CoatiSoftware / SourcetrailDB

Library to export Sourcetrail compatible database files for writing custom indexers
Apache License 2.0
287 stars 69 forks source link

SourcetrailDB

Library to export Sourcetrail compatible database files for writing custom indexers

CI Pipelines

Build status

Introduction

The SourcetrailDB project provides write access to Sourcetrail database files. You can use the SourcetrailDB project to write an indexer for an arbitrary programming language (or other kind of data, e.g. see poetry indexer example) and export a Sourcetrail database file that can be viewed and navigated within Sourcetrail.

"C++ Poetry Indexer"

Projects Using SourcetrailDB

The following list of projects already use SourcetrailDB API to extend the language support of Sourcetrail.

Adding new Language Support

When you plan on starting a Sourcetrail language package on your own, you can use whatever programming language you like for this task. The only requirement is: you need to make calls to SourectrailDB to enter the information your indexer records to the Sourcetrail database.

To get an overview on everything involved, please take a look at our Language Extension Guide.

Supported Language Bindings

Even though the core implementation is written in C++, this does not require you to write your indexer in C++ as well. Instead you can use a language binding (e.g. see SWIG). These language bindings are already available:

If the language of your choice is not covered by this list, feel free to open an issue or provide a pull request.

Versioning

The SourcetrailDB version format consists of three numbers in the format vXX.dbYY.pZZ.

You can find a complete list of available releases on the GitHub release page. If you are interested in the actual changes of every release, please take a look at the Changelog.

Contributing

If you like this project and want to get involved, there are lots of ways you can help:

To create a pull request, follow these steps:

Building

Take a look at Appveyor (for Windows) or Travis (for Linux and macOs) CI scripts for reference.

SourcetrailDB Core

Before building the core project, please make sure that you also have checked out the tags of this repository (they are relevant for deriving version number information during the build).

Requirements:

To build the SourcetrailDB core library libsourcetraildb.a:

$ cd path/to/SourcetrailDB
$ mkdir build
$ cd build
$ cmake ..
$ make lib_core

To run the tests, just build and execute the test executable:

$ make test_core
$ ./core/test_core

Perl Bindings

Requirements:

If you want to build the Perl bindings run:

$ cd path/to/SourcetrailDB
$ mkdir build
$ cd build
$ cmake -DBUILD_BINDINGS_PERL=ON ..
$ make sourcetraildb

Swig is configured to generate the Perl binding code as a pre-build event, so you don't need to bother with updating manually.

Python Bindings

Requirements:

If you want to build the Python bindings run:

$ cd path/to/SourcetrailDB
$ mkdir build
$ cd build
$ cmake -DBUILD_BINDINGS_PYTHON=ON ..
$ make _sourcetraildb

Swig is configured to generate the Python binding code as a pre-build event, so you don't need to bother with updating manually.

Java Bindings

Requirements:

If you want to build the Java bindings run:

$ cd path/to/SourcetrailDB
$ mkdir build
$ cd build
$ cmake -DBUILD_BINDINGS_JAVA=ON ..
$ make

Swig is configured to generate the Java binding code as a pre-build event, so you don't need to bother with updating manually.

C# Bindings

Requirements:

If you want to build the C# bindings run:

$ cd path/to/SourcetrailDB
$ mkdir build
$ cd build
$ cmake -DBUILD_BINDINGS_CSHARP=ON ..
$ make

Examples

The examples help you to understand SourcetrailDB usage in practice. Please take a look at each examples README file for build and use instructions. Each example also provides a Sourcetrail project file .srctrlprj showing you how to use a custom indexer directly from Sourcetrail (see Integrating with Sourcetrail).

SourcetrailDB API

SourcetrailDB tries to keep the API for writing to the Sourcetrail database as simple as possible. For the C++ core take a look at the list of methods of the SourcetrailDBWriter class for a complete overview of the provided write API. If you are using bindings for other languages, the code may look different but the set of provided functionality should stay the same.

Opening or Creating a Database

// create an instance of SourcetrailDBWriter
sourcetrail::SourcetrailDBWriter writer;

// open a database connection by passing a database file path (.srctrldb or .srctrldb_tmp)
// if the database file does not exist yet, it will be created in this location.
// a project file (.srctrlprj) will also be created, if it does not exist in this location.
writer.open("MyProject.srctrldb");

// record data...

// close the database connection.
writer.close();

Recording a Symbol

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

// records a named non-indexed symbol without type
writer.recordSymbol({ "::", { { "void", "foo", "()" } } });

writer.close();

"Recording a Symbol"

Recording Additional Symbol Information

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

sourcetrail::NameHierarchy name = { "::", { { "void", "foo", "()" } } };

// the returned symbolId will be used to record further information for the symbol
int symbolId = writer.recordSymbol(name);

// recording a unique name multiple times will always return the same symbol id
assert(symbolId == writer.recordSymbol(name));

// set symbol "explicit" to remove the "non-indexed" state.
writer.recordSymbolDefinitionKind(symbolId, sourcetrail::DEFINITION_EXPLICIT);

// set type to turn the symbol into "function"
writer.recordSymbolKind(symbolId, sourcetrail::SYMBOL_FUNCTION);

writer.close();

