feat: Parser & Definitions

.gitignore

@@ -20,3 +20,4 @@ zig-out/
 # Although this was renamed to .zig-cache, let's leave it here for a few
 # releases to make it less annoying to work with multiple branches.
 zig-cache/
+.aider*

src/definitions.zig

@@ -0,0 +1,72 @@
+const std = @import("std");
+
+pub const Function = struct {
+    name: []u8,
+    params: []u8,
+    return_type: []u8,
+    access_modifier: []u8,
+    documentation: []u8,
+    allocator: std.mem.Allocator,
+
+    pub fn init(
+        allocator: std.mem.Allocator,
+        name: []const u8,
+        params: []const u8,
+        return_type: []const u8,
+        access_modifier: []const u8,
+        documentation: []const u8,
+    ) !Function {
+        return .{
+            .name = try allocator.dupe(u8, name),
+            .params = try allocator.dupe(u8, params),
+            .return_type = try allocator.dupe(u8, return_type),
+            .access_modifier = try allocator.dupe(u8, access_modifier),
+            .documentation = try allocator.dupe(u8, documentation),
+            .allocator = allocator,
+        };
+    }
+
+    pub fn destroy(self: *Function) void {
+        self.allocator.free(self.name);
+        self.allocator.free(self.params);
+        self.allocator.free(self.return_type);
+        self.allocator.free(self.access_modifier);
+        self.allocator.free(self.documentation);
+    }
+
+    pub fn print(self: Function, writer: anytype) void {
+        writer.print("func {s}() -> {s};", .{ self.name, self.return_type });
+    }
+};
+
+pub const Property = struct {
+    name: []u8,
+    allocator: std.mem.Allocator,
+
+    pub fn init(allocator: std.mem.Allocator, name: []const u8) !Property {
+        return Property{
+            .name = try allocator.dupe(u8, name),
+            .allocator = allocator,
+        };
+    }
+
+    pub fn destroy(self: *Property) void {
+        self.allocator.free(self.name);
+    }
+};
+
+pub const Definition = union(enum) {
+    function: Function,
+
+    pub fn print(self: Definition, writer: anytype) !void {
+        switch (self) {
+            inline else => |case| return case.print(writer),
+        }
+    }
+
+    pub fn destroy(self: *Definition) void {
+        switch (self) {
+            inline else => |case| return case.destroy(),
+        }
+    }
+};

src/main.zig

@@ -1,36 +1,19 @@
 const std = @import("std");
 const ts = @import("tree-sitter");
 
+const Parser = @import("parser.zig");
+
 extern fn tree_sitter_zig() callconv(.C) *ts.Language;
 
 pub fn main() !void {
-    // Create a parser for the zig language
-    const language = tree_sitter_zig();
-    defer language.destroy();
+    var gpa = std.heap.GeneralPurposeAllocator(.{}){};
+    defer _ = gpa.deinit();
+    const allocator = gpa.allocator();
 
-    const parser = ts.Parser.create();
+    const file_path = "/Users/bogdanbuduroiu/development/aurelio-labs/semantic-router/semantic_router/route.py";
+
+    var parser = try Parser.create(allocator, file_path);
     defer parser.destroy();
-    try parser.setLanguage(language);
-
-    // Parse some source code and get the root node
-    const tree = parser.parseString("pub fn main() !void {}", null);
-    defer tree.?.destroy();
-
-    const node = tree.?.rootNode();
-    std.debug.assert(std.mem.eql(u8, node.kind(), "source_file"));
-    std.debug.print("{s}", .{node.kind()});
-
-    // Create a query and execute it
-    var error_offset: u32 = 0;
-    const query = try ts.Query.create(language, "name: (identifier) @name", &error_offset);
-    defer query.destroy();
-
-    const cursor = ts.QueryCursor.create();
-    defer cursor.destroy();
-    cursor.exec(query, node);
-
-    // Get the captured node of the first match
-    const match = cursor.nextMatch().?;
-    const capture = match.captures[0].node;
-    std.debug.assert(std.mem.eql(u8, capture.kind(), "identifier"));
+    const definitions = try parser.extractDefinitions();
+    _ = definitions; // autofix
 }

