ids instead of unreferenced()
- drop_unused now up-to-date for subsequent passes
closes #1476
2017-02-21 13:29:58 +08:00
alexlamsl
c525a2b190
fix duplicated test names
...
previously test cases with the same name would be skipped except for the last one
`test/run-test.js` will now report duplicated names as errors
closes #1461
2017-02-21 13:29:58 +08:00
kzc
6ffbecb72b
smarter const replacement taking name length into account
...
closes #1459
2017-02-21 13:29:58 +08:00
alexlamsl
f0ff6189be
clean up negate_iife
...
- remove extra tree scanning phase for `negate_iife`
- `negate_iife` now only deals with the narrowest form, i.e. IIFE sitting directly under `AST_SimpleStatement`
- `booleans`, `conditionals` etc. will now take care the rest via more accurate accounting
- `a(); void b();` => `a(); b();`
fixes #1288
closes #1451
2017-02-21 13:29:58 +08:00
alexlamsl
6b3c49e458
improve string concatenation
...
shuffle associative operations to minimise parentheses and aid other uglification efforts
closes #1454
2017-02-21 13:29:57 +08:00
alexlamsl
f584ca8d07
-c sequences=N suboptimal at N expression cutoff
...
N = 2:
a;
b;
c;
d;
was:
a, b;
c;
d;
now:
a, b;
c, d;
fixes #1455
closes #1457
2017-02-21 13:29:57 +08:00
alexlamsl
ae4db00991
tweak do-while loops
...
- `do{...}while(false)` => `{...}`
- clean up `AST_While` logic
closes #1452
2017-02-21 13:29:57 +08:00
alexlamsl
100307ab31
fixes & improvements to [].join()
...
fixes
- [a].join() => "" + a
- ["a", , "b"].join() => "a,,b"
- ["a", null, "b"].join() => "a,,b"
- ["a", undefined, "b"].join() => "a,,b"
improvements
- ["a", "b"].join(null) => "anullb"
- ["a", "b"].join(undefined) => "a,b"
- [a + "b", c].join("") => a + "b" + c
closes #1453
2017-02-21 13:29:57 +08:00
alexlamsl
148047fbbf
drop unused: toplevel, assign-only
...
- assign statement does not count towards variable usage by default
- only works with assignments on the same scope level as declaration
- can be disabled with `unused` set to "keep_assign"
- `toplevel` to drop unused top-level variables and/or functions
- `top_retain` to whitelist top-level exceptions
closes #1450
2017-02-21 13:29:57 +08:00
kzc
d11dca3cf9
fix stray else in compress with conditionals=false
...
closes #1449
2017-02-21 13:29:57 +08:00
alexlamsl
e5badb9541
enable typeof "undefined" for general use
...
move out of unsafe, guard corner case with screw_id8 instead
closes #1446
2017-02-18 19:01:42 +08:00
alexlamsl
fa668a28b4
fix corner case in keep_fnames
...
happens when inner function:
- just below top level
- not referenced
- `unused` is disabled
closes #1445
2017-02-18 19:00:54 +08:00
alexlamsl
11676f9d72
fix crash in unsafe replacement of undefined
...
remove extraneous call to AST_SymbolRef.reference()
closes #1443
2017-02-18 18:58:23 +08:00
Anthony Van de Gejuchte
dd31d12a91
Improve optimizing function() { if(c){return foo} bar();}
...
closes #1437
2017-02-18 18:56:18 +08:00
Alex Lam S.L
7f8d72d9d3
update test ( #1441 )
...
improved reduce_vars & binary operands produce more optimal results
2017-01-26 12:59:32 +01:00
Alex Lam S.L
1eaa211e09
fix mangling collision with keep_fnames ( #1431 )
...
* fix mangling collision with keep_fnames
fixes #1423
* pass mangle options to figure_out_scope()
bring command-line in line with minify()
2017-01-26 12:18:28 +01:00
Alex Lam S.L
0610c020b1
optimise binary operands with evaluate() ( #1427 )
...