"Recording Additional Symbol Information"

Recording Nested Symbol Names

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject_3.srctrldb");

// create a name hierarchy
sourcetrail::NameHierarchy name;

// specify the delimiter string - in many languages names can be nested
name.nameDelimiter = "::";

// add a name element for the parent
sourcetrail::NameElement parentElement;
parentElement.prefix = ""; // only shown in tooltip of the parent symbol
parentElement.name = "Bar";
parentElement.postfix = ""; // only shown in tooltip of the parent symbol
name.nameElements.push_back(parentElement);

// add a name element for the child
sourcetrail::NameElement childElement;
childElement.prefix = "void"; // only shown in tooltip
childElement.name = "bar";
childElement.postfix = "()"; // only shown in tooltip
name.nameElements.push_back(childElement);

// recording a name hierarchy with multiple elements automatically creates parent symbols, if still unknown
int childId = writer.recordSymbol(name);
writer.recordSymbolDefinitionKind(childId, sourcetrail::DEFINITION_EXPLICIT);
writer.recordSymbolKind(childId, sourcetrail::SYMBOL_METHOD);

writer.close();

"Recording Nested Symbol Names"

Recording Symbol Location

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

int symbolId = writer.recordSymbol({ "::", { { "", "Bar", "" }, { "void", "bar", "()" } } });

// record a file via it's absolute path
int fileId = writer.recordFile("C:/example/Bar.cpp");

// create a location
sourcetrail::SourceRange location;
location.fileId = fileId;
location.startLine = 8; // lines start at 1, not 0
location.startColumn = 7; // columns start at 1, not 0
location.endLine = 8;
location.endColumn = 9;

// adds a clickable location to the file for the specified symbol
writer.recordSymbolLocation(symbolId, location);

// the whole scope location will be visible and highlighted when the symbol is active
writer.recordSymbolScopeLocation(symbolId, { fileId, 8, 1, 11, 1 });

writer.close();

"Recording Symbol Location"

Recording References Between Symbols

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

int contextSymbolId = writer.recordSymbol({ "::", { { "", "Bar", "" }, { "void", "bar", "()" } } });
int referencedSymbolId = writer.recordSymbol({ "::", { { "void", "foo", "()" } } });

// edges always go from the context to the referenced symbol
int referenceId = writer.recordReference(contextSymbolId, referencedSymbolId, sourcetrail::ReferenceKind::CALL);

// add a location to the reference - highlights the location in the code view when clicking the edge in the graph view
int fileId = writer.recordFile("C:/example/Bar.cpp");
writer.recordReferenceLocation(referenceId, { fileId, 10, 3, 10, 5 });

writer.close();

"Recording References between Symbols"

Recording File Information for Syntax Highlighting

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

int fileId = writer.recordFile("C:/example/Bar.cpp");

// enable syntax highlighting using a <language>.rules file loaded from Sourcetrail's data directory
writer.recordFileLanguage(fileId, "cpp");

writer.close();

"Recording File Information for Syntax Highlighting"

Recording Local Symbols

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

// provide a unique local symbol name - the name is not displayed anywhere
int localId = writer.recordLocalSymbol("some_unique_name");
int fileId = writer.recordFile("C:/example/Foo.cpp");
writer.recordLocalSymbolLocation(localId, { fileId, 3, 6, 3, 6 });
writer.recordLocalSymbolLocation(localId, { fileId, 4, 2, 4, 2 });

writer.close();

"Recording Local Symbols"

Recording Comments

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

int fileId = writer.recordFile("C:/example/Bar.cpp");

// the recorded source range is atomic - it is always shown completely
int id = writer.recordCommentLocation({ fileId, 3, 2, 7, 4 });

writer.close();

"Recording Comments"

Recording Errors

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

int fileId = writer.recordFile("C:/example/Foo.cpp");

// store and show parsing and indexing errors
std::string message = "Really? You missed that \";\" again?";
bool fatal = false;
sourcetrail::SourceRange location = { fileId, 4, 4, 4, 4 };
int id = writer.recordError(message, fatal, location);

writer.close();

"Recording Errors"

Use Transactions for Speed

sourcetrail::SourcetrailDBWriter writer;
writer.open("MyProject.srctrldb");

// wrapping your code in transactions really speeds up database operations
writer.beginTransaction();

for (int i = 0; i < 1000; ++i)
{
    int id = writer.recordSymbol({ "::", { { "void", "foo" + std::to_string(i), "()" } } });
    if (id == 0) // if something goes wrong
    {
        // discard all the changes made within the transaction
        writer.rollbackTransaction();
        writer.close();
        return;
    }
}

// don't forget to commit your transaction to persistently write the changes
writer.commitTransaction();

writer.close();

Integrating with Sourcetrail

Applications using SourcetrailDB can be directly integrated with Sourcetrail by creating a project with a Custom Command Source Group. Choose Custom in the project selection dialog:

Custom Command Source Group Selection

Project Setup

A Custom Command Source Group defines:

"Custom Command Source Group Setup"

Indexing

Indexing the project works as usual, the specified Custom Command is executed with each source file passed in %{SOURCE_FILE_PATH}. The files in the Custom Command Source Group will be indexed after all other files.

Custom Command Source Group Indexing