feat(hogqlx): nested tags (#18213)

PostHog · Oct 27, 2023 · 4ea13f1 · 4ea13f1
1 parent f393cc7
commit 4ea13f1
Show file tree

Hide file tree

Showing 15 changed files with 1,276 additions and 892 deletions.
diff --git a/hogql_parser/HogQLParser.cpp b/hogql_parser/HogQLParser.cpp
diff --git a/hogql_parser/HogQLParser.h b/hogql_parser/HogQLParser.h
@@ -1457,17 +1457,46 @@ class  HogQLParser : public antlr4::Parser {
   class  HogqlxTagElementContext : public antlr4::ParserRuleContext {
   public:
     HogqlxTagElementContext(antlr4::ParserRuleContext *parent, size_t invokingState);
+
+    HogqlxTagElementContext() = default;
+    void copyFrom(HogqlxTagElementContext *context);
+    using antlr4::ParserRuleContext::copyFrom;
+
     virtual size_t getRuleIndex() const override;
+
+
+  };
+
+  class  HogqlxTagElementClosedContext : public HogqlxTagElementContext {
+  public:
+    HogqlxTagElementClosedContext(HogqlxTagElementContext *ctx);
+
     antlr4::tree::TerminalNode *LT();
     IdentifierContext *identifier();
     antlr4::tree::TerminalNode *SLASH();
     antlr4::tree::TerminalNode *GT();
     std::vector<HogqlxTagAttributeContext *> hogqlxTagAttribute();
     HogqlxTagAttributeContext* hogqlxTagAttribute(size_t i);
 
+    virtual std::any accept(antlr4::tree::ParseTreeVisitor *visitor) override;
+  };
+
+  class  HogqlxTagElementNestedContext : public HogqlxTagElementContext {
+  public:
+    HogqlxTagElementNestedContext(HogqlxTagElementContext *ctx);
+
+    std::vector<antlr4::tree::TerminalNode *> LT();
+    antlr4::tree::TerminalNode* LT(size_t i);
+    std::vector<IdentifierContext *> identifier();
+    IdentifierContext* identifier(size_t i);
+    std::vector<antlr4::tree::TerminalNode *> GT();
+    antlr4::tree::TerminalNode* GT(size_t i);
+    antlr4::tree::TerminalNode *SLASH();
+    std::vector<HogqlxTagAttributeContext *> hogqlxTagAttribute();
+    HogqlxTagAttributeContext* hogqlxTagAttribute(size_t i);
+    HogqlxTagElementContext *hogqlxTagElement();
 
     virtual std::any accept(antlr4::tree::ParseTreeVisitor *visitor) override;
-
   };
 
   HogqlxTagElementContext* hogqlxTagElement();

diff --git a/hogql_parser/HogQLParser.interp b/hogql_parser/HogQLParser.interp
diff --git a/hogql_parser/HogQLParserBaseVisitor.h b/hogql_parser/HogQLParserBaseVisitor.h
@@ -351,7 +351,11 @@ class  HogQLParserBaseVisitor : public HogQLParserVisitor {
     return visitChildren(ctx);
   }
 
-  virtual std::any visitHogqlxTagElement(HogQLParser::HogqlxTagElementContext *ctx) override {
+  virtual std::any visitHogqlxTagElementClosed(HogQLParser::HogqlxTagElementClosedContext *ctx) override {
+    return visitChildren(ctx);
+  }
+
+  virtual std::any visitHogqlxTagElementNested(HogQLParser::HogqlxTagElementNestedContext *ctx) override {
     return visitChildren(ctx);
   }
 

diff --git a/hogql_parser/HogQLParserVisitor.h b/hogql_parser/HogQLParserVisitor.h
@@ -187,7 +187,9 @@ class  HogQLParserVisitor : public antlr4::tree::AbstractParseTreeVisitor {
 
     virtual std::any visitColumnLambdaExpr(HogQLParser::ColumnLambdaExprContext *context) = 0;
 
-    virtual std::any visitHogqlxTagElement(HogQLParser::HogqlxTagElementContext *context) = 0;
+    virtual std::any visitHogqlxTagElementClosed(HogQLParser::HogqlxTagElementClosedContext *context) = 0;
+
+    virtual std::any visitHogqlxTagElementNested(HogQLParser::HogqlxTagElementNestedContext *context) = 0;
 
     virtual std::any visitHogqlxTagAttribute(HogQLParser::HogqlxTagAttributeContext *context) = 0;
 

diff --git a/hogql_parser/parser.cpp b/hogql_parser/parser.cpp
@@ -1244,7 +1244,7 @@ class HogQLParseTreeConverter : public HogQLParserBaseVisitor {
     );
   }
 