- remove call to evaluate() in is_constant() and let nested optimize() does its job instead
- reject RegExp in is_constant() and remove special case logic under collapse_vars
- operands to conditionals optimisation are now always evaluate()-ed
- throw error in constant_value() instead of returning undefined to catch possible bugs, similar to make_node_from_constant()
- optimise binary boolean operators under `evaluate` instead of `conditionals`
2017-01-26 12:16:50 +01:00
Alex Lam S.L
0d7d4918eb
augment evaluate to extract within objects ( #1425 )
...
- gated by `unsafe`
- replaces previous optimisation specific to String.length
- "123"[0] => 1
- [1, 2, 3][0] => 1
- [1, 2, 3].length => 3
- does not apply to objects with overridden prototype functions
2017-01-26 12:14:18 +01:00
alexlamsl
48284844a4
add missing LHS cases which global_defs should avoid
2017-01-19 21:06:28 +01:00
kzc
ec2e5fa3a2
Have minify() and tests use figure_out_scope() as uglifyjs CLI does
...
Clarify docs, help and tests for --support-ie8 and screw_ie8=false
2017-01-19 17:14:33 +01:00
Anthony Van de Gejuchte
da17766ddd
Add preventive test involving non-ascii function identifiers
2017-01-19 17:13:33 +01:00
Ashley (Scirra)
2a9989dd18
Add --mangle-props-debug and fix --mangle-props=unquoted collision
...
Patch by @AshleyScirra
Based on: PR #1316
Renamed the CLI debug option to --mangle-props-debug
Fixes : #1321 name collision in --mangle-props=unquoted
2016-11-29 20:25:39 +01:00
Anthony Van de Gejuchte
057de570e6
Pass mangle options to figure_out_scope before mangling in tests
2016-10-27 22:55:49 +02:00
Richard van Velzen
6389e52305
Remove console.log and add extra test case
2016-10-06 14:11:32 +02:00
Richard van Velzen
e05510f3bc
Add an option to wrap IIFEs in parenthesis
...
For #1307 .
2016-10-06 14:11:32 +02:00
kzc
fc9804b909
Fix (typeof side_effect()) in boolean context
...
Fixes #1289 with suggestion by @rvanvelzen
2016-10-06 13:50:11 +02:00
alexlamsl
4761d07e0b
Optimize unmodified variables
2016-10-01 11:36:11 +02:00
kzc
25fc02743a
Account for side effects in string + expr optimization
2016-09-01 09:24:56 -04:00
kzc
0bd8053524
implement optimization: (x = 2 * x) ---> (x *= 2)
2016-08-30 08:43:02 -04:00
Richard van Velzen
8430123e9d
Fix negate_iife transform to return a correct tree for nested IIFEs
...
Fix for #1256 , partially reverts d854523783
2016-08-17 11:55:59 +02:00
kzc
d854523783
Fix negate_iife regression #1254
2016-08-17 01:29:34 -04:00
kzc
de619ae5a6
Fix --mangle-props and --mangle-props=unquoted
...
Fixes : #1247
Fix --mangle-props and --name-cache inconsistency.
AST_Dot and AST_Sub properties are now mangled by --mangle-props
without regard to being used in an assignment statement.
Note: if --mangle-props is used then *all* javascript files used must
be uglified with the same mangle options.
Fix the ignore_quoted=true mangle option, also known as
`--mangle-props=unquoted`. If a given property is quoted anywhere
it will not be mangled in any quoted or non-quoted context.
2016-08-14 21:51:25 +02:00
kzc
86859f6d7e
Additional object literal property tests
2016-08-14 21:49:43 +02:00
kzc
fb049d3a81
Fix unneeded parens around unary args in new expression.
2016-08-14 21:38:38 +02:00
Anthony Van de Gejuchte
7eb52d2837
Keep const in own scope while compressing
...
