fix(upload): validate path

[cnlangzi]

Changes

Fixes

• it will be rejected if a file is uploaded inside a file

Tests

Tasks to complete before merging PR:

• Ensure system tests are passing. If not Run them manually to check for any regressions 📋
• Do any new system tests need added to test this change? do any existing system tests need updated? If so create a PR at 0chain/system_test
• Merge your system tests PR to master AFTER merging this PR

Associated PRs (Link as appropriate):

[lpoli]

@cnlangzi Actually this won't work. First of all, util.SplitPath does not return path's values for parent path but for file path i.e. for /a/b/c/d/e.txt it returns [a b c d e.txt] instead of [a b c d].

The other important missed point is, if there is already a file with path /a/b/c.txt and user uploaded a file with path /a/b/c.txt/d/e/f/g.txt then we cannot only check against parent path of the filepath. So we need to check for /a, /a/b, /a/b/c.txt, ....

I had considered this as well. If you make this changes then I will drop my PR.

[cnlangzi]

@cnlangzi Actually this won't work. First of all, util.SplitPath does not return path's values for parent path but for file path i.e. for /a/b/c/d/e.txt it returns [a b c d e.txt] instead of [a b c d].

The other important missed point is, if there is already a file with path /a/b/c.txt and user uploaded a file with path /a/b/c.txt/d/e/f/g.txt then we cannot only check against parent path of the filepath. So we need to check for /a, /a/b, /a/b/c.txt, ....

I had considered this as well. If you make this changes then I will drop my PR.

fixed both of them.

[peterlimg]

Perhaps we should return error when d is nil, or when index is >= length, otherwise users will have to check the code to know what happened when the returned value is "".

[cnlangzi]

fixed

[peterlimg]

Same as Path(), we'd better return errors on "" strings

[cnlangzi]

fixed

[peterlimg]

Adding cases for empty subpath perhaps, e.g. /dir1/dir2//dir3///dir4

[cnlangzi]

fixed

[cnlangzi]

@peterlimg let's close this PR. and @lpoli will fix this bug on his PR #571. please go to review his PR instead of it. thanks.

[cnlangzi]

sorry. it is #578. #571 seems is outdated.