tree 4957bd0ce1d763cb52fd552809fa9ae143990bc4
parent b368df5ca35484e686fe2b22ee75a8718f4839c8
author Anton Hansson <hansson@google.com> 1622721092 +0100
committer Anton Hansson <hansson@google.com> 1623243260 +0100

Refactor metalava html parsing

Metalava would previously parse html files that it discovered via
crawling the sourcepath argument. Refactor things such that if html
files are passed as explicit sources, those get parsed as well.

Note that metalava is still capable of crawling -sourcepath, if no
sources are passed explicitly. If so, it will now also pick up html
files during that crawl.

There's a bit of ugliness around how metalava discovers the package name
those html files are in: in reads random java files in the directory
containing the html file. This should probably be cleaned up by
migrating from package.html to package-info.java.

Bug: 153703940
Test: DocAnalayzerTest + SymlinkTest
Merged-In: Ic17dade2d834098a4fb4def89172757a86ae7f35
Change-Id: Ic17dade2d834098a4fb4def89172757a86ae7f35
(cherry picked from commit 01c25630d4244deaa1dcbd43683d5d041fa932a0)