- Fixes #1205
- Fix provided by @kzc
2016-07-15 13:20:52 +02:00
kzc
eb63fece2f
Fix mangle with option keep_fnames=true for Safari.
...
Fixes : #1202
2016-07-15 13:18:14 +02:00
kzc
02c638209e
Enable --screw-ie8 by default.
...
catch identifier is mangled correctly for ES5 standards-compliant JS engines by default.
Unconditionally use the ie8 if/do-while workaround whether or not --screw-ie8 is enabled.
To support non-standard ie8 javascript use: uglifyjs --support-ie8
2016-06-30 21:49:48 +02:00
Anthony Van de Gejuchte
3a7d53f3cf
Move OctalEscapeSequence to read_escape_char
...
This should simplify and improve implementation, make it easier to
implement template strings, and keep master a bit more in sync with
harmony.
Previous implementation wasn't broken, though the loop gave me the
impression it could read infinite numbers and annoyed me a bit. It was
also slightly unnecessary because the lookup involved only 3 characters.
2016-06-30 21:42:15 +02:00
kzc
ace8aaa0f4
Fix conditional expressions of form (x ? -1 : -1)
...
Fixes #1154 , #1153
2016-06-21 14:52:13 -04:00
kzc
0c003c92a8
Don't replace undefined, NaN and Infinity within with scope
2016-06-21 10:53:29 +02:00
Shrey Banga
e645ba84cf
Respect quote style in object literals
...
The option added in fbbaa42ee5 wasn't
being respected inside object literals, so quoted property names would
still be stripped out with this option.
This is mostly a corner-case, but useful when the output is passed to
something like the Closure compiler, where quoted property names can be
used to prevent mangling.
2016-06-19 21:13:31 +02:00
Anthony Van de Gejuchte
d7971ba0e4
Fix test262 failures related to <, <=, in and instanceof
...
Fixed-by: @kzc
2016-06-15 23:11:08 +02:00
Anthony Van de Gejuchte
6c8e001fee
Stop dropping args in new expressions
2016-06-12 17:17:17 +02:00
David Bau
f99b7b630d
Escape null characters as \0 unless followed by 0-7.
2016-06-12 14:32:32 +02:00
Anthony Van de Gejuchte
ea31da2455
Don't drop unused if scope uses with statement
...
Fix provided by @kzc
2016-06-12 14:30:28 +02:00
Anthony Van de Gejuchte
8287ef6781
Fix uglify attempting to rewrite invalid new expressions
2016-06-08 19:45:21 +02:00
kzc
09d5707a8a
collapse_vars: Do not consider RegExp literals to be constants
...
Fixes #1100
2016-05-27 00:03:51 -04:00
kzc
1e390269d4
Optimize if_return for single if/return cases.
...
Fixes #1089
2016-05-24 17:54:08 +02:00
Richard van Velzen
bc49dfd27a
Completely allow evaluating -0
2016-05-24 17:50:29 +02:00
Richard van Velzen
27eedbc302
Never produce -0 when evaluating expressions (like -"")
...
Fix for #1085 . The major case was already there, but more expressions can result in -0.
2016-05-17 22:34:38 +02:00
kzc
5f464b41e2
Simplify iife new fix
...
as suggested by @rvanvelzen.
Added a test for IIFEs in nested contexts.
2016-05-15 19:12:17 -04:00
kzc
bcc1318d4b
Do not apply negate_iife optimization to new expression
2016-05-09 03:19:28 -04:00
Richard van Velzen
e9224ab444
Add test cases for slightly more esoteric cases
2016-04-26 11:49:55 +02:00
Richard van Velzen
4d9a085687
Add test case for hoisting a single function
2016-04-26 11:43:03 +02:00
Richard van Velzen
4fe630431c
Hoist functions when reversing if (x) return; ... vs. if (!x) ...
...
Fixes #1052
2016-04-23 23:48:33 +02:00
kzc
c55dd5ed74
Add passes compress option. Fix duplicate compress warnings.
