feat: Add support for the Solidity language (#1616)

docs/docs/modules/indexes/text_splitters/examples/code.mdx

@@ -6,7 +6,7 @@ hide_table_of_contents: true
 
 LangChain supports a variety of different markup and programming language-specific text splitters to split your text based on language-specific syntax.
 This results in more semantically self-contained chunks that are more useful to a vector store or other retriever.
-Popular languages like JavaScript, Python, and Rust are supported as well as Latex, HTML, and Markdown.
+Popular languages like JavaScript, Python, Solidity, and Rust are supported as well as Latex, HTML, and Markdown.
 
 ## Usage
 

langchain/src/tests/code_text_splitter.test.ts

@@ -287,3 +287,32 @@ test("Rust code splitter", async () => {
     "}",
   ]);
 });
+
+test("Solidity code splitter", async () => {
+  const splitter = RecursiveCharacterTextSplitter.fromLanguage("sol", {
+    chunkSize: 16,
+    chunkOverlap: 0,
+  });
+  const code = `pragma solidity ^0.8.20;
+  contract HelloWorld {
+    function add(uint a, uint b) pure public returns(uint) {
+      return  a + b;
+    }
+  }
+  `;
+  const chunks = await splitter.splitText(code);
+  expect(chunks).toStrictEqual([
+    "pragma solidity",
+    "^0.8.20;",
+    "contract",
+    "HelloWorld {",
+    "function",
+    "add(uint a,",
+    "uint b) pure",
+    "public",
+    "returns(uint) {",
+    "return  a",
+    "+ b;",
+    "}\n  }",
+  ]);
+});

langchain/src/text_splitter.ts

@@ -222,6 +222,7 @@ export const SupportedTextSplitterLanguages = [
   "markdown",
   "latex",
   "html",
+  "sol",
 ] as const;
 
 export type SupportedTextSplitterLanguage =
@@ -617,6 +618,36 @@ export class RecursiveCharacterTextSplitter
         " ",
         "",
       ];
+    } else if (language === "sol") {
+      return [
+        // Split along compiler informations definitions
+        "\npragma ",
+        "\nusing ",
+        // Split along contract definitions
+        "\ncontract ",
+        "\ninterface ",
+        "\nlibrary ",
+        // Split along method definitions
+        "\nconstructor ",
+        "\ntype ",
+        "\nfunction ",
+        "\nevent ",
+        "\nmodifier ",
+        "\nerror ",
+        "\nstruct ",
+        "\nenum ",
+        // Split along control flow statements
+        "\nif ",
+        "\nfor ",
+        "\nwhile ",
+        "\ndo while ",
+        "\nassembly ",
+        // Split by the normal type of lines
+        "\n\n",
+        "\n",
+        " ",
+        "",
+      ];
     } else {
       throw new Error(`Language ${language} is not supported.`);
     }