Attribute source ranges are not normalized correctly

[jhy]

As reported in #1933:

---

Here's my PR on the Rascal project. It works! But... I have a missing Range on the attributes on the top node.

usethesource/rascal#1893

For testing purposes I was parsing the pom.xml file of the rascal project itself:

<project xmlns="http://maven.apache.org/POM/4.0.0" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd">
    <modelVersion>4.0.0</modelVersion>

    <groupId>org.rascalmpl</groupId>
    <artifactId>rascal</artifactId>
    <version>0.35.0-RC2-SNAPSHOT</version>
    <packaging>jar</packaging>
  ... SNIP!

The project node has an xsi:schemaLocation attribute.

Here's some output of the readXML function that will print the filename only if the attribute sources ranges are -1:

rascal>import IO;
ok
rascal>import lang::xml::IO;
ok
rascal>iprintln(readXML(|cwd:///pom.xml|, trackOrigins=true), lineLimit=-1)
"#document"(
  "project"(
    "modelversion"(
      "4.0.0",
      src=|cwd:///pom.xml|(209,14,<2,4>,<2,18>)),
    "groupid"(
      "org.rascalmpl",
      src=|cwd:///pom.xml|(249,9,<4,4>,<4,13>)),
    "artifactid"(
      "rascal",
      src=|cwd:///pom.xml|(286,12,<5,4>,<5,16>)),
    "version"(
      "0.35.0-RC2-SNAPSHOT",
      src=|cwd:///pom.xml|(322,9,<6,4>,<6,13>)),
    "packaging"(
      "jar",
      src=|cwd:///pom.xml|(365,11,<7,4>,<7,15>)),
    "scm"(
      "developerconnection"(
        "scm:git:ssh://git@github.com/usethesource/rascal.git",
        src=|cwd:///pom.xml|(411,21,<10,8>,<10,29>)),
      "tag"(
        "SNAPSHOT",
        src=|cwd:///pom.xml|(515,5,<11,8>,<11,13>)),
      src=|cwd:///pom.xml|(397,5,<9,4>,<9,9>)),
    "repositories"(
      ... SNIP!
      "dependency"(
        "groupid"(
          "com.ibm.icu",
          src=|cwd:///pom.xml|(18901,9,<441,12>,<441,21>)),
        "artifactid"(
          "icu4j",
          src=|cwd:///pom.xml|(18944,12,<442,12>,<442,24>)),
        "version"(
          "69.1",
          src=|cwd:///pom.xml|(18987,9,<443,12>,<443,21>)),
        src=|cwd:///pom.xml|(18876,12,<440,8>,<440,20>)),
      src=|cwd:///pom.xml|(15755,14,<351,4>,<351,18>)),
    src=|cwd:///pom.xml|(0,204,<1,0>,<1,204>),
    srcs=("schemalocation":|cwd:///pom.xml|),
    schemalocation="http://maven.apache.org/POM/4.0.0 http://maven.apache.org/xsd/maven-4.0.0.xsd"),
  src=|cwd:///pom.xml|)

Here we see the srcs map saying: ("schemalocation":|cwd:///pom.xml|), which is caused by the AttributeRange.valueRange() having a begin offset of -1. I tracked this with the Java debugger. Is this supposed to happen? Or is the top-node somehow different?

Originally posted by @jurgenvinju in #1933 (comment)

[jhy]

The issue is that the attribute source ranges are not being normalized per the settings of the parser. So in the above example, with attribute xsi:schemaLocation with preserve case=off, the range is tracked in the key xsi:schemaLocation, but the actual attribute is normalized to xsi:schemalocation.

https://github.com/usethesource/rascal/blob/16fe28156c7c57305c7c4f4702cba5d9ce93cb58/src/org/rascalmpl/library/lang/xml/IO.java#L92C5-L95

Fix is to normalize the Range keys when normalizing the Attribute names.