2016-04-19 20:05:33 +02:00
Richard van Velzen
4b4528ee05
Prevent endless recursion when evaluating self-referencing consts
...
Fix #1041
2016-04-13 15:03:31 +02:00
Richard van Velzen
187a0caf9d
Add base54.reset() to compress tests
...
Without this reset, char counts bleed to next tests. One test had a bad expect clause.
2016-04-12 20:08:09 +02:00
kzc
3907a5e3b2
Fix warnings for referenced non-hoisted functions.
...
Fixes #1034
Also added `expect_warnings` functionality to test framework.
2016-04-11 18:15:20 +02:00
kzc
98434258d0
Optimize ternaries with boolean consequent or alternative.
...
Fixes #511
2016-04-02 17:22:12 +02:00
kzc
07bb7262d0
Escape all ASCII control characters within strings when using ascii_only.
...
Fixes #1017 .
Tab characters within strings are now output as `\t` in all output modes.
2016-03-24 11:51:54 -04:00
philippsimon
ee6c9fabb7
Fix: Uglified Number.prototype functions on big numbers
2016-03-14 12:41:06 +01:00
kzc
11b0efdf84
boolean_expression ? true : false --> boolean_expression
2016-02-22 17:59:36 +01:00
kzc
5486b68850
Take operator || precendence into account for AST_If optimization.
...
Fixes #979 .
2016-02-21 12:05:02 -05:00
alexlamsl
6547437725
preserve ThisBinding for side_effects
2016-02-17 19:34:01 +01:00
Richard van Velzen
9662228f6a
Don't compress (0, eval)() to eval()
2016-02-16 19:00:48 +01:00
alexlamsl
31a9b05c96
Preserve ThisBinding in conditionals & collapse_vars
...
Fixes #973
2016-02-16 18:47:49 +01:00
kzc
929de2b0de
collapse_vars: fix if/else and ternary operator side effects
2016-01-28 12:17:06 -05:00
kzc
af2472d85e
collapse_vars: fix bug in repeated var defs of same name
2016-01-28 16:48:50 +01:00
kzc
0a38a688f9
fix bug in collapse_vars for right side of "||" and "&&"
2016-01-27 14:18:46 -05:00
kzc
f4c2ea37bf
Collapse single use var definitions
...
Fix #721
2016-01-27 11:48:15 +02:00
Richard van Velzen
b5a7197ae5
Merge pull request #928 from STRML/constPragma
...
Mark vars with /** @const */ pragma as consts so they can be eliminated.
2016-01-20 19:04:36 +01:00
Samuel Reed
1b703349cf
Tighten up @const regex.
2016-01-20 11:35:45 -06:00
Samuel Reed
f97da4294a
Use TreeWalker for more accurate @const results and update tests
2016-01-20 10:54:00 -06:00
Samuel Reed
8b71c6559b
Mark vars with /** @const */ pragma as consts so they can be eliminated.
...
Fixes older browser support for consts and allows more flexibility
in dead code removal.
2016-01-19 13:23:02 -06:00
Anthony Van de Gejuchte
26641f3fb2
Allow operator names as getters/setters
...
Fixes #919
Fix provided by @kzc
2016-01-19 19:28:51 +01:00
Anthony Van de Gejuchte
6605d15783
Never mangle arguments and keep them in their scope
...
Fixes #892
Helped-by: kzc
2016-01-14 19:45:52 +01:00
Mihai Bazon
fe4e9f9d97
Fix hoisting the var in ForIn
...
Close #913
2016-01-05 13:56:52 +02:00
Richard van Velzen
60c4030a4d
Merge pull request #874 from kzc/fix-conditionals
...
