| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Previously when comparing local phone numbers with
international it was possible to match the wrong number
because the first digit of the area code was being treated
as a skippable trunk digit, e.g:
"550-123-4567" would be considered equal to "+14501234567".
Since there are two different algorithms (loose/old and strict)
there are two solutions with the same basic goal:
Only ignore mismatches if there's an actual extra digit which
could possibly be the trunk digit, i.e: "0550-123-4567" is acceptable
as equal to "+15501234567". NB: The US not having a trunk digit
to begin with is a different issue entirely - the code is agnostic on
which country has which trunk digits!
For "loose" matching we achieve this by checking the length of the
mismatch.
For "strict" matching we keep track of the supposed trunk digit
and compare it against the current position in match. So for the above
example 5 != 4.
Several new unit tests were added (including replicating tests for
OldPhoneNumberUtils). I broke down the c++ tests into smaller methods
for additional readability.
NB: By adding more tests I uncovered that two digit trunk prefixes
were never handled correctly for the "strict" matching case. "Loose"
matching is much more robust. I commented out the "strict" test cases.
Test: All unit tests pass for PhoneNumberUtilsTest,
OldPhoneNumberUtilsTest and DatabaseGeneralTest.
Bug: 63179537
Bug: 63178542
Bug: 62916088
Change-Id: Idc2a1c2c2f64ed29995208503de4755c521e1c82
|
| |
|
|
|
|
|
|
|
| |
Removes dead use of deprecated ucol_getShortDefinitionString.
Converts PhoneNumberUtils tests to gtest.
Bug: http://b/31459453
Change-Id: I4662e25d711868e21afb0661602c64852f0cef65
|
| |
|
|
|
| |
Bug: 3210604
Change-Id: I6e9a20ec7887375b67c42a28cff54e2e9d399ea6
|
| |
|
|
| |
Change-Id: I11fc145464e29364dbf302c1e607a07031e1821d
|
| |
|
|
|
|
|
|
|
| |
Now developers have to explicitly set the third argument of PHONE_NUMBERS_COMPARE() into 1,
when they want to use "strict" phone number comparation algorithm, which was used in Donut.
In default, PHONE_NUMBER_COMPARE() now uses "loose" phone number comparation algorithm,
which had been used in Cupcake.
Internal issue id: 1892808
|
| |
|
|
|
|
| |
in PhoneNumberUtils.java.
Due to the Thailand's special case, the code became a little nastier than before, but I believe it is inevitable...
|
| |
|
|
|
|
|
|
| |
phone_number_compare() treated all alpha addresses as the same phone number.
- changed phone_number_compare() to handle alpha chars. Before this change the algorithm skipped over all non-dialable chars for
comparison, and thus treating "foo" and "bar" as the same addresses. now we don't skip any char that's alpha. This would treat
"foo" and "bar" as separate addresses, as well as treating "1-800-flowers" and "800-flowers" as the same number.
|
| |
|
|
| |
unexpectedly...
|
| |
|
|
| |
Thailand one...
|
| |
|
|
|
|
|
|
|
|
|
| |
many characters as possible, unlike the previous implementation.
Unlike the change 5480, this considers the NDD-problem mentioned in the internal issue 1949010.
This time, 1-650-555-1234 and 650-555-1234 match each other.
Added tests for the cases.
related internal issue: 1868702, 1949010
|
| |
|
|
|
|
|
|
| |
using as many characters as possible, unlike the previous implementation."
This reverts commit de43094477419f8a190a6f97b47d346827310a02.
related internal issue: 1949010
|
|
|
many characters as possible, unlike the previous implementation.
The new implementation considers country code seriously.
To confirm this change does not do something wrong, I also created another change (https://android-git.corp.google.com/g/Gerrit#change,4036).
All tests related to PhoneNumberUtils pass, including Thailand's ugly buggy behavior around country code handling :-P
Also Add tests for PhoneNumberUtils, which can be executed in the usual workstation like this:
> g++ -Wall external/sqlite/android/PhoneNumberUtils.cpp external/sqlite/android/PhoneNumberUtilsTest.cpp
> ./a.out
This change fixes the internal bug 1868702.
|
