fix(python): add docs + newline fix (#108)

* fix(python): add docs + newline fix

* fix: allow space after :

* revert: do not use single-line and multi-line

* fix: catche spaces only

Author: ShaunSHamilton

docs/python.md

@@ -1,8 +1,6 @@
 # Python
 
-## Browser
-
-Testing a function:
+## `getDef`
 
 ```javascript,mdbook-runnable,hidelines=#
 # {{#rustdoc_include tools/helpers.js:1}}
@@ -11,9 +9,9 @@ a = 1
 b = 2
 
 def add(x, y):
-  result = x + y
-  print(f"{x} + {y} = {result}")
-  return result
+    result = x + y
+    print(f"{x} + {y} = {result}")
+    return result
 
 `;
 
@@ -28,6 +26,60 @@ def add(x, y):
 }
 ```
 
+## `getBlock`
+
+```javascript,mdbook-runnable,hidelines=#
+# {{#rustdoc_include tools/helpers.js:1}}
+const code = `
+a = 1
+b = 2
+
+def add_or_subtract(a, b, add=True):
+    if add:
+      return a + b
+    else:
+      return a - b
+
+`;
+
+{
+  const equivalentPatterns = [
+    "if add",
+    /(if|elif) add/,
+  ];
+  for (const pattern of equivalentPatterns) {
+    const ifBlock = __helpers.python.getBlock(code, pattern);
+    const { block_body, block_indentation, block_condition } = ifBlock;
+    console.assert(block_indentation === 4);
+    console.assert(block_condition === "if add");
+    console.assert(block_body.match(/return a \+ b/));
+    console.log(ifBlock);
+  }
+}
+```
+
+## `removeComments`
+
+```javascript,mdbook-runnable,hidelines=#
+# {{#rustdoc_include tools/helpers.js:1}}
+// Note: Comment identifiers are escaped for docs markdown parser
+const code = `
+a = 1
+\# comment
+def b(d, e):
+    a = 2
+    \# comment
+    return a #comment
+`;
+{
+  const commentlessCode = __helpers.python.removeComments(code);
+  console.assert(commentlessCode === `\na = 1\n\ndef b(d, e):\n    a = 2\n    \n    return a \n`);
+  console.log(commentlessCode);
+}
+```
+
+## `__pyodide.runPython`
+
 Running the code of a singluar function to get the output:
 
 ```javascript,mdbook-runnable,hidelines=#
@@ -69,8 +121,8 @@ assert add(a, b) == 300
 
 ```python
 def
-  add(x, y):
-  return x + y
+    add(x, y):
+    return x + y
 ```
 
 - Python **does** allow newline characters between function arguments. E.g:
@@ -80,5 +132,5 @@ def add(
   x,
   y
 ):
-  return x + y
+    return x + y
 ```

docs/tools/helpers.js

@@ -2,8 +2,8 @@ const __helpers = {
   python: {
     getDef: (code, functionName) => {
       const regex = new RegExp(
-        `\\n(?<function_indentation> *?)def +${functionName} *\\((?<function_parameters>[^\\)]*)\\)\\s*:\\n(?<function_body>.*?)(?=\\n\\k<function_indentation>[\\w#]|$)`,
-        "s"
+        `^(?<function_indentation> *?)def +${functionName} *\\((?<function_parameters>[^\\)]*)\\)\\s*:\\n(?<function_body>.*?)(?=\\n\\k<function_indentation>[\\w#]|$)`,
+        "ms"
       );
 
       const matchedCode = regex.exec(code);
@@ -25,5 +25,39 @@ const __helpers = {
 
       return null;
     },
+    getBlock: (code, blockPattern) => {
+      const escapedBlockPattern =
+        blockPattern instanceof RegExp
+          ? blockPattern.source
+          : blockPattern.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
+
+      const regex = new RegExp(
+        `^(?<block_indentation> *?)(?<block_condition>${escapedBlockPattern})\\s*:\\n(?<block_body>.*?)(?=\\n\\k<block_indentation>[\\w#]|$)`,
+        "ms"
+      );
+
+      const matchedCode = regex.exec(code);
+      if (matchedCode) {
+        /* eslint-disable camelcase */
+        const { block_body, block_indentation, block_condition } =
+          matchedCode.groups;
+
+        const blockIndentationSansNewLine = block_indentation.replace(
+          /\n+/g,
+          ""
+        );
+        return {
+          block_body,
+          block_condition,
+          block_indentation: blockIndentationSansNewLine.length,
+        };
+        /* eslint-enable camelcase */
+      }
+
+      return null;
+    },
+    removeComments: (code) => {
+      return code.replace(/\/\/.*|\/\*[\s\S]*?\*\/|(#.*$)/gm, "");
+    },
   },
 };

lib/index.ts

@@ -99,7 +99,7 @@ const getIsDeclaredAfter = (styleRule: CSSStyleRule) => (selector: string) => {
 export module python {
   export function getDef(code: string, functionName: string) {
     const regex = new RegExp(
-      `\\n(?<function_indentation> *?)def +${functionName} *\\((?<function_parameters>[^\\)]*)\\)\\s*:\\n(?<function_body>.*?)(?=\\n\\k<function_indentation>[\\w#]|$)`,
+      `\\n?(?<function_indentation> *?)def +${functionName} *\\((?<function_parameters>[^\\)]*)\\)\\s*: *?\\n(?<function_body> +.*?)(?=\\n\\k<function_indentation>[\\w#]|$)`,
       "s"
     );
 
@@ -142,7 +142,7 @@ export module python {
         : blockPattern.replace(/[.*+?^${}()|[\]\\]/g, "\\$&");
 
     const regex = new RegExp(
-      `\\n(?<block_indentation> *?)(?<block_condition>${escapedBlockPattern})\\s*:\\n(?<block_body>.*?)(?=\\n\\k<block_indentation>[\\w#]|$)`,
+      `\\n?(?<block_indentation> *?)(?<block_condition>${escapedBlockPattern})\\s*: *?\\n(?<block_body> +.*?)(?=\\n\\k<block_indentation>[\\w#]|$)`,
       "s"
     );