#873 Fix `conditionals` optimizations with default compress options
2015-12-26 14:28:33 +01:00
Anthony Van de Gejuchte
0cabedc526
Disable loop optimization for parse-only tests
2015-12-18 19:20:56 +01:00
Anthony Van de Gejuchte
5cd26c005b
Add tests
2015-12-18 14:39:48 +01:00
kzc
774bda13cd
#873 Fix conditionals optimizations with default compress options
2015-11-24 13:27:50 -05:00
kzc
bd0ae6569f
return undefined optimization no longer uses return_void_0 option
2015-10-29 08:19:12 +01:00
kzc
841a661071
more tests for return undefined optimization
2015-10-29 08:19:12 +01:00
kzc
7491d07666
optimize return undefined and return void 0
2015-10-29 08:19:12 +01:00
Richard van Velzen
335e349314
Allow specification beautify options in tests
...
Caught an error in #847 as well - `output` wasn't passed anywhere which led to an exception. `options` was available though.
2015-10-28 20:50:01 +01:00
kzc
dff54a6552
Fix other operator output related to <!-- or -->
2015-10-13 01:17:10 -04:00
kzc
9f1f21b810
Output -- > instead of --> in expressions. Escape <!-- and --> within string literals.
2015-10-12 10:19:17 +03:00
kzc
0d952ae43d
add asm.js test
2015-10-07 10:00:28 +02:00
Ville Lautanala
5fd12451f9
Control keeping function arguments with a single option
2015-09-14 19:38:45 +02:00
Richard van Velzen
fcde6109b0
Fix bad parsing of new new x()() constructs
...
Fixes #739
2015-08-27 12:29:36 +03:00
Richard van Velzen
e3bd223dac
Don't change sequences that influence lexical binding in calls
...
Fixes #782
2015-08-25 10:53:35 +02:00
Christopher Chedeau
3ff0b9e0c9
[Fix] --define replaces SymbolRefs in LHS of assignments
...
See #208 for context
2015-08-10 11:22:36 -07:00
Richard van Velzen
759b3f7d6d
Fix mangling of property names which overwrite unmangleable properties
...
Fixes #747 .
2015-08-05 21:18:39 +02:00
Mihai Bazon
905b601178
Don't attempt to negate non-boolean AST_Binary
...
Fix #751
2015-07-22 16:55:55 +03:00
kzc
9d398d999c
spacing
2015-06-14 17:45:19 -04:00
kzc
fedb6191a1
optimizations for && and || where left side is constant expression
2015-06-11 23:22:38 -04:00
Mihai Bazon
a5b60217ce
Fix compressing conditionals
...
Only transform foo() ? EXP(x) : EXP(y) into EXP(foo() ? x : y) if EXP has no
side effects.
Fix #710
2015-05-18 13:56:04 +03:00
Mihai Bazon
3b14582d6b
Fix tests
2015-04-17 11:28:59 +03:00
Mihai Bazon
18c63ff3d8
Fix compression of conditionals
...
Don't move the condition on the right side of an assignment when
the left side may have side effects.
Fix #677
2015-04-13 17:29:48 +03:00
Mihai Bazon
ecfd881ac6
Keep unused function arguments by default
...
Discarding unused function arguments affects function.length, which can lead
to some hard to debug issues. This optimization is now done only in "unsafe
mode".
Fix #121
2015-03-20 10:28:51 +02:00
Richard van Velzen
61e850ceb5
Clean up unit test breakage
...
In 992b6b9fcc unit test broke (which I missed). This was due to undeclared variables not being side-effects free.
However, since they're really not side-effect free, just declare them in the test cases.
2015-02-11 21:27:21 +01:00
Richard van Velzen
992b6b9fcc
Fix invalid removal of left side in && and || compression
...
See #637 . This does not produce the optimal result, but it does prevent the removal of non-side-effect-free code.
2015-02-11 21:08:41 +01:00
Richard van Velzen
605362f89d
Drop all console statements properly
...
Because the base reference can be an member expression as well, we have to dig a bit deeper to find the leftmost base reference.
Fixes #451
2015-01-31 13:24:44 +01:00
Mihai Bazon
d36067cd35
Merge pull request #615 from avdg/unicode
...
