How to run in Firefox:
- Check out mozilla-unified (see Firefox dev setup under "Setting up your machine" for details, note that you can use git via git-cinnabar)
cd
to the checked out directory.- Run
git am <this-repo>/firefox/0001-WIP-Firefox-source-map-spec-tests.patch
- Run
mach build
(this builds the test too, you'll need to run it if you modify tests/harness) - Run
mach test devtools/client/shared/source-map-loader/
How to run in WebKit:
- Check out WebKit
cd
to the checked out WebKit directory.- Run
git am <this-repo>/webkit/0001-Add-harness-for-source-maps-spec-tests.patch
- Run
Tools/Scripts/build-webkit
(depending on the platform you may need to pass--gtk
or other flags) - Run
Tools/Scripts/run-webkit-tests LayoutTests/inspector/model/source-map-spec.html
(again, you may need--gtk
on Linux)
Mozilla source-map library:
- There is a branch for adding the spec tests to the package.
- Thoroughly test all aspects of the source maps spec that can be tested.
- Strictly follow the spec when determining test behavior.
- Encoding
- Source map must be a valid JSON document.
- Base64 VLQ
- VLQs should decode correctly
- A VLQ with a non-base64 character will fail to decode.
- A VLQ with one digit and no continuation digits should decode.
- A negative VLQ with the sign bit set to 1 should decode.
- A VLQ with non-zero continuation bits (and more than one digit) should decode.
- A VLQ with a non-zero continuation bit with no further digits should fail to decode.
- A VLQ should decode with the correct order of digits (least to most significant).
- A long VLQ with many trailing zero digits will decode.
- A VLQ exceeding the 32-bit size limit is invalid (note: the spec is unclear on the details of this limit)
- A VLQ at exactly the 32-bit size limit should be decoded (positive and negative).
- VLQs should decode correctly
- Basic format
version
field- Must be present
- Must be a number
- Must be 3
file
field- Optional, allow missing
- Must be a string? (spec is not clear)
sourceRoot
field- Optional, allow missing
- Must be a string? (spec is not clear)
sources
field- Must be present
- Must be an array
- Array elements must be
null
or a string
sourcesContent
field- Must be present
- Must be an array
- Array elements must be
null
or a string
names
field- Must be present (note: the spec implies this but implementations may not agree)
- Must be an array
- Array elements must be strings
mappings
field- Must be present
- Must be a string
- Empty string is valid
ignoreList
field- Optional, allow missing
- Must be an array
- Array elements must be numbers
- Elements must not be out of bounds for the
sources
list
- Extra unrecognized fields are allowed
- Index maps
- ? Must be mutually exclusive with non-index map?
file
field- Optional, allow missing
- Must be a string? (spec is not clear)
sections
field- Must be present
- Must be an array
- An empty sections array is valid
- Array elements are valid section objects
offset
field- Must be present
line
field- Must be present
- Must be a number
column
field- Must be present
- Must be a number
map
field- Must be present
- Must be an object
- Must be a valid source map
- Sections are in order (the spec is not 100% clear, but assumption is increasing numeric order, may need subtests)
- Sections are non-overlapping (the definition of overlap is not clear, may need subtests)
- Mappings format
- Each line is separated by ";"
- A line may consist of zero segments (e.g., ";;")
- Each line consists only of segments separated by ","
- Must have greater than zero fields (note: many implementations don't check)
- Must have 1, 4, or 5 fields
- The source index must not be out of bounds of the sources array
- The name index must not be out of bounds of the names array
- Absolute VLQ values must be non-negative
- The column must be non-negative
- The source index must be non-negative
- The original line must be non-negative
- The original column must be non-negative
- The name index must be non-negative
- Relative VLQ values must be non-negative after adding to previous value
- The column must be non-negative
- The source index must be non-negative
- The original line must be non-negative
- The original column must be non-negative
- The name index must be non-negative
- Ignore list
- An ignore list is optional, may be missing
- An ignore list can't be a non-array value
- An ignore list can be empty
- An ignore list entry must be a number
- An ignore list entry cannot be out-of-bounds of the sources array
- Ignore list entries are detected and are present
- Items not specified in the ignore list don't show up as ignored
- Mappings semantics
- A source map with no mappings does not map any position.
- A single field segment gets mapped to the correct line and column.
- A four field segment gets mapped to the correct line and column.
- A five field segment gets mapped to the correct line and column.
- When a name is present in a segment, it is correctly mapped.
- When a source is present in a segment, it is correctly mapped.
- The second occurence of a field segment in a line is mapped relative to the previous one.
- When a new line starts, the generated column field resets to zero rather than being relative to the previous line.
- For fields other than the generated column, a segment field that has occured once in a previous line is mapped relatively when it occurs in the next line.
- Ensure that a transitive source map mapping works as expected
- Index maps are correctly used in mappings
- An index map with one sub-map will map correctly.
- An index map with multiple sub-maps will map correctly, with appropriate offsets for the second and later sub-maps.
- Resolution of sources
- When
sourceRoot
is provided, it is prepended to anysources
entries and will be mapped with the full URL. - If the source URL is an absolute URL, it is resolved as an absolute URL.
- If the source URL is a relative URL, it is resolved relative to the source map path.
- When
- Wasm support
- Create versions of the tests that use a Wasm source.
TODO