Fixes issues #7, #10, #13. Prefix matching related (java equivalent == LookingAt() )

[loopfz]

In (at least) these 2 places, a regular match/FindIndex was used instead of enforcing matching from the start of the string, like LookingAt() does in the Java implementation.

So for maybeStripNationalPrefixAndCarrierCode, country prefixes (e.g. "0") were matched and truncated at any place in the number, and most cases were saved by the fact that the resulting number would be of an invalid size. But in some situations (like #13), the number legal size would be flexible enough to allow the incorrectly truncated number to go through.

Same thing for parsePrefixAsIdd. Furthermore, line 2553 find[1] was incorrect. You cannot make the assumption that len(find) > 1, so index 1 might be out of range. Java implem used groups(1) because groups(0) has a special meaning (== whole string), but it in fact used the first match, corresponding to find[0] here.

[loopfz]

Unsure why the tests fail now and not before, but the conditions seem wrong:

    if reflect.DeepEqual(getTestNumber("DE_NUMBER"), GetExampleNumber("DE")) {

    if reflect.DeepEqual(
            getTestNumber("DE_NUMBER"), GetExampleNumberForType("DE", FIXED_LINE)) {

Missing a NOT ?

[ttacon]

So those tests were taken from the Java version of libphonenumber so they need to pass in their original form. I don't have a ton of free time at the moment (work and all), but I'll try to get a chance to look deeper into this soon.

[loopfz]

Actually no, if you take a look at https://github.com/googlei18n/libphonenumber/blob/master/java/libphonenumber/test/com/google/i18n/phonenumbers/PhoneNumberUtilTest.java#L310 the Java tests assert equality, ie error on non-equality. The tests fix I did in my commit actually detect the issue in #13, and now they pass.

[loopfz]

By the way, I've been reading the Go & Java code side by side for a while now, really huge work you did there, congrats. It's actually very convenient too, the implem being so close to the Java one, it makes it much easier to detect small behavior differences (I have another pending fix I'm working on, some weird behavior on some AR numbers, will PR you later if interested).

[ttacon]

LOL. That's what I get for not drinking enough coffee...

[ttacon]

Ya, I would like to make the Go version idiomatic Go, but this library hasn't been on my plate recently so I haven't been able to make the changes I want to. If you've got other PR's fixing issues for this library, please keep them coming lol!