Fix probable boolean logic error

[mmcco]

• Type: bug fix

• I have read and understood the Contributing Guidelines?
• I have checked that another pull request for this purpose does not exist.
• I wrote some tests for this PR.

Small description of change: This if condition is always false, and it seems clear that the author meant to use || rather than &&.

Thanks for your time,
Michael McConville
University of Utah

[Ultimater]

Good eye! And thanks for repairing the test!

Interestingly enough, this change wouldn't affect the CSS minimizing logic whatsoever without a break; as the STATE_FREE and STATE_SELECTOR states currently have no differences and would still encounter:

p = cssmin_back_peek(parser);
if (p == ',' || p == '>' || p == ':') {
     c = 0;
} else {
     c = ' ';
}

In other words we could actually write:

case STATE_FREE:
case STATE_SELECTOR:

and it would be the equivalent of either before or after your change.

I think what the author intended was actually to trim ALL whitespace when in the STATE_FREE, without using looking around to make sure we're not in a selector:

if (c == ' ' || c == '\t' || c == '\n' || c == '\r') {
    c = 0;
    break;
}

Note the usage of my break.

Preserving the space is necessary on the selector for example:
div p{margin:2px 4px 5px 7px;}
However if we're in the STATE_FREE, we have not yet encountered the first character d of div, so it's safe to remove all whitespace until such a character is encountered.

While we're in here, it would be nice to change:
parser.state = 1; to parser.state = STATE_FREE;
This way it's easier to search for STATE_FREE and find where it's initially set.

Also last_state doesn't appear to be used whatsoever and would be nice to trash.

Probably some room to fix the parser code further as mentioned at the top of the file:

** cannot handle nested { blocks but will ignore aditional { in parens ()
** no in quote detection for ( or }

[Ultimater]

Oh wait, a slight difference:

if ((c == ' ' || c == '\t')) {
    p = cssmin_peek(parser);
    if (p == '{' || p == '\t' || p == ' ' || p == '>' || p == ',') {
    c = 0;
} else {
    p = cssmin_back_peek(parser);
    if (p == ',' || p == '>' || p == ':') {
        c = 0;
    } else {
        c = ' ';
    }
}

c would be set to 0 in STATE_FREE of your version before it runs into the above code in STATE_SELECTOR.
Yeah, so that would explain the need to adjust the test when it was working before.

I would sooner still see a break; though. If we're changing c, it's spaghetti code if we let the states spill through, specially after changing an important variable holding the character we're working on...

[sergeyklay]

@mmcco Could you please update change log?

[sergeyklay]

Thank you