Give parser more unicode support
2015-01-20 13:00:31 +02:00
Anthony Van de Gejuchte
f1b2134dd1
Add test
2015-01-20 00:31:44 +01:00
Tal Ater
a1a4c2ada7
Optimize conditionals where the consequent and alternative are both booleans and not equivalent
2015-01-13 18:27:21 +01:00
Richard van Velzen
42ecd42ac0
Replace the correct node when replacing in void sequences
...
Close #611 .
2015-01-12 17:09:34 +01:00
Mihai Bazon
a10f6a96d7
Merge pull request #482 from arty-name/inline-ng-inject
...
added @ngInject support for inline functions
2015-01-11 12:10:42 +02:00
Richard van Velzen
0d48af3f36
Add a "keep_fnames" option to the compressor to retain function expression names
...
See #552 . This is useful for stack traces.
2015-01-04 20:14:38 +01:00
Richard van Velzen
c75f5a1fd8
Fix #597
...
NaN and Infinity were replaced in the output generation, instead of
during compression. This could lead to results where `1/0` was inserted
without parens leading to invalid output.
The nodes are replaced in the compression step now, and the output
generation returns their regular names. This should not be a problem,
since they're already only constructed from the original name.
2014-12-31 12:23:00 +01:00
Mihai Bazon
01d19b4b52
Referencing a global is assumed to have side effects.
...
Close #550
2014-09-28 11:18:25 +03:00
Tal Ater
fb0ec720a4
Compress conditions that have side effects using sequences
2014-09-04 02:57:49 +03:00
Tal Ater
7971ed33d1
Added a test for else if
2014-09-03 01:35:30 +03:00
Tal Ater
885835a655
Compress conditional assignments where all possible outcomes are equivalant and condition has no side effects
2014-09-02 23:30:25 +03:00
Artemy Tregubenko
6006dd933d
added newline at the end of the file
2014-07-08 11:16:35 +02:00
Dan Wolff
8511e80f48
Evaluate "foo".length ==> 3
2014-07-01 11:06:51 +03:00
Artemy Tregubenko
524a8a42a4
added @ngInject support for inline functions
2014-05-11 14:01:08 +02:00
Mihai Bazon
37693d2812
Update tests.
2014-04-18 11:19:52 +03:00
Matt Basta
ac0086a745
Simplify nested conditionals if possible
2014-02-06 12:39:13 -08:00
Mihai Bazon
b521b4b926
Conditional/call optimization
...
foo ? bar(x) : bar(y) ==> bar(foo ? x : y)
2013-12-29 10:31:30 +02:00
Mihai Bazon
8f35a363d9
AST_Catch shouldn't really inherit from AST_Scope. Fix #363
...
I hereby acknowledge that figure_out_scope has become a mess.
2013-12-05 13:30:29 +02:00
Mihai Bazon
50b8d7272c
Fix faulty compression
...
`String(x + 5)` is not always the same as `x + "5"`. Overlooked that. :-(
Close #350
2013-11-20 21:13:16 +02:00
Mihai Bazon
eab99a1c3d
Better fix for #343
...
We can in fact lift sequences, but only if the operation is assignment and
the left-hand side has no side effects nor property access -- that should
guarantee that whatever we place before it cannot affect the sense of the
assignment.
Dropped contrived test case (too hard to support it now), added a more
meaningful one.
2013-11-06 10:48:48 +02:00
Richard van Velzen
785c6064cc
Disallow reversal where lhs has higher or equal precedence
...
Fixes #267
2013-10-29 21:37:36 +01:00
Mark Jaquith
9cd118ca3d
Add a unit test for issue-126
...
Add a unit test to test to test for aggressive parenthesis removal that causes functional changes.
2013-10-25 16:28:15 -04:00
Richard van Velzen
1a5a4bd631
Fix #269
...
Shorten most primitives where possible. Also optimize some edge cases.
2013-10-24 11:08:33 +02:00
Mihai Bazon
afdaeba37d
More attempts to determine when addition is associative
...