-  VISIT(HogqlxTagElement) {
+  VISIT(HogqlxTagElementClosed) {
     string kind = visitAsString(ctx->identifier());
     PyObject* tag_element = build_ast_node(
         "HogQLXTag", "{s:s#,s:N}",
@@ -1254,6 +1254,46 @@ class HogQLParseTreeConverter : public HogQLParserBaseVisitor {
     return tag_element;
   }
 
+  VISIT(HogqlxTagElementNested) {
+    string opening = visitAsString(ctx->identifier(0));
+    string closing = visitAsString(ctx->identifier(1));
+    if (opening != closing) {
+      throw HogQLSyntaxException("Opening and closing HogQLX tags must match. Got " + opening + " and " + closing);
+    }
+
+    auto tag_element_ctx = ctx->hogqlxTagElement();
+    auto tag_attribute_ctx = ctx->hogqlxTagAttribute();
+    PyObject* attributes = PyList_New(tag_attribute_ctx.size() + (tag_element_ctx ? 1 : 0));
+    bool found_source = false;
+    for (size_t i = 0; i < tag_attribute_ctx.size(); i++) {
+      PyObject* object = visitAsPyObject(tag_attribute_ctx[i]);
+      PyList_SET_ITEM(attributes, i, object);
+
+      PyObject* name = PyObject_GetAttrString(object, "name");
+      if (PyObject_RichCompareBool(name, PyUnicode_FromString("source"), Py_EQ)) {
+        found_source = true;
+      }
+      Py_DECREF(name);
+    }
+
+    if (tag_element_ctx) {
+        if (found_source) {
+            Py_DECREF(attributes);
+            throw HogQLSyntaxException("Nested HogQLX tags cannot have a source attribute");
+        }
+        PyList_SET_ITEM(attributes, tag_attribute_ctx.size(), build_ast_node(
+            "HogQLXAttribute", "{s:s#,s:N}", "name", "source", 6, "value", visitAsPyObject(ctx->hogqlxTagElement())
+        ));
+    }
+
+    PyObject* tag_element = build_ast_node(
+        "HogQLXTag", "{s:s#,s:N}",
+        "kind", opening.data(), opening.size(),
+        "attributes", attributes
+    );
+    return tag_element;
+  }
+
   VISIT(Placeholder) {
     string name = visitAsString(ctx->identifier());
     return build_ast_node("Placeholder", "{s:s#}", "field", name.data(), name.size());

diff --git a/hogql_parser/setup.py b/hogql_parser/setup.py
@@ -32,7 +32,7 @@
 
 setup(
     name="hogql_parser",
-    version="0.1.11",
+    version="0.1.12",
     url="https://github.com/PostHog/posthog/tree/master/hogql_parser",
     author="PostHog Inc.",
     author_email="[email protected]",

diff --git a/posthog/hogql/grammar/HogQLParser.g4 b/posthog/hogql/grammar/HogQLParser.g4
@@ -170,7 +170,10 @@ columnLambdaExpr:
     ;
 
 
-hogqlxTagElement: LT identifier hogqlxTagAttribute* SLASH GT;
+hogqlxTagElement
+    : LT identifier hogqlxTagAttribute* SLASH GT                                        # HogqlxTagElementClosed
+    | LT identifier hogqlxTagAttribute* GT hogqlxTagElement? LT SLASH identifier GT     # HogqlxTagElementNested
+    ;
 hogqlxTagAttribute
     :   identifier '=' STRING_LITERAL
     |   identifier '=' LBRACE columnExpr RBRACE
@@ -196,7 +199,7 @@ tableExpr
     | tableFunctionExpr                  # TableExprFunction
     | LPAREN selectUnionStmt RPAREN      # TableExprSubquery
     | tableExpr (alias | AS identifier)  # TableExprAlias
-    | hogqlxTagElement                         # TableExprTag
+    | hogqlxTagElement                   # TableExprTag
     | placeholder                        # TableExprPlaceholder
     ;
 tableFunctionExpr: identifier LPAREN tableArgList? RPAREN;

diff --git a/posthog/hogql/grammar/HogQLParser.interp b/posthog/hogql/grammar/HogQLParser.interp