src/parser.zig

@@ -0,0 +1,170 @@
+const std = @import("std");
+const ts = @import("tree-sitter");
+const Allocator = std.mem.Allocator;
+const MultiArrayList = std.MultiArrayList;
+const definitions = @import("definitions.zig");
+const Definition = definitions.Definition;
+const Function = definitions.Function;
+
+const Self = @This();
+parser: *ts.Parser,
+language_name: []const u8,
+source: []const u8,
+allocator: Allocator,
+
+pub fn create(allocator: Allocator, file_path: []const u8) !*Self {
+    const ext = std.fs.path.extension(file_path);
+
+    var parser = ts.Parser.create();
+    errdefer parser.destroy();
+
+    const language = try getLanguageForExtension(ext);
+    try parser.setLanguage(language);
+
+    const source = try std.fs.cwd().readFileAlloc(allocator, file_path, 1024 * 1024 * 10); // 10MB max
+    errdefer allocator.free(source);
+
+    const p = try allocator.create(Self);
+    p.* = .{
+        .parser = parser,
+        .source = source,
+        .allocator = allocator,
+        .language_name = "python",
+    };
+    return p;
+}
+
+pub fn destroy(self: *Self) void {
+    self.parser.destroy();
+    self.allocator.free(self.source);
+    self.allocator.destroy(self);
+}
+
+pub fn extractDefinitions(self: *Self) !MultiArrayList(Definition) {
+    var defs = MultiArrayList(Definition){};
+    defer defs.deinit(self.allocator);
+
+    // Parse the source code
+    const tree = self.parser.parseString(self.source, null);
+    if (tree == null) {
+        return error.ParseFailed;
+    }
+    defer tree.?.destroy();
+
+    const root_node = tree.?.rootNode();
+
+    // Get the appropriate query for this language
+    const query_string = try getQueryForLanguage(self.language_name);
+    var error_offset: u32 = 0;
+    const query = try ts.Query.create(self.parser.getLanguage() orelse tree_sitter_python(), query_string, &error_offset);
+    defer query.destroy();
+
+    // Execute the query
+    const cursor = ts.QueryCursor.create();
+    defer cursor.destroy();
+    cursor.exec(query, root_node);
+
+    while (cursor.nextMatch()) |match| {
+        for (match.captures) |capture| {
+            const capture_name = query.captureNameForId(capture.index) orelse "mock_caputer";
+            const node = capture.node;
+            const node_text = self.source[node.startByte()..node.endByte()];
+            const name = if (node.childByFieldName("name")) |name_node|
+                self.source[name_node.startByte()..name_node.endByte()]
+            else
+                node_text;
+
+            if (std.mem.eql(u8, capture_name, "function")) {
+                var func_def = try Function.init(self.allocator, name, "", "", "", "");
+                try defs.append(self.allocator, func_def);
+                defer func_def.destroy();
+            }
+        }
+    }
+
+    for (defs) |def| {
+        try def.print(std.debug);
+    }
+    return defs;
+}
+
+// Helper
+
+fn getLanguageForExtension(ext: []const u8) !*ts.Language {
+    if (std.mem.eql(u8, ext, ".zig")) {
+        return tree_sitter_zig();
+    } else if (std.mem.eql(u8, ext, ".c") or std.mem.eql(u8, ext, ".h")) {
+        return tree_sitter_c();
+    } else if (std.mem.eql(u8, ext, ".py")) {
+        return tree_sitter_python();
+    } else {
+        return error.UnsupportedLanguage;
+    }
+}
+
+fn getQueryForLanguage(language_name: []const u8) ![]const u8 {
+    // In a real implementation, this would load queries from files
+    if (std.mem.eql(u8, language_name, "python")) {
+        return 
+        \\;; Capture top-level functions, class, and method definitions
+        \\(module
+        \\  (expression_statement
+        \\    (assignment) @assignment
+        \\  )
+        \\)
+        \\(module
+        \\  (function_definition) @function
+        \\)
+        \\(module
+        \\  (decorated_definition
+        \\    definition: (function_definition) @function
+        \\  )
+        \\)
+        \\(module
+        \\  (class_definition
+        \\    body: (block
+        \\      (expression_statement
+        \\        (assignment) @class_assignment
+        \\      )
+        \\    )
+        \\  ) @class
+        \\)
+        \\(module
+        \\  (class_definition
+        \\    body: (block
+        \\      (function_definition) @method
+        \\    )
+        \\  ) @class
+        \\)
+        \\(module
+        \\  (class_definition
+        \\    body: (block
+        \\      (expression_statement 
+        \\        (string) @docstring
+        \\      )
+        \\    )
+        \\  ) @class
+        \\)
+        \\(module
+        \\  (class_definition
+        \\    body: (block
+        \\      (decorated_definition
+        \\        definition: (function_definition) @method
+        \\      )
+        \\    )
+        \\  ) @class
+        \\)
+        ;
+    } else {
+        return 
+        \\(function_definition name: (identifier) @function)
+        \\(class_definition name: (identifier) @class)
+        \\(method_definition name: (identifier) @method)
+        ;
+    }
+}
+
+// External C functions for tree-sitter languages
+extern fn tree_sitter_zig() callconv(.C) *ts.Language;
+extern fn tree_sitter_c() callconv(.C) *ts.Language;
+extern fn tree_sitter_python() callconv(.C) *ts.Language;