Position warning bug despite input having positional tags

[aalramadan]

Attempting to run nameCollector on a srcML archive input with --position errors with the following:

WARNING: srcml --position NOT used to generate input file.

The system tested is PowerShell-7.5.4:
https://github.com/PowerShell/PowerShell/archive/refs/tags/v7.5.4.zip

To replicate:
1] Download the PowerShell ZIP file.
2] Then run:

srcml PowerShell-7.5.4.zip -o PowerShell-7.5.4.xml --position
 ./nameCollector -i PowerShell-7.5.4.xml --csv -o results.csv

[aalramadan]

Found the issue. The issue is here:

virtual void startUnit(const char* localname, const char* prefix, const char* URI,
                       int numNamespaces, const struct srcsax_namespace * namespaces, int numAttributes,
                       const struct srcsax_attribute * attributes) {
// code

    //Check if srcml --position used to generate input
    bool positionNotUsed = true;
    for (int i=0; i<numNamespaces; ++i)
        if (std::string(namespaces[i].uri) == "http://www.srcML.org/srcML/position") positionNotUsed = false;
    if (positionNotUsed) std::cerr << "WARNING: srcml --position NOT used to generate input file." << std::endl;

// code
}

The current implementation checks for the position namespace only within startUnit (which runs for every file in the archive). However, the position namespace (http://www.srcML.org/srcML/position) is only declared only once at the root unit, not on every individual unit.

Fix is to move the check block to startRoot so that it is only checked once:

virtual void startRoot(const char* localname, const char* prefix, const char* URI,
                       int numNamespaces, const struct srcsax_namespace * namespaces, int numAttributes,
                       const struct srcsax_attribute * attributes) {
    //Check if srcml --position used to generate input
    bool positionNotUsed = true;
    for (int i=0; i<numNamespaces; ++i)
        if (std::string(namespaces[i].uri) == "http://www.srcML.org/srcML/position") positionNotUsed = false;
    if (positionNotUsed) std::cerr << "WARNING: srcml --position NOT used to generate input file." << std::endl;
}

[Joshua2000B]

I do not believe that is a complete fix for the problem, as this will only help with srcML Archives and not single-file srcML units. I will play around with this

[aalramadan]

It works on solitary units too. Just tested.