Somebody hit me with bug reports on this. :)
Refs #300
2013-09-22 15:26:10 +03:00
Mihai Bazon
037199bfe2
Actually let's move away those monsters from the evaluate function
...
ev() should do a single thing — evaluate constant expressions. if that's
not possible, just return the original node. it's not the best place for
partial evaluation there, instead doing it in the compress functions.
2013-09-22 15:00:42 +03:00
Mihai Bazon
583fac0a0f
More dirty handling of [ ... ].join() in unsafe mode
...
Close #300
2013-09-22 13:14:42 +03:00
Dan Wolff
e8158279ff
Evaluate [...].join() if possible: minor bugfix
...
Follow-up to 78e98d2 .
2013-09-22 11:34:30 +03:00
Mihai Bazon
78e98d2611
When unsafe is set, evaluate [...].join() if possible
...
Close #298
2013-09-19 18:20:45 +03:00
Dan Wolff
8d14efe818
Concatenate strings also on the right-hand side of an expression that cannot be evaluated. Fix #126
...
E.g. converts:
a+'Hello'+'World'
to
a+'HelloWorld'
2013-09-19 13:03:03 +03:00
Mihai Bazon
ed80b4a534
Move support for negate_iife in the compressor, rather than code generator
...
(the code generator doesn't maintain enough context to know whether
the return value is important or discarded)
Fixes #272
2013-08-20 17:45:52 +03:00
Mihai Bazon
4c4dc2137c
Don't drop unused setter argument.
...
Fix #257
2013-08-07 12:04:58 +03:00
David Glasser
b1febde3e9
Fix output for arrays whose last element is a hole: [1,,]
...
1529ab96 started to do this (by considering holes to be separate from
"undefined") but it still converted
[1,,] (length 2, last element hole, trailing comma)
to
[1,] (length 1, trailing comma)
Unfortunately the test suite doesn't really make this clear: the new test here
passes with or without this patch because run-tests.js beautifys the expected
output (in "make_code"), which does the incorrect transformation! If you make
some manual change to arrays.js to make the test fail and see the INPUT and
OUTPUT, then you can see that without this fix, [1,,] -> [1,], and with this fix
it stays [1,,].
2013-07-18 15:39:22 +03:00
Mihai Bazon
d56ebd7d7b
Fix a["1_1"]
...
Close #204
2013-05-14 10:42:34 +03:00
Trey Griffith
1e9f98aa51
add a test for zero-length string in is_identifier_string, which is used in property compression. Also added a test exercising the change.
2013-05-08 22:43:20 +03:00
Mihai Bazon
a6ed2c84ac
Better fix for equality of typeof ... against "undefined"
2013-05-08 16:22:48 +03:00
Justin Lau
a1958aad56
Fixed typeof undefined optimization and updated related test case to
...
accomodates the sort behaviour changes made in commit
mishoo/UglifyJS2@aebafad41e .
Signed-off-by: Justin Lau <justin@tclau.com>
2013-05-08 16:22:48 +03:00
Justin Lau
672699613e
Added test cases for #104 .
...
Signed-off-by: Justin Lau <justin@tclau.com>
2013-05-08 16:22:48 +03:00
Justin Lau
fcd544cc10
Added test scenario with unicode in properties name.
...
Signed-off-by: Justin Lau <justin@tclau.com>
2013-05-06 01:26:33 +08:00
Justin Lau
8227e8795b
Added scenario in test case where properties shouldn't be accessed with
...
dotted syntax even with screw_ie8 option.
Signed-off-by: Justin Lau <justin@tclau.com>
2013-05-05 22:08:13 +08:00
Michael Ficarra
5af144522a
fixes #189 : use dotted member access when --screw-ie8 option given
2013-04-20 15:11:05 -05:00
Roman Bataev
b956e5f1d9
Add tests for typeof evaluation
2013-04-03 22:34:19 -04:00
Mihai Bazon
dac6efb43d
Drop last default: if it's the last branch and empty
...
Close #141
2013-03-01 13:12:03 +02:00